id
stringlengths
3
8
content
stringlengths
100
981k
117
from pytest import raises from discopy.cartesian import * def test_Box_repr(): f = Box('f', 1, 2, lambda x: (x, x)) assert "Box('f', 1, 2" in repr(f) def test_Function_str(): f = Function(2, 1, lambda x, y: x + y) assert 'Function(dom=2, cod=1,' in str(f) def test_Function_call(): f = Swap(2, 1) values = (2, 3) with raises(TypeError) as err: f(*values) assert str(err.value) == messages.expected_input_length(f, values) def test_Function_then(): f, g = Function(2, 1, lambda x, y: x + y), Function(1, 1, lambda x: x + 1) assert Function.id(2).then(*(f, g))(20, 21) == 42 def test_Function_then_err(): f = Function(2, 1, lambda x, y: x + y) g = (lambda x: x, ) with raises(TypeError) as err: f >> g assert str(err.value) == messages.type_err(Function, g) g = Function.id(2) with raises(AxiomError) as err: f >> g assert str(err.value) == messages.does_not_compose(f, g) def test_Function_tensor(): assert Function.id(3)(1, 2, 3)\ == Function.id(0).tensor(*(3 * [Function.id(1)]))(1, 2, 3) def test_Function_tensor_err(): f = Function(2, 1, lambda x, y: x + y) g = (lambda x: x, ) with raises(TypeError) as err: f @ g assert str(err.value) == messages.type_err(Function, g)
136
import os import sys import shutil cwd_path = os.getcwd() sys.path.append(os.path.join(os.path.dirname(cwd_path), 'rt-thread', 'tools')) # BSP dist function def dist_do_building(BSP_ROOT, dist_dir): from mkdist import bsp_copy_files import rtconfig library_dir = os.path.join(dist_dir, 'libraries') print("=> copy nrf52 bsp libraries") library_path = os.path.join(os.path.dirname(BSP_ROOT), 'libraries') bsp_copy_files(library_path, library_dir)
149
from itertools import groupby class Solution: def countAndSay(self, n): def gen(s): return "".join(str(len(list(g))) + k for k, g in groupby(s)) s, i = "1", 1 while i < n: s = gen(s) i += 1 return s
157
import json import aiohttp async def request(url, payload=None, params=None, headers=None): headers = {'content-type': 'application/json', **(headers or {})} data = payload and json.dumps(payload) async with aiohttp.ClientSession() as client: async with client.post( url, data=data, params=params, headers=headers) as resp: # TODO: Check response status json_response = await resp.json() return json_response async def get_updates(base_url, timeout, offset): params = { 'timeout': timeout, 'offset': offset } return await request(f'{base_url}/getUpdates', params=params) async def send_message(base_url, chat_id, text, reply_markup=None): payload = { 'chat_id': chat_id, 'text': text } if reply_markup is not None: payload['reply_markup'] = reply_markup return await request(f'{base_url}/sendMessage', payload) async def answer_callback_query( base_url, callback_query_id, text, show_alert, url=None, cache_time=None): payload = { 'callback_query_id': callback_query_id, 'text': text, 'show_alert': show_alert } if url is not None: payload['url'] = url if cache_time is not None: payload['cache_time'] = cache_time return await request(f'{base_url}/answerCallbackQuery', payload)
185
import traceback from pprint import pformat from threading import Thread import itchat import logging from wxpy.chat import Chat from wxpy.chats import Chats from wxpy.friend import Friend from wxpy.group import Group from wxpy.message import MessageConfigs, Messages, Message, MessageConfig from wxpy.mp import MP from wxpy.response import ResponseError from wxpy.user import User from wxpy.utils.constants import SYSTEM from wxpy.utils.tools import handle_response, get_user_name, wrap_user_name, ensure_list logger = logging.getLogger('wxpy') class Robot(object): """ 机器人对象,用于登陆和操作微信账号,涵盖大部分 Web 微信的功能 """ def __init__( self, save_path=None, console_qr=False, qr_path=None, qr_callback=None, login_callback=None, logout_callback=None ): """ :param save_path: | 用于保存或载入登陆状态的文件路径,例如: 'wxpy.pkl',为空则不尝试载入。 | 填写本参数后,可在短时间内重新载入登陆状态,避免重复扫码,失效时会重新要求登陆 :param console_qr: 在终端中显示登陆二维码,需要安装 Pillow 模块 :param qr_path: 保存二维码的路径 :param qr_callback: 获得二维码时的回调,接收参数: uuid, status, qrcode :param login_callback: 登陆时的回调,接收参数同上 :param logout_callback: 登出时的回调,接收参数同上 """ self.core = itchat.Core() itchat.instanceList.append(self) self.core.auto_login( hotReload=bool(save_path), statusStorageDir=save_path, enableCmdQR=console_qr, picDir=qr_path, qrCallback=qr_callback, loginCallback=login_callback, exitCallback=logout_callback ) self.message_configs = MessageConfigs(self) self.messages = Messages(robot=self) self.file_helper = Chat(wrap_user_name('filehelper')) self.file_helper.robot = self self.file_helper.nick_name = '文件传输助手' self.self = Chat(self.core.loginInfo['User']) self.self.robot = self self.save_path = save_path def __repr__(self): return '<{}: {}>'.format(self.__class__.__name__, self.self.name) @handle_response() def logout(self): """ 登出当前账号 """ return self.core.logout() @property def alive(self): """ 当前的登陆状态 :return: 若为登陆状态,则为 True,否则为 False """ return self.core.alive @alive.setter def alive(self, value): self.core.alive = value def dump_login_status(self, save_path=None): return self.core.dump_login_status(save_path or self.save_path) # chats def except_self(self, chats_or_dicts): """ 从聊天对象合集或用户字典列表中排除自身 :param chats_or_dicts: 聊天对象合集或用户字典列表 :return: 排除自身后的列表 """ return list(filter(lambda x: get_user_name(x) != self.self.user_name, chats_or_dicts)) def chats(self, update=False): """ 获取所有聊天对象 :param update: 是否更新 :return: 聊天对象合集 """ return Chats(self.friends(update) + self.groups(update) + self.mps(update), self) def friends(self, update=False): """ 获取所有好友 :param update: 是否更新 :return: 聊天对象合集 """ @handle_response(Friend) def do(): return self.core.get_friends(update=update) ret = do() ret.source = self return ret @handle_response(Group) def groups(self, update=False, contact_only=False): """ 获取所有群聊 :param update: 是否更新 :param contact_only: 是否限于保存为联系人的群聊 :return: 群聊合集 """ return self.core.get_chatrooms(update=update, contactOnly=contact_only) @handle_response(MP) def mps(self, update=False): """ 获取所有公众号 :param update: 是否更新 :return: 聊天对象合集 """ return self.core.get_mps(update=update) @handle_response(User) def user_details(self, user_or_users, chunk_size=50): """ 获取单个或批量获取多个用户的详细信息(地区、性别、签名等),但不可用于群聊成员 :param user_or_users: 单个或多个用户对象或 user_name :param chunk_size: 分配请求时的单批数量,目前为 50 :return: 单个或多个用户用户的详细信息 """ def chunks(): total = ensure_list(user_or_users) for i in range(0, len(total), chunk_size): yield total[i:i + chunk_size] @handle_response() def process_one_chunk(_chunk): return self.core.update_friend(userName=get_user_name(_chunk)) if isinstance(user_or_users, (list, tuple)): ret = list() for chunk in chunks(): chunk_ret = process_one_chunk(chunk) if isinstance(chunk_ret, list): ret += chunk_ret else: ret.append(chunk_ret) return ret else: return process_one_chunk(user_or_users) def search(self, name=None, **attributes): """ 在所有类型的聊天对象中进行搜索 :param name: 名称 (可以是昵称、备注等) :param attributes: 属性键值对,键可以是 sex(性别), province(省份), city(城市) 等。例如可指定 province='广东' :return: 匹配的聊天对象合集 """ return self.chats().search(name, **attributes) # add / create @handle_response() def add_friend(self, user, verify_content=''): """ 添加用户为好友 :param user: 用户对象或用户名 :param verify_content: 验证说明信息 """ return self.core.add_friend( userName=get_user_name(user), status=2, verifyContent=verify_content, autoUpdate=True ) @handle_response() def accept_friend(self, user, verify_content=''): """ 接受用户为好友 :param user: 用户对象或用户名 :param verify_content: 验证说明信息 """ # Todo: 验证好友接口可用性,并在接受好友时直接返回新好友 return self.core.add_friend( userName=get_user_name(user), status=3, verifyContent=verify_content, autoUpdate=True ) def create_group(self, users, topic=None): """ 创建一个新的群聊 :param users: 用户列表 :param topic: 群名称 :return: 若建群成功,返回一个新的群聊对象 """ @handle_response() def request(): return self.core.create_chatroom( memberList=wrap_user_name(users), topic=topic or '' ) ret = request() user_name = ret.get('ChatRoomName') if user_name: return Group(self.core.update_chatroom(userName=user_name)) else: raise ResponseError('Failed to create group:\n{}'.format(pformat(ret))) # messages def _process_message(self, msg): """ 处理接收到的消息 """ if not self.alive: return func, run_async = self.message_configs.get_func(msg) if not func: return def process(): # noinspection PyBroadException try: ret = func(msg) if ret is not None: if isinstance(ret, (tuple, list)): self.core.send( msg=str(ret[0]), toUserName=msg.chat.user_name, mediaId=ret[1] ) else: self.core.send( msg=str(ret), toUserName=msg.chat.user_name ) except: logger.warning( 'An error occurred in registered function, ' 'use `Robot().start(debug=True)` to show detailed information') logger.debug(traceback.format_exc()) if run_async: Thread(target=process).start() else: process() def register( self, chats=None, msg_types=None, except_self=True, run_async=True, enabled=True ): """ 装饰器:用于注册消息配置 :param chats: 单个或列表形式的多个聊天对象或聊天类型,为空时匹配所有聊天对象 :param msg_types: 单个或列表形式的多个消息类型,为空时匹配所有消息类型 (SYSTEM 类消息除外) :param except_self: 排除自己在手机上发送的消息 :param run_async: 异步执行配置的函数,可提高响应速度 :param enabled: 当前配置的默认开启状态,可事后动态开启或关闭 """ def register(func): self.message_configs.append(MessageConfig( robot=self, func=func, chats=chats, msg_types=msg_types, except_self=except_self, run_async=run_async, enabled=enabled )) return func return register def start(self, block=True): """ 开始监听和处理消息 :param block: 是否堵塞线程,为 False 时将在新的线程中运行 """ def listen(): logger.info('{} Auto-reply started.'.format(self)) try: while self.alive: msg = Message(self.core.msgList.get(), self) if msg.type is not SYSTEM: self.messages.append(msg) self._process_message(msg) except KeyboardInterrupt: logger.info('KeyboardInterrupt received, ending...') self.alive = False if self.core.useHotReload: self.dump_login_status() logger.info('Bye.') if block: listen() else: t = Thread(target=listen, daemon=True) t.start()
201
import requests from allauth.socialaccount.providers.oauth2.views import ( OAuth2Adapter, OAuth2CallbackView, OAuth2LoginView, ) from .provider import DropboxOAuth2Provider class DropboxOAuth2Adapter(OAuth2Adapter): provider_id = DropboxOAuth2Provider.id access_token_url = "https://api.dropbox.com/oauth2/token" authorize_url = "https://www.dropbox.com/oauth2/authorize" profile_url = "https://api.dropbox.com/2/users/get_current_account" redirect_uri_protocol = "https" def complete_login(self, request, app, token, **kwargs): response = requests.post( self.profile_url, headers={"Authorization": "Bearer %s" % (token.token,)}, ) response.raise_for_status() return self.get_provider().sociallogin_from_response(request, response.json()) oauth_login = OAuth2LoginView.adapter_view(DropboxOAuth2Adapter) oauth_callback = OAuth2CallbackView.adapter_view(DropboxOAuth2Adapter)
274
import logging from collections import Counter from django.core.management.base import BaseCommand from django.db.models import Q from TWLight.applications.models import Application from TWLight.resources.models import Partner from TWLight.applications.signals import Reminder from TWLight.users.models import Editor logger = logging.getLogger(__name__) class Command(BaseCommand): def handle(self, *args, **options): # This is not DRY. Originally, this pulled the queryset from # TWLight.applications.views.ListApplicationsView.get_queryset(). # But that now expects a request object. So, we did a copy/paste. # We're actually getting apps with a status of PENDING or QUESTION # or APPROVED, and their corresponding user preferences being True # for partners with a status of AVAILABLE. all_apps = ( Application.objects.filter( Q( partner__coordinator__editor__user__userprofile__pending_app_reminders=True ) & Q(status=Application.PENDING) | Q( partner__coordinator__editor__user__userprofile__discussion_app_reminders=True ) & Q(status=Application.QUESTION) | Q( partner__coordinator__editor__user__userprofile__approved_app_reminders=True ) & Q(status=Application.APPROVED), partner__status__in=[Partner.AVAILABLE], editor__isnull=False, ) .exclude(editor__user__groups__name="restricted") .order_by("status", "partner", "date_created") ) # A deduplicated dict of coordinators from the pending app queryset, along # with a count of how many total pending apps they have coordinators = Counter( all_apps.values_list( "partner__coordinator__editor", "partner__coordinator__email", "partner__coordinator__editor__user__userprofile__lang", ) ) for coordinator, count in list(coordinators.items()): try: # We create a dictionary with the three status codes # we'd want to send emails for, and their corresponding # counts. app_status_and_count = { Application.PENDING: all_apps.filter( status=Application.PENDING, partner__coordinator__editor=coordinator[0], ).count(), Application.QUESTION: all_apps.filter( status=Application.QUESTION, partner__coordinator__editor=coordinator[0], ).count(), Application.APPROVED: all_apps.filter( status=Application.APPROVED, partner__coordinator__editor=coordinator[0], ).count(), } editor = Editor.objects.get(id=coordinator[0]) except Editor.DoesNotExist: logger.info( "Editor {} does not exist; skipping.".format(coordinator[0]) ) break # Only bother with the signal if we have a coordinator email. if coordinator[1]: Reminder.coordinator_reminder.send( sender=self.__class__, app_status_and_count=app_status_and_count, coordinator_wp_username=editor.wp_username, coordinator_email=coordinator[1], coordinator_lang=coordinator[2], )
330
from tools.geofunc import GeoFunc import pandas as pd import json def getData(index): '''报错数据集有(空心):han,jakobs1,jakobs2 ''' '''形状过多暂时未处理:shapes、shirt、swim、trousers''' name=["ga","albano","blaz1","blaz2","dighe1","dighe2","fu","han","jakobs1","jakobs2","mao","marques","shapes","shirts","swim","trousers"] print("开始处理",name[index],"数据集") '''暂时没有考虑宽度,全部缩放来表示''' scale=[100,0.5,100,100,20,20,20,10,20,20,0.5,20,50] print("缩放",scale[index],"倍") df = pd.read_csv("data/"+name[index]+".csv") polygons=[] for i in range(0,df.shape[0]): for j in range(0,df['num'][i]): poly=json.loads(df['polygon'][i]) GeoFunc.normData(poly,scale[index]) polygons.append(poly) return polygons
340
import FWCore.ParameterSet.Config as cms # # module to make the MaxSumPtWMass jet combination # findTtSemiLepJetCombMaxSumPtWMass = cms.EDProducer("TtSemiLepJetCombMaxSumPtWMass", ## jet input jets = cms.InputTag("selectedPatJets"), ## lepton input leps = cms.InputTag("selectedPatMuons"), ## maximum number of jets to be considered maxNJets = cms.int32(4), ## nominal WMass parameter (in GeV) wMass = cms.double(80.4), ## use b-tagging two distinguish between light and b jets useBTagging = cms.bool(False), ## choose algorithm for b-tagging bTagAlgorithm = cms.string("trackCountingHighEffBJetTags"), ## minimum b discriminator value required for b jets and ## maximum b discriminator value allowed for non-b jets minBDiscBJets = cms.double(1.0), maxBDiscLightJets = cms.double(3.0) )
351
from vyper import ast as vy_ast def test_output_class(): old_node = vy_ast.parse_to_ast("foo = 42") new_node = vy_ast.Int.from_node(old_node, value=666) assert isinstance(new_node, vy_ast.Int) def test_source(): old_node = vy_ast.parse_to_ast("foo = 42") new_node = vy_ast.Int.from_node(old_node, value=666) assert old_node.src == new_node.src assert old_node.node_source_code == new_node.node_source_code def test_kwargs(): old_node = vy_ast.parse_to_ast("42").body[0].value new_node = vy_ast.Int.from_node(old_node, value=666) assert old_node.value == 42 assert new_node.value == 666 def test_compare_nodes(): old_node = vy_ast.parse_to_ast("foo = 42") new_node = vy_ast.Int.from_node(old_node, value=666) assert not vy_ast.compare_nodes(old_node, new_node) def test_new_node_has_no_parent(): old_node = vy_ast.parse_to_ast("foo = 42") new_node = vy_ast.Int.from_node(old_node, value=666) assert new_node._parent is None assert new_node._depth == 0
377
import os from functools import wraps from os.path import join as join_path from dash import Dash from flask import make_response, render_template_string, redirect excluded_resources_endpoints = ( 'static', '_dash_assets.static', '/_favicon.ico', '/login', '/logout', '/_user', '/auth') def add_routes(app, authorizer): """Adds authentication endpoints to a flask app. Decorates other endpoints to grant access. The endpoints are: * /login * Method: GET * /logout * Method: GET * Erases cookies * /auth * Method: GET * Validates cookies if present or header authentication * Header: 'Authorization: DASHBOARD-AUTH username=([^/]*)/password=([^/]*)' * Sets cookies on login * Rejects unauthorized users Parameters ---------- app: flask.Flask or dash.Dash The flask or dash application excluded_resources_endpoints: tuple(str) Tuple with endpoints where access must not be checked. """ def login(): ok, _ = authorizer.validate() if ok: return make_response(redirect('/'), 307) return render_template_string(login_template) def logout(): _, response = authorizer.clean_cookie() return response def auth(): _, response = authorizer.validate() return response def authorize_endpoint(function): @wraps(function) def authorized_function(*args, **kwargs): ok, response = authorizer.validate() if ok: return function(*args, **kwargs) return response return authorized_function if isinstance(app, Dash): app = app.server login_template = load_template('login.html') app.add_url_rule('/auth', '/auth', auth) app.add_url_rule('/login', '/login', login) app.add_url_rule('/logout', '/logout', logout) for endpoint, function in app.view_functions.items(): if endpoint not in excluded_resources_endpoints: app.view_functions[endpoint] = authorize_endpoint(function) def load_template(filename): """Loads the login html template.""" pyfile_path = os.path.dirname(os.path.abspath(__file__)) path = join_path(pyfile_path, 'templates', filename) with open(path, 'r') as f: return f.read().strip()
432
from robotpy_ext.control.toggle import Toggle from robotpy_ext.misc.precise_delay import NotifierDelay class FakeJoystick: def __init__(self): self._pressed = [False] * 2 def getRawButton(self, num): return self._pressed[num] def press(self, num): self._pressed[num] = True def release(self, num): self._pressed[num] = False def test_toggle(): joystick = FakeJoystick() toggleButton = Toggle(joystick, 0) toggleButton2 = Toggle(joystick, 1) assert toggleButton.off joystick.press(0) assert toggleButton.on assert toggleButton2.off joystick.release(0) assert toggleButton.on joystick.press(0) assert toggleButton.off joystick.release(0) assert toggleButton.off joystick.press(1) assert toggleButton.off assert toggleButton2.on def test_toggle_debounce(): # TODO: use simulated time delay = NotifierDelay(0.5) joystick = FakeJoystick() toggleButton = Toggle(joystick, 1, 0.1) assert toggleButton.off joystick.press(1) assert toggleButton.on joystick.release(1) joystick.press(1) joystick.release(1) assert toggleButton.on delay.wait() assert toggleButton.on joystick.press(1) assert toggleButton.off
505
from test_plus.test import TestCase from ...administrative_units.factories import AdministrativeUnitFactory from ...cases.factories import CaseFactory from ...channels.factories import ChannelFactory from ...events.factories import EventFactory from ...features.factories import FeatureFactory, FeatureOptionFactory from ...generic.tests.test_views import ReadOnlyViewSetMixin from ...institutions.factories import InstitutionFactory from ...letters.factories import DocumentTypeFactory, ReferenceNumberFactory from ...search.tests.mixins import SearchQueryMixin from ...tags.factories import TagFactory from ...users.factories import UserFactory class AdministrativeUnitAutocompleteViewSetTestCase( ReadOnlyViewSetMixin, SearchQueryMixin, TestCase ): basename = "autocomplete_administrative_unit" factory_class = AdministrativeUnitFactory def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["name"], self.obj.name) class CaseAutocompleteViewSetTestCase(ReadOnlyViewSetMixin, SearchQueryMixin, TestCase): basename = "autocomplete_case" factory_class = CaseFactory def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["name"], self.obj.name) class ChannelAutocompleteViewSetTestCase( ReadOnlyViewSetMixin, SearchQueryMixin, TestCase ): basename = "autocomplete_channel" factory_class = ChannelFactory def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["name"], self.obj.name) class DocumentTypeAutocompleteViewSetTestCase( ReadOnlyViewSetMixin, SearchQueryMixin, TestCase ): basename = "autocomplete_document_type" factory_class = DocumentTypeFactory def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["name"], self.obj.name) class ReferenceNumberAutocompleteViewSetTestCase( ReadOnlyViewSetMixin, SearchQueryMixin, TestCase ): basename = "autocomplete_reference_number" factory_class = ReferenceNumberFactory def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["name"], self.obj.name) class EventAutocompleteViewSetTestCase( ReadOnlyViewSetMixin, SearchQueryMixin, TestCase ): basename = "autocomplete_event" factory_class = EventFactory def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["name"], self.obj.name) class FeatureAutocompleteViewSetTestCase( ReadOnlyViewSetMixin, SearchQueryMixin, TestCase ): basename = "autocomplete_feature" factory_class = FeatureFactory def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["name"], self.obj.name) class FeatureOptionAutocompleteViewSetTestCase( ReadOnlyViewSetMixin, SearchQueryMixin, TestCase ): basename = "autocomplete_feature_option" factory_class = FeatureOptionFactory def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["name"], self.obj.name) class InstitutionAutocompleteViewSetTestCase( ReadOnlyViewSetMixin, SearchQueryMixin, TestCase ): basename = "autocomplete_institution" factory_class = InstitutionFactory def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["name"], self.obj.name) class TagAutocompleteViewSetTestCase(ReadOnlyViewSetMixin, SearchQueryMixin, TestCase): basename = "autocomplete_tag" factory_class = TagFactory def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["name"], self.obj.name) class UserAutocompleteViewSetTestCase(ReadOnlyViewSetMixin, SearchQueryMixin, TestCase): basename = "autocomplete_user" factory_class = UserFactory initial_count = 1 def validate_item(self, item): self.assertEqual(item["id"], self.obj.id) self.assertEqual(item["username"], self.obj.username)
509
import json import re import responses from werkzeug.test import Client from werkzeug.wrappers import Response from satosa.proxy_server import make_app from satosa.satosa_config import SATOSAConfig class TestConsent: def test_full_flow(self, satosa_config_dict, consent_module_config): api_url = "https://consent.example.com/api" redirect_url = "https://consent.example.com/redirect" consent_module_config["config"]["api_url"] = api_url consent_module_config["config"]["redirect_url"] = redirect_url satosa_config_dict["MICRO_SERVICES"].append(consent_module_config) # application test_client = Client(make_app(SATOSAConfig(satosa_config_dict)), Response) # incoming auth req http_resp = test_client.get("/{}/{}/request".format(satosa_config_dict["BACKEND_MODULES"][0]["name"], satosa_config_dict["FRONTEND_MODULES"][0]["name"])) assert http_resp.status_code == 200 verify_url_re = re.compile(r"{}/verify/\w+".format(api_url)) with responses.RequestsMock() as rsps: # fake no previous consent consent_request_url_re = re.compile(r"{}/creq/\w+".format(api_url)) rsps.add(responses.GET, verify_url_re, status=401) rsps.add(responses.GET, consent_request_url_re, "test_ticket", status=200) # incoming auth resp http_resp = test_client.get("/{}/response".format(satosa_config_dict["BACKEND_MODULES"][0]["name"])) assert http_resp.status_code == 302 assert http_resp.headers["Location"].startswith(redirect_url) with responses.RequestsMock() as rsps: # fake consent rsps.add(responses.GET, verify_url_re, json.dumps({"foo": "bar"}), status=200) # incoming consent response http_resp = test_client.get("/consent/handle_consent") assert http_resp.status_code == 200
511
import argparse import glob import os import pickle from pathlib import Path import numpy as np from PIL import Image from tqdm import tqdm from src.align.align_trans import get_reference_facial_points, warp_and_crop_face # sys.path.append("../../") from src.align.detector import detect_faces if __name__ == "__main__": parser = argparse.ArgumentParser(description="face alignment") parser.add_argument( "-source_root", "--source_root", help="specify your source dir", default="../../data/fiw-videos/new-processed/", type=str, ) parser.add_argument( "-dest_root", "--dest_root", help="specify your destination dir", default="../../data/fiw-videos/new-processed/", type=str, ) parser.add_argument( "-crop_size", "--crop_size", help="specify size of aligned faces, align and crop with padding", default=112, type=int, ) args = parser.parse_args() source_root = args.source_root # specify your source dir dest_root = args.dest_root # specify your destination dir crop_size = ( args.crop_size ) # specify size of aligned faces, align and crop with padding scale = crop_size / 112.0 reference = get_reference_facial_points(default_square=True) * scale cwd = os.getcwd() # delete '.DS_Store' existed in the source_root os.chdir(source_root) os.system("find . -name '*.DS_Store' -type f -delete") os.chdir(cwd) imfiles = [ f for f in glob.glob(f"{source_root}F????/MID*/faces/msceleb*") if Path(f).is_file() ] # images = {imfile.replace(source_root, ''): Image.open(imfile) for imfile in imfiles} meta = {} # for subfolder in tqdm(os.listdir(source_root)): for imfile in tqdm(imfiles): ref = imfile.replace(source_root, "") print("Processing\t{}".format(imfile)) img = Image.open(imfile) try: # Handle exception bbs, landmarks = detect_faces(img) except Exception: print("{} is discarded due to exception!".format(imfile)) continue ref = imfile.replace(source_root, "") ndetections = len(landmarks) if ( ndetections == 0 ): # If the landmarks cannot be detected, the img will be discarded print("{} is discarded due to non-detected landmarks!".format(imfile)) meta[ref] = [] continue li_meta = [] for i in range(ndetections): im_meta = {} im_meta["face"] = i im_meta["landmarks"] = landmarks[i] im_meta["bb"] = bbs[i] facial5points = [[landmarks[i][j], landmarks[i][j + 5]] for j in range(5)] warped_face = warp_and_crop_face( np.array(img), facial5points, reference, crop_size=(crop_size, crop_size), ) img_warped = Image.fromarray(warped_face) image_name = imfile.replace("images", "cropped").replace( ".jpg", "-{:02d}.jpg".format(i) ) # im_meta['ref'] = "/".join(image_name.split('/')[-5:]) img_warped.save(image_name) li_meta.append(im_meta) meta[ref] = li_meta with open(source_root + "cropped-meta.pkl", "wb") as f: pickle.dump(meta, f)
513
import collections class ReadOnlyDict(collections.MutableMapping): def __init__(self, store): self.store = store def __getitem__(self, key): return self.store[key] def __setitem__(self, key, value): raise TypeError('Cannot modify ReadOnlyDict') def __delitem__(self, key): raise TypeError('Cannot modify ReadOnlyDict') def __iter__(self): return iter(self.store) def __len__(self): return len(self.store) def __str__(self): return 'ReadOnlyDict(%s)' % self.store def __repr__(self): return 'ReadOnlyDict(%r)' % self.store
522
import os import unittest import torch import torch.distributed as dist from torch.multiprocessing import Process import torch.nn as nn from machina.optims import DistributedAdamW def init_processes(rank, world_size, function, backend='tcp'): os.environ['MASTER_ADDR'] = '127.0.0.1' os.environ['MASTER_PORT'] = '29500' dist.init_process_group(backend, rank=rank, world_size=world_size) function(rank, world_size) class TestDistributedAdamW(unittest.TestCase): def test_step(self): def _run(rank, world_size): model = nn.Linear(10, 1) optimizer = DistributedAdamW( model.parameters()) optimizer.zero_grad() loss = model(torch.ones(10).float()) loss.backward() optimizer.step() processes = [] world_size = 4 for rank in range(world_size): p = Process(target=init_processes, args=(rank, world_size, _run)) p.start() processes.append(p) for p in processes: p.join()
532
import datetime import io import json_tricks import logging import os from os.path import (abspath, basename, dirname, exists, expanduser, join, realpath, relpath, splitext) import re import shutil import sys from traits.api import (Any, Dict, Enum, HasTraits, Instance, List, Long, Str) from whoosh import fields, qparser, query from whoosh.util.times import datetime_to_long, long_to_datetime from .common import get_project_dir from .media import Media, MediaData, get_media_data from .directory import Directory from . import processor logger = logging.getLogger(__name__) if sys.version_info[0] > 2: unicode = str string_types = (str,) import csv else: string_types = (basestring,) import backports.csv as csv INT = fields.NUMERIC(numtype=int) FLOAT = fields.NUMERIC(numtype=float) def get_file_saved_time(path): dt = datetime.datetime.fromtimestamp(os.stat(path).st_ctime) return dt.ctime() def _get_sample(fname): sample = '' with io.open(fname, 'r', newline='', encoding='utf-8') as fp: sample += fp.readline() + fp.readline() return sample def _get_csv_headers(fname): sample = _get_sample(fname) sniffer = csv.Sniffer() has_header = sniffer.has_header(sample) dialect = sniffer.sniff(sample) with io.open(fname, 'r', newline='', encoding='utf-8') as fp: reader = csv.reader(fp, dialect) header = next(reader) return has_header, header, dialect class TagInfo(HasTraits): name = Str type = Enum("string", "text", "int", "float", "bool") default = Any def __repr__(self): return 'TagInfo(%r, %r)' % (self.name, self.type) def _default_default(self): map = {"string": "", "text": "", "int": 0, "float": 0.0, "bool": False} return map[self.type] def open_file(fname_or_file, mode='rb'): if hasattr(fname_or_file, 'read'): return fname_or_file else: return open(fname_or_file, mode) def sanitize_name(name): name = name.lower() name = re.sub(r'\s+', '_', name) return re.sub(r'\W+', '', name) def get_non_existing_filename(fname): if exists(fname): base, ext = splitext(basename(fname)) return join(dirname(fname), base + '_a' + ext) else: return fname COMMON_TAGS = dict( file_name='string', path='string', relpath='string', ctime='string', mtime='string', size='int', type='string' ) def _cleanup_query(q, tag_types): type_map = dict(float=FLOAT.from_bytes, int=INT.from_bytes) for term in q.leaves(): if isinstance(term, query.Term): if isinstance(term.text, (str, unicode, bytes)): fieldtype = tag_types[term.fieldname] if fieldtype in type_map: term.text = type_map[fieldtype](term.text) else: term.text = term.text.lower() elif isinstance(term, query.Phrase): term.words = [x.lower() for x in term.words] def _check_value(value, expr): if isinstance(expr, string_types): return expr in value.lower() else: return expr == value def _check_range(x, term): result = True if term.start is not None: if term.startexcl: result &= x > term.start else: result &= x >= term.start if term.end is not None and result: if term.endexcl: result &= x < term.end else: result &= x <= term.end return result def _check_date_range(x, term): result = True if term.startdate is not None: result &= x >= term.start if term.enddate is not None and result: result &= x <= term.end return result def _search_media(expr, m_key, get_tag): """Given search expression, index to media, and a getter to get the attribute check if the media matches expression. """ if expr.is_leaf(): if isinstance(expr, query.Term): attr = expr.fieldname return _check_value(get_tag(m_key, attr), expr.text) elif isinstance(expr, query.Phrase): attr = expr.fieldname text = " ".join(expr.words) return _check_value(get_tag(m_key, attr), text) elif isinstance(expr, query.DateRange): if expr.fieldname == 'ctime': value = get_tag(m_key, 'ctime_') elif expr.fieldname == 'mtime': value = get_tag(m_key, 'mtime_') return _check_date_range(value, expr) elif isinstance(expr, query.NumericRange): attr = expr.fieldname return _check_range(get_tag(m_key, attr), expr) else: print("Unsupported term: %r" % expr) return False else: if isinstance(expr, query.And): result = True for child in expr.children(): result &= _search_media(child, m_key, get_tag) if not result: break return result elif isinstance(expr, query.Or): result = False for child in expr.children(): result |= _search_media(child, m_key, get_tag) if result: break return result elif isinstance(expr, query.Not): subquery = list(expr.children())[0] return not _search_media(subquery, m_key, get_tag) else: print("Unsupported term: %r" % expr) return False class Project(HasTraits): name = Str description = Str path = Str root = Instance(Directory) tags = List(TagInfo) _media = Dict(Str, Media) extensions = List(Str) processors = List(processor.FactoryBase) number_of_files = Long # Path where the project data is saved. save_file = Str last_save_time = Str _data = Dict _tag_data = Dict _relpath2index = Dict() _query_parser = Instance(qparser.QueryParser) def add_tags(self, tags): tags = list(self.tags) + tags self.update_tags(tags) def update_tags(self, new_tags): old_tags = self.tags new_tag_names = set(tag.name for tag in new_tags) tag_info = dict((tag.name, tag.type) for tag in old_tags) removed = [] added = [] for tag in new_tags: if tag.name not in tag_info: added.append(tag) elif tag_info[tag.name] != tag.type: removed.append(tag) added.append(tag) for tag in old_tags: if tag.name not in new_tag_names: removed.append(tag) for tag in removed: del self._tag_data[tag.name] n_entries = len(self._relpath2index) for tag in added: self._tag_data[tag.name] = [tag.default]*n_entries # The above can be the first time when self._tag_data is accessed, when # creating a new project for example. In this case, # self.__tag_data_default is called, so if self.tags is set then the # removed tags will not exist in _tag_data causing an error. So we only # set self.tags below. self.tags = new_tags # Update the cached media for m in self._media.values(): for tag in removed: del m.tags[tag.name] for tag in added: m.tags[tag.name] = tag.default self._query_parser = self._make_query_parser() def copy(self): """Make a copy of this project. This does not copy the data but only the tags, extensions and the other settings of the project. This will not copy any of the processor states but only their settings. """ name = self.name + ' copy' p = Project(name=name) traits = ['description', 'extensions', 'path', 'processors', 'tags'] p.copy_traits(self, traits, copy='deep') # Clear out the _done information from the processors for proc in p.processors: proc._done.clear() return p # #### CRUD interface to the data #### def update(self, media_data, tags=None): """Create/update the internal data given the media data and tags. Parameters ---------- f: vixen.directory.File instance tags: dict """ relpath = media_data.relpath if not self.has_media(relpath): index = len(self._relpath2index) self._relpath2index[relpath] = index for key in MediaData._fields: self._data[key].append(None) for tag in self.tags: self._tag_data[tag.name].append(tag.default) index = self._relpath2index[relpath] for i, key in enumerate(MediaData._fields): self._data[key][index] = media_data[i] if tags: for key, value in tags.items(): self._tag_data[key][index] = value media = self._media.get(relpath) if media is not None: media.update(media_data, tags) def get(self, relpath): """Given the relative path of some media, return a Media instance. """ if relpath in self._media: return self._media[relpath] else: data = {} index = self._relpath2index[relpath] for key in MediaData._fields: data[key] = self._data[key][index] tags = {} for key in self._tag_data: tags[key] = self._tag_data[key][index] media = Media.from_data(MediaData(**data), tags) media.on_trait_change(self._media_tag_handler, 'tags_items') self._media[relpath] = media return media def remove(self, relpaths): """Given a list of relative path of some media, remove them from the database. """ relpath2index = self._relpath2index indices = [(x, relpath2index[x]) for x in relpaths] for relpath, index in sorted(indices, reverse=True): last = len(relpath2index) - 1 if index == last: self._delete_record(last, relpath) else: self._replace_with_last_record(index, last) self._delete_record(last, relpath) def has_media(self, relpath): """Returns True if the media data is available. """ return relpath in self._relpath2index def keys(self): """Return all the keys for the media relative paths.""" return self._relpath2index.keys() def _get_media_attr(self, index, attr): """Given an index to the media, return its value. """ if attr in self._data: return self._data[attr][index] elif attr in self._tag_data: return self._tag_data[attr][index] # #### End of CRUD interface to the data #### def clean(self): """Scan the project and remove any dead entries. This is useful when you remove or rename files. This does not refresh the directory tree or set the number of files. It simply cleans up the db of files that no longer exist. """ logger.info('Cleaning project: %s', self.name) root_path = self.path to_remove = [] relpath2index = self._relpath2index for rpath in list(relpath2index.keys()): fname = os.path.join(root_path, rpath) if not os.path.exists(fname): to_remove.append(rpath) self.remove(to_remove) def export_csv(self, fname, cols=None): """Export metadata to a csv file. If `cols` are not specified, it writes out all the useful metadata. Parameters ----------- fname: str: a path to the csv file to dump. cols: sequence: a sequence of columns to write. """ logger.info('Exporting CSV: %s', fname) all_keys = ((set(MediaData._fields) | set(self._tag_data.keys())) - set(('ctime_', 'mtime_'))) if cols is None: cols = all_keys cols = list(sorted(cols)) data_cols = set([x for x in cols if x in self._data]) with io.open(fname, 'w', newline='', encoding='utf-8') as of: # Write the header. writer = csv.writer(of) writer.writerow(cols) for i in range(len(self._relpath2index)): line = [] for col in cols: if col in data_cols: elem = self._data[col][i] else: elem = self._tag_data[col][i] line.append(elem) writer.writerow(line) def import_csv(self, fname): """Read tag information from given CSV filename. Returns the success status and the error message if any. Note that this only applies tags for column headers with known tags. Unknown tags are not added. Parameters ---------- fname : str Input filename. """ logger.info('Importing tags from: %s', fname) has_header, header, dialect = _get_csv_headers(fname) if not has_header: return False, "The CSV file does not appear to have a header." if 'path' not in header: msg = "The CSV file does not have a 'path' column." return False, msg tags = {x: header.index(x.name) for x in self.tags if x.name in header} path_idx = header.index('path') TRUE = ('1', 't', 'true', 'y', 'yes') type_map = { 'bool': lambda x: x.lower() in TRUE, 'string': lambda x: x, 'text': lambda x: x, 'int': int, 'float': float } count = 0 total = 0 with io.open(fname, 'r', newline='', encoding='utf-8') as fp: reader = csv.reader(fp, dialect) next(reader) # Skip header for record in reader: total += 1 path = record[path_idx] rpath = relpath(path, self.path) index = self._relpath2index.get(rpath, None) media = self._media.get(rpath) if index is not None: count += 1 for tag, header_index in tags.items(): data = record[header_index] try: value = type_map[tag.type](data) if media is not None: media.tags[tag.name] = value else: self._tag_data[tag.name][index] = value except ValueError: pass msg = "Read tags for %d paths out of %d entries." % (count, total) if count == 0 and total > 0: msg += ("\nPlease check that your path column matches " "the media paths.") return False, msg else: msg += ("\nPlease check the imported tags and make sure you " "save the project.") return True, msg def load(self, fp=None): """Load media info from opened file object. """ if fp is None: if not exists(self.save_file): return fp = open_file(self.save_file, 'rb') else: fp = open_file(fp, 'rb') data = json_tricks.load( fp, preserve_order=False, ignore_comments=False ) fp.close() self.name = data.get('name', '') self.description = data.get('description', '') self.path = data.get('path') self.tags = [TagInfo(name=x[0], type=x[1]) for x in data['tags']] self.processors = [processor.load(x) for x in data.get('processors', [])] version = data.get('version') if version == 1: self._read_version1_media(data['media']) else: self._data = data['media_data'] self._tag_data = data['tag_data'] self._relpath2index = data['relpath2index'] root = Directory() root.__setstate__(data.get('root')) self.extensions = root.extensions self.root = root self.number_of_files = len(self._relpath2index) def save(self): """Save current media info to a file object """ if len(self.save_file) > 0: self.save_as(self.save_file) self._update_last_save_time() else: raise IOError("No valid save file set.") def save_as(self, fp): """Save copy to specified path. """ fp = open_file(fp, 'wb') tags = [(t.name, t.type) for t in self.tags] root = self.root.__getstate__() processors = [processor.dump(x) for x in self.processors] data = dict( version=2, path=self.path, name=self.name, description=self.description, tags=tags, media_data=self._data, tag_data=self._tag_data, relpath2index=self._relpath2index, root=root, processors=processors ) json_tricks.dump(data, fp, compression=True) fp.close() logger.info('Saved project: %s', self.name) def scan(self, refresh=False): """Find all the media recursively inside the root directory. This will not clobber existing records but will add any new ones. """ self._setup_root() def _scan(dir): for f in dir.files: if not self.has_media(f.relpath) or refresh: data = get_media_data(f.path, f.relpath) self.update(data) for d in dir.directories: if refresh: d.refresh() _scan(d) if refresh: self.root.refresh() _scan(self.root) self.number_of_files = len(self._relpath2index) def search(self, q): """A generator which yields the (filename, relpath) for each file satisfying the search query. """ logger.info('Searching for %s', q) try: parsed_q = self._query_parser.parse(q) except Exception: logger.warn("Invalid search expression: %s", q) print("Invalid search expression: %s" % q) return tag_types = self._get_tag_types() _cleanup_query(parsed_q, tag_types) for key, index in self._relpath2index.items(): if _search_media(parsed_q, index, self._get_media_attr): yield basename(key), key def refresh(self): logger.info('Refreshing project: %s', self.name) self.clean() self.scan(refresh=True) # #### Private protocol ################################################ def _setup_root(self): path = abspath(expanduser(self.path)) root = self.root if root is None or realpath(root.path) != realpath(path): self.root = Directory(path=path, extensions=self.extensions) def _tags_default(self): return [TagInfo(name='completed', type='bool')] def _save_file_default(self): if len(self.name) > 0: fname = sanitize_name(self.name) + '.vxn' d = get_project_dir() return get_non_existing_filename(join(d, fname)) else: return '' def _update_last_save_time(self): self.last_save_time = get_file_saved_time(self.save_file) def _last_save_time_default(self): if exists(self.save_file): return get_file_saved_time(self.save_file) else: return '' def _name_changed(self, name): if len(name) > 0: old_save_file = self.save_file old_dir = dirname(old_save_file) new_save_file = join(old_dir, sanitize_name(name) + '.vxn') if new_save_file != old_save_file: self.save_file = new_save_file if exists(old_save_file): shutil.move(old_save_file, self.save_file) def _extensions_changed(self, ext): if self.root is not None: self.root.extensions = ext def _extensions_items_changed(self): if self.root is not None: self.root.extensions = self.extensions def _get_tag_types(self): result = dict(COMMON_TAGS) result.update(dict((t.name, t.type) for t in self.tags)) return result def _make_schema(self): from whoosh.fields import BOOLEAN, DATETIME, TEXT, Schema kw = dict( type=TEXT, file_name=TEXT, path=TEXT, mtime=DATETIME, ctime=DATETIME, size=INT ) type_to_field = dict( string=TEXT, text=TEXT, int=INT, float=FLOAT, bool=BOOLEAN ) for tag in self.tags: kw[tag.name] = type_to_field[tag.type] return Schema(**kw) def _make_query_parser(self): schema = self._make_schema() qp = qparser.QueryParser('path', schema=schema) qp.add_plugin(qparser.GtLtPlugin()) from whoosh.qparser.dateparse import DateParserPlugin qp.add_plugin(DateParserPlugin()) return qp def __query_parser_default(self): return self._make_query_parser() def __data_default(self): data = {} for key in MediaData._fields: data[key] = [] return data def __tag_data_default(self): tags = {} for key in self.tags: tags[key.name] = [] return tags def _media_tag_handler(self, obj, tname, old, new): index = self._relpath2index[obj.relpath] for tag in new.changed: self._tag_data[tag][index] = obj.tags[tag] def _read_version1_media(self, media): data = self.__data_default() tag_data = self.__tag_data_default() relpath2index = {} keymap = dict.fromkeys(MediaData._fields) for k in keymap: keymap[k] = k keymap['_ctime'] = 'ctime_' keymap['_mtime'] = 'mtime_' for index, (key, m) in enumerate(media): relpath2index[key] = index tags = m.pop('tags') for tname, v in tags.items(): tag_data[tname].append(v) for k, v in m.items(): data[keymap[k]].append(v) if 'file_name' not in m: data['file_name'].append(basename(key)) data['mtime_'] = [datetime_to_long(x) for x in data['mtime_']] data['ctime_'] = [datetime_to_long(x) for x in data['ctime_']] self._data = data self._tag_data = tag_data self._relpath2index = relpath2index def _delete_record(self, index, relpath): for key in MediaData._fields: del self._data[key][index] for key in self._tag_data: del self._tag_data[key][index] if relpath in self._media: del self._media[relpath] del self._relpath2index[relpath] def _replace_with_last_record(self, index, last): _data = self._data _tag_data = self._tag_data for key in MediaData._fields: _data[key][index] = _data[key][last] for key in self._tag_data: _tag_data[key][index] = _tag_data[key][last] last_relpath = _data['relpath'][last] self._relpath2index[last_relpath] = index def _save_as_v1(self, fp): """Save copy to specified path. This mainly exists for testing and making sure we still read the old saved files. """ def _rewrite_dir(state): "Rewrite directories in the old format." state['files'] = [x[0] for x in state['files']] state['directories'] = [_rewrite_dir(d) for d in state['directories']] state.pop('relpath') state.pop('name') return state fp = open_file(fp, 'wb') media = [(key, self.get(key).to_dict()) for key in self._relpath2index] tags = [(t.name, t.type) for t in self.tags] root = _rewrite_dir(self.root.__getstate__()) processors = [processor.dump(x) for x in self.processors] for k, m in media: m['_ctime'] = long_to_datetime(m['_ctime']) m['_mtime'] = long_to_datetime(m['_mtime']) data = dict( version=1, path=self.path, name=self.name, description=self.description, tags=tags, media=media, root=root, processors=processors ) json_tricks.dump(data, fp, compression=True) fp.close() logger.info('Saved project: %s', self.name)
534
from io import StringIO from unittest import TestCase from dropSQL.parser.streams import * class StreamTestCase(TestCase): def test(self): s = '12' cs = Characters(StringIO(s)) ch = cs.peek().ok() self.assertEqual(ch, '1') ch = cs.peek().ok() self.assertEqual(ch, '1') ch = cs.next().ok() self.assertEqual(ch, '1') ch = cs.next().ok() self.assertEqual(ch, '2') r = cs.next() self.assertFalse(r) self.assertTrue(r.err()) r = cs.next() self.assertFalse(r) cs.back() r = cs.next() self.assertTrue(r) self.assertEqual(r.ok(), '2') cs.back(2) r = cs.next() self.assertTrue(r) self.assertEqual(r.ok(), '1')
538
MANIFEST = { "hilt": { "h1": { "offsets": {"blade": 0, "button": {"x": (8, 9), "y": (110, 111)}}, "colours": { "primary": (216, 216, 216), # d8d8d8 "secondary": (141, 141, 141), # 8d8d8d "tertiary": (180, 97, 19), # b46113 }, "length": 24, "materials": "Alloy metal/Salvaged materials", }, "h2": { "offsets": {"blade": 20, "button": {"x": (8, 8), "y": (100, 105)}}, "colours": { "primary": (112, 112, 112), # 707070 "secondary": (0, 0, 0), # 000000 "tertiary": (212, 175, 55), # 000000 }, "length": 24, "materials": "Alloy metal and carbon composite", }, "h3": { "offsets": {"blade": 0, "button": {"x": (10, 10), "y": (100, 118)}}, "colours": { "primary": (157, 157, 157), # 707070 "secondary": (0, 0, 0), # 000000 "tertiary": (180, 97, 19), # b46113 }, "length": 24, "materials": "Alloy metal", }, "h4": { "offsets": {"blade": 7, "button": {"x": (8, 9), "y": (92, 100)}}, "colours": { "primary": (0, 0, 0), # 000000 "secondary": (157, 157, 157), # 9d9d9d "tertiary": (180, 97, 19), # b46113 }, "length": 13, "materials": "Alloy metal", }, "h5": { "offsets": {"blade": 0, "button": {"x": (8, 8), "y": (92, 105)}}, "colours": { "primary": (111, 111, 111), # 6f6f6f "secondary": (0, 0, 0), # 000000 "tertiary": (180, 97, 19), # b46113 }, "length": 24, "materials": "Alloy metal", }, "h6": { "offsets": {"blade": 2, "button": {"x": (8, 9), "y": (112, 113)}}, "colours": { "primary": (120, 120, 120), # 787878 "secondary": (0, 0, 0), # 000000 "tertiary": (180, 97, 19), # b46113 }, "length": 22, "materials": "Alloy metal/Salvaged materials", }, "h7": { "offsets": {"blade": 0, "button": {"x": (8, 9), "y": (105, 113)}}, "colours": { "primary": (192, 192, 192), # c0c0c0 "secondary": (255, 215, 0), # ffd700 "tertiary": (0, 0, 0), # 000000 }, "length": 22, "materials": "Alloy metal and Gold", }, "h8": { "offsets": {"blade": 0, "button": {"x": (8, 9), "y": (100, 111)}}, "colours": { "primary": (216, 216, 216), # d8d8d8 "secondary": (180, 97, 19), # b46113 "tertiary": (0, 0, 0), # 000000 }, "length": 24, "materials": "Alloy metal/Copper", }, }, "blade": { "b1": {"colour": "Red", "crystal": "Adegan crystal", "type": "Sith"}, "b2": {"colour": "Blue", "crystal": "Zophis crystal", "type": "Jedi"}, "b3": {"colour": "Green", "crystal": "Nishalorite stone", "type": "Jedi"}, "b4": {"colour": "Yellow", "crystal": "Kimber stone", "type": "Jedi"}, "b5": {"colour": "White", "crystal": "Dragite gem", "type": "Jedi"}, "b6": {"colour": "Purple", "crystal": "Krayt dragon pearl", "type": "Jedi"}, "b7": {"colour": "Blue/Green", "crystal": "Dantari crystal", "type": "Jedi"}, "b8": { "colour": "Orange", "crystal": ["Ilum crystal", "Ultima Pearl"], "type": "Sith", }, "b9": { "colour": "Black", "crystal": "Obsidian", "type": ["Jedi", "Mandalorian"], }, }, "pommel": { "p1": {"length": 5,}, "p2": {"length": 14,}, "p3": {"length": 3,}, "p4": {"length": 8,}, "p5": {"length": 5,}, "p6": {"length": 5,}, "p7": {"length": 8,}, }, # These are lightsabers for a specific Jedi or Sith. Should use their name instead of "unique_urls": {""}, }
545
import json from cisco_sdwan_policy.BaseObject import BaseObject class Application(BaseObject): def __init__(self,name,app_list,is_app_family,id=None,reference=None,**kwargs): self.type = "appList" self.id = id self.name = name self.references = reference self.app_family=is_app_family self._entries = app_list self.url = "template/policy/list/app" super().__init__(**kwargs) self.modified=False def get_entries(self): return self._entries def set_entries(self,entries): self.modified=True self._entries=entries @classmethod def from_json(cls,jsonfile,**kwargs): id = jsonfile["listId"] name = jsonfile["name"] references = jsonfile.get("references") if len(jsonfile["entries"])>0 and jsonfile["entries"][0].get("app"): appFamily=False entries = [i["app"] for i in jsonfile["entries"]] else: if not jsonfile["entries"][0].get("appFamily"): return None else: appFamily=True entries = [i["appFamily"] for i in jsonfile["entries"]] return cls(name,entries,appFamily,id,references,**kwargs) def to_json(self): return { "name":self.name, "description":"Desc Not Required", "type":"app", "entries":[ {"appFamily" if self.app_family else "app":i} for i in self._entries] }
552
from functools import partial from corpustools.corpus.classes import Word from corpustools.symbolsim.edit_distance import edit_distance from corpustools.symbolsim.khorsi import khorsi from corpustools.symbolsim.phono_edit_distance import phono_edit_distance from corpustools.symbolsim.phono_align import Aligner from corpustools.multiproc import filter_mp, score_mp def _is_edit_distance_neighbor(w, query, sequence_type, max_distance): w_len = len(getattr(w, sequence_type)) query_len = len(getattr(query, sequence_type)) if w_len > query_len+max_distance: return False if w_len < query_len-max_distance: return False return edit_distance(getattr(w, sequence_type), getattr(query, sequence_type), sequence_type, max_distance) <= max_distance def _is_phono_edit_distance_neighbor(w, query, sequence_type, specifier, max_distance): return phono_edit_distance(getattr(w, sequence_type), getattr(query, sequence_type), sequence_type, specifier) <= max_distance def _is_khorsi_neighbor(w, query, freq_base, sequence_type, max_distance): return khorsi(getattr(w, sequence_type), getattr(query, sequence_type), freq_base, sequence_type, max_distance) >= max_distance def neighborhood_density_all_words(corpus_context, tierdict, tier_type = None, sequence_type = None, algorithm = 'edit_distance', max_distance = 1, output_format = 'spelling', num_cores = -1, settable_attr = None, collapse_homophones = False, stop_check = None, call_back = None): """Calculate the neighborhood density of all words in the corpus and adds them as attributes of the words. Parameters ---------- corpus_context : CorpusContext Context manager for a corpus algorithm : str The algorithm used to determine distance max_distance : float, optional Maximum edit distance from the queried word to consider a word a neighbor. stop_check : callable, optional Optional function to check whether to gracefully terminate early call_back : callable, optional Optional function to supply progress information during the function settable_attr: string Name of attribute that neighbourhood density results will be assigned to """ function = partial(neighborhood_density, corpus_context, tierdict = tierdict, tier_type = tier_type, sequence_type = sequence_type, algorithm = algorithm, max_distance = max_distance, collapse_homophones = collapse_homophones) if call_back is not None: call_back('Calculating neighborhood densities...') call_back(0,len(corpus_context)) cur = 0 results = dict() last_value_removed = None last_key_removed = None if num_cores == -1 or num_cores == 1: for w in corpus_context: if stop_check is not None and stop_check(): return if last_value_removed: tierdict[last_key_removed].append(last_value_removed) w_sequence = getattr(w, corpus_context.sequence_type) last_key_removed = str(w_sequence) for i, item in enumerate(tierdict[last_key_removed]): if str(item) == str(w): last_value_removed = tierdict[last_key_removed].pop(i) break res = neighborhood_density(corpus_context, w, tierdict, tier_type = tier_type, sequence_type = sequence_type, algorithm = algorithm, max_distance = max_distance, collapse_homophones = collapse_homophones) results[str(w)] = [getattr(r, output_format) for r in res[1]] setattr(w.original, settable_attr.name, res[0]) # for w in corpus_context: # if stop_check is not None and stop_check(): # return # cur += 1 # call_back(cur) # res = function(w) # results[str(w)] = [getattr(r, output_format) for r in res[1]] # setattr(w.original, settable_attr.name, res[0]-1) # #the -1 is to account for the fact that words are counted as their own neighbour, and this is incorrect # #subtracting 1 here is easier than fixing the neighbourhood density algorithm else: iterable = ((w,) for w in corpus_context) neighbors = score_mp(iterable, function, num_cores, call_back, stop_check, chunk_size = 1) for n in neighbors: #Have to look up the key, then look up the object due to how #multiprocessing pickles objects setattr(corpus_context.corpus.find(corpus_context.corpus.key(n[0])), #corpus_context.attribute.name, n[1][0]) settable_attr.name, n[1][0]) return results def neighborhood_density(corpus_context, query, tierdict, algorithm = 'edit_distance', max_distance = 1, collapse_homophones = False, force_quadratic = False, file_type = None, tier_type=None, sequence_type = None, stop_check = None, call_back = None): """Calculate the neighborhood density of a particular word in the corpus. Parameters ---------- corpus_context : CorpusContext Context manager for a corpus query : Word The word whose neighborhood density to calculate. algorithm : str The algorithm used to determine distance max_distance : float, optional Maximum edit distance from the queried word to consider a word a neighbor force_quadratic : bool Force use of the less efficient quadratic algorithm even when finding edit distance of 1 neighborhoods stop_check : callable, optional Optional function to check whether to gracefully terminate early call_back : callable, optional Optional function to supply progress information during the function Returns ------- tuple(int, set) Tuple of the number of neighbors and the set of neighbor Words. """ matches = [] query = ensure_query_is_word(query, corpus_context, corpus_context.sequence_type, tier_type) if call_back is not None: call_back('Finding neighbors for {}...'.format(query)) call_back(0,len(corpus_context)) cur = 0 if algorithm == 'edit_distance' and max_distance == 1 and not force_quadratic: return fast_neighborhood_density(corpus_context, query, corpus_context.sequence_type, tier_type, tierdict, file_type=file_type, collapse_homophones=collapse_homophones) if algorithm == 'edit_distance': is_neighbor = partial(_is_edit_distance_neighbor, sequence_type = corpus_context.sequence_type, max_distance = max_distance) elif algorithm == 'phono_edit_distance': is_neighbor = partial(_is_phono_edit_distance_neighbor, specifier = corpus_context.specifier, sequence_type = corpus_context.sequence_type, max_distance = max_distance) elif algorithm == 'khorsi': freq_base = corpus_context.get_frequency_base() is_neighbor = partial(_is_khorsi_neighbor, freq_base = freq_base, sequence_type = corpus_context.sequence_type, max_distance = max_distance) for w in corpus_context: if stop_check is not None and stop_check(): return if call_back is not None: cur += 1 if cur % 10 == 0: call_back(cur) if not is_neighbor(w, query): continue matches.append(w) neighbors = set(matches)-set([query]) return (len(neighbors), neighbors) def fast_neighborhood_density(corpus_context, query, sequence_type, tier_type, tierdict, file_type=None, trans_delimiter='.', collapse_homophones = False): """Generates all neighbors of edit distance <= 1 and searches for them in corpus_context. Will be faster than neighborhood_density when: n > m * (1 + s), where n: number of words in corpus m: length of query s: size of segment inventory """ neighbors = list() query = ensure_query_is_word(query, corpus_context, sequence_type, tier_type, file_type=file_type) for candidate in generate_neighbor_candidates(corpus_context, query, sequence_type): if tier_type.att_type == 'tier': cand_str = trans_delimiter.join(candidate) else: cand_str = ''.join(candidate) if cand_str in tierdict: for w in tierdict[cand_str]: w_sequence = getattr(w, sequence_type) if collapse_homophones and any(getattr(word, sequence_type) == w_sequence for word in neighbors): continue else: neighbors.append(w) return (len(neighbors), neighbors) def generate_neighbor_candidates(corpus_context, query, sequence_type): sequence = getattr(query, sequence_type) yield [str(c) for c in sequence] for i in range(len(sequence)): yield [str(c) for c in sequence[:i]] + [str(c) for c in sequence[i+1:]] # deletion for char in corpus_context.inventory: if str(char) not in ['#', sequence[i]]: yield [str(c) for c in sequence[:i]] + [str(char)] + [str(c) for c in sequence[i:]] # insertion yield [str(c) for c in sequence[:i]] + [str(char)] + [str(c) for c in sequence[i+1:]] # substitution for char in corpus_context.inventory: # final pass to get insertion at len+1 if str(char) not in ['#', sequence[i]]: yield [str(c) for c in sequence[:]] + [str(char)] # insertion def find_mutation_minpairs_all_words(corpus_context, tierdict, tier_type = None, num_cores = -1, collapse_homophones = False, stop_check = None, call_back = None): function = partial(find_mutation_minpairs, corpus_context, tier_type=tier_type, collapse_homophones = collapse_homophones) if call_back is not None: call_back('Calculating neighborhood densities...') call_back(0,len(corpus_context)) cur = 0 results = dict() last_value_removed = None last_key_removed = None if num_cores == -1 or num_cores == 1: for w in corpus_context: if stop_check is not None and stop_check(): return if last_value_removed: tierdict[last_key_removed].append(last_value_removed) w_sequence = getattr(w, corpus_context.sequence_type) last_key_removed = str(w_sequence) for i, item in enumerate(tierdict[last_key_removed]): if str(item) == str(w): last_value_removed = tierdict[last_key_removed].pop(i) break res = find_mutation_minpairs(corpus_context, w, tier_type=tier_type, collapse_homophones = collapse_homophones) results[str(w)] = res[1] setattr(w.original, corpus_context.attribute.name, res[0]) # for w in corpus_context: # if stop_check is not None and stop_check(): # return # cur += 1 # call_back(cur) # res = function(w) # results[str(w)] = res[1]#[str(r) for r in res[1]] # setattr(w.original, corpus_context.attribute.name, res[0]) else: iterable = ((w,) for w in corpus_context) neighbors = score_mp(iterable, function, num_cores, call_back, stop_check, chunk_size= 1) for n in neighbors: #Have to look up the key, then look up the object due to how #multiprocessing pickles objects setattr(corpus_context.corpus.find(corpus_context.corpus.key(n[0])), corpus_context.attribute.name, n[1][0]) return results def find_mutation_minpairs(corpus_context, query, tier_type = None, collapse_homophones = False, stop_check = None, call_back = None): """Find all minimal pairs of the query word based only on segment mutations (not deletions/insertions) Parameters ---------- corpus_context : CorpusContext Context manager for a corpus query : Word The word whose minimal pairs to find stop_check : callable or None Optional function to check whether to gracefully terminate early call_back : callable or None Optional function to supply progress information during the function Returns ------- list The found minimal pairs for the queried word """ matches = [] sequence_type = corpus_context.sequence_type query = ensure_query_is_word(query, corpus_context, corpus_context.sequence_type, tier_type) if call_back is not None: call_back('Finding neighbors...') call_back(0,len(corpus_context)) cur = 0 al = Aligner(features_tf=False, ins_penalty=float('inf'), del_penalty=float('inf'), sub_penalty=1) for w in corpus_context: w_sequence = getattr(w, sequence_type) query_sequence = getattr(query, sequence_type) if stop_check is not None and stop_check(): return if call_back is not None: cur += 1 if cur % 10 == 0: call_back(cur) if (len(w_sequence) > len(query_sequence)+1 or len(w_sequence) < len(query_sequence)-1): continue m = al.make_similarity_matrix(query_sequence, w_sequence) if m[-1][-1]['f'] != 1: continue w_sequence = getattr(w, sequence_type) if collapse_homophones and any(getattr(m, sequence_type) == w_sequence for m in matches): continue else: #matches.append(str(w_sequence)) matches.append(w) matches = [m.spelling for m in matches] neighbors = list(set(matches)-set([str(query_sequence)])) return (len(neighbors), neighbors) def ensure_query_is_word(query, corpus, sequence_type, tier_type, trans_delimiter='.', file_type=None): if isinstance(query, Word): query_word = query else: if tier_type.att_type == 'spelling': if file_type == sequence_type: query_word = Word(**{sequence_type: list(query)}) else: query_word = query.replace(trans_delimiter, '') query_word = Word(**{sequence_type: list(query_word)}) elif tier_type.att_type == 'tier': if file_type == sequence_type: query_with̠td = '.'.join(query) if '.' not in query else query for entry in corpus: corpus_word_with_td = str(getattr(entry, sequence_type)) if query_with̠td == corpus_word_with_td: # if a word in corpus has the same transcription return entry # that word in the corpus is to be referred to. # the following should be run if no word found in corpus with the transcription new_query = parse(query, trans_delimiter) query_word = Word(**{sequence_type: new_query}) else: # if file contains spelling try: query_word = corpus.corpus.find(query) except KeyError: # if the word in the file can't be found in the corpus new_query = parse(query, trans_delimiter) query_word = Word(**{sequence_type: list(new_query)}) return query_word def parse(word, delimiter): return word.split(delimiter) if delimiter in word else list(word)
571
from CommonServerPython import * ''' IMPORTS ''' import re import requests # Disable insecure warnings requests.packages.urllib3.disable_warnings() ''' GLOBALS/PARAMS ''' VENDOR = 'Have I Been Pwned? V2' MAX_RETRY_ALLOWED = demisto.params().get('max_retry_time', -1) API_KEY = demisto.params().get('api_key') USE_SSL = not demisto.params().get('insecure', False) BASE_URL = 'https://haveibeenpwned.com/api/v3' HEADERS = { 'hibp-api-key': API_KEY, 'user-agent': 'DBOT-API', 'Content-Type': 'application/json', 'Accept': 'application/json' } DEFAULT_DBOT_SCORE_EMAIL = 2 if demisto.params().get('default_dbot_score_email') == 'SUSPICIOUS' else 3 DEFAULT_DBOT_SCORE_DOMAIN = 2 if demisto.params().get('default_dbot_score_domain') == 'SUSPICIOUS' else 3 SUFFIXES = { "email": '/breachedaccount/', "domain": '/breaches?domain=', "username": '/breachedaccount/', "paste": '/pasteaccount/', "email_truncate_verified": '?truncateResponse=false&includeUnverified=true', "domain_truncate_verified": '&truncateResponse=false&includeUnverified=true', "username_truncate_verified": '?truncateResponse=false&includeUnverified=true' } RETRIES_END_TIME = datetime.min ''' HELPER FUNCTIONS ''' def http_request(method, url_suffix, params=None, data=None): while True: res = requests.request( method, BASE_URL + url_suffix, verify=USE_SSL, params=params, data=data, headers=HEADERS ) if res.status_code != 429: # Rate limit response code break if datetime.now() > RETRIES_END_TIME: return_error('Max retry time has exceeded.') wait_regex = re.search(r'\d+', res.json()['message']) if wait_regex: wait_amount = wait_regex.group() else: demisto.error('failed extracting wait time will use default (5). Res body: {}'.format(res.text)) wait_amount = 5 if datetime.now() + timedelta(seconds=int(wait_amount)) > RETRIES_END_TIME: return_error('Max retry time has exceeded.') time.sleep(int(wait_amount)) if res.status_code == 404: return None if not res.status_code == 200: if not res.status_code == 401: demisto.error( 'Error in API call to Pwned Integration [%d]. Full text: %s' % (res.status_code, res.text)) return_error('Error in API call to Pwned Integration [%d] - %s' % (res.status_code, res.reason)) return None return res.json() def html_description_to_human_readable(breach_description): """ Converting from html description to hr :param breach_description: Description of breach from API response :return: Description string that altered HTML urls to clickable urls for better readability in war-room """ html_link_pattern = re.compile('<a href="(.+?)"(.+?)>(.+?)</a>') patterns_found = html_link_pattern.findall(breach_description) for link in patterns_found: html_actual_address = link[0] html_readable_name = link[2] link_from_desc = '[' + html_readable_name + ']' + '(' + html_actual_address + ')' breach_description = re.sub(html_link_pattern, link_from_desc, breach_description, count=1) return breach_description def data_to_markdown(query_type, query_arg, api_res, api_paste_res=None): records_found = False md = '### Have I Been Pwned query for ' + query_type.lower() + ': *' + query_arg + '*\n' if api_res: records_found = True for breach in api_res: verified_breach = 'Verified' if breach['IsVerified'] else 'Unverified' md += '#### ' + breach['Title'] + ' (' + breach['Domain'] + '): ' + str(breach['PwnCount']) + \ ' records breached [' + verified_breach + ' breach]\n' md += 'Date: **' + breach['BreachDate'] + '**\n\n' md += html_description_to_human_readable(breach['Description']) + '\n' md += 'Data breached: **' + ','.join(breach['DataClasses']) + '**\n' if api_paste_res: records_found = True pastes_list = [] for paste_breach in api_paste_res: paste_entry = \ { 'Source': paste_breach['Source'], 'Title': paste_breach['Title'], 'ID': paste_breach['Id'], 'Date': '', 'Amount of emails in paste': str(paste_breach['EmailCount']) } if paste_breach['Date']: paste_entry['Date'] = paste_breach['Date'].split('T')[0] pastes_list.append(paste_entry) md += tableToMarkdown('The email address was found in the following "Pastes":', pastes_list, ['ID', 'Title', 'Date', 'Source', 'Amount of emails in paste']) if not records_found: md += 'No records found' return md def create_dbot_score_dictionary(indicator_value, indicator_type, dbot_score): return { 'Indicator': indicator_value, 'Type': indicator_type, 'Vendor': VENDOR, 'Score': dbot_score } def create_context_entry(context_type, context_main_value, comp_sites, comp_pastes, malicious_score): context_dict = dict() # dict if context_type == 'email': context_dict['Address'] = context_main_value else: context_dict['Name'] = context_main_value context_dict['Pwned-V2'] = { 'Compromised': { 'Vendor': VENDOR, 'Reporters': ', '.join(comp_sites + comp_pastes) } } if malicious_score == 3: context_dict['Malicious'] = add_malicious_to_context(context_type) return context_dict def add_malicious_to_context(malicious_type): return { 'Vendor': VENDOR, 'Description': 'The ' + malicious_type + ' has been compromised' } def email_to_entry_context(email, api_email_res, api_paste_res): dbot_score = 0 comp_email = dict() # type: dict comp_sites = sorted([item['Title'] for item in api_email_res]) comp_pastes = sorted(set(item['Source'] for item in api_paste_res)) if len(comp_sites) > 0: dbot_score = DEFAULT_DBOT_SCORE_EMAIL email_context = create_context_entry('email', email, comp_sites, comp_pastes, DEFAULT_DBOT_SCORE_EMAIL) comp_email[outputPaths['email']] = email_context comp_email['DBotScore'] = create_dbot_score_dictionary(email, 'email', dbot_score) return comp_email def domain_to_entry_context(domain, api_res): comp_sites = [item['Title'] for item in api_res] comp_sites = sorted(comp_sites) comp_domain = dict() # type: dict dbot_score = 0 if len(comp_sites) > 0: dbot_score = DEFAULT_DBOT_SCORE_DOMAIN domain_context = create_context_entry('domain', domain, comp_sites, [], DEFAULT_DBOT_SCORE_DOMAIN) comp_domain[outputPaths['domain']] = domain_context comp_domain['DBotScore'] = create_dbot_score_dictionary(domain, 'domain', dbot_score) return comp_domain def set_retry_end_time(): global RETRIES_END_TIME if MAX_RETRY_ALLOWED != -1: RETRIES_END_TIME = datetime.now() + timedelta(seconds=int(MAX_RETRY_ALLOWED)) ''' COMMANDS + REQUESTS FUNCTIONS ''' def test_module(args_dict): """ If the http request was successful the test will return OK :return: 3 arrays of outputs """ http_request('GET', SUFFIXES.get("username", '') + 'test') return ['ok'], [None], [None] def pwned_email_command(args_dict): """ Executing the pwned request for emails list, in order to support list input, the function returns 3 lists of outputs :param args_dict: the demisto argument - in this case the email list is needed :return: 3 arrays of outputs """ email_list = argToList(args_dict.get('email', '')) api_email_res_list, api_paste_res_list = pwned_email(email_list) md_list = [] ec_list = [] for email, api_email_res, api_paste_res in zip(email_list, api_email_res_list, api_paste_res_list): md_list.append(data_to_markdown('Email', email, api_email_res, api_paste_res)) ec_list.append(email_to_entry_context(email, api_email_res or [], api_paste_res or [])) return md_list, ec_list, api_email_res_list def pwned_email(email_list): """ Executing the http requests :param email_list: the email list that needed for the http requests :return: 2 arrays of http requests outputs """ api_email_res_list = [] api_paste_res_list = [] for email in email_list: email_suffix = SUFFIXES.get("email") + email + SUFFIXES.get("email_truncate_verified") paste_suffix = SUFFIXES.get("paste") + email api_email_res_list.append(http_request('GET', url_suffix=email_suffix)) api_paste_res_list.append(http_request('GET', url_suffix=paste_suffix)) return api_email_res_list, api_paste_res_list def pwned_domain_command(args_dict): """ Executing the pwned request for domains list, in order to support list input, the function returns 3 lists of outputs :param args_dict: the demisto argument - in this case the domain list is needed :return: 3 arrays of outputs """ domain_list = argToList(args_dict.get('domain', '')) api_res_list = pwned_domain(domain_list) md_list = [] ec_list = [] for domain, api_res in zip(domain_list, api_res_list): md_list.append(data_to_markdown('Domain', domain, api_res)) ec_list.append(domain_to_entry_context(domain, api_res or [])) return md_list, ec_list, api_res_list def pwned_domain(domain_list): """ Executing the http request :param domain_list: the domains list that needed for the http requests :return: an array of http requests outputs """ api_res_list = [] for domain in domain_list: suffix = SUFFIXES.get("domain") + domain + SUFFIXES.get("domain_truncate_verified") api_res_list.append(http_request('GET', url_suffix=suffix)) return api_res_list def pwned_username_command(args_dict): """ Executing the pwned request for usernames list, in order to support list input, the function returns 3 lists of outputs :param args_dict: the demisto argument - in this case the username list is needed :return: 3 arrays of outputs """ username_list = argToList(args_dict.get('username', '')) api_res_list = pwned_username(username_list) md_list = [] ec_list = [] for username, api_res in zip(username_list, api_res_list): md_list.append(data_to_markdown('Username', username, api_res)) ec_list.append(domain_to_entry_context(username, api_res or [])) return md_list, ec_list, api_res_list def pwned_username(username_list): """ Executing the http request :param username_list: the username list that needed for the http requests :return: an array of http requests outputs """ api_res_list = [] for username in username_list: suffix = SUFFIXES.get("username") + username + SUFFIXES.get("username_truncate_verified") api_res_list.append(http_request('GET', url_suffix=suffix)) return api_res_list command = demisto.command() LOG('Command being called is: {}'.format(command)) try: handle_proxy() set_retry_end_time() commands = { 'test-module': test_module, 'email': pwned_email_command, 'pwned-email': pwned_email_command, 'domain': pwned_domain_command, 'pwned-domain': pwned_domain_command, 'pwned-username': pwned_username_command } if command in commands: md_list, ec_list, api_email_res_list = commands[command](demisto.args()) for md, ec, api_paste_res in zip(md_list, ec_list, api_email_res_list): return_outputs(md, ec, api_paste_res) # Log exceptions except Exception as e: return_error(str(e))
596
from itertools import product import numpy as np import pytest from alibi_detect.utils.discretizer import Discretizer x = np.random.rand(10, 4) n_features = x.shape[1] feature_names = [str(_) for _ in range(n_features)] categorical_features = [[], [1, 3]] percentiles = [list(np.arange(25, 100, 25)), list(np.arange(10, 100, 10))] tests = list(product(categorical_features, percentiles)) n_tests = len(tests) @pytest.fixture def cats_and_percentiles(request): cat, perc = tests[request.param] return cat, perc @pytest.mark.parametrize('cats_and_percentiles', list(range(n_tests)), indirect=True) def test_discretizer(cats_and_percentiles): cat, perc = cats_and_percentiles disc = Discretizer(x, cat, feature_names, perc) to_disc = list(disc.names.keys()) assert len(to_disc) == (x.shape[1] - len(cat)) x_disc = disc.discretize(x) for k, v in disc.names.items(): assert len(v) <= len(perc) + 1 assert callable(disc.lambdas[k]) assert (x_disc[:, k].min() == 0).all() assert (x_disc[:, k].max() == len(perc)).all() for i in range(x.shape[1]): if i not in to_disc: assert (x_disc[:, i] == x[:, i]).all()
600
import hashlib from typing import TypeVar, Union import redis from openff.toolkit.topology import Molecule from openff.bespokefit.executor.services.qcgenerator import worker from openff.bespokefit.schema.tasks import HessianTask, OptimizationTask, Torsion1DTask from openff.bespokefit.utilities.molecule import canonical_order_atoms _T = TypeVar("_T", HessianTask, OptimizationTask, Torsion1DTask) def _canonicalize_task(task: _T) -> _T: task = task.copy(deep=True) # Ensure the SMILES has a canonical ordering to help ensure cache hits. canonical_molecule = canonical_order_atoms( Molecule.from_smiles(task.smiles, allow_undefined_stereo=True) ) if isinstance(task, Torsion1DTask): map_to_atom_index = { j: i for i, j in canonical_molecule.properties["atom_map"].items() } central_atom_indices = sorted( map_to_atom_index[task.central_bond[i]] for i in (0, 1) ) canonical_molecule.properties["atom_map"] = { atom_index: (i + 1) for i, atom_index in enumerate(central_atom_indices) } canonical_smiles = canonical_molecule.to_smiles( isomeric=True, explicit_hydrogens=True, mapped=True ) task.central_bond = (1, 2) else: canonical_smiles = canonical_molecule.to_smiles( isomeric=True, explicit_hydrogens=True, mapped=False ) task.smiles = canonical_smiles return task def cached_compute_task( task: Union[HessianTask, OptimizationTask, Torsion1DTask], redis_connection: redis.Redis, ) -> str: """Checks to see if a QC task has already been executed and if not send it to a worker. """ if isinstance(task, Torsion1DTask): compute = worker.compute_torsion_drive elif isinstance(task, OptimizationTask): compute = worker.compute_optimization elif isinstance(task, HessianTask): compute = worker.compute_hessian else: raise NotImplementedError() # Canonicalize the task to improve the cache hit rate. task = _canonicalize_task(task) task_hash = hashlib.sha512(task.json().encode()).hexdigest() task_id = redis_connection.hget("qcgenerator:task-ids", task_hash) if task_id is not None: return task_id.decode() task_id = compute.delay(task_json=task.json()).id redis_connection.hset("qcgenerator:types", task_id, task.type) # Make sure to only set the hash after the type is set in case the connection # goes down before this information is entered and subsequently discarded. redis_connection.hset("qcgenerator:task-ids", task_hash, task_id) return task_id
605
import contextlib from datetime import date from datetime import datetime from datetime import timezone from functools import wraps from io import BytesIO from itertools import count from typing import Any from typing import Dict from typing import Sequence import pytest from dateutil.parser import parse as parse_date from dateutil.relativedelta import relativedelta from django import forms from django.core.exceptions import ValidationError from django.template.loader import render_to_string from django.urls import reverse from freezegun import freeze_time from lxml import etree from common.models.records import TrackedModel from common.renderers import counter_generator from common.serializers import validate_taric_xml_record_order from common.util import TaricDateRange from common.util import get_accessor from common.util import get_field_tuple INTERDEPENDENT_IMPORT_IMPLEMENTED = True UPDATE_IMPORTER_IMPLEMENTED = True EXPORT_REFUND_NOMENCLATURE_IMPLEMENTED = False COMMODITIES_IMPLEMENTED = True MEURSING_TABLES_IMPLEMENTED = False PARTIAL_TEMPORARY_STOP_IMPLEMENTED = False UTC = timezone.utc requires_commodities = pytest.mark.skipif( not COMMODITIES_IMPLEMENTED, reason="Commodities not implemented", ) requires_export_refund_nomenclature = pytest.mark.skipif( not EXPORT_REFUND_NOMENCLATURE_IMPLEMENTED, reason="Export refund nomenclature not implemented", ) requires_meursing_tables = pytest.mark.skipif( not MEURSING_TABLES_IMPLEMENTED, reason="Meursing tables not implemented", ) requires_partial_temporary_stop = pytest.mark.skipif( not PARTIAL_TEMPORARY_STOP_IMPLEMENTED, reason="Partial temporary stop not implemented", ) requires_interdependent_import = pytest.mark.skipif( not INTERDEPENDENT_IMPORT_IMPLEMENTED, reason="Interdependent imports not implemented", ) requires_update_importer = pytest.mark.skipif( not UPDATE_IMPORTER_IMPLEMENTED, reason="Requires Updating importers to be implemented", ) @contextlib.contextmanager def raises_if(exception, expected): try: yield except exception: if not expected: raise else: if expected: pytest.fail(f"Did not raise {exception}") def check_validator(validate, value, expected_valid): try: validate(value) except ValidationError: if expected_valid: pytest.fail(f'Unexpected validation error for value "{value}"') except Exception: raise else: if not expected_valid: pytest.fail(f'Expected validation error for value "{value}"') def make_duplicate_record(factory, identifying_fields=None): """Creates two records using the passed factory that are duplicates of each other and returns the record created last.""" existing = factory.create() # allow overriding identifying_fields if identifying_fields is None: identifying_fields = list(factory._meta.model.identifying_fields) return factory.create( **dict(get_field_tuple(existing, field) for field in identifying_fields) ) def make_non_duplicate_record(factory, identifying_fields=None): """Creates two records using the passed factory that are not duplicates of each other and returns the record created last.""" existing = factory.create() not_duplicate = factory.create() if identifying_fields is None: identifying_fields = list(factory._meta.model.identifying_fields) assert any( get_field_tuple(existing, f) != get_field_tuple(not_duplicate, f) for f in identifying_fields ) return not_duplicate def get_checkable_data(model: TrackedModel, ignore=frozenset()): """ Returns a dict representing the model's data ignoring any automatically set fields and fields with names passed to `ignore`. The returned data will contain the identifying fields for any linked models rather than internal PKs. For example: get_checkable_data(FootnoteDescriptionFactory(), ignore={"sid"}) # { # "description": "My sample footnote text", # "described_footnote": { # "footnote_type__footnote_type_id": "FN" # "footnote_id": "123", # }, # } """ checked_field_names = {f.name for f in model.copyable_fields} - ignore data = { name: getattr(model, get_accessor(model._meta.get_field(name))) for name in checked_field_names } identifying_fields = { name: data[name].get_identifying_fields() for name in checked_field_names if hasattr(data[name], "get_identifying_fields") } data.update(identifying_fields) return data def assert_records_match( expected: TrackedModel, imported: TrackedModel, ignore=frozenset(), ): """ Asserts that every value for every field in the imported model is the same as the data in the expected model. System fields that will change from model to model are not checked. Any field names given to `ignore` will also not be checked. """ expected_data = get_checkable_data(expected, ignore=ignore) imported_data = get_checkable_data(imported, ignore=ignore) assert expected_data == imported_data def assert_many_records_match( expected: Sequence[TrackedModel], imported: Sequence[TrackedModel], ignore=frozenset(), ): """ Asserts that every value for every field in the imported models is the same as the data in the expected models, and that the count of both is equal. System fields that will change from model to model are not checked. Any field names given to `ignore` will also not be checked. """ expected_data = [get_checkable_data(e, ignore=ignore) for e in expected] imported_data = [get_checkable_data(i, ignore=ignore) for i in imported] assert expected_data == imported_data _transaction_counter = count(start=1) def generate_test_import_xml(obj: dict) -> BytesIO: xml = render_to_string( template_name="workbaskets/taric/transaction_detail.xml", context={ "envelope_id": next(_transaction_counter), "tracked_models": [obj], "transaction_id": next(_transaction_counter), "message_counter": counter_generator(), "counter_generator": counter_generator, }, ) return BytesIO(xml.encode()) def taric_xml_record_codes(xml): """Yields tuples of (record_code, subrecord_code)""" records = xml.xpath(".//*[local-name() = 'record']") codes = etree.XPath( ".//*[local-name()='record.code' or local-name()='subrecord.code']/text()", ) return [tuple(codes(record)) for record in records] def validate_taric_xml( factory=None, instance=None, factory_kwargs=None, check_order=True, ): def decorator(func): def wraps( api_client, taric_schema, approved_transaction, valid_user, *args, **kwargs, ): if not factory and not instance: raise AssertionError( "Either a factory or an object instance need to be provided", ) if factory and instance: raise AssertionError( "Either a factory or an object instance need to be provided - not both.", ) current_instance = instance or factory.create( transaction=approved_transaction, **factory_kwargs or {} ) api_client.force_login(user=valid_user) response = api_client.get( reverse( "workbaskets:workbasket-detail", kwargs={"pk": approved_transaction.workbasket.pk}, ), {"format": "xml"}, ) assert response.status_code == 200 content = response.content xml = etree.XML(content) taric_schema.validate(xml) assert not taric_schema.error_log, f"XML errors: {taric_schema.error_log}" if check_order: validate_taric_xml_record_order(xml) kwargs = {"xml": xml, **kwargs} func( *args, **kwargs, ) return wraps return decorator class Dates: deltas = { "normal": (relativedelta(), relativedelta(months=+1)), "earlier": (relativedelta(years=-1), relativedelta(years=-1, months=+1)), "later": ( relativedelta(years=+1, months=+1, days=+1), relativedelta(years=+1, months=+2), ), "big": (relativedelta(years=-2), relativedelta(years=+2, days=+1)), "adjacent": (relativedelta(days=+1), relativedelta(months=+1)), "adjacent_earlier": (relativedelta(months=-1), relativedelta(days=-1)), "adjacent_later": (relativedelta(months=+1, days=+1), relativedelta(months=+2)), "adjacent_no_end": (relativedelta(months=+1, days=+1), None), "adjacent_even_later": ( relativedelta(months=+2, days=+1), relativedelta(months=+3), ), "adjacent_earlier_big": ( relativedelta(years=-2, months=-2), relativedelta(years=-2), ), "adjacent_later_big": ( relativedelta(months=+1, days=+1), relativedelta(years=+2, months=+2), ), "overlap_normal": ( relativedelta(days=+15), relativedelta(days=+14, months=+1, years=+1), ), "overlap_normal_earlier": ( relativedelta(months=-1, days=+14), relativedelta(days=+14), ), "overlap_normal_same_year": ( relativedelta(days=+15), relativedelta(days=+14, months=+1), ), "overlap_big": (relativedelta(years=+1), relativedelta(years=+3, days=+2)), "after_big": ( relativedelta(years=+3, months=+1), relativedelta(years=+3, months=+2), ), "backwards": (relativedelta(months=+1), relativedelta(days=+1)), "starts_with_normal": (relativedelta(), relativedelta(days=+14)), "ends_with_normal": (relativedelta(days=+14), relativedelta(months=+1)), "current": (relativedelta(weeks=-4), relativedelta(weeks=+4)), "future": (relativedelta(weeks=+10), relativedelta(weeks=+20)), "no_end": (relativedelta(), None), "normal_first_half": (relativedelta(), relativedelta(days=+14)), } @property def now(self): return self.datetime_now.date() @property def datetime_now(self): return datetime.now(tz=UTC).replace(hour=0, minute=0, second=0, microsecond=0) def __getattr__(self, name): if name in self.deltas: start, end = self.deltas[name] start = self.now + start if end is not None: end = self.now + end return TaricDateRange(start, end) raise AttributeError(name) @classmethod def short_before(cls, dt): return TaricDateRange( dt + relativedelta(months=-1), dt + relativedelta(days=-14), ) @classmethod def medium_before(cls, dt): return TaricDateRange( dt + relativedelta(months=-1), dt + relativedelta(days=-1), ) @classmethod def short_after(cls, dt): return TaricDateRange( dt + relativedelta(days=+14), dt + relativedelta(months=+1), ) @classmethod def short_overlap(cls, dt): return TaricDateRange( dt + relativedelta(months=-1), dt + relativedelta(months=+1), ) @classmethod def no_end_before(cls, dt): return TaricDateRange( dt + relativedelta(months=-1), None, ) def only_applicable_after(cutoff): """ Decorator which asserts that a test fails after a specified cutoff date. :param cutoff: A date string, or datetime object before which the test should fail. """ cutoff = parse_date(cutoff) def decorator(fn): @wraps(fn) def do_test(*args, **kwargs): # test should pass normally fn(*args, **kwargs) # test should fail before cutoff with freeze_time(cutoff + relativedelta(days=-1)): try: fn(*args, **kwargs) except pytest.fail.Exception: pass except Exception: raise else: pytest.fail(f"Rule applied before {cutoff:%Y-%m-%d}") return True return do_test return decorator def validity_period_post_data(start: date, end: date) -> Dict[str, int]: """ Construct a POST data fragment for the validity period start and end dates of a ValidityPeriodForm from the given date objects, eg: >>> validity_period_post_data( >>> datetime.date(2021, 1, 2), >>> datetime.date(2022, 3, 4), >>> ) { "start_date_0": 1, "start_date_1": 2, "start_date_2": 2021, "end_date_0": 4, "end_date_1": 3, "end_date_2": 2022, } """ return { f"{name}_{i}": part for name, date in (("start_date", start), ("end_date", end)) for i, part in enumerate([date.day, date.month, date.year]) } def get_form_data(form: forms.ModelForm) -> Dict[str, Any]: """Returns a dictionary of the fields that the form will put onto a page and their current values, taking account of any fields that have sub-fields and hence result in multiple HTML <input> objects.""" data = {**form.initial} for field in form.rendered_fields: value = data[field] if field in data else form.fields[field].initial if hasattr(form.fields[field].widget, "decompress"): # If the widget can be decompressed, then it is not just a simple # value and has some internal structure. So we need to generate one # form item per decompressed value and append the name with _0, _1, # etc. This mirrors the MultiValueWidget in django/forms/widgets.py. if field in data: del data[field] value = form.fields[field].widget.decompress(value) data.update( **{f"{field}_{i}": v for i, v in enumerate(value) if v is not None} ) elif value is not None: data.setdefault(field, value) return data
608
import logging import unittest from pyinstrument import Profiler from nuplan.planning.scenario_builder.nuplan_db.test.nuplan_scenario_test_utils import get_test_nuplan_scenario from nuplan.planning.simulation.history.simulation_history_buffer import SimulationHistoryBuffer from nuplan.planning.simulation.observation.idm_agents import IDMAgents from nuplan.planning.simulation.simulation_time_controller.simulation_iteration import SimulationIteration logger = logging.getLogger(__name__) logging.basicConfig(level=logging.INFO) class TestProfileIDM(unittest.TestCase): """ Profiling test for IDM agents. """ def setUp(self) -> None: """ Inherited, see super class. """ self.n_repeat_trials = 1 self.display_results = True self.scenario = get_test_nuplan_scenario() def test_profile_idm_agent_observation(self) -> None: """Profile IDMAgents.""" profiler = Profiler(interval=0.0001) profiler.start() # How many times to repeat runtime test for _ in range(self.n_repeat_trials): observation = IDMAgents( target_velocity=10, min_gap_to_lead_agent=0.5, headway_time=1.5, accel_max=1.0, decel_max=2.0, scenario=self.scenario, ) for step in range(self.scenario.get_number_of_iterations() - 1): iteration = SimulationIteration(time_point=self.scenario.get_time_point(step), index=step) next_iteration = SimulationIteration(time_point=self.scenario.get_time_point(step + 1), index=step + 1) buffer = SimulationHistoryBuffer.initialize_from_list( 1, [self.scenario.get_ego_state_at_iteration(step)], [self.scenario.get_tracked_objects_at_iteration(step)], next_iteration.time_point.time_s - iteration.time_point.time_s, ) observation.update_observation(iteration, next_iteration, buffer) profiler.stop() if self.display_results: logger.info(profiler.output_text(unicode=True, color=True)) if __name__ == "__main__": unittest.main()
623
import math from vp import geom_tools def horizon_error(ground_truth_horizon, detected_horizon, image_dims): """Calculates error in a detected horizon. This measures the max distance between the detected horizon line and the ground truth horizon line, within the image's x-axis, and normalized by image height. Args: ground_truth_horizon: Tuple with (slope, intercept) for the GT horizon line. detected_horizon: Tuple with (slope, intercept) for the detected horizon line. image_dims: Tuple of integers, (width, height) of the image, in pixels. Returns: Float, or None if a horizon is missing altogether. """ if ground_truth_horizon is None or detected_horizon is None: return None def gt(x): return ground_truth_horizon[0] * x + ground_truth_horizon[1] def dt(x): return detected_horizon[0] * x + detected_horizon[1] width, height = image_dims return max(abs(gt(0) - dt(0)), abs(gt(width) - dt(width))) / height def vp_direction_error(ground_truth_vps, detected_vps, image_dims): """Measures error in direction from center of detected vanishing points. Each detected VP is matched with its closest unclaimed ground truth VP. Args: ground_truth_vps: List of ground truth VP point tuples. detected_vps: List of detected VP point tuples. image_dims: Tuple of integers, (width, height) of the image, in pixels. Returns: List with float degrees of error for each ground truth VP. Error is None for missing VPs. """ principal_point = (image_dims[0] // 2, image_dims[1] // 2) point_pair_dists = [] for gt_vp in ground_truth_vps: for dt_vp in detected_vps: gt_angle = geom_tools.get_line_angle(( principal_point[0], principal_point[1], gt_vp[0], gt_vp[1])) dt_angle = geom_tools.get_line_angle(( principal_point[0], principal_point[1], dt_vp[0], dt_vp[1])) angle_diff = 180 - abs(abs(gt_angle - dt_angle) - 180) point_pair_dists.append((angle_diff, gt_vp, dt_vp)) point_pair_dists = sorted(point_pair_dists, key=lambda k: k[0]) gt_vp_to_error = {} seen_dt_vps = set() for distance, gt_vp, dt_vp in point_pair_dists: if gt_vp in gt_vp_to_error or dt_vp in seen_dt_vps: continue gt_vp_to_error[gt_vp] = distance seen_dt_vps.add(dt_vp) return [gt_vp_to_error.get(gt, None) for gt in ground_truth_vps] def location_accuracy_error(ground_truth_vps, detected_vps): """Measures average error in the location of detected vanishing points. "Missed" or "extra" VPs do not count against the score. Based on log distance of detected vp from ground truth vp. Args: ground_truth_vps: List of ground truth VP point tuples. detected_vps: List of detected VP point tuples. Returns: Float, error. """ if len(ground_truth_vps) == 0 or len(detected_vps) == 0: return 0 point_pair_dists = [] for gt_vp in ground_truth_vps: for dt_vp in detected_vps: distance = geom_tools.point_to_point_dist(gt_vp, dt_vp) point_pair_dists.append((distance, gt_vp, dt_vp)) sorted(point_pair_dists, key=lambda k: k[0]) seen_gt_vps = set() seen_dt_vps = set() total_error = 0 for distance, gt_vp, dt_vp in point_pair_dists: if gt_vp in seen_gt_vps or dt_vp in seen_dt_vps: continue seen_gt_vps.add(gt_vp) seen_dt_vps.add(dt_vp) if distance > 0: total_error += math.log(distance) return total_error / min(len(detected_vps), len(ground_truth_vps)) def num_model_detection_error(ground_truth_vps, detected_vps): """Measures error in the number of detected vanishing points. Returns: Integer, positive when there are too many VPs, negative when there are too few. """ return len(detected_vps) - len(ground_truth_vps)
629
import torch import torchvision.transforms as transforms from torch.utils.data import Dataset import glob from PIL import Image import random class SUN397EncodableDataset(Dataset): """SUN397 encodable dataset class""" def __init__(self, train=True): super().__init__() path = 'data/SUN397/train/*/*.jpg' if train else 'data/SUN397/test/*/*.jpg' self.data = list(glob.glob(path)) random.shuffle(self.data) cats = list(set([path.split("/")[3] for path in self.data])) cats.sort() self.labels = torch.LongTensor([cats.index(path.split("/")[3]) for path in self.data]) self.preprocessor = transforms.Compose([ transforms.Resize((224, 224)), transforms.ToTensor(), transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) ]) self.device = torch.device("cuda:0" if torch.cuda.is_available() else "cpu") def __getitem__(self, idx): if torch.is_tensor(idx): idx = idx.tolist() if len(self.encoded_data) == 0: return self.preprocessor(Image.open(self.data[idx]).convert('RGB')), self.labels[idx] return self.encoded_data[idx], self.labels[idx] def __len__(self): return len(self.labels) def num_classes(self): return int(max(self.labels) + 1)
643
from PhysicsTools.Heppy.analyzers.core.Analyzer import Analyzer from PhysicsTools.Heppy.analyzers.core.AutoHandle import AutoHandle from PhysicsTools.Heppy.physicsobjects.Tau import Tau from PhysicsTools.HeppyCore.utils.deltar import deltaR, matchObjectCollection3 import PhysicsTools.HeppyCore.framework.config as cfg class TauAnalyzer( Analyzer ): def __init__(self, cfg_ana, cfg_comp, looperName ): super(TauAnalyzer,self).__init__(cfg_ana,cfg_comp,looperName) #---------------------------------------- # DECLARATION OF HANDLES OF LEPTONS STUFF #---------------------------------------- def declareHandles(self): super(TauAnalyzer, self).declareHandles() self.handles['taus'] = AutoHandle( ('slimmedTaus',''),'std::vector<pat::Tau>') def beginLoop(self, setup): super(TauAnalyzer,self).beginLoop(setup) self.counters.addCounter('events') count = self.counters.counter('events') count.register('all events') count.register('has >=1 tau at preselection') count.register('has >=1 selected taus') count.register('has >=1 other taus') #------------------ # MAKE LEPTON LISTS #------------------ def makeTaus(self, event): event.inclusiveTaus = [] event.selectedTaus = [] event.otherTaus = [] #get all alltaus = map( Tau, self.handles['taus'].product() ) #make inclusive taus for tau in alltaus: tau.associatedVertex = event.goodVertices[0] if len(event.goodVertices)>0 else event.vertices[0] tau.lepVeto = False tau.idDecayMode = tau.tauID("decayModeFinding") tau.idDecayModeNewDMs = tau.tauID("decayModeFindingNewDMs") if hasattr(self.cfg_ana, 'inclusive_decayModeID') and self.cfg_ana.inclusive_decayModeID and not tau.tauID(self.cfg_ana.inclusive_decayModeID): continue tau.inclusive_lepVeto = False if self.cfg_ana.inclusive_vetoLeptons: for lep in event.selectedLeptons: if deltaR(lep.eta(), lep.phi(), tau.eta(), tau.phi()) < self.cfg_ana.inclusive_leptonVetoDR: tau.inclusive_lepVeto = True if tau.inclusive_lepVeto: continue if self.cfg_ana.inclusive_vetoLeptonsPOG: if not tau.tauID(self.cfg_ana.inclusive_tauAntiMuonID): tau.inclusive_lepVeto = True if not tau.tauID(self.cfg_ana.inclusive_tauAntiElectronID): tau.inclusive_lepVeto = True if tau.inclusive_lepVeto: continue if tau.pt() < self.cfg_ana.inclusive_ptMin: continue if abs(tau.eta()) > self.cfg_ana.inclusive_etaMax: continue if abs(tau.dxy()) > self.cfg_ana.inclusive_dxyMax or abs(tau.dz()) > self.cfg_ana.inclusive_dzMax: continue def id3(tau,X): """Create an integer equal to 1-2-3 for (loose,medium,tight)""" return tau.tauID(X%"Loose") + tau.tauID(X%"Medium") + tau.tauID(X%"Tight") def id5(tau,X): """Create an integer equal to 1-2-3-4-5 for (very loose, loose, medium, tight, very tight)""" return id3(tau, X) + tau.tauID(X%"VLoose") + tau.tauID(X%"VTight") def id6(tau,X): """Create an integer equal to 1-2-3-4-5-6 for (very loose, loose, medium, tight, very tight, very very tight)""" return id5(tau, X) + tau.tauID(X%"VVTight") tau.idMVA = id6(tau, "by%sIsolationMVArun2v1DBoldDMwLT") tau.idMVANewDM = id6(tau, "by%sIsolationMVArun2v1DBnewDMwLT") tau.idCI3hit = id3(tau, "by%sCombinedIsolationDeltaBetaCorr3Hits") tau.idAntiMu = tau.tauID("againstMuonLoose3") + tau.tauID("againstMuonTight3") tau.idAntiE = id5(tau, "againstElectron%sMVA6") #print "Tau pt %5.1f: idMVA2 %d, idCI3hit %d, %s, %s" % (tau.pt(), tau.idMVA2, tau.idCI3hit, tau.tauID(self.cfg_ana.tauID), tau.tauID(self.cfg_ana.tauLooseID)) if tau.tauID(self.cfg_ana.inclusive_tauID): event.inclusiveTaus.append(tau) for tau in event.inclusiveTaus: tau.loose_lepVeto = False if self.cfg_ana.loose_vetoLeptons: for lep in event.selectedLeptons: if deltaR(lep.eta(), lep.phi(), tau.eta(), tau.phi()) < self.cfg_ana.loose_leptonVetoDR: tau.loose_lepVeto = True if self.cfg_ana.loose_vetoLeptonsPOG: if not tau.tauID(self.cfg_ana.loose_tauAntiMuonID): tau.loose_lepVeto = True if not tau.tauID(self.cfg_ana.loose_tauAntiElectronID): tau.loose_lepVeto = True if tau.tauID(self.cfg_ana.loose_decayModeID) and \ tau.pt() > self.cfg_ana.loose_ptMin and abs(tau.eta()) < self.cfg_ana.loose_etaMax and \ abs(tau.dxy()) < self.cfg_ana.loose_dxyMax and abs(tau.dz()) < self.cfg_ana.loose_dzMax and \ tau.tauID(self.cfg_ana.loose_tauID) and not tau.loose_lepVeto: event.selectedTaus.append(tau) else: event.otherTaus.append(tau) event.inclusiveTaus.sort(key = lambda l : l.pt(), reverse = True) event.selectedTaus.sort(key = lambda l : l.pt(), reverse = True) event.otherTaus.sort(key = lambda l : l.pt(), reverse = True) self.counters.counter('events').inc('all events') if len(event.inclusiveTaus): self.counters.counter('events').inc('has >=1 tau at preselection') if len(event.selectedTaus): self.counters.counter('events').inc('has >=1 selected taus') if len(event.otherTaus): self.counters.counter('events').inc('has >=1 other taus') def matchTaus(self, event): match = matchObjectCollection3(event.inclusiveTaus, event.gentaus, deltaRMax = 0.5) for lep in event.inclusiveTaus: gen = match[lep] lep.mcMatchId = 1 if gen else 0 lep.genp = gen def process(self, event): self.readCollections( event.input ) self.makeTaus(event) if not self.cfg_comp.isMC: return True if hasattr(event, 'gentaus'): self.matchTaus(event) return True # Find the definitions of the tau ID strings here: # http://cmslxr.fnal.gov/lxr/source/PhysicsTools/PatAlgos/python/producersLayer1/tauProducer_cfi.py setattr(TauAnalyzer,"defaultConfig",cfg.Analyzer( class_object = TauAnalyzer, # inclusive very loose hadronic tau selection inclusive_ptMin = 18, inclusive_etaMax = 9999, inclusive_dxyMax = 1000., inclusive_dzMax = 0.4, inclusive_vetoLeptons = False, inclusive_leptonVetoDR = 0.4, inclusive_decayModeID = "decayModeFindingNewDMs", # ignored if not set or "" inclusive_tauID = "decayModeFindingNewDMs", inclusive_vetoLeptonsPOG = False, # If True, the following two IDs are required inclusive_tauAntiMuonID = "", inclusive_tauAntiElectronID = "", # loose hadronic tau selection loose_ptMin = 18, loose_etaMax = 9999, loose_dxyMax = 1000., loose_dzMax = 0.2, loose_vetoLeptons = True, loose_leptonVetoDR = 0.4, loose_decayModeID = "decayModeFindingNewDMs", # ignored if not set or "" loose_tauID = "byLooseCombinedIsolationDeltaBetaCorr3Hits", loose_vetoLeptonsPOG = False, # If True, the following two IDs are required loose_tauAntiMuonID = "againstMuonLoose3", loose_tauAntiElectronID = "againstElectronLooseMVA5" ) )
650
from django.db import models from .query import BookQuerySet class Book(models.Model): objects = BookQuerySet.as_manager() title = models.CharField(max_length=50) publication_date = models.DateTimeField() author = models.ForeignKey('Author') genres = models.ManyToManyField('Genre') class Author(models.Model): name = models.CharField(max_length=50) nationality = models.ForeignKey('Nation', null=True) class Genre(models.Model): name = models.CharField(max_length=50) class Nation(models.Model): name = models.CharField(max_length=50) demonym = models.CharField(max_length=50)
666
def ips_between(start, end): calc = lambda n, m: (int(end.split(".")[n]) - int(start.split(".")[n])) * m return calc(0, 256 * 256 * 256) + calc(1, 256 * 256) + calc(2, 256) + calc(3, 1)
667
from abc import ABCMeta, abstractmethod import os from vmaf.tools.misc import make_absolute_path, run_process from vmaf.tools.stats import ListStats __copyright__ = "Copyright 2016-2018, Netflix, Inc." __license__ = "Apache, Version 2.0" import re import numpy as np import ast from vmaf import ExternalProgramCaller, to_list from vmaf.config import VmafConfig, VmafExternalConfig from vmaf.core.executor import Executor from vmaf.core.result import Result from vmaf.tools.reader import YuvReader class FeatureExtractor(Executor): """ FeatureExtractor takes in a list of assets, and run feature extraction on them, and return a list of corresponding results. A FeatureExtractor must specify a unique type and version combination (by the TYPE and VERSION attribute), so that the Result generated by it can be identified. A derived class of FeatureExtractor must: 1) Override TYPE and VERSION 2) Override _generate_result(self, asset), which call a command-line executable and generate feature scores in a log file. 3) Override _get_feature_scores(self, asset), which read the feature scores from the log file, and return the scores in a dictionary format. For an example, follow VmafFeatureExtractor. """ __metaclass__ = ABCMeta @property @abstractmethod def ATOM_FEATURES(self): raise NotImplementedError def _read_result(self, asset): result = {} result.update(self._get_feature_scores(asset)) executor_id = self.executor_id return Result(asset, executor_id, result) @classmethod def get_scores_key(cls, atom_feature): return "{type}_{atom_feature}_scores".format( type=cls.TYPE, atom_feature=atom_feature) @classmethod def get_score_key(cls, atom_feature): return "{type}_{atom_feature}_score".format( type=cls.TYPE, atom_feature=atom_feature) def _get_feature_scores(self, asset): # routine to read the feature scores from the log file, and return # the scores in a dictionary format. log_file_path = self._get_log_file_path(asset) atom_feature_scores_dict = {} atom_feature_idx_dict = {} for atom_feature in self.ATOM_FEATURES: atom_feature_scores_dict[atom_feature] = [] atom_feature_idx_dict[atom_feature] = 0 with open(log_file_path, 'rt') as log_file: for line in log_file.readlines(): for atom_feature in self.ATOM_FEATURES: re_template = "{af}: ([0-9]+) ([a-zA-Z0-9.-]+)".format(af=atom_feature) mo = re.match(re_template, line) if mo: cur_idx = int(mo.group(1)) assert cur_idx == atom_feature_idx_dict[atom_feature] # parse value, allowing NaN and inf val = float(mo.group(2)) if np.isnan(val) or np.isinf(val): val = None atom_feature_scores_dict[atom_feature].append(val) atom_feature_idx_dict[atom_feature] += 1 continue len_score = len(atom_feature_scores_dict[self.ATOM_FEATURES[0]]) assert len_score != 0 for atom_feature in self.ATOM_FEATURES[1:]: assert len_score == len(atom_feature_scores_dict[atom_feature]), \ "Feature data possibly corrupt. Run cleanup script and try again." feature_result = {} for atom_feature in self.ATOM_FEATURES: scores_key = self.get_scores_key(atom_feature) feature_result[scores_key] = atom_feature_scores_dict[atom_feature] return feature_result class VmafFeatureExtractor(FeatureExtractor): TYPE = "VMAF_feature" # VERSION = '0.1' # vmaf_study; Anush's VIF fix # VERSION = '0.2' # expose vif_num, vif_den, adm_num, adm_den, anpsnr # VERSION = '0.2.1' # expose vif num/den of each scale # VERSION = '0.2.2' # adm abs-->fabs, corrected border handling, uniform reading with option of offset for input YUV, updated VIF corner case # VERSION = '0.2.2b' # expose adm_den/num_scalex # VERSION = '0.2.3' # AVX for VMAF convolution; update adm features by folding noise floor into per coef # VERSION = '0.2.4' # Fix a bug in adm feature passing scale into dwt_quant_step # VERSION = '0.2.4b' # Modify by adding ADM noise floor outside cube root; add derived feature motion2 VERSION = '0.2.4c' # Modify by moving motion2 to c code ATOM_FEATURES = ['vif', 'adm', 'ansnr', 'motion', 'motion2', 'vif_num', 'vif_den', 'adm_num', 'adm_den', 'anpsnr', 'vif_num_scale0', 'vif_den_scale0', 'vif_num_scale1', 'vif_den_scale1', 'vif_num_scale2', 'vif_den_scale2', 'vif_num_scale3', 'vif_den_scale3', 'adm_num_scale0', 'adm_den_scale0', 'adm_num_scale1', 'adm_den_scale1', 'adm_num_scale2', 'adm_den_scale2', 'adm_num_scale3', 'adm_den_scale3', ] DERIVED_ATOM_FEATURES = ['vif_scale0', 'vif_scale1', 'vif_scale2', 'vif_scale3', 'vif2', 'adm2', 'adm3', 'adm_scale0', 'adm_scale1', 'adm_scale2', 'adm_scale3', ] ADM2_CONSTANT = 0 ADM_SCALE_CONSTANT = 0 def _generate_result(self, asset): # routine to call the command-line executable and generate feature # scores in the log file. quality_width, quality_height = asset.quality_width_height log_file_path = self._get_log_file_path(asset) yuv_type=self._get_workfile_yuv_type(asset) ref_path=asset.ref_workfile_path dis_path=asset.dis_workfile_path w=quality_width h=quality_height logger = self.logger ExternalProgramCaller.call_vmaf_feature(yuv_type, ref_path, dis_path, w, h, log_file_path, logger) @classmethod def _post_process_result(cls, result): # override Executor._post_process_result result = super(VmafFeatureExtractor, cls)._post_process_result(result) # adm2 = # (adm_num + ADM2_CONSTANT) / (adm_den + ADM2_CONSTANT) adm2_scores_key = cls.get_scores_key('adm2') adm_num_scores_key = cls.get_scores_key('adm_num') adm_den_scores_key = cls.get_scores_key('adm_den') result.result_dict[adm2_scores_key] = list( (np.array(result.result_dict[adm_num_scores_key]) + cls.ADM2_CONSTANT) / (np.array(result.result_dict[adm_den_scores_key]) + cls.ADM2_CONSTANT) ) # vif_scalei = vif_num_scalei / vif_den_scalei, i = 0, 1, 2, 3 vif_num_scale0_scores_key = cls.get_scores_key('vif_num_scale0') vif_den_scale0_scores_key = cls.get_scores_key('vif_den_scale0') vif_num_scale1_scores_key = cls.get_scores_key('vif_num_scale1') vif_den_scale1_scores_key = cls.get_scores_key('vif_den_scale1') vif_num_scale2_scores_key = cls.get_scores_key('vif_num_scale2') vif_den_scale2_scores_key = cls.get_scores_key('vif_den_scale2') vif_num_scale3_scores_key = cls.get_scores_key('vif_num_scale3') vif_den_scale3_scores_key = cls.get_scores_key('vif_den_scale3') vif_scale0_scores_key = cls.get_scores_key('vif_scale0') vif_scale1_scores_key = cls.get_scores_key('vif_scale1') vif_scale2_scores_key = cls.get_scores_key('vif_scale2') vif_scale3_scores_key = cls.get_scores_key('vif_scale3') result.result_dict[vif_scale0_scores_key] = list( (np.array(result.result_dict[vif_num_scale0_scores_key]) / np.array(result.result_dict[vif_den_scale0_scores_key])) ) result.result_dict[vif_scale1_scores_key] = list( (np.array(result.result_dict[vif_num_scale1_scores_key]) / np.array(result.result_dict[vif_den_scale1_scores_key])) ) result.result_dict[vif_scale2_scores_key] = list( (np.array(result.result_dict[vif_num_scale2_scores_key]) / np.array(result.result_dict[vif_den_scale2_scores_key])) ) result.result_dict[vif_scale3_scores_key] = list( (np.array(result.result_dict[vif_num_scale3_scores_key]) / np.array(result.result_dict[vif_den_scale3_scores_key])) ) # vif2 = # ((vif_num_scale0 / vif_den_scale0) + (vif_num_scale1 / vif_den_scale1) + # (vif_num_scale2 / vif_den_scale2) + (vif_num_scale3 / vif_den_scale3)) / 4.0 vif_scores_key = cls.get_scores_key('vif2') result.result_dict[vif_scores_key] = list( ( (np.array(result.result_dict[vif_num_scale0_scores_key]) / np.array(result.result_dict[vif_den_scale0_scores_key])) + (np.array(result.result_dict[vif_num_scale1_scores_key]) / np.array(result.result_dict[vif_den_scale1_scores_key])) + (np.array(result.result_dict[vif_num_scale2_scores_key]) / np.array(result.result_dict[vif_den_scale2_scores_key])) + (np.array(result.result_dict[vif_num_scale3_scores_key]) / np.array(result.result_dict[vif_den_scale3_scores_key])) ) / 4.0 ) # adm_scalei = adm_num_scalei / adm_den_scalei, i = 0, 1, 2, 3 adm_num_scale0_scores_key = cls.get_scores_key('adm_num_scale0') adm_den_scale0_scores_key = cls.get_scores_key('adm_den_scale0') adm_num_scale1_scores_key = cls.get_scores_key('adm_num_scale1') adm_den_scale1_scores_key = cls.get_scores_key('adm_den_scale1') adm_num_scale2_scores_key = cls.get_scores_key('adm_num_scale2') adm_den_scale2_scores_key = cls.get_scores_key('adm_den_scale2') adm_num_scale3_scores_key = cls.get_scores_key('adm_num_scale3') adm_den_scale3_scores_key = cls.get_scores_key('adm_den_scale3') adm_scale0_scores_key = cls.get_scores_key('adm_scale0') adm_scale1_scores_key = cls.get_scores_key('adm_scale1') adm_scale2_scores_key = cls.get_scores_key('adm_scale2') adm_scale3_scores_key = cls.get_scores_key('adm_scale3') result.result_dict[adm_scale0_scores_key] = list( (np.array(result.result_dict[adm_num_scale0_scores_key]) + cls.ADM_SCALE_CONSTANT) / (np.array(result.result_dict[adm_den_scale0_scores_key]) + cls.ADM_SCALE_CONSTANT) ) result.result_dict[adm_scale1_scores_key] = list( (np.array(result.result_dict[adm_num_scale1_scores_key]) + cls.ADM_SCALE_CONSTANT) / (np.array(result.result_dict[adm_den_scale1_scores_key]) + cls.ADM_SCALE_CONSTANT) ) result.result_dict[adm_scale2_scores_key] = list( (np.array(result.result_dict[adm_num_scale2_scores_key]) + cls.ADM_SCALE_CONSTANT) / (np.array(result.result_dict[adm_den_scale2_scores_key]) + cls.ADM_SCALE_CONSTANT) ) result.result_dict[adm_scale3_scores_key] = list( (np.array(result.result_dict[adm_num_scale3_scores_key]) + cls.ADM_SCALE_CONSTANT) / (np.array(result.result_dict[adm_den_scale3_scores_key]) + cls.ADM_SCALE_CONSTANT) ) # adm3 = \ # (((adm_num_scale0 + ADM_SCALE_CONSTANT) / (adm_den_scale0 + ADM_SCALE_CONSTANT)) # + ((adm_num_scale1 + ADM_SCALE_CONSTANT) / (adm_den_scale1 + ADM_SCALE_CONSTANT)) # + ((adm_num_scale2 + ADM_SCALE_CONSTANT) / (adm_den_scale2 + ADM_SCALE_CONSTANT)) # + ((adm_num_scale3 + ADM_SCALE_CONSTANT) / (adm_den_scale3 + ADM_SCALE_CONSTANT))) / 4.0 adm3_scores_key = cls.get_scores_key('adm3') result.result_dict[adm3_scores_key] = list( ( ((np.array(result.result_dict[adm_num_scale0_scores_key]) + cls.ADM_SCALE_CONSTANT) / (np.array(result.result_dict[adm_den_scale0_scores_key]) + cls.ADM_SCALE_CONSTANT)) + ((np.array(result.result_dict[adm_num_scale1_scores_key]) + cls.ADM_SCALE_CONSTANT) / (np.array(result.result_dict[adm_den_scale1_scores_key]) + cls.ADM_SCALE_CONSTANT)) + ((np.array(result.result_dict[adm_num_scale2_scores_key]) + cls.ADM_SCALE_CONSTANT) / (np.array(result.result_dict[adm_den_scale2_scores_key]) + cls.ADM_SCALE_CONSTANT)) + ((np.array(result.result_dict[adm_num_scale3_scores_key]) + cls.ADM_SCALE_CONSTANT) / (np.array(result.result_dict[adm_den_scale3_scores_key]) + cls.ADM_SCALE_CONSTANT)) ) / 4.0 ) # validate for feature in cls.DERIVED_ATOM_FEATURES: assert cls.get_scores_key(feature) in result.result_dict return result class VifFrameDifferenceFeatureExtractor(FeatureExtractor): TYPE = "VifDiff_feature" VERSION = '0.1' ATOM_FEATURES = ['vifdiff', 'vifdiff_num', 'vifdiff_den', 'vifdiff_num_scale0', 'vifdiff_den_scale0', 'vifdiff_num_scale1', 'vifdiff_den_scale1', 'vifdiff_num_scale2', 'vifdiff_den_scale2', 'vifdiff_num_scale3', 'vifdiff_den_scale3', ] DERIVED_ATOM_FEATURES = ['vifdiff_scale0', 'vifdiff_scale1', 'vifdiff_scale2', 'vifdiff_scale3', ] ADM2_CONSTANT = 0 ADM_SCALE_CONSTANT = 0 def _generate_result(self, asset): # routine to call the command-line executable and generate feature # scores in the log file. quality_width, quality_height = asset.quality_width_height log_file_path = self._get_log_file_path(asset) yuv_type=self._get_workfile_yuv_type(asset) ref_path=asset.ref_workfile_path dis_path=asset.dis_workfile_path w=quality_width h=quality_height logger = self.logger ExternalProgramCaller.call_vifdiff_feature(yuv_type, ref_path, dis_path, w, h, log_file_path, logger) @classmethod def _post_process_result(cls, result): # override Executor._post_process_result result = super(VifFrameDifferenceFeatureExtractor, cls)._post_process_result(result) # vifdiff_scalei = vifdiff_num_scalei / vifdiff_den_scalei, i = 0, 1, 2, 3 vifdiff_num_scale0_scores_key = cls.get_scores_key('vifdiff_num_scale0') vifdiff_den_scale0_scores_key = cls.get_scores_key('vifdiff_den_scale0') vifdiff_num_scale1_scores_key = cls.get_scores_key('vifdiff_num_scale1') vifdiff_den_scale1_scores_key = cls.get_scores_key('vifdiff_den_scale1') vifdiff_num_scale2_scores_key = cls.get_scores_key('vifdiff_num_scale2') vifdiff_den_scale2_scores_key = cls.get_scores_key('vifdiff_den_scale2') vifdiff_num_scale3_scores_key = cls.get_scores_key('vifdiff_num_scale3') vifdiff_den_scale3_scores_key = cls.get_scores_key('vifdiff_den_scale3') vifdiff_scale0_scores_key = cls.get_scores_key('vifdiff_scale0') vifdiff_scale1_scores_key = cls.get_scores_key('vifdiff_scale1') vifdiff_scale2_scores_key = cls.get_scores_key('vifdiff_scale2') vifdiff_scale3_scores_key = cls.get_scores_key('vifdiff_scale3') result.result_dict[vifdiff_scale0_scores_key] = list( (np.array(result.result_dict[vifdiff_num_scale0_scores_key]) / np.array(result.result_dict[vifdiff_den_scale0_scores_key])) ) result.result_dict[vifdiff_scale1_scores_key] = list( (np.array(result.result_dict[vifdiff_num_scale1_scores_key]) / np.array(result.result_dict[vifdiff_den_scale1_scores_key])) ) result.result_dict[vifdiff_scale2_scores_key] = list( (np.array(result.result_dict[vifdiff_num_scale2_scores_key]) / np.array(result.result_dict[vifdiff_den_scale2_scores_key])) ) result.result_dict[vifdiff_scale3_scores_key] = list( (np.array(result.result_dict[vifdiff_num_scale3_scores_key]) / np.array(result.result_dict[vifdiff_den_scale3_scores_key])) ) # validate for feature in cls.DERIVED_ATOM_FEATURES: assert cls.get_scores_key(feature) in result.result_dict return result class PsnrFeatureExtractor(FeatureExtractor): TYPE = "PSNR_feature" VERSION = "1.0" ATOM_FEATURES = ['psnr'] def _generate_result(self, asset): # routine to call the command-line executable and generate quality # scores in the log file. quality_width, quality_height = asset.quality_width_height log_file_path = self._get_log_file_path(asset) yuv_type=self._get_workfile_yuv_type(asset) ref_path=asset.ref_workfile_path dis_path=asset.dis_workfile_path w=quality_width h=quality_height logger = self.logger ExternalProgramCaller.call_psnr(yuv_type, ref_path, dis_path, w, h, log_file_path, logger) class MomentFeatureExtractor(FeatureExtractor): TYPE = "Moment_feature" # VERSION = "1.0" # call executable VERSION = "1.1" # python only ATOM_FEATURES = ['ref1st', 'ref2nd', 'dis1st', 'dis2nd', ] DERIVED_ATOM_FEATURES = ['refvar', 'disvar', ] def _generate_result(self, asset): # routine to call the command-line executable and generate feature # scores in the log file. quality_w, quality_h = asset.quality_width_height ref_scores_mtx = None with YuvReader(filepath=asset.ref_workfile_path, width=quality_w, height=quality_h, yuv_type=self._get_workfile_yuv_type(asset)) as ref_yuv_reader: scores_mtx_list = [] i = 0 for ref_yuv in ref_yuv_reader: ref_y = ref_yuv[0] firstm = ref_y.mean() secondm = ref_y.var() + firstm**2 scores_mtx_list.append(np.hstack(([firstm], [secondm]))) i += 1 ref_scores_mtx = np.vstack(scores_mtx_list) dis_scores_mtx = None with YuvReader(filepath=asset.dis_workfile_path, width=quality_w, height=quality_h, yuv_type=self._get_workfile_yuv_type(asset)) as dis_yuv_reader: scores_mtx_list = [] i = 0 for dis_yuv in dis_yuv_reader: dis_y = dis_yuv[0] firstm = dis_y.mean() secondm = dis_y.var() + firstm**2 scores_mtx_list.append(np.hstack(([firstm], [secondm]))) i += 1 dis_scores_mtx = np.vstack(scores_mtx_list) assert ref_scores_mtx is not None and dis_scores_mtx is not None log_dict = {'ref_scores_mtx': ref_scores_mtx.tolist(), 'dis_scores_mtx': dis_scores_mtx.tolist()} log_file_path = self._get_log_file_path(asset) with open(log_file_path, 'wt') as log_file: log_file.write(str(log_dict)) def _get_feature_scores(self, asset): # routine to read the feature scores from the log file, and return # the scores in a dictionary format. log_file_path = self._get_log_file_path(asset) with open(log_file_path, 'rt') as log_file: log_str = log_file.read() log_dict = ast.literal_eval(log_str) ref_scores_mtx = np.array(log_dict['ref_scores_mtx']) dis_scores_mtx = np.array(log_dict['dis_scores_mtx']) _, num_ref_features = ref_scores_mtx.shape assert num_ref_features == 2 # ref1st, ref2nd _, num_dis_features = dis_scores_mtx.shape assert num_dis_features == 2 # dis1st, dis2nd feature_result = {} feature_result[self.get_scores_key('ref1st')] = list(ref_scores_mtx[:, 0]) feature_result[self.get_scores_key('ref2nd')] = list(ref_scores_mtx[:, 1]) feature_result[self.get_scores_key('dis1st')] = list(dis_scores_mtx[:, 0]) feature_result[self.get_scores_key('dis2nd')] = list(dis_scores_mtx[:, 1]) return feature_result @classmethod def _post_process_result(cls, result): # override Executor._post_process_result result = super(MomentFeatureExtractor, cls)._post_process_result(result) # calculate refvar and disvar from ref1st, ref2nd, dis1st, dis2nd refvar_scores_key = cls.get_scores_key('refvar') ref1st_scores_key = cls.get_scores_key('ref1st') ref2nd_scores_key = cls.get_scores_key('ref2nd') disvar_scores_key = cls.get_scores_key('disvar') dis1st_scores_key = cls.get_scores_key('dis1st') dis2nd_scores_key = cls.get_scores_key('dis2nd') get_var = lambda m: m[1] - m[0] * m[0] result.result_dict[refvar_scores_key] = \ to_list(map(get_var, zip(result.result_dict[ref1st_scores_key], result.result_dict[ref2nd_scores_key]))) result.result_dict[disvar_scores_key] = \ to_list(map(get_var, zip(result.result_dict[dis1st_scores_key], result.result_dict[dis2nd_scores_key]))) # validate for feature in cls.DERIVED_ATOM_FEATURES: assert cls.get_scores_key(feature) in result.result_dict return result class SsimFeatureExtractor(FeatureExtractor): TYPE = "SSIM_feature" # VERSION = "1.0" VERSION = "1.1" # fix OPT_RANGE_PIXEL_OFFSET = 0 ATOM_FEATURES = ['ssim', 'ssim_l', 'ssim_c', 'ssim_s'] def _generate_result(self, asset): # routine to call the command-line executable and generate quality # scores in the log file. quality_width, quality_height = asset.quality_width_height log_file_path = self._get_log_file_path(asset) yuv_type=self._get_workfile_yuv_type(asset) ref_path=asset.ref_workfile_path dis_path=asset.dis_workfile_path w=quality_width h=quality_height logger = self.logger ExternalProgramCaller.call_ssim(yuv_type, ref_path, dis_path, w, h, log_file_path, logger) class MsSsimFeatureExtractor(FeatureExtractor): TYPE = "MS_SSIM_feature" # VERSION = "1.0" VERSION = "1.1" # fix OPT_RANGE_PIXEL_OFFSET = 0 ATOM_FEATURES = ['ms_ssim', 'ms_ssim_l_scale0', 'ms_ssim_c_scale0', 'ms_ssim_s_scale0', 'ms_ssim_l_scale1', 'ms_ssim_c_scale1', 'ms_ssim_s_scale1', 'ms_ssim_l_scale2', 'ms_ssim_c_scale2', 'ms_ssim_s_scale2', 'ms_ssim_l_scale3', 'ms_ssim_c_scale3', 'ms_ssim_s_scale3', 'ms_ssim_l_scale4', 'ms_ssim_c_scale4', 'ms_ssim_s_scale4', ] def _generate_result(self, asset): # routine to call the command-line executable and generate quality # scores in the log file. quality_width, quality_height = asset.quality_width_height log_file_path = self._get_log_file_path(asset) yuv_type=self._get_workfile_yuv_type(asset) ref_path=asset.ref_workfile_path dis_path=asset.dis_workfile_path w=quality_width h=quality_height logger = self.logger ExternalProgramCaller.call_ms_ssim(yuv_type, ref_path, dis_path, w, h, log_file_path, logger)
677
from sklearn.linear_model import LogisticRegression from fightchurn.listings.chap8.listing_8_2_logistic_regression import prepare_data, save_regression_model from fightchurn.listings.chap8.listing_8_2_logistic_regression import save_regression_summary, save_dataset_predictions def regression_cparam(data_set_path, C_param): X,y = prepare_data(data_set_path) retain_reg = LogisticRegression( C=C_param, penalty='l1', solver='liblinear', fit_intercept=True) retain_reg.fit(X, y) c_ext = '_c{:.3f}'.format(C_param) save_regression_summary(data_set_path,retain_reg,ext=c_ext) save_regression_model(data_set_path,retain_reg,ext=c_ext) save_dataset_predictions(data_set_path,retain_reg,X,ext=c_ext)
696
from operator import attrgetter import logging import os import shutil import subprocess import pyfastaq import pymummer from cluster_vcf_records import vcf_record from varifier import utils # We only want the .snps file from the dnadiff script from MUMmer. From reading # the docs inspecting that script, we need to run these commands: # # nucmer --maxmatch --delta out.delta ref.fasta query.fasta # delta-filter -1 out.delta > out.1delta # show-snps -rlTHC out.1delta > out.snps # # This is instead of just running show-snps, which runs several other commands # in addition to making the snps file. def _run_dnadiff_one_split(ref_fasta, query_fasta, outfile, threads=1, maxmatch=True): delta = f"{outfile}.tmp.delta" delta_1 = f"{outfile}.tmp.1delta" subprocess.check_output(f"rm -f {delta} {delta_1}", shell=True) maxmatch_opt = "--maxmatch" if maxmatch else "" commands = [ f"nucmer --threads {threads} {maxmatch_opt} --delta {delta} {ref_fasta} {query_fasta}", f"delta-filter -1 {delta} > {delta_1}", f"show-snps -rlTHC {delta_1} > {outfile}", ] for command in commands: logging.info("Start run command: " + command) subprocess.check_output(command, shell=True) logging.info("Finish run command: " + command) os.unlink(delta) os.unlink(delta_1) def _run_dnadiff( ref_fasta, query_fasta, outfile, split_query=False, debug=False, threads=1, maxmatch=True, ): if not split_query: _run_dnadiff_one_split( ref_fasta, query_fasta, outfile, threads=threads, maxmatch=maxmatch ) else: tmp_snp_files = [] seq_reader = pyfastaq.sequences.file_reader(query_fasta) for seq in seq_reader: prefix = f"{outfile}.tmp.split.{len(tmp_snp_files)}" tmp_fasta = f"{prefix}.fasta" with open(tmp_fasta, "w") as f: print(seq, file=f) snp_file = f"{prefix}.snps" _run_dnadiff_one_split( ref_fasta, tmp_fasta, snp_file, threads=threads, maxmatch=maxmatch ) os.unlink(tmp_fasta) tmp_snp_files.append(snp_file) with open(outfile, "wb") as f_out: for snp_file in tmp_snp_files: with open(snp_file, "rb") as f_in: shutil.copyfileobj(f_in, f_out) if not debug: os.unlink(snp_file) def _snps_file_to_vcf(snps_file, query_fasta, outfile): """Loads the .snps file made by dnadiff. query_fasta = fasta file of query sequences. Writes a new VCF file unmerged records.""" vcf_records = {} variants = pymummer.snp_file.get_all_variants(snps_file) query_seqs = utils.file_to_dict_of_seqs(query_fasta) for variant in variants: # If the variant is reversed, it means that either the ref or query had to be # reverse complemented when aligned by mummer. Need to do the appropriate # reverse (complement) fixes so the VCF has the correct REF and ALT sequences if variant.reverse: qry_seq = pyfastaq.sequences.Fasta("x", variant.qry_base) qry_seq.revcomp() variant.qry_base = "".join(reversed(qry_seq.seq)) ref_seq = pyfastaq.sequences.Fasta("x", variant.ref_base) ref_seq.revcomp() variant.ref_base = ref_seq.seq if variant.var_type == pymummer.variant.SNP: new_record = vcf_record.VcfRecord( "\t".join( [ variant.qry_name, str(variant.qry_start + 1), ".", variant.qry_base, variant.ref_base, ".", ".", "SVTYPE=DNADIFF_SNP", "GT", "1/1", ] ) ) elif variant.var_type == pymummer.variant.DEL: # The query has sequence missing, compared to the # reference. We're making VCF records w.r.t. the # query, so this is an insertion. So need to # get the nucleotide before the insertion as well. new_record = vcf_record.VcfRecord( "\t".join( [ variant.qry_name, str(variant.qry_start + 1), ".", query_seqs[variant.qry_name][variant.qry_start], query_seqs[variant.qry_name][variant.qry_start] + variant.ref_base, ".", ".", "SVTYPE=DNADIFF_INS", "GT", "1/1", ] ) ) elif variant.var_type == pymummer.variant.INS: # The ref has sequence missing, compared to the # query. We're making VCF records w.r.t. the # query, so this is a deletion. So need to # get the nucleotide before the deletion as well. new_record = vcf_record.VcfRecord( "\t".join( [ variant.qry_name, str(variant.qry_start), ".", query_seqs[variant.qry_name][variant.qry_start - 1] + variant.qry_base, query_seqs[variant.qry_name][variant.qry_start - 1], ".", ".", "SVTYPE=DNADIFF_DEL", "GT", "1/1", ] ) ) else: raise Exception("Unknown variant type: " + str(variant)) assert ( new_record.REF == query_seqs[new_record.CHROM][ new_record.POS : new_record.POS + len(new_record.REF) ] ) if new_record.CHROM not in vcf_records: vcf_records[new_record.CHROM] = [] vcf_records[new_record.CHROM].append(new_record) for vcf_list in vcf_records.values(): vcf_list.sort(key=attrgetter("POS")) with open(outfile, "w") as f: print("##fileformat=VCFv4.2", file=f) for seq in query_seqs.values(): print(f"##contig=<ID={seq.id},length={len(seq)}>", file=f) print("#CHROM\tPOS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\tFORMAT\tsample", file=f) for key, vcf_list in sorted(vcf_records.items()): for record in vcf_list: print(record, file=f) def make_truth_vcf( ref_fasta, truth_fasta, outfile, debug=False, split_ref=False, threads=1, maxmatch=True, ): snps_file = f"{outfile}.tmp.snps" _run_dnadiff( truth_fasta, ref_fasta, snps_file, split_query=split_ref, debug=debug, threads=threads, maxmatch=maxmatch, ) _snps_file_to_vcf(snps_file, ref_fasta, outfile) if not debug: os.unlink(snps_file)
747
import os import numpy as np import tensorflow as tf def get_train_data(train_dir, batch_size): train_images = np.load(os.path.join(train_dir, 'train_images.npy')) train_labels = np.load(os.path.join(train_dir, 'train_labels.npy')) print('train_images', train_images.shape, 'train_labels', train_labels.shape) dataset_train = tf.data.Dataset.from_tensor_slices((train_images, train_labels)) dataset_train = dataset_train.repeat().shuffle(10000).batch(batch_size) return dataset_train def get_val_data(val_dir): test_images = np.load(os.path.join(val_dir, 'validation_images.npy')) test_labels = np.load(os.path.join(val_dir, 'validation_labels.npy')) print('validation_images', test_images.shape, 'validation_labels', test_labels.shape) dataset_test = tf.data.Dataset.from_tensor_slices((test_images, test_labels)) return dataset_test
837
import SimpleXMLRPCServer import sys import logging from K8055Controller import K8055Controller logging.basicConfig() controller_log = logging.getLogger("Controller") class Controller: def __init__(self): self.k8055 = K8055Controller() controller_log.debug("initialized") def reset(self): self.k8055.reset() controller_log.debug("reset") return 0 def turn_on(self, i): self.k8055.turn_on(i) controller_log.debug('turned on %i' % (i)) return 0 def turn_off(self, i): self.k8055.turn_off(i) controller_log.debug('turned off %i' % (i)) return 0 def set_analog(self, i, level): if (i == 1): self.k8055.set_analog1(level) else: self.k8055.set_analog2(level) return 0 controller = Controller() server = SimpleXMLRPCServer.SimpleXMLRPCServer(("d6349.mysql.zone.ee", 7000)) server.register_instance(controller) server.serve_forever()
845
from freight.api.serializer import serialize from freight.testutils import TestCase class UserSerializerTest(TestCase): def test_simple(self): user = self.create_user() result = serialize(user) assert result["id"] == str(user.id) assert result["name"] == user.name
855
import os import pytest import torch from hivemind import RemoteExpert from hivemind.moe.server import background_server CUSTOM_EXPERTS_PATH = os.path.join(os.path.dirname(__file__), "test_utils", "custom_networks.py") @pytest.mark.forked def test_custom_expert(hid_dim=16): with background_server( expert_cls="perceptron", num_experts=2, device="cpu", hidden_dim=hid_dim, num_handlers=2, no_dht=True, custom_module_path=CUSTOM_EXPERTS_PATH, ) as (server_endpoint, _): expert0 = RemoteExpert("expert.0", server_endpoint) expert1 = RemoteExpert("expert.1", server_endpoint) for batch_size in (1, 4): batch = torch.randn(batch_size, hid_dim) output0 = expert0(batch) output1 = expert1(batch) loss = output0.sum() loss.backward() loss = output1.sum() loss.backward() @pytest.mark.forked def test_multihead_expert(hid_dim=16): with background_server( expert_cls="multihead", num_experts=2, device="cpu", hidden_dim=hid_dim, num_handlers=2, no_dht=True, custom_module_path=CUSTOM_EXPERTS_PATH, ) as (server_endpoint, _): expert0 = RemoteExpert("expert.0", server_endpoint) expert1 = RemoteExpert("expert.1", server_endpoint) for batch_size in (1, 4): batch = ( torch.randn(batch_size, hid_dim), torch.randn(batch_size, 2 * hid_dim), torch.randn(batch_size, 3 * hid_dim), ) output0 = expert0(*batch) output1 = expert1(*batch) loss = output0.sum() loss.backward() loss = output1.sum() loss.backward()
905
from typing import Dict, Optional, List, Any import torch import torch.nn.functional as F from allennlp.data import Vocabulary from allennlp.models.model import Model from allennlp.modules import FeedForward, TextFieldEmbedder, Seq2SeqEncoder from allennlp.nn import InitializerApplicator, RegularizerApplicator from allennlp.nn import util from allennlp.training.metrics import CategoricalAccuracy, F1Measure from overrides import overrides @Model.register("text_classifier") class TextClassifier(Model): """ Implements a basic text classifier: 1) Embed tokens using `text_field_embedder` 2) Seq2SeqEncoder, e.g. BiLSTM 3) Append the first and last encoder states 4) Final feedforward layer Optimized with CrossEntropyLoss. Evaluated with CategoricalAccuracy & F1. """ def __init__(self, vocab: Vocabulary, text_field_embedder: TextFieldEmbedder, text_encoder: Seq2SeqEncoder, classifier_feedforward: FeedForward, verbose_metrics: False, initializer: InitializerApplicator = InitializerApplicator(), regularizer: Optional[RegularizerApplicator] = None, ) -> None: super(TextClassifier, self).__init__(vocab, regularizer) self.text_field_embedder = text_field_embedder self.num_classes = self.vocab.get_vocab_size("labels") self.text_encoder = text_encoder self.classifier_feedforward = classifier_feedforward self.prediction_layer = torch.nn.Linear(self.classifier_feedforward.get_output_dim() , self.num_classes) self.label_accuracy = CategoricalAccuracy() self.label_f1_metrics = {} self.verbose_metrics = verbose_metrics for i in range(self.num_classes): self.label_f1_metrics[vocab.get_token_from_index(index=i, namespace="labels")] = F1Measure(positive_label=i) self.loss = torch.nn.CrossEntropyLoss() self.pool = lambda text, mask: util.get_final_encoder_states(text, mask, bidirectional=True) initializer(self) @overrides def forward(self, text: Dict[str, torch.LongTensor], label: torch.IntTensor = None, metadata: List[Dict[str, Any]] = None) -> Dict[str, torch.Tensor]: """ Parameters ---------- text : Dict[str, torch.LongTensor] From a ``TextField`` label : torch.IntTensor, optional (default = None) From a ``LabelField`` metadata : ``List[Dict[str, Any]]``, optional, (default = None) Metadata containing the original tokenization of the premise and hypothesis with 'premise_tokens' and 'hypothesis_tokens' keys respectively. Returns ------- An output dictionary consisting of: label_logits : torch.FloatTensor A tensor of shape ``(batch_size, num_labels)`` representing unnormalised log probabilities of the label. label_probs : torch.FloatTensor A tensor of shape ``(batch_size, num_labels)`` representing probabilities of the label. loss : torch.FloatTensor, optional A scalar loss to be optimised. """ embedded_text = self.text_field_embedder(text) mask = util.get_text_field_mask(text) encoded_text = self.text_encoder(embedded_text, mask) pooled = self.pool(encoded_text, mask) ff_hidden = self.classifier_feedforward(pooled) logits = self.prediction_layer(ff_hidden) class_probs = F.softmax(logits, dim=1) output_dict = {"logits": logits} if label is not None: loss = self.loss(logits, label) output_dict["loss"] = loss # compute F1 per label for i in range(self.num_classes): metric = self.label_f1_metrics[self.vocab.get_token_from_index(index=i, namespace="labels")] metric(class_probs, label) self.label_accuracy(logits, label) return output_dict @overrides def decode(self, output_dict: Dict[str, torch.Tensor]) -> Dict[str, torch.Tensor]: class_probabilities = F.softmax(output_dict['logits'], dim=-1) output_dict['class_probs'] = class_probabilities return output_dict def get_metrics(self, reset: bool = False) -> Dict[str, float]: metric_dict = {} sum_f1 = 0.0 for name, metric in self.label_f1_metrics.items(): metric_val = metric.get_metric(reset) if self.verbose_metrics: metric_dict[name + '_P'] = metric_val[0] metric_dict[name + '_R'] = metric_val[1] metric_dict[name + '_F1'] = metric_val[2] sum_f1 += metric_val[2] names = list(self.label_f1_metrics.keys()) total_len = len(names) average_f1 = sum_f1 / total_len metric_dict['average_F1'] = average_f1 metric_dict['accuracy'] = self.label_accuracy.get_metric(reset) return metric_dict
907
import stl_path class MyNDRPlugin(): def __init__(self): pass def pre_iteration(self, finding_max_rate, run_results=None, **kwargs): """ Function ran before each iteration. :parameters: finding_max_rate: boolean Indicates whether we are running for the first time, trying to find the max rate. In this is the case, the run_results will be None. run_results: dict A dictionary that contains the following keys: queue_full_percentage: Percentage of packets that are queued. drop_rate_percentage: Percentage of packets that were dropped. rate_tx_bps: TX rate in bps. rate_rx_bps: RX rate in bps. tx_util: TX utilization percentage. latency: Latency groups. cpu_util: CPU utilization percentage. tx_pps: TX in pps. rx_pps: RX in pps. tx_bps: TX in bps. rx_bps: RX in bps. bw_per_core: Bandwidth per core. rate_p: Running rate in percentage out of max. total_tx_L1: Total TX L1. total_rx_L1: Total RX L1. iteration: Description of iteration (not necessarily a number) Pay attention: The rate is of the upcoming iteration. All the rest are of the previous iteration. kwargs: dict List of tunables passed as parameters. """ # Pre iteration function. This function will run before TRex transmits to the DUT. # Could use this to better prepare the DUT, for example define shapers, policers, increase buffers and queues. # You can receive tunables in the command line, through the kwargs argument. pass def post_iteration(self, finding_max_rate, run_results, **kwargs): """ Function ran after each iteration. :parameters: finding_max_rate: boolean Indicates whether we are running for the first time, trying to find the max rate. If this is the case, some values of run_results (like iteration for example) are not relevant. run_results: dict A dictionary that contains the following keys: queue_full_percentage: Percentage of packets that are queued. drop_rate_percentage: Percentage of packets that were dropped. rate_tx_bps: TX rate in bps. rate_rx_bps: RX rate in bps. tx_util: TX utilization percentage. latency: Latency groups. cpu_util: CPU utilization percentage. tx_pps: TX in pps. rx_pps: RX in pps. tx_bps: TX in bps. rx_bps: RX in bps. bw_per_core: Bandwidth per core. rate_p: Running rate in percentage out of max. total_tx_L1: Total TX L1. total_rx_L1: Total RX L1. iteration: Description of iteration (not necessarily a number) kwargs: dict List of tunables passed as parameters. :returns: bool: should stop the benchmarking or not. """ # Post iteration function. This function will run after TRex transmits to the DUT. # Could use this to decide if to continue the benchmark after querying the DUT post run. The DUT might be overheated or any other thing that might make you want to stop the run. # You can receive tunables in the command line, through the kwargs argument. should_stop = False return should_stop # dynamic load of python module def register(): return MyNDRPlugin()
1014
from lemur import database def rotate_certificate(endpoint, new_cert): """ Rotates a certificate on a given endpoint. :param endpoint: :param new_cert: :return: """ # ensure that certificate is available for rotation endpoint.source.plugin.update_endpoint(endpoint, new_cert) endpoint.certificate = new_cert database.update(endpoint)
1017
from gtrain import Model import numpy as np import tensorflow as tf class NetForHypinv(Model): """ Implementaion of the crutial function for the HypINV algorithm. Warning: Do not use this class but implement its subclass, for example see FCNetForHypinv """ def __init__(self, weights): self.eval_session = None self.grad_session = None self.initial_x = None self.center = None self.weights = weights self.out_for_eval = None #(going to be filled in build_for_eval method) self.boundary_out_for_eval = None self.trained_x = None self.training_class_index = None self.x = None # tf variable for inversion (going to be filled in build method) self.x_for_eval = None self.out = None self.boundary_out = None # list of tf tensorf for each class of softmax class vs others output self.loss = None self.boundary_loss = None self.t = None #target self.boundary_t = None self.x1 = None # this attribute is used of purposes of modified loss function def __del__(self): # close arr sessions if self.eval_session: self.eval_session.close() if self.grad_session: self.grad_session.close() def set_initial_x(self, initial_x): # sets starting point for the search of the closest point self.initial_x = initial_x def set_center(self, center): # sets center point self.center = center / np.linalg.norm(center) def set_x1(self, x1): # sets x1 to which we want to found the cosest point x0 self.x1 = x1 def has_modified_loss(self): pass # if uses modified loss then it returns true def set_initial_x_in_session(self, x, session=None): # sets initial x in certain session if session is None: self.set_initial_x(x) else: pass # overide this method def eval(self, x): if len(x.shape) == 1: x = x.reshape((1,len(x))) if not self.eval_session: self.eval_session = tf.Session() with self.eval_session.as_default(): self.build_for_eval() self.eval_session.run(tf.global_variables_initializer()) return self.eval_session.run(self.out_for_eval, {self.x_for_eval: x}) def boundary_eval(self, x, class_index): # evaluates binary classificaitons class_index and other classes if not self.eval_session: self.eval_session = tf.Session() with self.eval_session.as_default(): self.build_for_eval() self.eval_session.run(tf.global_variables_initializer()) return self.eval_session.run(self.boundary_out_for_eval[class_index], {self.x_for_eval: x}) def get_boundary_gradient(self, x, class_index): # computes gradient of the boundary for specified class_index if not self.grad_session: self.grad_session = tf.Session() with self.grad_session.as_default(): self.build_for_eval() self.grad = list() for i in range(len(self.weights[0][-1][0])): self.grad.append(tf.gradients(self.boundary_out_for_eval[i], [self.x_for_eval])[0]) self.grad_x = self.x_for_eval return self.grad_session.run(self.grad[class_index], {self.grad_x: x}) def build_for_eval(self): # build model for evaluation pass #override this method (fill self.out_for_eval) def train_ended(self, session): self.trained_x = session.run(self.x) def build(self): # build model for training pass #override this method (fill self.x, self.out) def set_train_class(self, class_index): # sets class of the x1 self.training_class_index = class_index # overided methods from gtrain.Model def get_loss(self): if self.training_class_index is None: return self.loss else: return self.boundary_loss[self.training_class_index] def get_hits(self): return self.get_loss() def get_count(self): return self.get_loss() def get_train_summaries(self): return [] def get_dev_summaries(self): return [] def get_placeholders(self): if self.training_class_index is None: return [self.t] else: return [self.boundary_t] #________________________________________EXAMPLES_OF_NetForHypinv_CLASS_____________________________________________ class FCNetForHypinv(NetForHypinv): """ Implementation of multi layer perceptron to by used in HypINV rule extraction algorithm """ def __init__(self, weights, function=tf.sigmoid, use_modified_loss=False, mu = 0.01): """ :param weights: saved as [list of weights for layers][0 weight, 1 bias] :param function: tf function for propagation. For example tf.nn.sigmoid, tf.atan :param use_modified_loss: weather the modified loss should be used :param mu: factor of the penalty terms that specified the distance between x0 and x1 and the distance x1 from the boundary """ super(FCNetForHypinv, self).__init__(weights) self.function = function self.layer_sizes = [len(self.weights[0][0])] for bias in weights[1]: self.layer_sizes.append(len(bias)) self.num_classes = self.layer_sizes[-1] self.initial_x = np.zeros([1, self.layer_sizes[0]]) self.use_modified_loss = use_modified_loss self.mu = mu def build(self): with tf.name_scope("Input"): if self.center is not None: self.point_weights = tf.Variable(self.center.reshape((1, len(self.center))), dtype=tf.float64, trainable=False, name="Boundary_point") init_factor = self.center init_factor[init_factor!=0] = self.initial_x[init_factor!=0] / self.center[init_factor!=0] self.factor = tf.Variable(init_factor.reshape((1, len(self.center))), dtype=tf.float64, name="factor") else: self.point_weights = tf.Variable(self.initial_x.reshape((1, len(self.initial_x))), dtype=tf.float64, trainable=False, name="Boundary_point") self.factor = tf.Variable(np.ones((1, len(self.center))), dtype=tf.float64, name="factor") self.x = self.point_weights * self.factor with tf.name_scope("Target"): if self.use_modified_loss: x1_constant = tf.constant(self.x1.reshape((1, len(self.x1))), dtype=tf.float64) self.t = tf.placeholder(tf.float64, shape=[None, self.num_classes], name="Target_output") self.boundary_t = tf.placeholder(tf.float64, shape=[None, 2], name="Target_boundary_output") with tf.name_scope("FC_net"): flowing_x = self.x for i, _ in enumerate(self.weights[0]): with tf.name_scope("layer_{}".format(i)): W = tf.constant(self.weights[0][i], name="Weight_{}".format(i), dtype=tf.float64) b = tf.constant(self.weights[1][i], name="Bias_{}".format(i), dtype=tf.float64) flowing_x = self.function(tf.nn.xw_plus_b(flowing_x, W, b)) y = flowing_x self.out = tf.nn.softmax(y) with tf.name_scope("Binary_class_output"): self.boundary_out = list() for i in range(self.num_classes): mask = True+np.zeros(self.num_classes, dtype=np.bool) mask[i] = False x0 = self.out[:,i] x1 = tf.reduce_max(tf.boolean_mask(self.out, mask, axis=1), axis=1) s = x0+x1 out = tf.stack([x0/s, x1/s], axis=1) self.boundary_out.append(out) with tf.name_scope("Loss_functions"): self.loss = tf.reduce_mean( tf.nn.l2_loss(self.out-self.t), name="loss") with tf.name_scope("Binary_class_loss"): self.boundary_loss = list() if self.use_modified_loss: for i in range(self.num_classes): self.boundary_loss.append( tf.reduce_mean(tf.nn.l2_loss(self.boundary_out[i]-self.boundary_t)) + self.mu * tf.reduce_mean(tf.nn.l2_loss(self.x - x1_constant)) ) else: for i in range(self.num_classes): self.boundary_loss.append( tf.reduce_mean(tf.nn.l2_loss(self.boundary_out[i] - self.boundary_t)) ) def set_initial_x_in_session(self, x, session=None): if session is None: self.set_initial_x(x) else: if self.center is None: session.run([ self.point_weights.assign(x.reshape((1, len(x)))), self.factor.assign(np.ones((1, len(x)))) ]) else: init_factor = self.center init_factor[init_factor!=0] = x[init_factor!=0] / self.center[init_factor!=0] session.run(self.factor.assign(init_factor.reshape((1,len(init_factor))))) def build_for_eval(self): with tf.name_scope("eInput"): self.x_for_eval = tf.placeholder(tf.float32, shape=[None, len(self.weights[0][0])])#tf.Variable(tf.constant(self.initial_x), name="Boundary_point") with tf.name_scope("eFC_net"): flowing_x = self.x_for_eval for i, _ in enumerate(self.weights[0]): W = tf.constant(self.weights[0][i], name="eWeight_{}".format(i)) b = tf.constant(self.weights[1][i], name="eBias_{}".format(i)) flowing_x = self.function(tf.nn.xw_plus_b(flowing_x, W, b), name="elayer_{}".format(i)) y = flowing_x self.out_for_eval = tf.nn.softmax(y) with tf.name_scope("Binary_class_output"): self.boundary_out_for_eval = list() for i in range(self.num_classes): mask = True+np.zeros(self.num_classes, dtype=np.bool) mask[i] = False x0 = self.out_for_eval[:, i] x1 = tf.reduce_max(tf.boolean_mask(self.out_for_eval, mask, axis=1), axis=1) s = x0+x1 out = tf.stack([x0/s, x1/s], axis=1) self.boundary_out_for_eval.append(out) def has_modified_loss(self): return self.use_modified_loss def name(self): return "Hypinv_FC_net_{}".format("-".join([str(ls) for ls in self.layer_sizes])) class FCNetForHypinvBinary(FCNetForHypinv): """ Implementation of multi layer perceptron to by used in HypINV rule extraction algorithm The task is simplified to the binary classificaiton base_class_index against the other classes """ def __init__(self, weights, base_class_index, function=tf.sigmoid, use_modified_loss=False, mu = 0.01): """ :param weights: saved as [list of weights for layers][0 weight, 1 bias] :param base_class_index: an index of the class which is used as the base class :param function: tf function for propagation. For example tf.nn.sigmoid, tf.atan :param use_modified_loss: weather the modified loss should be used :param mu: factor of the penalty terms that specified the distance between x0 and x1 and the distance x1 from the boundary """ super(FCNetForHypinvBinary, self).__init__(weights) self.base_class_index = base_class_index self.function = function self.layer_sizes = [len(self.weights[0][0])] for bias in weights[1]: self.layer_sizes.append(len(bias)) self.num_classes = self.layer_sizes[-1] self.initial_x = np.zeros([1, self.layer_sizes[0]]) self.use_modified_loss = use_modified_loss self.mu = mu def build(self): with tf.name_scope("Input"): self.init_point = tf.Variable(self.initial_x.reshape((1, len(self.initial_x))), dtype=tf.float64, trainable=False, name="Boundary_point") self.factor = tf.Variable(np.ones((1, len(self.initial_x))), dtype=tf.float64, name="factor") self.x = self.init_point * self.factor with tf.name_scope("Target"): if self.use_modified_loss: x1_constant = tf.constant(self.x1.reshape((1, len(self.x1))), dtype=tf.float64) self.t = tf.placeholder(tf.float64, shape=[None, 2], name="Target_output") self.boundary_t = tf.placeholder(tf.float64, shape=[None, 2], name="Target_boundary_output") with tf.name_scope("FC_net"): flowing_x = self.x for i, _ in enumerate(self.weights[0]): with tf.name_scope("layer_{}".format(i)): W = tf.constant(self.weights[0][i], name="Weight_{}".format(i), dtype=tf.float64) b = tf.constant(self.weights[1][i], name="Bias_{}".format(i), dtype=tf.float64) flowing_x = self.function(tf.nn.xw_plus_b(flowing_x, W, b)) y = flowing_x full_out = tf.nn.softmax(y) with tf.name_scope("Binary_class_output"): self.boundary_out = list() mask = True+np.zeros(self.num_classes, dtype=np.bool) mask[self.base_class_index] = False x0 = full_out[:,self.base_class_index] x1 = tf.reduce_max(tf.boolean_mask(full_out, mask, axis=1), axis=1) s = x0+x1 self.out = tf.stack([x0/s, x1/s], axis=1) self.boundary_out.append(self.out) self.boundary_out.append(tf.stack([x1/s, x0/s], axis=1)) with tf.name_scope("Loss_functions"): self.loss = tf.reduce_mean( tf.nn.l2_loss(self.out-self.t), name="loss") with tf.name_scope("Binary_class_loss"): self.boundary_loss = list() if self.use_modified_loss: for i in range(2): self.boundary_loss.append( tf.reduce_mean(tf.nn.l2_loss(self.boundary_out[i]-self.boundary_t)) + self.mu * tf.reduce_mean(tf.nn.l2_loss(self.x - x1_constant)) ) else: for i in range(2): self.boundary_loss.append( tf.reduce_mean(tf.nn.l2_loss(self.boundary_out[i] - self.boundary_t)) ) def build_for_eval(self): with tf.name_scope("eInput"): self.x_for_eval = tf.placeholder(tf.float32, shape=[None, len(self.weights[0][0])])#tf.Variable(tf.constant(self.initial_x), name="Boundary_point") with tf.name_scope("eFC_net"): flowing_x = self.x_for_eval for i, _ in enumerate(self.weights[0]): W = tf.constant(self.weights[0][i], name="eWeight_{}".format(i)) b = tf.constant(self.weights[1][i], name="eBias_{}".format(i)) flowing_x = self.function(tf.nn.xw_plus_b(flowing_x, W, b), name="elayer_{}".format(i)) y = flowing_x full_out = tf.nn.softmax(y) with tf.name_scope("Binary_class_output"): self.boundary_out_for_eval = list() mask = True+np.zeros(self.num_classes, dtype=np.bool) mask[self.base_class_index] = False x0 = full_out[:, self.base_class_index] x1 = tf.reduce_max(tf.boolean_mask(full_out, mask, axis=1), axis=1) s = x0+x1 self.out_for_eval = tf.stack([x0/s, x1/s], axis=1) self.boundary_out_for_eval.append(self.out_for_eval) self.boundary_out_for_eval.append(tf.stack([x1/s, x0/s], axis=1)) def get_boundary_gradient(self, x, class_index): if not self.grad_session: self.grad_session = tf.Session() with self.grad_session.as_default(): self.build_for_eval() self.grad = list() for i in range(2): self.grad.append(tf.gradients(self.boundary_out_for_eval[i], [self.x_for_eval])[0]) self.grad_x = self.x_for_eval return self.grad_session.run(self.grad[class_index], {self.grad_x: x}) def has_modified_loss(self): return self.use_modified_loss def name(self): return "Hypinv_FC_net_{}".format("-".join([str(ls) for ls in self.layer_sizes]))
1018
from ..factory import Method class getChatMember(Method): chat_id = None # type: "int53" user_id = None # type: "int32"
1023
import numpy as np from sklearn.metrics import roc_curve, auc def compute_error_auc(op_str, gt, pred, prob): # classification error pred_int = (pred > prob).astype(np.int) class_acc = (pred_int == gt).mean() * 100.0 # ROC - area under curve fpr, tpr, thresholds = roc_curve(gt, pred) roc_auc = auc(fpr, tpr) print op_str, ', class acc = %.3f, ROC AUC = %.3f' % (class_acc, roc_auc) #return class_acc, roc_auc def calc_average_precision(recall, precision): precision[np.isnan(precision)] = 0 recall[np.isnan(recall)] = 0 # pascal'12 way mprec = np.hstack((0, precision, 0)) mrec = np.hstack((0, recall, 1)) for ii in range(mprec.shape[0]-2, -1,-1): mprec[ii] = np.maximum(mprec[ii], mprec[ii+1]) inds = np.where(np.not_equal(mrec[1:], mrec[:-1]))[0]+1 ave_prec = ((mrec[inds] - mrec[inds-1])*mprec[inds]).sum() return ave_prec def remove_end_preds(nms_pos_o, nms_prob_o, gt_pos_o, durations, win_size): # this filters out predictions and gt that are close to the end # this is a bit messy because of the shapes of gt_pos_o nms_pos = [] nms_prob = [] gt_pos = [] for ii in range(len(nms_pos_o)): valid_time = durations[ii] - win_size gt_cur = gt_pos_o[ii] if gt_cur.shape[0] > 0: gt_pos.append(gt_cur[:, 0][gt_cur[:, 0] < valid_time][..., np.newaxis]) else: gt_pos.append(gt_cur) valid_preds = nms_pos_o[ii] < valid_time nms_pos.append(nms_pos_o[ii][valid_preds]) nms_prob.append(nms_prob_o[ii][valid_preds, 0][..., np.newaxis]) return nms_pos, nms_prob, gt_pos def prec_recall_1d(nms_pos_o, nms_prob_o, gt_pos_o, durations, detection_overlap, win_size, remove_eof=True): """ nms_pos, nms_prob, and gt_pos are lists of numpy arrays specifying detection position, detection probability and GT position. Each list entry is a different file. Each entry in nms_pos is an array of length num_entries. For nms_prob and gt_pos its an array of size (num_entries, 1). durations is a array of the length of the number of files with each entry containing that file length in seconds. detection_overlap determines if a prediction is counted as correct or not. win_size is used to ignore predictions and ground truth at the end of an audio file. returns precision: fraction of retrieved instances that are relevant. recall: fraction of relevant instances that are retrieved. """ if remove_eof: # filter out the detections in both ground truth and predictions that are too # close to the end of the file - dont count them during eval nms_pos, nms_prob, gt_pos = remove_end_preds(nms_pos_o, nms_prob_o, gt_pos_o, durations, win_size) else: nms_pos = nms_pos_o nms_prob = nms_prob_o gt_pos = gt_pos_o # loop through each file true_pos = [] # correctly predicts the ground truth false_pos = [] # says there is a detection but isn't for ii in range(len(nms_pos)): num_preds = nms_pos[ii].shape[0] if num_preds > 0: # check to make sure it contains something num_gt = gt_pos[ii].shape[0] # for each set of predictions label them as true positive or false positive (i.e. 1-tp) tp = np.zeros(num_preds) distance_to_gt = np.abs(gt_pos[ii].ravel()-nms_pos[ii].ravel()[:, np.newaxis]) within_overlap = (distance_to_gt <= detection_overlap) # remove duplicate detections - assign to valid detection with highest prob for jj in range(num_gt): inds = np.where(within_overlap[:, jj])[0] # get the indices of all valid predictions if inds.shape[0] > 0: max_prob = np.argmax(nms_prob[ii][inds]) selected_pred = inds[max_prob] within_overlap[selected_pred, :] = False tp[selected_pred] = 1 # set as true positives true_pos.append(tp) false_pos.append(1 - tp) # calc precision and recall - sort confidence in descending order # PASCAL style conf = np.concatenate(nms_prob)[:, 0] num_gt = np.concatenate(gt_pos).shape[0] inds = np.argsort(conf)[::-1] true_pos_cat = np.concatenate(true_pos)[inds].astype(float) false_pos_cat = np.concatenate(false_pos)[inds].astype(float) # i.e. 1-true_pos_cat if (conf == conf[0]).sum() == conf.shape[0]: # all the probability values are the same therefore we will not sweep # the curve and instead will return a single value true_pos_sum = true_pos_cat.sum() false_pos_sum = false_pos_cat.sum() recall = np.asarray([true_pos_sum / float(num_gt)]) precision = np.asarray([(true_pos_sum / (false_pos_sum + true_pos_sum))]) elif inds.shape[0] > 0: # otherwise produce a list of values true_pos_cum = np.cumsum(true_pos_cat) false_pos_cum = np.cumsum(false_pos_cat) recall = true_pos_cum / float(num_gt) precision = (true_pos_cum / (false_pos_cum + true_pos_cum)) return precision, recall
1044
from typing import List from pybm import PybmConfig from pybm.command import CLICommand from pybm.config import get_reporter_class from pybm.exceptions import PybmError from pybm.reporters import BaseReporter from pybm.status_codes import ERROR, SUCCESS from pybm.util.path import get_subdirs class CompareCommand(CLICommand): """ Report benchmark results from specified sources. """ usage = "pybm compare <run> <anchor-ref> <compare-refs> [<options>]\n" def __init__(self): super(CompareCommand, self).__init__(name="compare") self.config = PybmConfig.load() def add_arguments(self): self.parser.add_argument( "run", type=str, metavar="<run>", help="Benchmark run to report results for. " "To report the preceding run, use the " '"latest" keyword. To report results ' "of the n-th preceding run " "(i.e., n runs ago), " 'use the "latest^{n}" syntax.', ) self.parser.add_argument( "refs", nargs="+", metavar="<refs>", help="Benchmarked refs to compare. The first " "given ref will be treated as the " "anchor ref, relative to which all " "differences are reported. An error is " "raised if any of the given " "refs are not present in the run.", ) reporter: BaseReporter = get_reporter_class(config=self.config) reporter_args = reporter.additional_arguments() if reporter_args: reporter_name = self.config.get_value("reporter.name") reporter_group_desc = ( f"Additional options from configured reporter class {reporter_name!r}" ) reporter_group = self.parser.add_argument_group(reporter_group_desc) # add builder-specific options into the group for arg in reporter_args: reporter_group.add_argument(arg.pop("flags"), **arg) def run(self, args: List[str]) -> int: if not args: self.parser.print_help() return ERROR self.add_arguments() options = self.parser.parse_args(args) reporter: BaseReporter = get_reporter_class(config=self.config) # TODO: Parse run to fit schema run = options.run refs: List[str] = options.refs result_dir = reporter.result_dir # TODO: Make this dynamic to support other run identifiers result = sorted(get_subdirs(result_dir))[-1] result_path = result_dir / result if result_path.exists(): reporter.compare( *refs, result=result, target_filter=options.target_filter, benchmark_filter=options.benchmark_filter, context_filter=options.context_filter, ) else: raise PybmError( f"No benchmark results found for the requested run {run!r}." ) return SUCCESS
1057
import network def conncb(task): print("[{}] Connected".format(task)) def disconncb(task): print("[{}] Disconnected".format(task)) def subscb(task): print("[{}] Subscribed".format(task)) def pubcb(pub): print("[{}] Published: {}".format(pub[0], pub[1])) def datacb(msg): print("[{}] Data arrived from topic: {}, Message:\n".format(msg[0], msg[1]), msg[2]) mqtt = network.mqtt("loboris", "mqtt://loboris.eu", user="wifimcu", password="<PASSWORD>", cleansession=True, connected_cb=conncb, disconnected_cb=disconncb, subscribed_cb=subscb, published_cb=pubcb, data_cb=datacb) # secure connection requires more memory and may not work # mqtts = network.mqtt("eclipse", "mqtts//iot.eclipse.org", cleansession=True, connected_cb=conncb, disconnected_cb=disconncb, subscribed_cb=subscb, published_cb=pubcb, data_cb=datacb) # wsmqtt = network.mqtt("eclipse", "ws://iot.eclipse.org:80/ws", cleansession=True, data_cb=datacb) mqtt.start() #mqtt.config(lwt_topic='status', lwt_msg='Disconected') ''' # Wait until status is: (1, 'Connected') mqtt.subscribe('test') mqtt.publish('test', 'Hi from Micropython') mqtt.stop() ''' # ================== # ThingSpeak example # ================== import network def datacb(msg): print("[{}] Data arrived from topic: {}, Message:\n".format(msg[0], msg[1]), msg[2]) thing = network.mqtt("thingspeak", "mqtt://mqtt.thingspeak.com", user="anyName", password="<PASSWORD>", cleansession=True, data_cb=datacb) # or secure connection #thing = network.mqtt("thingspeak", "mqtts://mqtt.thingspeak.com", user="anyName", password="<PASSWORD>", cleansession=True, data_cb=datacb) thingspeakChannelId = "123456" # enter Thingspeak Channel ID thingspeakChannelWriteApiKey = "ThingspeakWriteAPIKey" # EDIT - enter Thingspeak Write API Key thingspeakFieldNo = 1 thingSpeakChanelFormat = "json" pubchan = "channels/{:s}/publish/{:s}".format(thingspeakChannelId, thingspeakChannelWriteApiKey) pubfield = "channels/{:s}/publish/fields/field{}/{:s}".format(thingspeakChannelId, thingspeakFieldNo, thingspeakChannelWriteApiKey) subchan = "channels/{:s}/subscribe/{:s}/{:s}".format(thingspeakChannelId, thingSpeakChanelFormat, thingspeakChannelWriteApiKey) subfield = "channels/{:s}/subscribe/fields/field{}/{:s}".format(thingspeakChannelId, thingspeakFieldNo, thingspeakChannelWriteApiKey) thing.start() tmo = 0 while thing.status()[0] != 2: utime.sleep_ms(100) tmo += 1 if tmo > 80: print("Not connected") break # subscribe to channel thing.subscribe(subchan) # subscribe to field thing.subscribe(subfield) # publish to channel # Payload can include any of those fields separated b< ';': # "field1=value;field2=value;...;field8=value;latitude=value;longitude=value;elevation=value;status=value" thing.publish(pubchan, "field1=25.2;status=On line") # Publish to field thing.publish(pubfield, "24.5")
1094
import math def g_path_regularize(fake_img, latents, mean_path_length, decay=0.01): noise = torch.randn_like(fake_img) / math.sqrt( fake_img.shape[2] * fake_img.shape[3] ) grad, = autograd.grad( outputs=(fake_img * noise).sum(), inputs=latents, create_graph=True ) path_lengths = torch.sqrt(grad.pow(2).sum(2).mean(1)) path_mean = mean_path_length + decay * (path_lengths.mean() - mean_path_length) path_penalty = (path_lengths - path_mean).pow(2).mean() return path_penalty, path_mean.detach(), path_lengths
1104
def init(): global brightness global calibration_mode brightness = 500 calibration_mode = False
1116
def solution(A): total = sum(A) m = float('inf') left_sum = 0 for n in A[:-1]: left_sum += n v = abs(total - 2*left_sum) if v < m: m = v return m
1148
description = 'PGAA setup with XYZOmega sample table' group = 'basic' sysconfig = dict( datasinks = ['mcasink', 'chnsink', 'csvsink', 'livesink'] ) includes = [ 'system', 'reactor', 'nl4b', 'pressure', 'sampletable', 'pilz', 'detector', 'collimation', ] devices = dict( mcasink = device('nicos_mlz.pgaa.devices.MCASink', settypes = {'point'}, detectors = ['_60p', 'LEGe'], ), chnsink = device('nicos_mlz.pgaa.devices.CHNSink', settypes = {'point'}, detectors = ['_60p', 'LEGe'], ), csvsink = device('nicos_mlz.pgaa.devices.CSVDataSink', settypes = {'point'}, ), ) startupcode = """ SetDetectors('_60p', 'LEGe') SetEnvironment(chamber_pressure) printinfo("============================================================") printinfo("Welcome to the NICOS PGAI demo setup.") printinfo("============================================================") """
1160
import io import logging import json import numpy import torch import numpy as np from tqdm import tqdm from clie.inputters import constant from clie.objects import Sentence from torch.utils.data import Dataset from torch.utils.data.sampler import Sampler logger = logging.getLogger(__name__) def load_word_embeddings(file): embeddings_index = {} fin = io.open(file, 'r', encoding='utf-8', newline='\n', errors='ignore') n, d = map(int, fin.readline().split()) for i, line in tqdm(enumerate(fin), total=n): tokens = line.rstrip().split(' ') v = numpy.array(tokens[1:], dtype=float) embeddings_index[tokens[0]] = v return embeddings_index # ------------------------------------------------------------------------------ # Data loading # ------------------------------------------------------------------------------ def load_data(filename, src_lang, tgt_lang, knn_file, knn_size, max_examples=-1): examples = [] wrong_subj_pos, wrong_obj_pos = 0, 0 with open(filename) as f: data = json.load(f) knn_dict = None if knn_file: with open(knn_file) as f: knn_dict = json.load(f) for idx, ex in enumerate(tqdm(data, total=len(data))): sentence = Sentence(ex['id']) sentence.language = src_lang sentence.words = ex['token'] sentence.pos = ex['stanford_pos'] sentence.ner = ex['stanford_ner'] sentence.deprel = ex['stanford_deprel'] sentence.head = [int(x) for x in ex['stanford_head']] sentence.subj_type = ex['subj_type'] sentence.obj_type = ex['obj_type'] sentence.relation = ex['relation'] if ex['subj_end'] - ex['subj_start'] < 0: # we swap the start and end index wrong_subj_pos += 1 sentence.subject = [ex['subj_end'], ex['subj_start']] else: sentence.subject = [ex['subj_start'], ex['subj_end']] if ex['obj_end'] - ex['obj_start'] < 0: # we swap the start and end index wrong_obj_pos += 1 sentence.object = [ex['obj_end'], ex['obj_start']] else: sentence.object = [ex['obj_start'], ex['obj_end']] # store KNN word info if knn_dict: sentence.tgt_lang = tgt_lang knn_words = [] for w in ex['token']: w = '!{}_{}'.format(src_lang, w) if w in knn_dict: assert len(knn_dict[w]) == knn_size knn_words.append(knn_dict[w]) else: knn_words.append([constant.UNK_WORD] * knn_size) sentence.knn_words = knn_words examples.append(sentence) if max_examples != -1 and len(examples) > max_examples: break if wrong_subj_pos > 0 or wrong_obj_pos > 0: logger.info('{} and {} wrong subject and object positions found!'.format( wrong_subj_pos, wrong_obj_pos)) return examples def vectorize(ex, model, iseval): """Torchify a single example.""" words = ['!{}_{}'.format(ex.language, w) for w in ex.words] words = [model.word_dict[w] for w in words] knn_word = None if ex.knn_words: knn_word = [[model.word_dict[w] for w in knn] for knn in ex.knn_words] knn_word = torch.LongTensor(knn_word) word = torch.LongTensor(words) pos = torch.LongTensor([model.pos_dict[p] for p in ex.pos]) ner = torch.LongTensor([model.ner_dict[n] for n in ex.ner]) deprel = torch.LongTensor([model.deprel_dict[d] for d in ex.deprel]) assert any([x == 0 for x in ex.head]) head = torch.LongTensor(ex.head) subj_position = torch.LongTensor(ex.subj_position) obj_position = torch.LongTensor(ex.obj_position) type = [0] * len(ex.words) ttype = model.type_dict[ex.subj_type] start, end = ex.subject type[start: end + 1] = [ttype] * (end - start + 1) atype = model.type_dict[ex.obj_type] start, end = ex.object type[start: end + 1] = [atype] * (end - start + 1) type = torch.LongTensor(type) return { 'id': ex.id, 'language': ex.language, 'word': word, 'pos': pos, 'ner': ner, 'deprel': deprel, 'type': type, 'head': head, 'subject': ex.subj_text, 'object': ex.obj_text, 'subject_pos': subj_position, 'object_pos': obj_position, 'relation': model.label_dict[ex.relation], 'knn_word': knn_word } def batchify(batch): """Gather a batch of individual examples into one batch.""" # batch is a list of vectorized examples batch_size = len(batch) ids = [ex['id'] for ex in batch] language = [ex['language'] for ex in batch] use_knn = batch[0]['knn_word'] is not None # NOTE. batch[0]['knn_word'] is a 2d list knn_size = len(batch[0]['knn_word'][0]) if use_knn else 0 # --------- Prepare Code tensors --------- max_len = max([ex['word'].size(0) for ex in batch]) # Batch Code Representations len_rep = torch.LongTensor(batch_size).fill_(constant.PAD) word_rep = torch.LongTensor(batch_size, max_len).fill_(constant.PAD) head_rep = torch.LongTensor(batch_size, max_len).fill_(constant.PAD) subject_pos_rep = torch.LongTensor(batch_size, max_len).fill_(constant.PAD) object_pos_rep = torch.LongTensor(batch_size, max_len).fill_(constant.PAD) pos_rep = torch.LongTensor(batch_size, max_len).fill_(constant.PAD) ner_rep = torch.LongTensor(batch_size, max_len).fill_(constant.PAD) deprel_rep = torch.LongTensor(batch_size, max_len).fill_(constant.PAD) type_rep = torch.LongTensor(batch_size, max_len).fill_(constant.PAD) labels = torch.LongTensor(batch_size) subject = [] object = [] knn_rep = None if use_knn: knn_rep = torch.LongTensor(batch_size, max_len, knn_size).fill_(constant.PAD) for i, ex in enumerate(batch): len_rep[i] = ex['word'].size(0) labels[i] = ex['relation'] word_rep[i, :len_rep[i]] = ex['word'] head_rep[i, :len_rep[i]] = ex['head'] subject_pos_rep[i, :len_rep[i]] = ex['subject_pos'] object_pos_rep[i, :len_rep[i]] = ex['object_pos'] pos_rep[i, :len_rep[i]] = ex['pos'] ner_rep[i, :len_rep[i]] = ex['ner'] deprel_rep[i, :len_rep[i]] = ex['deprel'] type_rep[i, :len_rep[i]] = ex['type'] subject.append(ex['subject']) object.append(ex['object']) if use_knn: knn_rep[i, :len_rep[i]] = ex['knn_word'] return { 'ids': ids, 'language': language, 'batch_size': batch_size, 'len_rep': len_rep, 'word_rep': word_rep, 'knn_rep': knn_rep, 'head_rep': head_rep, 'subject': subject, 'object': object, 'subject_pos_rep': subject_pos_rep, 'object_pos_rep': object_pos_rep, 'labels': labels, 'pos_rep': pos_rep, 'ner_rep': ner_rep, 'deprel_rep': deprel_rep, 'type_rep': type_rep } class ACE05Dataset(Dataset): def __init__(self, examples, model, evaluation=False): self.model = model self.examples = examples self.evaluation = evaluation def __len__(self): return len(self.examples) def __getitem__(self, index): return vectorize(self.examples[index], self.model, iseval=self.evaluation) def lengths(self): return [len(ex.words) for ex in self.examples] class SortedBatchSampler(Sampler): def __init__(self, lengths, batch_size, shuffle=True): self.lengths = lengths self.batch_size = batch_size self.shuffle = shuffle def __iter__(self): lengths = np.array( [(-l, np.random.random()) for l in self.lengths], dtype=[('l1', np.int_), ('rand', np.float_)] ) indices = np.argsort(lengths, order=('l1', 'rand')) batches = [indices[i:i + self.batch_size] for i in range(0, len(indices), self.batch_size)] if self.shuffle: np.random.shuffle(batches) return iter([i for batch in batches for i in batch]) def __len__(self): return len(self.lengths)
1161
from distutils.extension import Extension cmdclass = {} try: # with Cython from Cython.Build import build_ext cmdclass["build_ext"] = build_ext module_src = "cgranges/python/cgranges.pyx" except ImportError: # without Cython module_src = "cgranges/python/cgranges.c" def build(setup_kwargs): """ This function is mandatory in order to build the extensions. """ setup_kwargs.update( { "ext_modules": [ Extension( "cgranges", sources=[module_src, "cgranges/cgranges.c"], depends=[ "cgranges/cgranges.h", "cgranges/khash.h", "cgranges/python/cgranges.pyx" ], include_dirs=["cgranges"] ) ], "cmdclass": cmdclass } )
1206
def unlock(m): return m.lower().translate( str.maketrans( 'abcdefghijklmnopqrstuvwxyz', '22233344455566677778889999' ) )
1319
import ssg.utils def preprocess(data, lang): data["arg_name_value"] = data["arg_name"] + "=" + data["arg_value"] if lang == "oval": # escape dot, this is used in oval regex data["escaped_arg_name_value"] = data["arg_name_value"].replace(".", "\\.") # replace . with _, this is used in test / object / state ids data["sanitized_arg_name"] = ssg.utils.escape_id(data["arg_name"]) return data
1342
from __future__ import absolute_import, division, print_function, unicode_literals from unittest import TestCase from beast.env.ReadEnvFile import read_env_file from beast.util import Terminal Terminal.CAN_CHANGE_COLOR = False JSON = """ { "FOO": "foo", "BAR": "bar bar bar", "CPPFLAGS": "-std=c++11 -frtti -fno-strict-aliasing -DWOMBAT" }""" ENV = """ # An env file. FOO=foo export BAR="bar bar bar" CPPFLAGS=-std=c++11 -frtti -fno-strict-aliasing -DWOMBAT # export BAZ=baz should be ignored. """ RESULT = { 'FOO': 'foo', 'BAR': 'bar bar bar', 'CPPFLAGS': '-std=c++11 -frtti -fno-strict-aliasing -DWOMBAT', } BAD_ENV = ENV + """ This line isn't right. NO SPACES IN NAMES="valid value" """ class test_ReadEnvFile(TestCase): def test_read_json(self): self.assertEqual(read_env_file(JSON), RESULT) def test_read_env(self): self.assertEqual(read_env_file(ENV), RESULT) def test_read_env_error(self): errors = [] self.assertEqual(read_env_file(BAD_ENV, errors.append), RESULT) self.assertEqual(errors, [ "WARNING: Didn't understand the following environment file lines:", "11. >>> This line isn't right.", '12. >>> NO SPACES IN NAMES="valid value"'])
1436
import requests from bs4 import BeautifulSoup import urllib.request import os import random import time def html(url): user_agents = [ 'Mozilla/5.0 (Windows; U; Windows NT 5.1; it; rv:1.8.1.11) Gecko/20071127 Firefox/2.0.0.11', 'Opera/9.25 (Windows NT 5.1; U; en)', 'Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.1.4322; .NET CLR 2.0.50727)', 'Mozilla/5.0 (compatible; Konqueror/3.5; Linux) KHTML/3.5.5 (like Gecko) (Kubuntu)', 'Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.0.12) Gecko/20070731 Ubuntu/dapper-security Firefox/1.5.0.12', 'Lynx/2.8.5rel.1 libwww-FM/2.14 SSL-MM/1.4.1 GNUTLS/1.2.9', "Mozilla/5.0 (X11; Linux i686) AppleWebKit/535.7 (KHTML, like Gecko) Ubuntu/11.04 Chromium/16.0.912.77 Chrome/16.0.912.77 Safari/535.7", "Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:10.0) Gecko/20100101 Firefox/10.0 "] user_agent = random.choice(user_agents) headers = { 'User-Agent': user_agent, 'Accept-Encoding': 'gzip'} req = requests.get(url=url, headers=headers) html_doc = req.text soup = BeautifulSoup(html_doc, "html.parser") times = soup.select("time") views = soup.select("p.label-key > b") active_str = str(views[2]) active = active_str[active_str.find("title=\"") + 7:active_str.find("Z")] answers = soup.select("#answers-header > div > h2 >span") question_content = soup.select("div.post-text") tags = soup.select("#question > div.post-layout > div.postcell.post-layout--right > " "div.post-taglist.grid.gs4.gsy.fd-column > div >a") title = soup.select("h1 >a") tags_str = "" item = [] for tag in tags: tags_str += tag.get_text() + "," answer_contetnts = [] for i in range(1, len(question_content)): answer_contetnts.append(question_content[i]) for i in range(len(times)): if len(times[i].get_text()) > 1: asked_time = times[i].get("datetime").replace("T", " ") item.append(title[ 0].get_text()) # title views answersnum asked_time tag_str active_time quest_content_ text answer_content_list item.append(views[1].get_text()) item.append(answers[0].get_text()) item.append(asked_time) item.append(tags_str) item.append(active) item.append(question_content[0]) item.append(answer_contetnts) print(item) # updatetosql(item) def updatetosql(item): ansers_text = "[split]".join(item[7]) updatesql = "UPDATE `t_stackoverflow_question` " \ "SET `tags`='%s', `views`='%s', `answers_num`='%s', `asked_time`='%s', `last_active_time`='%s', `question_content`='%s', `answers_contetnt`='%s' " \ "WHERE (`question_id`='%s') " \ % (item[4], item[1], item[2], item[3], item[5], item[6], ansers_text, item[0],) pass if __name__ == '__main__': html("https://stackoverflow.com/questions/50119673/nginx-fast-cgi-cache-on-error-page-404")
1469
import numpy as np from skimage.transform import resize from skimage import measure from skimage.measure import regionprops class OCROnObjects(): def __init__(self, license_plate): character_objects = self.identify_boundary_objects(license_plate) self.get_regions(character_objects, license_plate) def identify_boundary_objects(self, a_license_plate): labelImage = measure.label(a_license_plate) character_dimensions = (0.4*a_license_plate.shape[0], 0.85*a_license_plate.shape[0], 0.04*a_license_plate.shape[1], 0.15*a_license_plate.shape[1]) minHeight, maxHeight, minWidth, maxWidth = character_dimensions regionLists = regionprops(labelImage) return regionLists def get_regions(self, character_objects, a_license_plate): """ used to map out regions where the license plate charcters are the principle of connected component analysis and labelling were used Parameters: ----------- a_license_plate: 2D numpy binary image of the license plate Returns: -------- a dictionary containing the index fullscale: 3D array containig 2D array of each character columnsVal: 1D array the starting column of each character coordinates: """ cord = [] counter=0 column_list = [] character_dimensions = (0.35*a_license_plate.shape[0], 0.60*a_license_plate.shape[0], 0.05*a_license_plate.shape[1], 0.15*a_license_plate.shape[1]) minHeight, maxHeight, minWidth, maxWidth = character_dimensions for regions in character_objects: minimumRow, minimumCol, maximumRow, maximumCol = regions.bbox character_height = maximumRow - minimumRow character_width = maximumCol - minimumCol roi = a_license_plate[minimumRow:maximumRow, minimumCol:maximumCol] if character_height > minHeight and character_height < maxHeight and character_width > minWidth and character_width < maxWidth: if counter == 0: samples = resize(roi, (20,20)) cord.append(regions.bbox) counter += 1 elif counter == 1: roismall = resize(roi, (20,20)) samples = np.concatenate((samples[None,:,:], roismall[None,:,:]), axis=0) cord.append(regions.bbox) counter+=1 else: roismall = resize(roi, (20,20)) samples = np.concatenate((samples[:,:,:], roismall[None,:,:]), axis=0) cord.append(regions.bbox) column_list.append(minimumCol) if len(column_list) == 0: self.candidates = {} else: self.candidates = { 'fullscale': samples, 'coordinates': np.array(cord), 'columnsVal': column_list } return self.candidates
1485
from paddle.vision.transforms import ( ToTensor, RandomHorizontalFlip, RandomResizedCrop, SaturationTransform, Compose, HueTransform, BrightnessTransform, ContrastTransform, RandomCrop, Normalize, RandomRotation ) from paddle.vision.datasets import Cifar100 from paddle.io import DataLoader from paddle.optimizer.lr import CosineAnnealingDecay, MultiStepDecay, LinearWarmup import random from resnet20 import * import paddle # supernet trainning 基于paddleslim模型压缩包 # https://github.com/PaddlePaddle/PaddleSlim 欢迎大家多多star from paddleslim.nas.ofa.convert_super import Convert, supernet from paddleslim.nas.ofa import OFA, RunConfig, DistillConfig from paddleslim.nas.ofa.utils import utils channel_list = [] for i in range(1, 21): if 0 < i <= 7: # channel_list.append(random.choice([ 4, 8, 12, 16])) channel_list.append(16) elif 7 < i <= 13: # channel_list.append(random.choice([ 4, 8, 12, 16, 20, 24, 28, 32])) channel_list.append(32) elif 13 < i <= 19: # channel_list.append(random.choice([ 4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 48, 52, 56,60, 64])) channel_list.append(64) else: # channel_list.append(random.choice([ 4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 48, 52, 56,60, 64])) channel_list.append(64) net = ResNet20(100, channel_list) net2 = ResNet20(100, channel_list) net2.set_state_dict(paddle.load('./pretrained_model/resnet20.pdparams')) channel_optional = [] for i in range(0, 23): if i <= 7: channel_optional.append([4, 8, 12, 16]) # channel_optional.append([12, 16]) elif 7 < i <= 14: channel_optional.append([4, 8, 12, 16, 20, 24, 28, 32]) # channel_optional.append([20, 24, 28, 32]) elif 14 < i <= 21: channel_optional.append( [4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 48, 52, 56, 60, 64]) # channel_optional.append([36, 40, 44, 48, 52, 56,60, 64]) else: channel_optional.append( [4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 48, 52, 56, 60, 64]) # channel_optional.append([36, 40, 44, 48, 52, 56,60, 64]) distill_config = DistillConfig(teacher_model=net2) sp_net_config = supernet(channel=channel_optional) sp_model = Convert(sp_net_config).convert(net) ofa_net = OFA(sp_model, distill_config=distill_config) ofa_net.set_task('channel') model = paddle.Model(ofa_net) MAX_EPOCH = 300 LR = 0.1 WEIGHT_DECAY = 5e-4 MOMENTUM = 0.9 BATCH_SIZE = 128 CIFAR_MEAN = [0.5071, 0.4865, 0.4409] CIFAR_STD = [0.1942, 0.1918, 0.1958] DATA_FILE = './data/data76994/cifar-100-python.tar.gz' model.prepare( paddle.optimizer.Momentum( learning_rate=LinearWarmup( CosineAnnealingDecay(LR, MAX_EPOCH), 2000, 0., LR), momentum=MOMENTUM, parameters=model.parameters(), weight_decay=WEIGHT_DECAY), CrossEntropyLoss(), paddle.metric.Accuracy(topk=(1, 5))) transforms = Compose([ RandomCrop(32, padding=4), RandomApply(BrightnessTransform(0.1)), RandomApply(ContrastTransform(0.1)), RandomHorizontalFlip(), RandomRotation(15), ToArray(), Normalize(CIFAR_MEAN, CIFAR_STD), ]) val_transforms = Compose([ToArray(), Normalize(CIFAR_MEAN, CIFAR_STD)]) train_set = Cifar100(DATA_FILE, mode='train', transform=transforms) test_set = Cifar100(DATA_FILE, mode='test', transform=val_transforms) callbacks = [LRSchedulerM(), callbacks.VisualDL('vis_logs/ofa_resnet20')] model.fit( train_set, test_set, epochs=MAX_EPOCH, batch_size=BATCH_SIZE, save_dir='checkpoints', save_freq=100, shuffle=True, num_workers=4, verbose=1, callbacks=callbacks, )
1525
import numpy as np import cv2 import os import json import glob from PIL import Image, ImageDraw plate_diameter = 25 #cm plate_depth = 1.5 #cm plate_thickness = 0.2 #cm def Max(x, y): if (x >= y): return x else: return y def polygons_to_mask(img_shape, polygons): mask = np.zeros(img_shape, dtype=np.uint8) mask = Image.fromarray(mask) xy = list(map(tuple, polygons)) ImageDraw.Draw(mask).polygon(xy=xy, outline=1, fill=1) mask = np.array(mask, dtype=bool) return mask def mask2box(mask): index = np.argwhere(mask == 1) rows = index[:, 0] clos = index[:, 1] left_top_r = np.min(rows) left_top_c = np.min(clos) right_bottom_r = np.max(rows) right_bottom_c = np.max(clos) return [left_top_c, left_top_r, right_bottom_c, right_bottom_r] def get_bbox(points, h, w): polygons = points mask = polygons_to_mask([h,w], polygons) return mask2box(mask) def get_scale(points, img, lowest): bbox = get_bbox(points, img.shape[0], img.shape[1]) diameter = (bbox[2]-bbox[0]+1+bbox[3]-bbox[1]+1)/2 len_per_pix = plate_diameter/float(diameter) avg = 0 k = 0 for point in points: avg += img[point[1]][point[0]] k += 1 avg = avg/float(k) depth = lowest - avg depth_per_pix = plate_depth/depth return len_per_pix, depth_per_pix def cal_volume(points, img, len_per_pix, depth_per_pix, lowest): volume = 0.0 bbox = get_bbox(points, img.shape[0], img.shape[1]) points = np.array(points) shape = points.shape points = points.reshape(shape[0], 1, shape[1]) for i in range(bbox[0], bbox[2]+1): for j in range(bbox[1], bbox[3]+1): if (cv2.pointPolygonTest(points, (i,j), False) >= 0): volume += Max(0, (lowest - img[j][i]) * depth_per_pix - plate_thickness) * len_per_pix * len_per_pix return volume def get_volume(img, json_path): lowest = np.max(img) vol_dict = {} #print(lowest) len_per_pix = 0.0 depth_per_pix = 0.0 with open(json_path, 'r') as json_file: data = json.load(json_file) for shape in data['shapes']: if (shape['label'] == "plate"): len_per_pix, depth_per_pix = get_scale(shape['points'], img, lowest) #print(len_per_pix, depth_per_pix) break for shape in data['shapes']: label = shape['label'] if (label == "plate"): continue points = shape['points'] volume = cal_volume(points, img, len_per_pix, depth_per_pix, lowest) if (label in vol_dict): vol_dict[label] += volume else: vol_dict[label] = volume return vol_dict img = cv2.imread("out.png",0) print(get_volume(img,"test.json"))
1529
from __future__ import absolute_import from __future__ import division from __future__ import print_function import cntk as C import numpy as np from .common import floatx, epsilon, image_dim_ordering, image_data_format from collections import defaultdict from contextlib import contextmanager import warnings C.set_global_option('align_axis', 1) b_any = any dev = C.device.use_default_device() if dev.type() == 0: warnings.warn( 'CNTK backend warning: GPU is not detected. ' 'CNTK\'s CPU version is not fully optimized,' 'please run with GPU to get better performance.') # A learning phase is a bool tensor used to run Keras models in # either train mode (learning_phase == 1) or test mode (learning_phase == 0). # LEARNING_PHASE_PLACEHOLDER is the placeholder for dynamic learning phase _LEARNING_PHASE_PLACEHOLDER = C.constant(shape=(), dtype=np.float32, value=1.0, name='_keras_learning_phase') # static learning phase flag, if it is not 0 or 1, we will go with dynamic learning phase tensor. _LEARNING_PHASE = -1 _UID_PREFIXES = defaultdict(int) # cntk doesn't support gradient as symbolic op, to hook up with keras model, # we will create gradient as a constant placeholder, here use this global # map to keep the mapping from grad placeholder to parameter grad_parameter_dict = {} NAME_SCOPE_STACK = [] @contextmanager def name_scope(name): global NAME_SCOPE_STACK NAME_SCOPE_STACK.append(name) yield NAME_SCOPE_STACK.pop() def get_uid(prefix=''): _UID_PREFIXES[prefix] += 1 return _UID_PREFIXES[prefix] def learning_phase(): # If _LEARNING_PHASE is not 0 or 1, return dynamic learning phase tensor return _LEARNING_PHASE if _LEARNING_PHASE in {0, 1} else _LEARNING_PHASE_PLACEHOLDER def set_learning_phase(value): global _LEARNING_PHASE if value not in {0, 1}: raise ValueError('CNTK Backend: Set learning phase ' 'with value %s is not supported, ' 'expected 0 or 1.' % value) _LEARNING_PHASE = value def clear_session(): """Reset learning phase flag for cntk backend. """ global _LEARNING_PHASE global _LEARNING_PHASE_PLACEHOLDER _LEARNING_PHASE = -1 _LEARNING_PHASE_PLACEHOLDER.value = np.asarray(1.0) def in_train_phase(x, alt, training=None): global _LEARNING_PHASE if training is None: training = learning_phase() uses_learning_phase = True else: uses_learning_phase = False # CNTK currently don't support cond op, so here we use # element_select approach as workaround. It may have # perf issue, will resolve it later with cntk cond op. if callable(x) and isinstance(x, C.cntk_py.Function) is False: x = x() if callable(alt) and isinstance(alt, C.cntk_py.Function) is False: alt = alt() if training is True: x._uses_learning_phase = uses_learning_phase return x else: # if _LEARNING_PHASE is static if isinstance(training, int) or isinstance(training, bool): result = x if training == 1 or training is True else alt else: result = C.element_select(training, x, alt) result._uses_learning_phase = uses_learning_phase return result def in_test_phase(x, alt, training=None): return in_train_phase(alt, x, training=training) def _convert_string_dtype(dtype): # cntk only support float32 and float64 if dtype == 'float32': return np.float32 elif dtype == 'float64': return np.float64 else: # cntk only running with float, # try to cast to float to run the model return np.float32 def _convert_dtype_string(dtype): if dtype == np.float32: return 'float32' elif dtype == np.float64: return 'float64' else: raise ValueError('CNTK Backend: Unsupported dtype: %s. ' 'CNTK only supports float32 and ' 'float64.' % dtype) def variable(value, dtype=None, name=None, constraint=None): """Instantiates a variable and returns it. # Arguments value: Numpy array, initial value of the tensor. dtype: Tensor type. name: Optional name string for the tensor. constraint: Optional projection function to be applied to the variable after an optimizer update. # Returns A variable instance (with Keras metadata included). """ if dtype is None: dtype = floatx() if name is None: name = '' if isinstance( value, C.variables.Constant) or isinstance( value, C.variables.Parameter): value = value.value # we don't support init parameter with symbolic op, so eval it first as # workaround if isinstance(value, C.cntk_py.Function): value = eval(value) shape = value.shape if hasattr(value, 'shape') else () if hasattr(value, 'dtype') and value.dtype != dtype and len(shape) > 0: value = value.astype(dtype) # TODO: remove the conversion when cntk supports int32, int64 # https://docs.microsoft.com/en-us/python/api/cntk.variables.parameter dtype = 'float32' if 'int' in str(dtype) else dtype v = C.parameter(shape=shape, init=value, dtype=dtype, name=_prepare_name(name, 'variable')) v._keras_shape = v.shape v._uses_learning_phase = False v.constraint = constraint return v def bias_add(x, bias, data_format=None): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) dims = len(x.shape) if dims > 0 and x.shape[0] == C.InferredDimension: dims -= 1 bias_dims = len(bias.shape) if bias_dims != 1 and bias_dims != dims: raise ValueError('Unexpected bias dimensions %d, ' 'expected 1 or %d dimensions' % (bias_dims, dims)) if dims == 4: if data_format == 'channels_first': if bias_dims == 1: shape = (bias.shape[0], 1, 1, 1) else: shape = (bias.shape[3],) + bias.shape[:3] elif data_format == 'channels_last': if bias_dims == 1: shape = (1, 1, 1, bias.shape[0]) else: shape = bias.shape elif dims == 3: if data_format == 'channels_first': if bias_dims == 1: shape = (bias.shape[0], 1, 1) else: shape = (bias.shape[2],) + bias.shape[:2] elif data_format == 'channels_last': if bias_dims == 1: shape = (1, 1, bias.shape[0]) else: shape = bias.shape elif dims == 2: if data_format == 'channels_first': if bias_dims == 1: shape = (bias.shape[0], 1) else: shape = (bias.shape[1],) + bias.shape[:1] elif data_format == 'channels_last': if bias_dims == 1: shape = (1, bias.shape[0]) else: shape = bias.shape else: shape = bias.shape return x + reshape(bias, shape) def eval(x): if isinstance(x, C.cntk_py.Function): return x.eval() elif isinstance(x, C.variables.Constant) or isinstance(x, C.variables.Parameter): return x.value else: raise ValueError('CNTK Backend: `eval` method on ' '`%s` type is not supported. ' 'CNTK only supports `eval` with ' '`Function`, `Constant` or ' '`Parameter`.' % type(x)) def placeholder( shape=None, ndim=None, dtype=None, sparse=False, name=None, dynamic_axis_num=1): if dtype is None: dtype = floatx() if not shape: if ndim: shape = tuple([None for _ in range(ndim)]) dynamic_dimension = C.FreeDimension if _get_cntk_version() >= 2.2 else C.InferredDimension cntk_shape = [dynamic_dimension if s is None else s for s in shape] cntk_shape = tuple(cntk_shape) if dynamic_axis_num > len(cntk_shape): raise ValueError('CNTK backend: creating placeholder with ' '%d dimension is not supported, at least ' '%d dimensions are needed.' % (len(cntk_shape, dynamic_axis_num))) if name is None: name = '' cntk_shape = cntk_shape[dynamic_axis_num:] x = C.input( shape=cntk_shape, dtype=_convert_string_dtype(dtype), is_sparse=sparse, name=name) x._keras_shape = shape x._uses_learning_phase = False x._cntk_placeholder = True return x def is_placeholder(x): """Returns whether `x` is a placeholder. # Arguments x: A candidate placeholder. # Returns Boolean. """ return hasattr(x, '_cntk_placeholder') and x._cntk_placeholder def is_keras_tensor(x): if not is_tensor(x): raise ValueError('Unexpectedly found an instance of type `' + str(type(x)) + '`. ' 'Expected a symbolic tensor instance.') return hasattr(x, '_keras_history') def is_tensor(x): return isinstance(x, (C.variables.Constant, C.variables.Variable, C.variables.Parameter, C.ops.functions.Function)) def shape(x): shape = list(int_shape(x)) num_dynamic = _get_dynamic_axis_num(x) non_dyn_shape = [] for i in range(len(x.shape)): if shape[i + num_dynamic] is None: non_dyn_shape.append(x.shape[i]) else: non_dyn_shape.append(shape[i + num_dynamic]) return shape[:num_dynamic] + non_dyn_shape def is_sparse(tensor): return tensor.is_sparse def int_shape(x): if hasattr(x, '_keras_shape'): return x._keras_shape shape = x.shape if hasattr(x, 'dynamic_axes'): dynamic_shape = [None for a in x.dynamic_axes] shape = tuple(dynamic_shape) + shape return shape def ndim(x): shape = int_shape(x) return len(shape) def _prepare_name(name, default): prefix = '_'.join(NAME_SCOPE_STACK) if name is None or name == '': return prefix + '/' + default return prefix + '/' + name def constant(value, dtype=None, shape=None, name=None): if dtype is None: dtype = floatx() if shape is None: shape = () np_value = value * np.ones(shape) const = C.constant(np_value, dtype=dtype, name=_prepare_name(name, 'constant')) const._keras_shape = const.shape const._uses_learning_phase = False return const def random_binomial(shape, p=0.0, dtype=None, seed=None): # use numpy workaround now if seed is None: # ensure that randomness is conditioned by the Numpy RNG seed = np.random.randint(10e7) np.random.seed(seed) if dtype is None: dtype = np.float32 else: dtype = _convert_string_dtype(dtype) size = 1 for _ in shape: if _ is None: raise ValueError('CNTK Backend: randomness op with ' 'dynamic shape is not supported now. ' 'Please provide fixed dimension ' 'instead of `None`.') size *= _ binomial = np.random.binomial(1, p, size).astype(dtype).reshape(shape) return variable(value=binomial, dtype=dtype) def random_uniform(shape, minval=0.0, maxval=1.0, dtype=None, seed=None): for _ in shape: if _ is None: raise ValueError('CNTK Backend: randomness op with ' 'dynamic shape is not supported now. ' 'Please provide fixed dimension ' 'instead of `None`.') return random_uniform_variable(shape, minval, maxval, dtype, seed) def random_uniform_variable(shape, low, high, dtype=None, name=None, seed=None): if dtype is None: dtype = floatx() if seed is None: # ensure that randomness is conditioned by the Numpy RNG seed = np.random.randint(10e3) if dtype is None: dtype = np.float32 else: dtype = _convert_string_dtype(dtype) if name is None: name = '' scale = (high - low) / 2 p = C.parameter( shape, init=C.initializer.uniform( scale, seed=seed), dtype=dtype, name=name) return variable(value=p.value + low + scale) def random_normal_variable( shape, mean, scale, dtype=None, name=None, seed=None): if dtype is None: dtype = floatx() if seed is None: # ensure that randomness is conditioned by the Numpy RNG seed = np.random.randint(10e7) if dtype is None: dtype = np.float32 else: dtype = _convert_string_dtype(dtype) if name is None: name = '' return C.parameter( shape=shape, init=C.initializer.normal( scale=scale, seed=seed), dtype=dtype, name=name) def random_normal(shape, mean=0.0, stddev=1.0, dtype=None, seed=None): if dtype is None: dtype = floatx() for _ in shape: if _ is None: raise ValueError('CNTK Backend: randomness op with ' 'dynamic shape is not supported now. ' 'Please provide fixed dimension ' 'instead of `None`.') # how to apply mean and stddev return random_normal_variable(shape=shape, mean=mean, scale=1.0, seed=seed) def truncated_normal(shape, mean=0.0, stddev=1.0, dtype=None, seed=None): if seed is None: seed = np.random.randint(1, 10e6) if dtype is None: dtype = np.float32 else: dtype = _convert_string_dtype(dtype) return C.parameter( shape, init=C.initializer.truncated_normal( stddev, seed=seed), dtype=dtype) def dtype(x): return _convert_dtype_string(x.dtype) def zeros(shape, dtype=None, name=None): if dtype is None: dtype = floatx() ctype = _convert_string_dtype(dtype) return variable(value=np.zeros(shape, ctype), dtype=dtype, name=name) def ones(shape, dtype=None, name=None): if dtype is None: dtype = floatx() ctype = _convert_string_dtype(dtype) return variable(value=np.ones(shape, ctype), dtype=dtype, name=name) def eye(size, dtype=None, name=None): if dtype is None: dtype = floatx() return variable(np.eye(size), dtype, name) def zeros_like(x, dtype=None, name=None): return x * 0 def ones_like(x, dtype=None, name=None): return zeros_like(x) + 1 def count_params(x): for _ in x.shape: if _ == C.InferredDimension or _ == C.FreeDimension: raise ValueError('CNTK backend: `count_params` with dynamic ' 'shape is not supported. Please provide ' 'fixed dimension instead of `None`.') return np.prod(int_shape(x)) def cast(x, dtype): # cntk calculate everything in float, so don't need case from bool / int return x def dot(x, y): if len(x.shape) > 2 or len(y.shape) > 2: y_shape = int_shape(y) if len(y_shape) > 2: permutation = [len(y_shape) - 2] permutation += list(range(len(y_shape) - 2)) permutation += [len(y_shape) - 1] y = C.transpose(y, perm=permutation) return C.times(x, y, len(y_shape) - 1) else: return C.times(x, y) def batch_dot(x, y, axes=None): x_shape = int_shape(x) y_shape = int_shape(y) if isinstance(axes, int): axes = (axes, axes) if axes is None: # behaves like tf.batch_matmul as default axes = [len(x_shape) - 1, len(y_shape) - 2] if b_any([isinstance(a, (list, tuple)) for a in axes]): raise ValueError('Multiple target dimensions are not supported. ' + 'Expected: None, int, (int, int), ' + 'Provided: ' + str(axes)) if len(x_shape) == 2 and len(y_shape) == 2: if axes[0] == axes[1]: result = sum(x * y, axis=axes[0], keepdims=True) return result if axes[0] == 1 else transpose(result) else: return sum(x * transpose(y), axis=axes[0], keepdims=True) else: if len(y_shape) == 2: y = expand_dims(y) normalized_axis = [] normalized_axis.append(_normalize_axis(axes[0], x)[0]) normalized_axis.append(_normalize_axis(axes[1], y)[0]) # transpose i = normalized_axis[0] while i < len(x.shape) - 1: x = C.swapaxes(x, i, i + 1) i += 1 i = normalized_axis[1] while i > 0: y = C.swapaxes(y, i, i - 1) i -= 1 result = C.times(x, y, output_rank=(len(y.shape) - 1) if len(y.shape) > 1 else 1) if len(y_shape) == 2: result = squeeze(result, -1) return result def transpose(x): return C.swapaxes(x, 0, 1) def gather(reference, indices): # There is a bug in cntk gather op which may cause crash. # We have made a fix but not catched in CNTK 2.1 release. # Will update with gather op in next release if _get_cntk_version() >= 2.2: return C.ops.gather(reference, indices) else: num_classes = reference.shape[0] one_hot_matrix = C.ops.one_hot(indices, num_classes) return C.times(one_hot_matrix, reference, output_rank=len(reference.shape) - 1) def _remove_dims(x, axis, keepdims=False): if keepdims is False and isinstance(axis, list): # sequence axis is removed by default, so don't need reshape on it reduce_axes = [] for a in axis: if isinstance(a, C.Axis) is False: reduce_axes.append(a) return _reshape_dummy_dim(x, reduce_axes) else: if isinstance(axis, list): has_seq = False for a in axis: if isinstance(a, C.Axis): has_seq = True break if has_seq: nones = _get_dynamic_axis_num(x) x = expand_dims(x, nones) return x def max(x, axis=None, keepdims=False): axis = _normalize_axis(axis, x) output = _reduce_on_axis(x, axis, 'reduce_max') return _remove_dims(output, axis, keepdims) def min(x, axis=None, keepdims=False): axis = _normalize_axis(axis, x) output = _reduce_on_axis(x, axis, 'reduce_min') return _remove_dims(output, axis, keepdims) def sum(x, axis=None, keepdims=False): axis = _normalize_axis(axis, x) output = _reduce_on_axis(x, axis, 'reduce_sum') return _remove_dims(output, axis, keepdims) def prod(x, axis=None, keepdims=False): axis = _normalize_axis(axis, x) output = _reduce_on_axis(x, axis, 'reduce_prod') return _remove_dims(output, axis, keepdims) def logsumexp(x, axis=None, keepdims=False): return log(sum(exp(x), axis=axis, keepdims=keepdims)) def var(x, axis=None, keepdims=False): m = mean(x, axis, keepdims=True) devs_squared = C.square(x - m) return mean(devs_squared, axis=axis, keepdims=keepdims) def std(x, axis=None, keepdims=False): return C.sqrt(var(x, axis=axis, keepdims=keepdims)) def expand_dims(x, axis=-1): shape = list(int_shape(x)) nones = _get_dynamic_axis_num(x) index = axis if axis >= 0 else len(shape) + 1 shape.insert(index, 1) new_shape = shape[nones:] new_shape = tuple( [C.InferredDimension if _ is None else _ for _ in new_shape]) result = C.reshape(x, new_shape) if index < nones: result._keras_shape = shape return result def squeeze(x, axis): if isinstance(axis, tuple): axis = list(axis) if not isinstance(axis, list): axis = [axis] shape = list(int_shape(x)) _axis = [] for _ in axis: if isinstance(_, int): _axis.append(_ if _ >= 0 else _ + len(shape)) if len(_axis) == 0: return x nones = _get_dynamic_axis_num(x) for _ in sorted(_axis, reverse=True): del shape[_] new_shape = shape[nones:] new_shape = tuple([C.InferredDimension if _ == C.FreeDimension else _ for _ in new_shape]) return C.reshape(x, new_shape) def tile(x, n): if isinstance(n, int): n = (n,) elif isinstance(n, list): n = tuple(n) shape = int_shape(x) num_dynamic_axis = _get_dynamic_axis_num(x) # Padding the axis if len(n) < len(shape): n = tuple([1 for _ in range(len(shape) - len(n))]) + n if len(n) != len(shape): raise NotImplementedError i = num_dynamic_axis for i, rep in enumerate(n): if i >= num_dynamic_axis and shape[i] is not None: tmp = [x] * rep x = C.splice(*tmp, axis=i - num_dynamic_axis) i += 1 return x def _normalize_axis(axis, x): shape = int_shape(x) ndim = len(shape) nones = _get_dynamic_axis_num(x) if nones > ndim: raise ValueError('CNTK Backend: tensor with keras shape: `%s` has ' '%d cntk dynamic axis, this is not expected, please ' 'double check the keras shape history.' % (str(shape), nones)) # Current cntk does not support shape like (1, batch). so using the workaround # here to mapping the correct axis. Will remove this tricky after we add support # in native cntk op cntk_axis = [] dynamic_axis_index = 0 for i in range(ndim): if shape[i] is None and dynamic_axis_index < nones: cntk_axis.append(x.dynamic_axes[dynamic_axis_index]) dynamic_axis_index += 1 else: cntk_axis.append(i - dynamic_axis_index) if dynamic_axis_index < nones: i = 0 while dynamic_axis_index < nones: cntk_axis[i] = x.dynamic_axes[dynamic_axis_index] i += 1 dynamic_axis_index += 1 while i < len(cntk_axis): cntk_axis[i] -= nones i += 1 if isinstance(axis, tuple): _axis = list(axis) elif isinstance(axis, int): _axis = [axis] elif isinstance(axis, list): _axis = list(axis) else: _axis = axis if isinstance(_axis, list): for i, a in enumerate(_axis): if a is not None and a < 0: _axis[i] = (a % ndim) if _axis[i] is not None: _axis[i] = cntk_axis[_axis[i]] else: if _axis is None: _axis = C.Axis.all_axes() return _axis def _reshape_dummy_dim(x, axis): shape = list(x.shape) _axis = [_ + len(shape) if _ < 0 else _ for _ in axis] if shape.count(C.InferredDimension) > 1 or shape.count(C.FreeDimension) > 1: result = x for index in sorted(_axis, reverse=True): result = C.reshape(result, shape=(), begin_axis=index, end_axis=index + 1) return result else: for index in sorted(_axis, reverse=True): del shape[index] shape = [C.InferredDimension if _ == C.FreeDimension else _ for _ in shape] return C.reshape(x, shape) def mean(x, axis=None, keepdims=False): axis = _normalize_axis(axis, x) output = _reduce_on_axis(x, axis, 'reduce_mean') return _remove_dims(output, axis, keepdims) def any(x, axis=None, keepdims=False): reduce_result = sum(x, axis, keepdims=keepdims) any_matrix = C.element_select( reduce_result, ones_like(reduce_result), zeros_like(reduce_result)) if len(reduce_result.shape) == 0 and _get_dynamic_axis_num(x) == 0: return C.reduce_sum(any_matrix) else: return any_matrix def all(x, axis=None, keepdims=False): reduce_result = prod(x, axis, keepdims=keepdims) all_matrix = C.element_select( reduce_result, ones_like(reduce_result), zeros_like(reduce_result)) if len(reduce_result.shape) == 0 and _get_dynamic_axis_num(x) == 0: return C.reduce_sum(all_matrix) else: return all_matrix def classification_error(target, output, axis=-1): return C.ops.reduce_mean( C.equal( argmax( output, axis=-1), argmax( target, axis=-1)), axis=C.Axis.all_axes()) def argmax(x, axis=-1): axis = [axis] axis = _normalize_axis(axis, x) output = C.ops.argmax(x, axis=axis[0]) return _reshape_dummy_dim(output, axis) def argmin(x, axis=-1): axis = [axis] axis = _normalize_axis(axis, x) output = C.ops.argmin(x, axis=axis[0]) return _reshape_dummy_dim(output, axis) def square(x): return C.square(x) def abs(x): return C.abs(x) def sqrt(x): return C.sqrt(x) def exp(x): return C.exp(x) def log(x): return C.log(x) def round(x): return C.round(x) def sigmoid(x): return C.sigmoid(x) def sign(x): return x / C.abs(x) def pow(x, a): return C.pow(x, a) def clip(x, min_value, max_value): if max_value is not None and max_value < min_value: max_value = min_value if max_value is None: max_value = np.inf if min_value is None: min_value = -np.inf return C.clip(x, min_value, max_value) def binary_crossentropy(target, output, from_logits=False): if from_logits: output = C.sigmoid(output) output = C.clip(output, epsilon(), 1.0 - epsilon()) output = -target * C.log(output) - (1.0 - target) * C.log(1.0 - output) return output def get_variable_shape(x): return int_shape(x) def update(x, new_x): return C.assign(x, new_x) def moving_average_update(variable, value, momentum): return C.assign(variable, variable * momentum + value * (1. - momentum)) def update_add(x, increment): result = x + increment return C.assign(x, result) def gradients(loss, variables): # cntk does not support gradients as symbolic op, # to hook up with keras model # we will return a constant as place holder, the cntk learner will apply # the gradient during training. global grad_parameter_dict if isinstance(variables, list) is False: variables = [variables] grads = [] for v in variables: g = C.constant(0, shape=v.shape, name='keras_grad_placeholder') grads.append(g) grad_parameter_dict[g] = v return grads def equal(x, y): return C.equal(x, y) def not_equal(x, y): return C.not_equal(x, y) def greater(x, y): return C.greater(x, y) def greater_equal(x, y): return C.greater_equal(x, y) def less(x, y): return C.less(x, y) def less_equal(x, y): return C.less_equal(x, y) def maximum(x, y): return C.element_max(x, y) def minimum(x, y): return C.element_min(x, y) def sin(x): return C.sin(x) def cos(x): return C.cos(x) def normalize_batch_in_training(x, gamma, beta, reduction_axes, epsilon=1e-3): if gamma is None: if beta is None: gamma = ones_like(x) else: gamma = ones_like(beta) if beta is None: if gamma is None: beta = zeros_like(x) else: beta = zeros_like(gamma) mean, variant = _moments(x, _normalize_axis(reduction_axes, x)) if sorted(reduction_axes) == list(range(ndim(x)))[:-1]: normalized = batch_normalization( x, mean, variant, beta, gamma, epsilon) else: # need broadcasting target_shape = [] x_shape = int_shape(x) # skip the batch axis for axis in range(1, ndim(x)): if axis in reduction_axes: target_shape.append(1) if ndim(gamma) > axis: gamma = C.reduce_mean(gamma, axis - 1) beta = C.reduce_mean(beta, axis - 1) else: target_shape.append(x_shape[axis]) broadcast_mean = C.reshape(mean, target_shape) broadcast_var = C.reshape(variant, target_shape) broadcast_gamma = C.reshape(gamma, target_shape) broadcast_beta = C.reshape(beta, target_shape) normalized = batch_normalization( x, broadcast_mean, broadcast_var, broadcast_beta, broadcast_gamma, epsilon) return normalized, mean, variant def _moments(x, axes=None, shift=None, keep_dims=False): _axes = tuple(axes) if shift is None: shift = x # Compute true mean while keeping the dims for proper broadcasting. for axis in _axes: shift = C.reduce_mean(shift, axis=axis) shift = C.stop_gradient(shift) shifted_mean = C.minus(x, shift) for axis in _axes: shifted_mean = C.reduce_mean(shifted_mean, axis=axis) variance_mean = C.square(C.minus(x, shift)) for axis in _axes: variance_mean = C.reduce_mean(variance_mean, axis=axis) variance = C.minus(variance_mean, C.square(shifted_mean)) mean = C.plus(shifted_mean, shift) if not keep_dims: mean = squeeze(mean, _axes) variance = squeeze(variance, _axes) return mean, variance def batch_normalization(x, mean, var, beta, gamma, epsilon=1e-3): # The mean / var / beta / gamma may be processed by broadcast # so it may have an extra batch axis with 1, it is not needed # in cntk, need to remove those dummy axis. if ndim(mean) == ndim(x) and shape(mean)[0] == 1: mean = _reshape_dummy_dim(mean, [0]) if ndim(var) == ndim(x) and shape(var)[0] == 1: var = _reshape_dummy_dim(var, [0]) if gamma is None: gamma = ones_like(var) elif ndim(gamma) == ndim(x) and shape(gamma)[0] == 1: gamma = _reshape_dummy_dim(gamma, [0]) if beta is None: beta = zeros_like(mean) elif ndim(beta) == ndim(x) and shape(beta)[0] == 1: beta = _reshape_dummy_dim(beta, [0]) return (x - mean) / (C.sqrt(var) + epsilon) * gamma + beta def concatenate(tensors, axis=-1): if len(tensors) == 0: return None axis = [axis] axis = _normalize_axis(axis, tensors[0]) return C.splice(*tensors, axis=axis[0]) def flatten(x): return reshape(x, (-1,)) def reshape(x, shape): shape = tuple([C.InferredDimension if _ == C.FreeDimension else _ for _ in shape]) if isinstance(x, C.variables.Parameter): return C.reshape(x, shape) else: num_dynamic_axis = _get_dynamic_axis_num(x) if num_dynamic_axis == 1 and len(shape) > 0 and shape[0] == -1: # collapse axis with batch axis if b_any(_ == C.InferredDimension for _ in x.shape) or b_any( _ == C.FreeDimension for _ in x.shape): warnings.warn( 'Warning: CNTK backend does not support ' 'collapse of batch axis with inferred dimension. ' 'The reshape did not take place.') return x return _reshape_batch(x, shape) else: # no collapse, then first need to padding the shape if num_dynamic_axis >= len(shape): i = 0 while i < len(shape): if shape[i] is None or shape[i] == -1: i += 1 else: break shape = tuple([-1 for _ in range(num_dynamic_axis - i)]) + shape new_shape = list(shape) new_shape = new_shape[num_dynamic_axis:] new_shape = [C.InferredDimension if _ is None else _ for _ in new_shape] return C.reshape(x, new_shape) def permute_dimensions(x, pattern): dims = len(int_shape(x)) num_dynamic_axis = _get_dynamic_axis_num(x) if isinstance(pattern, list): current_layout = [i for i in range(dims)] else: current_layout = tuple([i for i in range(dims)]) if num_dynamic_axis > 0 and pattern[:num_dynamic_axis] != current_layout[:num_dynamic_axis]: raise ValueError('CNTK backend: the permute pattern %s ' 'requested permute on dynamic axis, ' 'which is not supported. Please do permute ' 'on static axis.' % pattern) axis = list(pattern) axis = axis[num_dynamic_axis:] axis = _normalize_axis(axis, x) return C.transpose(x, axis) def resize_images(x, height_factor, width_factor, data_format): if data_format == 'channels_first': output = repeat_elements(x, height_factor, axis=2) output = repeat_elements(output, width_factor, axis=3) return output elif data_format == 'channels_last': output = repeat_elements(x, height_factor, axis=1) output = repeat_elements(output, width_factor, axis=2) return output else: raise ValueError('CNTK Backend: Invalid data_format:', data_format) def resize_volumes(x, depth_factor, height_factor, width_factor, data_format): if data_format == 'channels_first': output = repeat_elements(x, depth_factor, axis=2) output = repeat_elements(output, height_factor, axis=3) output = repeat_elements(output, width_factor, axis=4) return output elif data_format == 'channels_last': output = repeat_elements(x, depth_factor, axis=1) output = repeat_elements(output, height_factor, axis=2) output = repeat_elements(output, width_factor, axis=3) return output else: raise ValueError('CNTK Backend: Invalid data_format:', data_format) def repeat_elements(x, rep, axis): axis = _normalize_axis(axis, x) axis = axis[0] slices = [] shape = x.shape i = 0 while i < shape[axis]: tmp = C.ops.slice(x, axis, i, i + 1) for _ in range(rep): slices.append(tmp) i += 1 return C.splice(*slices, axis=axis) def repeat(x, n): # this is a workaround for recurrent layer # if n is inferred dimension, # we can't figure out how to repeat it in cntk now # return the same x to take cntk broadcast feature # to make the recurrent layer work. # need to be fixed in GA. if n is C.InferredDimension or n is C.FreeDimension: return x index = 1 - _get_dynamic_axis_num(x) if index < 0 or index > 1: raise NotImplementedError new_shape = list(x.shape) new_shape.insert(index, 1) new_shape = tuple(new_shape) x = C.reshape(x, new_shape) temp = [x] * n return C.splice(*temp, axis=index) def tanh(x): return C.tanh(x) def _static_rnn(step_function, inputs, initial_states, go_backwards=False, mask=None, constants=None, unroll=False, input_length=None): shape = int_shape(inputs) dims = len(shape) uses_learning_phase = False if dims < 3: raise ValueError('Input should be at least 3D.') # if the second axis is static axis, CNTK will do unroll by default if shape[1] is None: raise ValueError('CNTK Backend: the input of static rnn ' 'has shape `%s`, the second axis ' 'is not static. If you want to run ' 'rnn with non-static axis, please try ' 'dynamic rnn with sequence axis.' % shape) if constants is None: constants = [] if mask is not None: mask_shape = int_shape(mask) if len(mask_shape) == dims - 1: mask = expand_dims(mask) nones = _get_dynamic_axis_num(inputs) states = tuple(initial_states) outputs = [] time_axis = 1 - nones if nones > 0 else 1 if go_backwards: i = shape[1] - 1 while i >= 0: current = C.ops.slice(inputs, time_axis, i, i + 1) # remove dummy dimension current = squeeze(current, time_axis) output, new_states = step_function( current, tuple(states) + tuple(constants)) if getattr(output, '_uses_learning_phase', False): uses_learning_phase = True if mask is not None: mask_slice = C.ops.slice(mask, time_axis, i, i + 1) mask_slice = squeeze(mask_slice, time_axis) if len(outputs) == 0: prev_output = zeros_like(output) else: prev_output = outputs[-1] output = C.ops.element_select(mask_slice, output, prev_output) return_states = [] for s, n_s in zip(states, new_states): return_states.append( C.ops.element_select( mask_slice, n_s, s)) new_states = return_states outputs.append(output) states = new_states i -= 1 else: i = 0 while i < shape[1]: current = C.ops.slice(inputs, time_axis, i, i + 1) # remove dummy dimension current = squeeze(current, 1) output, new_states = step_function( current, tuple(states) + tuple(constants)) if getattr(output, '_uses_learning_phase', False): uses_learning_phase = True if mask is not None: mask_slice = C.ops.slice(mask, time_axis, i, i + 1) mask_slice = squeeze(mask_slice, 1) if len(outputs) == 0: prev_output = zeros_like(output) else: prev_output = outputs[-1] output = C.ops.element_select(mask_slice, output, prev_output) return_states = [] for s, n_s in zip(states, new_states): return_states.append( C.ops.element_select( mask_slice, n_s, s)) new_states = return_states outputs.append(output) states = new_states[:len(states)] i += 1 i = 1 # add the time_step axis back final_output = expand_dims(outputs[0], 1) last_output = outputs[0] while i < len(outputs): # add the time_step axis back output_slice = expand_dims(outputs[i], 1) final_output = C.splice(final_output, output_slice, axis=time_axis) last_output = outputs[i] i += 1 last_output._uses_learning_phase = uses_learning_phase return last_output, final_output, states def rnn(step_function, inputs, initial_states, go_backwards=False, mask=None, constants=None, unroll=False, input_length=None): shape = int_shape(inputs) dims = len(shape) global uses_learning_phase uses_learning_phase = False if dims < 3: raise ValueError('CNTK Backend: the input of rnn has only rank %d ' 'Need at least rank 3 to run RNN.' % dims) if _get_dynamic_axis_num(inputs) == 0 or unroll: return _static_rnn( step_function, inputs, initial_states, go_backwards, mask, constants, unroll, input_length) if constants is None: constants = [] num_time_step = shape[1] if num_time_step is None and not has_seq_axis(inputs): num_time_step = inputs.shape[0] initial = [] for s in initial_states: if _get_dynamic_axis_num(s) == 0: if hasattr(C, 'to_batch'): initial.append(C.to_batch(s)) else: initial.append(C.user_function(ConvertToBatch(s))) else: initial.append(s) need_convert = not has_seq_axis(inputs) if go_backwards and need_convert is False: raise NotImplementedError('CNTK Backend: `go_backwards` is not supported with ' 'variable-length sequences. Please specify a ' 'static length for your sequences.') rnn_inputs = inputs if need_convert: if go_backwards: rnn_inputs = reverse(rnn_inputs, 1) rnn_inputs = C.to_sequence(rnn_inputs) rnn_constants = [] for constant in constants: if isinstance(constant, list): new_c = [] for c in constant: if _get_dynamic_axis_num(c) == 1: new_c.append(C.sequence.broadcast_as(c, rnn_inputs)) else: new_c.append(c) rnn_constants.append(new_c) else: if _get_dynamic_axis_num(constant) == 1: rnn_constants.append(C.sequence.broadcast_as(constant, rnn_inputs)) else: rnn_constants.append(constant) else: rnn_constants = constants if mask is not None and not has_seq_axis(mask): if go_backwards: mask = reverse(mask, 1) if len(int_shape(mask)) == 2: mask = expand_dims(mask) mask = C.to_sequence_like(mask, rnn_inputs) states = tuple(initial) with C.default_options(axis_offset=1): def _recurrence(x, states, m): # create place holder place_holders = [C.placeholder(dynamic_axes=x.dynamic_axes) for _ in states] past_values = [] for s, p in zip(states, place_holders): past_values.append(C.sequence.past_value(p, s)) new_output, new_states = step_function( x, tuple(past_values) + tuple(rnn_constants)) if getattr(new_output, '_uses_learning_phase', False): global uses_learning_phase uses_learning_phase = True if m is not None: new_states = [C.element_select(m, n, s) for n, s in zip(new_states, past_values)] n_s = [] for o, p in zip(new_states, place_holders): n_s.append(o.replace_placeholders({p: o.output})) if len(n_s) > 0: new_output = n_s[0] return new_output, n_s final_output, final_states = _recurrence(rnn_inputs, states, mask) last_output = C.sequence.last(final_output) last_states = [C.sequence.last(s) for s in final_states] if need_convert: final_output = C.sequence.unpack(final_output, 0, no_mask_output=True) if num_time_step is not None and num_time_step is not C.FreeDimension: final_output = _reshape_sequence(final_output, num_time_step) f_stats = [] for l_s, i_s in zip(last_states, initial_states): if _get_dynamic_axis_num(i_s) == 0 and _get_dynamic_axis_num(l_s) == 1: if hasattr(C, 'unpack_batch'): f_stats.append(C.unpack_batch(l_s)) else: f_stats.append(C.user_function(ConvertToStatic(l_s, batch_size=i_s.shape[0]))) else: f_stats.append(l_s) last_output._uses_learning_phase = uses_learning_phase return last_output, final_output, f_stats def has_seq_axis(x): return hasattr(x, 'dynamic_axes') and len(x.dynamic_axes) > 1 def l2_normalize(x, axis=None): axis = [axis] axis = _normalize_axis(axis, x) norm = C.sqrt(C.reduce_sum(C.square(x), axis=axis[0])) return x / norm def hard_sigmoid(x): x = (0.2 * x) + 0.5 x = C.clip(x, 0.0, 1.0) return x def conv1d(x, kernel, strides=1, padding='valid', data_format=None, dilation_rate=1): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) if padding == 'causal': # causal (dilated) convolution: left_pad = dilation_rate * (kernel.shape[0] - 1) x = temporal_padding(x, (left_pad, 0)) padding = 'valid' if data_format == 'channels_last': x = C.swapaxes(x, 0, 1) kernel = C.swapaxes(kernel, 0, 2) padding = _preprocess_border_mode(padding) strides = [strides] x = C.convolution( kernel, x, strides=tuple(strides), auto_padding=[ False, padding]) if data_format == 'channels_last': x = C.swapaxes(x, 0, 1) return x def conv2d(x, kernel, strides=(1, 1), padding='valid', data_format=None, dilation_rate=(1, 1)): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) x = _preprocess_conv2d_input(x, data_format) kernel = _preprocess_conv2d_kernel(kernel, data_format) padding = _preprocess_border_mode(padding) if dilation_rate == (1, 1): strides = (1,) + strides x = C.convolution( kernel, x, strides, auto_padding=[ False, padding, padding]) else: assert dilation_rate[0] == dilation_rate[1] assert strides == (1, 1), 'Invalid strides for dilated convolution' x = C.convolution( kernel, x, strides=dilation_rate[0], auto_padding=[ False, padding, padding]) return _postprocess_conv2d_output(x, data_format) def separable_conv1d(x, depthwise_kernel, pointwise_kernel, strides=1, padding='valid', data_format=None, dilation_rate=1): raise NotImplementedError def separable_conv2d(x, depthwise_kernel, pointwise_kernel, strides=(1, 1), padding='valid', data_format=None, dilation_rate=(1, 1)): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) x = _preprocess_conv2d_input(x, data_format) depthwise_kernel = _preprocess_conv2d_kernel(depthwise_kernel, data_format) depthwise_kernel = C.reshape(C.transpose(depthwise_kernel, (1, 0, 2, 3)), (-1, 1) + depthwise_kernel.shape[2:]) pointwise_kernel = _preprocess_conv2d_kernel(pointwise_kernel, data_format) padding = _preprocess_border_mode(padding) if dilation_rate == (1, 1): strides = (1,) + strides x = C.convolution(depthwise_kernel, x, strides=strides, auto_padding=[False, padding, padding], groups=x.shape[0]) x = C.convolution(pointwise_kernel, x, strides=(1, 1, 1), auto_padding=[False]) else: if dilation_rate[0] != dilation_rate[1]: raise ValueError('CNTK Backend: non-square dilation_rate is ' 'not supported.') if strides != (1, 1): raise ValueError('Invalid strides for dilated convolution') x = C.convolution(depthwise_kernel, x, strides=dilation_rate[0], auto_padding=[False, padding, padding]) x = C.convolution(pointwise_kernel, x, strides=(1, 1, 1), auto_padding=[False]) return _postprocess_conv2d_output(x, data_format) def depthwise_conv2d(x, depthwise_kernel, strides=(1, 1), padding='valid', data_format=None, dilation_rate=(1, 1)): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) x = _preprocess_conv2d_input(x, data_format) depthwise_kernel = _preprocess_conv2d_kernel(depthwise_kernel, data_format) depthwise_kernel = C.reshape(C.transpose(depthwise_kernel, (1, 0, 2, 3)), (-1, 1) + depthwise_kernel.shape[2:]) padding = _preprocess_border_mode(padding) if dilation_rate == (1, 1): strides = (1,) + strides x = C.convolution(depthwise_kernel, x, strides=strides, auto_padding=[False, padding, padding], groups=x.shape[0]) else: if dilation_rate[0] != dilation_rate[1]: raise ValueError('CNTK Backend: non-square dilation_rate is ' 'not supported.') if strides != (1, 1): raise ValueError('Invalid strides for dilated convolution') x = C.convolution(depthwise_kernel, x, strides=dilation_rate[0], auto_padding=[False, padding, padding], groups=x.shape[0]) return _postprocess_conv2d_output(x, data_format) def conv3d(x, kernel, strides=(1, 1, 1), padding='valid', data_format=None, dilation_rate=(1, 1, 1)): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) x = _preprocess_conv3d_input(x, data_format) kernel = _preprocess_conv3d_kernel(kernel, data_format) padding = _preprocess_border_mode(padding) strides = strides + (strides[0],) x = C.convolution( kernel, x, strides, auto_padding=[ False, padding, padding, padding]) return _postprocess_conv3d_output(x, data_format) def conv3d_transpose(x, kernel, output_shape, strides=(1, 1, 1), padding='valid', data_format=None): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) x = _preprocess_conv3d_input(x, data_format) kernel = _preprocess_conv3d_kernel(kernel, data_format) padding = _preprocess_border_mode(padding) strides = (1,) + strides # cntk output_shape does not include batch axis output_shape = output_shape[1:] # in keras2, need handle output shape in different format if data_format == 'channels_last': shape = list(output_shape) shape[0] = output_shape[3] shape[1] = output_shape[0] shape[2] = output_shape[1] shape[3] = output_shape[2] output_shape = tuple(shape) x = C.convolution_transpose( kernel, x, strides, auto_padding=[ False, padding, padding, padding], output_shape=output_shape) return _postprocess_conv3d_output(x, data_format) def pool2d(x, pool_size, strides=(1, 1), padding='valid', data_format=None, pool_mode='max'): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) padding = _preprocess_border_mode(padding) strides = strides pool_size = pool_size x = _preprocess_conv2d_input(x, data_format) if pool_mode == 'max': x = C.pooling( x, C.MAX_POOLING, pool_size, strides, auto_padding=[padding]) elif pool_mode == 'avg': x = C.pooling( x, C.AVG_POOLING, pool_size, strides, auto_padding=[padding]) else: raise ValueError('Invalid pooling mode: ' + str(pool_mode)) return _postprocess_conv2d_output(x, data_format) def pool3d(x, pool_size, strides=(1, 1, 1), padding='valid', data_format=None, pool_mode='max'): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) padding = _preprocess_border_mode(padding) x = _preprocess_conv3d_input(x, data_format) if pool_mode == 'max': x = C.pooling( x, C.MAX_POOLING, pool_size, strides, auto_padding=[padding]) elif pool_mode == 'avg': x = C.pooling( x, C.AVG_POOLING, pool_size, strides, auto_padding=[padding]) else: raise ValueError('Invalid pooling mode: ' + str(pool_mode)) return _postprocess_conv3d_output(x, data_format) def relu(x, alpha=0., max_value=None): if alpha != 0.: negative_part = C.relu(-x) x = C.relu(x) if max_value is not None: x = C.clip(x, 0.0, max_value) if alpha != 0.: x -= alpha * negative_part return x def dropout(x, level, noise_shape=None, seed=None): if level < 0. or level >= 1: raise ValueError('CNTK Backend: Invalid dropout level %s, ' 'must be in interval [0, 1].' % level) return C.dropout(x, level) def batch_flatten(x): # cntk's batch axis is not in shape, # so just flatten all the dim in x.shape dim = np.prod(x.shape) x = C.reshape(x, (-1,)) x._keras_shape = (None, dim) return x def softmax(x, axis=-1): return C.softmax(x, axis=axis) def softplus(x): return C.softplus(x) def softsign(x): return x / (1 + C.abs(x)) def categorical_crossentropy(target, output, from_logits=False): if from_logits: result = C.cross_entropy_with_softmax(output, target) # cntk's result shape is (batch, 1), while keras expect (batch, ) return C.reshape(result, ()) else: # scale preds so that the class probas of each sample sum to 1 output /= C.reduce_sum(output, axis=-1) # avoid numerical instability with epsilon clipping output = C.clip(output, epsilon(), 1.0 - epsilon()) return -sum(target * C.log(output), axis=-1) def sparse_categorical_crossentropy(target, output, from_logits=False): target = C.one_hot(target, output.shape[-1]) target = C.reshape(target, output.shape) return categorical_crossentropy(target, output, from_logits) class Function(object): def __init__(self, inputs, outputs, updates=[], **kwargs): self.placeholders = inputs self.trainer = None self.unrelated_updates = None self.updates = updates if len(updates) > 0: assert len(outputs) > 0 self.loss = outputs[0] # need group update by gradient place holder u_ops = [] unrelated_updates = [] for update in updates: if isinstance(update, tuple): if len(update) != 2: raise NotImplementedError else: u = C.assign(update[0], update[1]) else: u = update if len(u.arguments) == 0: u_ops.append(u) else: unrelated_updates.append(u) update_func = C.combine([u.output for u in u_ops]) grads = update_func.find_all_with_name('keras_grad_placeholder') u_list = [] p_list = [] for g in grads: if g in grad_parameter_dict: p_list.append(grad_parameter_dict[g]) u_list.append(g) else: raise ValueError( 'CNTK backend: when constructing trainer, ' 'found gradient node `%s` which is not ' 'related to any parameters in the model. ' 'Please double check how the gradient node ' 'is constructed.' % g) if len(u_list) > 0: learner = C.cntk_py.universal_learner(p_list, u_list, update_func) criterion = ( outputs[0], outputs[1]) if len(outputs) > 1 else ( outputs[0], ) self.trainer = C.trainer.Trainer( outputs[0], criterion, [learner]) self.trainer_output = tuple([f.output for f in criterion]) elif len(u_ops) > 0: unrelated_updates.extend(u_ops) if len(unrelated_updates) > 0: self.unrelated_updates = C.combine([_.output for _ in unrelated_updates]) if self.trainer is None: self.metrics_outputs = [f.output for f in outputs] self.metrics_func = C.combine(self.metrics_outputs) # cntk only could handle loss and 1 metric in trainer, for metrics more # than 2, need manual eval elif len(outputs) > 2: self.metrics_outputs = [f.output for f in outputs[2:]] self.metrics_func = C.combine(self.metrics_outputs) else: self.metrics_func = None @staticmethod def _is_input_shape_compatible(input, placeholder): if hasattr(input, 'shape') and hasattr(placeholder, 'shape'): num_dynamic = get_num_dynamic_axis(placeholder) input_shape = input.shape[num_dynamic:] placeholder_shape = placeholder.shape for i, p in zip(input_shape, placeholder_shape): if i != p and p != C.InferredDimension and p != C.FreeDimension: return False return True def __call__(self, inputs): global _LEARNING_PHASE_PLACEHOLDER global _LEARNING_PHASE assert isinstance(inputs, (list, tuple)) feed_dict = {} for tensor, value in zip(self.placeholders, inputs): # cntk only support calculate on float, do auto cast here if (hasattr(value, 'dtype') and value.dtype != np.float32 and value.dtype != np.float64): value = value.astype(np.float32) if tensor == _LEARNING_PHASE_PLACEHOLDER: _LEARNING_PHASE_PLACEHOLDER.value = np.asarray(value) else: # in current version cntk can't support input with variable # length. Will support it in next release. if not self._is_input_shape_compatible(value, tensor): raise ValueError('CNTK backend: The placeholder has been resolved ' 'to shape `%s`, but input shape is `%s`. Currently ' 'CNTK can not take variable length inputs. Please ' 'pass inputs that have a static shape.' % (str(tensor.shape), str(value.shape))) feed_dict[tensor] = value updated = [] if self.trainer is not None: input_dict = {} for argument in self.loss.arguments: if argument in feed_dict: input_dict[argument] = feed_dict[argument] else: raise ValueError( 'CNTK backend: argument %s is not found in inputs. ' 'Please double check the model and inputs in ' '`train_function`.' % argument.name) result = self.trainer.train_minibatch( input_dict, self.trainer_output) assert(len(result) == 2) outputs = result[1] for o in self.trainer_output: updated.append(outputs[o]) if self.metrics_func is not None: input_dict = {} for argument in self.metrics_func.arguments: if argument in feed_dict: input_dict[argument] = feed_dict[argument] else: raise ValueError('CNTK backend: metrics argument %s ' 'is not found in inputs. Please double ' 'check the model and inputs.' % argument.name) # Some ops (like dropout) won't be applied during "eval" in cntk. # They only evaluated in training phase. To make it work, call # "forward" method to let cntk know we want to evaluate them.from # But the assign ops won't be executed under this mode, that's why # we need this check. if (self.unrelated_updates is None and (_LEARNING_PHASE_PLACEHOLDER.value == 1.0 or _LEARNING_PHASE == 1)): _, output_values = self.metrics_func.forward( input_dict, self.metrics_func.outputs, (self.metrics_func.outputs[0],), as_numpy=False) else: output_values = self.metrics_func.eval(input_dict, as_numpy=False) if isinstance(output_values, dict): for o in self.metrics_outputs: value = output_values[o] v = value.asarray() updated.append(v) else: v = output_values.asarray() for o in self.metrics_outputs: updated.append(v) if self.unrelated_updates is not None: input_dict = {} for argument in self.unrelated_updates.arguments: if argument in feed_dict: input_dict[argument] = feed_dict[argument] else: raise ValueError( 'CNTK backend: assign ops argument %s ' 'is not found in inputs. Please double ' 'check the model and inputs.' % argument.name) self.unrelated_updates.eval(input_dict, as_numpy=False) return updated def function(inputs, outputs, updates=[], **kwargs): return Function(inputs, outputs, updates=updates, **kwargs) def temporal_padding(x, padding=(1, 1)): assert len(padding) == 2 num_dynamic_axis = _get_dynamic_axis_num(x) base_shape = x.shape if num_dynamic_axis > 0: assert len(base_shape) == 2 if hasattr(C, 'pad'): x = C.pad(x, pattern=[padding, (0, 0)]) else: x = _padding(x, padding, 0) else: assert len(base_shape) == 3 if hasattr(C, 'pad'): x = C.pad(x, pattern=[(0, 0), padding, (0, 0)]) else: x = _padding(x, padding, 1) return x def _padding(x, pattern, axis): base_shape = x.shape if b_any([dim < 0 for dim in base_shape]): raise ValueError('CNTK Backend: padding input tensor with ' 'shape `%s` contains non-specified dimension, ' 'which is not supported. Please give fixed ' 'dimension to enable padding.' % base_shape) if pattern[0] > 0: prefix_shape = list(base_shape) prefix_shape[axis] = pattern[0] prefix_shape = tuple(prefix_shape) x = C.splice(C.constant(value=0, shape=prefix_shape), x, axis=axis) base_shape = x.shape if pattern[1] > 0: postfix_shape = list(base_shape) postfix_shape[axis] = pattern[1] postfix_shape = tuple(postfix_shape) x = C.splice(x, C.constant(value=0, shape=postfix_shape), axis=axis) return x def spatial_2d_padding(x, padding=((1, 1), (1, 1)), data_format=None): assert len(padding) == 2 assert len(padding[0]) == 2 assert len(padding[1]) == 2 if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) num_dynamic_axis = _get_dynamic_axis_num(x) base_shape = x.shape if data_format == 'channels_first': if num_dynamic_axis > 0: assert len(base_shape) == 3 if hasattr(C, 'pad'): x = C.pad(x, pattern=[[0, 0], list(padding[0]), list(padding[1])]) else: x = _padding(x, padding[0], 1) x = _padding(x, padding[1], 2) else: assert len(base_shape) == 4 if hasattr(C, 'pad'): x = C.pad(x, pattern=[[0, 0], [0, 0], list(padding[0]), list(padding[1])]) else: x = _padding(x, padding[0], 2) x = _padding(x, padding[1], 3) else: if num_dynamic_axis > 0: assert len(base_shape) == 3 if hasattr(C, 'pad'): x = C.pad(x, pattern=[list(padding[0]), list(padding[1]), [0, 0]]) else: x = _padding(x, padding[0], 0) x = _padding(x, padding[1], 1) else: assert len(base_shape) == 4 if hasattr(C, 'pad'): x = C.pad(x, pattern=[[0, 0], list(padding[0]), list(padding[1]), [0, 0]]) else: x = _padding(x, padding[0], 1) x = _padding(x, padding[1], 2) return x def spatial_3d_padding(x, padding=((1, 1), (1, 1), (1, 1)), data_format=None): assert len(padding) == 3 assert len(padding[0]) == 2 assert len(padding[1]) == 2 assert len(padding[2]) == 2 if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) num_dynamic_axis = _get_dynamic_axis_num(x) base_shape = x.shape if data_format == 'channels_first': if num_dynamic_axis > 0: assert len(base_shape) == 4 if hasattr(C, 'pad'): x = C.pad(x, pattern=[[0, 0], list(padding[0]), list(padding[1]), list(padding[2])]) else: x = _padding(x, padding[0], 1) x = _padding(x, padding[1], 2) x = _padding(x, padding[2], 3) else: assert len(base_shape) == 5 if hasattr(C, 'pad'): x = C.pad(x, pattern=[[0, 0], [0, 0], list(padding[0]), list(padding[1]), list(padding[2])]) else: x = _padding(x, padding[0], 2) x = _padding(x, padding[1], 3) x = _padding(x, padding[2], 4) else: if num_dynamic_axis > 0: assert len(base_shape) == 4 if hasattr(C, 'pad'): x = C.pad(x, pattern=[list(padding[0]), list(padding[1]), list(padding[2]), [0, 0]]) else: x = _padding(x, padding[0], 0) x = _padding(x, padding[1], 1) x = _padding(x, padding[2], 2) else: assert len(base_shape) == 5 if hasattr(C, 'pad'): x = C.pad(x, pattern=[[0, 0], list(padding[0]), list(padding[1]), list(padding[2]), [0, 0]]) else: x = _padding(x, padding[0], 1) x = _padding(x, padding[1], 2) x = _padding(x, padding[2], 3) return x def one_hot(indices, num_classes): return C.one_hot(indices, num_classes) def get_value(x): if isinstance( x, C.variables.Parameter) or isinstance( x, C.variables.Constant): return x.value else: return eval(x) def batch_get_value(xs): result = [] for x in xs: if (isinstance(x, C.variables.Parameter) or isinstance(x, C.variables.Constant)): result.append(x.value) else: result.append(eval(x)) return result def set_value(x, value): if (isinstance(x, C.variables.Parameter) or isinstance(x, C.variables.Constant)): if isinstance(value, (float, int)): value = np.full(x.shape, value, dtype=floatx()) x.value = value else: raise NotImplementedError def print_tensor(x, message=''): return C.user_function( LambdaFunc(x, when=lambda x: True, execute=lambda x: print(message))) def batch_set_value(tuples): for t in tuples: x = t[0] value = t[1] if isinstance(value, np.ndarray) is False: value = np.asarray(value) if isinstance(x, C.variables.Parameter): x.value = value else: raise NotImplementedError def stop_gradient(variables): if isinstance(variables, (list, tuple)): return map(C.stop_gradient, variables) else: return C.stop_gradient(variables) def switch(condition, then_expression, else_expression): ndim_cond = ndim(condition) ndim_expr = ndim(then_expression) if ndim_cond > ndim_expr: raise ValueError('Rank of condition should be less' ' than or equal to rank of then and' ' else expressions. ndim(condition)=' + str(ndim_cond) + ', ndim(then_expression)' '=' + str(ndim_expr)) elif ndim_cond < ndim_expr: shape_expr = int_shape(then_expression) ndim_diff = ndim_expr - ndim_cond for i in range(ndim_diff): condition = expand_dims(condition) condition = tile(condition, shape_expr[ndim_cond + i]) return C.element_select(condition, then_expression, else_expression) def elu(x, alpha=1.): res = C.elu(x) if alpha == 1: return res else: return C.element_select(C.greater(x, 0), res, alpha * res) def in_top_k(predictions, targets, k): _targets = C.one_hot(targets, predictions.shape[-1]) result = C.classification_error(predictions, _targets, topN=k) return 1 - C.reshape(result, shape=()) def conv2d_transpose(x, kernel, output_shape, strides=(1, 1), padding='valid', data_format=None): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) x = _preprocess_conv2d_input(x, data_format) kernel = _preprocess_conv2d_kernel(kernel, data_format) padding = _preprocess_border_mode(padding) strides = (1,) + strides # cntk output_shape does not include batch axis output_shape = output_shape[1:] # in keras2, need handle output shape in different format if data_format == 'channels_last': shape = list(output_shape) shape[0] = output_shape[2] shape[1] = output_shape[0] shape[2] = output_shape[1] output_shape = tuple(shape) x = C.convolution_transpose( kernel, x, strides, auto_padding=[ False, padding, padding], output_shape=output_shape) return _postprocess_conv2d_output(x, data_format) def identity(x, name=None): if name is None: name = '%s_alias' % x.name return C.alias(x, name=name) def _preprocess_conv2d_input(x, data_format): if data_format == 'channels_last': # TF uses the last dimension as channel dimension, # instead of the 2nd one. # TH input shape: (samples, input_depth, rows, cols) # TF input shape: (samples, rows, cols, input_depth) x = C.transpose(x, (2, 0, 1)) return x def _preprocess_conv2d_kernel(kernel, data_format): # As of Keras 2.0.0, all kernels are normalized # on the format `(rows, cols, input_depth, depth)`, # independently of `data_format`. # CNTK expects `(depth, input_depth, rows, cols)`. kernel = C.transpose(kernel, (3, 2, 0, 1)) return kernel def _preprocess_border_mode(padding): if padding == 'same': padding = True elif padding == 'valid': padding = False else: raise ValueError('Invalid border mode: ' + str(padding)) return padding def _postprocess_conv2d_output(x, data_format): if data_format == 'channels_last': x = C.transpose(x, (1, 2, 0)) return x def _preprocess_conv3d_input(x, data_format): if data_format == 'channels_last': # TF uses the last dimension as channel dimension, # instead of the 2nd one. # TH input shape: (samples, input_depth, conv_dim1, conv_dim2, conv_dim3) # TF input shape: (samples, conv_dim1, conv_dim2, conv_dim3, # input_depth) x = C.transpose(x, (3, 0, 1, 2)) return x def _preprocess_conv3d_kernel(kernel, dim_ordering): kernel = C.transpose(kernel, (4, 3, 0, 1, 2)) return kernel def _postprocess_conv3d_output(x, dim_ordering): if dim_ordering == 'channels_last': x = C.transpose(x, (1, 2, 3, 0)) return x def _get_dynamic_axis_num(x): if hasattr(x, 'dynamic_axes'): return len(x.dynamic_axes) else: return 0 def _contain_seqence_axis(x): if _get_dynamic_axis_num(x) > 1: return x.dynamic_axes[1] == C.Axis.default_dynamic_axis() else: return False def get_num_dynamic_axis(x): return _get_dynamic_axis_num(x) def _reduce_on_axis(x, axis, reduce_fun_name): if isinstance(axis, list): for a in axis: if isinstance(a, C.Axis) \ and a != C.Axis.default_batch_axis() \ and hasattr(C.sequence, reduce_fun_name): x = getattr(C.sequence, reduce_fun_name)(x, a) else: x = getattr(C, reduce_fun_name)(x, a) else: x = getattr(C, reduce_fun_name)(x, axis) return x def _reshape_sequence(x, time_step): tmp_shape = list(int_shape(x)) tmp_shape[1] = time_step return reshape(x, tmp_shape) def local_conv1d(inputs, kernel, kernel_size, strides, data_format=None): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) stride = strides[0] kernel_shape = int_shape(kernel) output_length, feature_dim, filters = kernel_shape xs = [] for i in range(output_length): slice_length = slice(i * stride, i * stride + kernel_size[0]) xs.append(reshape(inputs[:, slice_length, :], (-1, 1, feature_dim))) x_aggregate = concatenate(xs, axis=1) # transpose kernel to output_filters first, to apply broadcast weight = permute_dimensions(kernel, (2, 0, 1)) # Shape: (batch, filters, output_length, input_length * kernel_size) output = x_aggregate * weight # Shape: (batch, filters, output_length) output = sum(output, axis=3) # Shape: (batch, output_length, filters) return permute_dimensions(output, (0, 2, 1)) def local_conv2d(inputs, kernel, kernel_size, strides, output_shape, data_format=None): if data_format is None: data_format = image_data_format() if data_format not in {'channels_first', 'channels_last'}: raise ValueError('Unknown data_format ' + str(data_format)) stride_row, stride_col = strides output_row, output_col = output_shape kernel_shape = int_shape(kernel) _, feature_dim, filters = kernel_shape xs = [] for i in range(output_row): for j in range(output_col): slice_row = slice(i * stride_row, i * stride_row + kernel_size[0]) slice_col = slice(j * stride_col, j * stride_col + kernel_size[1]) if data_format == 'channels_first': xs.append(reshape(inputs[:, :, slice_row, slice_col], (-1, 1, feature_dim))) else: xs.append(reshape(inputs[:, slice_row, slice_col, :], (-1, 1, feature_dim))) x_aggregate = concatenate(xs, axis=1) # transpose kernel to put filters first weight = permute_dimensions(kernel, (2, 0, 1)) # shape: batch, filters, output_length, input_length * kernel_size output = x_aggregate * weight # shape: batch, filters, output_length output = sum(output, axis=3) # shape: batch, filters, row, col output = reshape(output, (-1, filters, output_row, output_col)) if data_format == 'channels_last': # shape: batch, row, col, filters output = permute_dimensions(output, (0, 2, 3, 1)) return output def reverse(x, axes): if isinstance(axes, int): axes = [axes] cntk_axes = _normalize_axis(axes, x) begin_index = [0 for _ in cntk_axes] end_index = [0 for _ in cntk_axes] strides = [-1 for _ in cntk_axes] return C.slice(x, cntk_axes, begin_index, end_index, strides) def _reshape_batch(x, shape): # there is a bug in cntk 2.1's unpack_batch implementation if hasattr(C, 'unpack_batch') and _get_cntk_version() >= 2.2: const_a = C.unpack_batch(x) const_a = C.reshape(const_a, shape) return C.to_batch(const_a) else: return C.user_function(ReshapeBatch(x, shape[1:])) def _get_cntk_version(): version = C.__version__ if version.endswith('+'): version = version[:-1] # for hot fix, ignore all the . except the first one. if len(version) > 2 and version[1] == '.': version = version[:2] + version[2:].replace('.', '') try: return float(version) except: warnings.warn( 'CNTK backend warning: CNTK version not detected. ' 'Will using CNTK 2.0 GA as default.') return float(2.0) class ReshapeBatch(C.ops.functions.UserFunction): def __init__(self, input, shape, name='reshape_with_batch'): super(ReshapeBatch, self).__init__([input], as_numpy=False, name=name) self.from_shape = input.shape self.target_shape = shape def infer_outputs(self): batch_axis = C.Axis.default_batch_axis() return [ C.output_variable( self.target_shape, self.inputs[0].dtype, [batch_axis])] def forward(self, arguments, device=None, outputs_to_retain=None): num_element = arguments.shape()[0] * np.prod(np.asarray(self.from_shape)) num_static_element = np.prod(np.asarray(self.target_shape)) num_batch = int(num_element / num_static_element) result = arguments.data().as_shape((num_batch,) + self.target_shape) return None, C.cntk_py.Value(result) def backward(self, state, root_gradients): grad_array_view = root_gradients.data() num_element = root_gradients.shape()[0] * np.prod(np.asarray(self.target_shape)) num_static_element = np.prod(np.asarray(self.from_shape)) num_old_batch = int(num_element / num_static_element) return C.cntk_py.Value( grad_array_view.as_shape( (num_old_batch,) + self.from_shape)) class ConvertToBatch(C.ops.functions.UserFunction): """Converts input first axis to CNTK batch axis. We may introduce this operation in CNTK native implementation later. # Arguments inputs: a cntk variable (parameter/constant) name: name of this node """ def __init__(self, input, name='convert_to_batch'): super(ConvertToBatch, self).__init__([input], as_numpy=False, name=name) def infer_outputs(self): batch_axis = C.Axis.default_batch_axis() return [ C.output_variable( self.inputs[0].shape[1:], self.inputs[0].dtype, [batch_axis])] def forward(self, arguments, device=None, outputs_to_retain=None): return None, C.cntk_py.Value(arguments.data()) def backward(self, state, root_gradients): return C.cntk_py.Value(root_gradients.data()) class ConvertToStatic(C.ops.functions.UserFunction): """Converts input first axis to CNTK static axis. We may introduce this operation in CNTK native implementation later. # Arguments inputs: a cntk tensor which has batch axis batch_size: size of batch axis. name: name of this node. """ def __init__(self, input, batch_size, name='convert_to_static'): super(ConvertToStatic, self).__init__([input], as_numpy=False, name=name) self.target_shape = (batch_size,) + input.shape def infer_outputs(self): return [ C.output_variable( self.target_shape, self.inputs[0].dtype, [])] def forward(self, arguments, device=None, outputs_to_retain=None): return None, C.cntk_py.Value(arguments.data()) def backward(self, state, root_gradients): return C.cntk_py.Value(root_gradients.data()) class LambdaFunc(C.ops.functions.UserFunction): def __init__(self, arg, when=lambda arg: True, execute=lambda arg: print(arg), name=''): self.when = when self.execute = execute super(LambdaFunc, self).__init__([arg], name=name) def infer_outputs(self): return [ C.output_variable( self.inputs[0].shape, self.inputs[0].dtype, self.inputs[0].dynamic_axes)] def forward(self, argument, device=None, outputs_to_retain=None): if self.when(argument): self.execute(argument) return None, argument def backward(self, state, root_gradients): return root_gradients
1534
from contextlib import contextmanager from django.db import DatabaseError from ..core.tracing import traced_atomic_transaction @contextmanager def transaction_with_commit_on_errors(): """Perform transaction and raise an error in any occurred.""" error = None with traced_atomic_transaction(): try: yield except DatabaseError: raise except Exception as e: error = e if error: raise error
1561
import requests import redis import json import ast import sys import time import urllib import re import sys from threading import Thread from concurrent.futures import ThreadPoolExecutor import argparse def get_options(): parser = argparse.ArgumentParser() parser.add_argument("-i", "--sentinel-ip", default='127.0.0.1', help="Sentinel IP") parser.add_argument("-p", "--sentinel-port", default="16379", help="Sentinel Port") parser.add_argument("-v", "--redis-password", default=None, help="Redis AUTH Password") parser.add_argument("-n", "--sentinel-cluster-name", default='scality-s3', help="Redis cluster name") parser.add_argument("-b", "--bucketd-addr", default='http://127.0.0.1:9000', help="URL of the bucketd server") return parser.parse_args() def safe_print(content): print("{0}".format(content)) class askRedis(): def __init__(self, ip="127.0.0.1", port="16379", sentinel_cluster_name="scality-s3", password=<PASSWORD>): self._password = password r = redis.Redis(host=ip, port=port, db=0, password=password) self._ip, self._port = r.sentinel_get_master_addr_by_name(sentinel_cluster_name) def read(self, resource, name): r = redis.Redis(host=self._ip, port=self._port, db=0, password=self._password) res = 's3:%s:%s:storageUtilized:counter' % (resource, name) total_size = r.get(res) res = 's3:%s:%s:numberOfObjects:counter' % (resource, name) files = r.get(res) try: return {'files': int(files), "total_size": int(total_size)} except Exception as e: return {'files': 0, "total_size": 0} class S3ListBuckets(): def __init__(self, host='127.0.0.1:9000'): self.bucketd_host = host def run(self): docs = [] url = "%s/default/bucket/users..bucket" % self.bucketd_host session = requests.Session() r = session.get(url, timeout=30) if r.status_code == 200: payload = json.loads(r.text) for keys in payload['Contents']: key = keys["key"] r1 = re.match("(\w+)..\|..(\w+.*)", key) docs.append(r1.groups()) return docs return(self.userid, self.bucket, user, files, total_size) if __name__ == '__main__': options = get_options() redis_conf = dict( ip=options.sentinel_ip, port=options.sentinel_port, sentinel_cluster_name=options.sentinel_cluster_name, password=options.redis_password ) P = S3ListBuckets(options.bucketd_addr) listbuckets = P.run() userids = set([x for x, y in listbuckets]) executor = ThreadPoolExecutor(max_workers=1) for userid, bucket in listbuckets: U = askRedis(**redis_conf) data = U.read('buckets', bucket) content = "Account:%s|Bucket:%s|NumberOFfiles:%s|StorageCapacity:%s " % ( userid, bucket, data["files"], data["total_size"]) executor.submit(safe_print, content) data = U.read('buckets', 'mpuShadowBucket'+bucket) content = "Account:%s|Bucket:%s|NumberOFfiles:%s|StorageCapacity:%s " % ( userid, 'mpuShadowBucket'+bucket, data["files"], data["total_size"]) executor.submit(safe_print, content) executor.submit(safe_print, "") for userid in sorted(userids): U = askRedis(**redis_conf) data = U.read('accounts', userid) content = "Account:%s|NumberOFfiles:%s|StorageCapacity:%s " % ( userid, data["files"], data["total_size"]) executor.submit(safe_print, content)
1569
from pprint import pprint import yaml import netmiko import paramiko def send_cmd_with_prompt(device, command, *, wait_for, confirmation): if type(wait_for) == str: wait_for = [wait_for] if type(confirmation) == str: confirmation = [confirmation] with netmiko.Netmiko(**device) as ssh: ssh.enable() result = ssh.send_command_timing( command, strip_prompt=False, strip_command=False ) for wait, confirm in zip(wait_for, confirmation): if wait in result: result += ssh.send_command_timing( confirm, strip_prompt=False, strip_command=False ) return result if __name__ == "__main__": with open("devices.yaml") as f: devices = yaml.safe_load(f) r1 = devices[0] out = send_cmd_with_prompt( r1, "copy run start", wait_for="Destination filename", confirmation="\n" ) print(out) """ R1#copy run start Destination filename [startup-config]? Building configuration... [OK] R1# """
1588
from dataclasses import dataclass @dataclass class PayloadSender: phone: int name: str @dataclass class PayloadBaseModel: sender: PayloadSender payload_id: str
1660
import filters as f from iota import TransactionHash, Address from iota.commands import FilterCommand, RequestFilter, ResponseFilter from iota.filters import Trytes __all__ = [ 'GetNodeInfoCommand', ] class GetNodeInfoCommand(FilterCommand): """ Executes `getNodeInfo` command. See :py:meth:`iota.api.StrictIota.get_node_info`. """ command = 'getNodeInfo' def get_request_filter(self): return GetNodeInfoRequestFilter() def get_response_filter(self): return GetNodeInfoResponseFilter() class GetNodeInfoRequestFilter(RequestFilter): def __init__(self) -> None: # ``getNodeInfo`` does not accept any parameters. # Using a filter here just to enforce that the request is empty. super(GetNodeInfoRequestFilter, self).__init__({}) class GetNodeInfoResponseFilter(ResponseFilter): def __init__(self) -> None: super(GetNodeInfoResponseFilter, self).__init__({ 'coordinatorAddress': f.ByteString(encoding='ascii') | Trytes(Address), 'latestMilestone': f.ByteString(encoding='ascii') | Trytes(TransactionHash), 'latestSolidSubtangleMilestone': f.ByteString(encoding='ascii') | Trytes(TransactionHash), })
1670
try: from public_config import * except ImportError: pass PORT = 9028 SERVICE_NAME = 'interface'
1684
from PyQt5 import QtWidgets, uic from Factory import Factory from Dialogs.DialogMacros import turn_into_free_point, free_point_checkbox from Fill.ListWidget import fill_listWidget_with_data, set_selected_id_in_listWidget import Constant as c class RegularPolygonDialog(QtWidgets.QDialog): def __init__(self, scene, data): """Construct RegularPolygonDialog.""" super(RegularPolygonDialog, self).__init__() self.ui = uic.loadUi('regularpolygon.ui', self) self.scene = scene self.sides = 3 self.free_point = False self.data = data self.ui.buttonBox.accepted.connect(self.accepted) self.ui.buttonBox.rejected.connect(self.rejected) self.ui.sides_slider.valueChanged.connect(self.hslider_sides_func) self.ui.checkBox.stateChanged.connect(lambda x: free_point_checkbox(self, x)) def hslider_sides_func(self, value): """Be slider callback function to set sides.""" self.sides = value self.ui.sides_spin.setValue(value) def accepted(self): """Create new regular polygon with settings.""" A, B = self.data angle = -(self.sides - 2) * 180 / self.sides polygon = [A, B] for _ in range(self.sides - 2): item = Factory.create_empty_item('point', c.Point.Definition.ROTATION) definition = {'A': A, 'B': B, 'angle': angle} id_ = Factory.next_id(item, definition, self.scene.project_data.items) item.item["id"] = id_ item.item["definition"] = definition if self.free_point: item = turn_into_free_point(item, self.scene) self.scene.project_data.add(item) A = B B = item.item["id"] polygon.append(item.item["id"]) item = Factory.create_empty_item('polygon', None) definition = polygon item.item["id"] = Factory.next_id(item, definition, self.scene.project_data.items) item.item["definition"] = definition self.scene.project_data.add(item) self.scene.project_data.recompute_canvas(*self.scene.init_canvas_dims) current_row_old = self.scene.ui.listWidget.currentRow() fill_listWidget_with_data(self.scene.project_data, self.scene.ui.listWidget, self.scene.current_tab_idx) set_selected_id_in_listWidget(self.scene, current_row_old) self.scene.edit.add_undo_item(self.scene) def rejected(self): """Add no new regular polygon.""" pass
1695
import asyncio import logging import synapse.exc as s_exc import synapse.lib.types as s_types import synapse.lib.module as s_module import synapse.lib.version as s_version logger = logging.getLogger(__name__) class Cpe23Str(s_types.Str): ''' CPE 2.3 Formatted String https://nvlpubs.nist.gov/nistpubs/Legacy/IR/nistir7695.pdf (Section 6.2) cpe:2.3: part : vendor : product : version : update : edition : language : sw_edition : target_sw : target_hw : other * = "any" - = N/A ''' def __init__(self, modl, name, info, opts): opts['lower'] = True s_types.Str.__init__(self, modl, name, info, opts) def _splitCpe23(self, text): part = '' parts = [] genr = iter(text) try: while True: c = next(genr) if c == '\\': c += next(genr) if c == ':': parts.append(part) part = '' continue part += c except StopIteration: parts.append(part) return parts def _normPyStr(self, valu): if not valu.startswith('cpe:2.3:'): mesg = 'CPE 2.3 string is expected to start with "cpe:2.3:"' raise s_exc.BadTypeValu(valu=valu, mesg=mesg) text, info = s_types.Str._normPyStr(self, valu) parts = self._splitCpe23(text) if len(parts) != 13: mesg = f'CPE 2.3 string has {len(parts)} parts, expected 13.' raise s_exc.BadTypeValu(valu=valu, mesg=mesg) subs = { 'part': parts[2], 'vendor': parts[3], 'product': parts[4], 'version': parts[5], 'update': parts[6], 'edition': parts[7], 'language': parts[8], 'sw_edition': parts[9], 'target_sw': parts[10], 'target_hw': parts[11], 'other': parts[12], } return ':'.join(parts), {'subs': subs} class SemVer(s_types.Int): ''' Provides support for parsing a semantic version string into its component parts. This normalizes a version string into an integer to allow version ordering. Prerelease information is disregarded for integer comparison purposes, as we cannot map an arbitrary pre-release version into a integer value Major, minor and patch levels are represented as integers, with a max width of 20 bits. The comparable integer value representing the semver is the bitwise concatenation of the major, minor and patch levels. Prerelease and build information will be parsed out and available as strings if that information is present. ''' def postTypeInit(self): s_types.Int.postTypeInit(self) self.setNormFunc(str, self._normPyStr) self.setNormFunc(int, self._normPyInt) def _normPyStr(self, valu): valu = valu.strip() if not valu: raise s_exc.BadTypeValu(valu=valu, name=self.name, mesg='No text left after stripping whitespace') subs = s_version.parseSemver(valu) if subs is None: raise s_exc.BadTypeValu(valu=valu, name=self.name, mesg='Unable to parse string as a semver.') valu = s_version.packVersion(subs.get('major'), subs.get('minor'), subs.get('patch')) return valu, {'subs': subs} def _normPyInt(self, valu): if valu < 0: raise s_exc.BadTypeValu(valu=valu, name=self.name, mesg='Cannot norm a negative integer as a semver.') if valu > s_version.mask60: raise s_exc.BadTypeValu(valu=valu, name=self.name, mesg='Cannot norm a integer larger than 1152921504606846975 as a semver.') major, minor, patch = s_version.unpackVersion(valu) valu = s_version.packVersion(major, minor, patch) subs = {'major': major, 'minor': minor, 'patch': patch} return valu, {'subs': subs} def repr(self, valu): major, minor, patch = s_version.unpackVersion(valu) valu = s_version.fmtVersion(major, minor, patch) return valu loglevels = ( (10, 'debug'), (20, 'info'), (30, 'notice'), (40, 'warning'), (50, 'err'), (60, 'crit'), (70, 'alert'), (80, 'emerg'), ) class ItModule(s_module.CoreModule): async def initCoreModule(self): self.model.form('it:dev:str').onAdd(self._onFormItDevStr) self.model.form('it:dev:pipe').onAdd(self._onFormMakeDevStr) self.model.form('it:dev:mutex').onAdd(self._onFormMakeDevStr) self.model.form('it:dev:regkey').onAdd(self._onFormMakeDevStr) self.model.prop('it:prod:softver:arch').onSet(self._onPropSoftverArch) self.model.prop('it:prod:softver:vers').onSet(self._onPropSoftverVers) self.model.prop('it:prod:softver:software').onSet(self._onPropSoftverSoft) def bruteVersionStr(self, valu): ''' Brute force the version out of a string. Args: valu (str): String to attempt to get version information for. Notes: This first attempts to parse strings using the it:semver normalization before attempting to extract version parts out of the string. Returns: int, dict: The system normalized version integer and a subs dictionary. ''' try: valu, info = self.core.model.type('it:semver').norm(valu) subs = info.get('subs') return valu, subs except s_exc.BadTypeValu: # Try doing version part extraction by noming through the string subs = s_version.parseVersionParts(valu) if subs is None: raise s_exc.BadTypeValu(valu=valu, name='bruteVersionStr', mesg='Unable to brute force version parts out of the string') if subs: valu = s_version.packVersion(subs.get('major'), subs.get('minor', 0), subs.get('patch', 0)) return valu, subs async def _onFormItDevStr(self, node): await node.set('norm', node.ndef[1]) async def _onFormMakeDevStr(self, node): pprop = node.ndef[1] await node.snap.addNode('it:dev:str', pprop) async def _onPropSoftverSoft(self, node, oldv): # Check to see if name is available and set it if possible prop = node.get('software') if prop: opts = {'vars': {'soft': prop}} nodes = await node.snap.nodes('it:prod:soft=$soft', opts=opts) if nodes: name = nodes[0].get('name') if name: await node.set('software:name', name) async def _onPropSoftverArch(self, node, oldv): # make it:dev:str for arch prop = node.get('arch') if prop: await node.snap.addNode('it:dev:str', prop) async def _onPropSoftverVers(self, node, oldv): # Set vers:norm and make it's normed valu prop = node.get('vers') if not prop: return await node.set('vers:norm', prop) # Make it:dev:str from version str await node.snap.addNode('it:dev:str', prop) # form the semver properly or bruteforce parts try: valu, subs = self.bruteVersionStr(prop) await node.set('semver', valu) for k, v in subs.items(): await node.set(f'semver:{k}', v) except asyncio.CancelledError: # pragma: no cover raise except Exception: logger.exception('Failed to brute force version string [%s]', prop) def getModelDefs(self): modl = { 'ctors': ( ('it:semver', 'synapse.models.infotech.SemVer', {}, { 'doc': 'Semantic Version type.', }), ('it:sec:cpe', 'synapse.models.infotech.Cpe23Str', {}, { 'doc': 'A NIST CPE 2.3 Formatted String', }), ), 'types': ( ('it:hostname', ('str', {'strip': True, 'lower': True}), { 'doc': 'The name of a host or system.', }), ('it:host', ('guid', {}), { 'doc': 'A GUID that represents a host or system.' }), ('it:log:event', ('guid', {}), { 'doc': 'A GUID representing an individual log event.', 'interfaces': ('it:host:activity',), }), ('it:network', ('guid', {}), { 'doc': 'A GUID that represents a logical network.' }), ('it:domain', ('guid', {}), { 'doc': 'A logical boundary of authentication and configuration such as a windows domain.' }), ('it:account', ('guid', {}), { 'doc': 'A GUID that represents an account on a host or network.' }), ('it:group', ('guid', {}), { 'doc': 'A GUID that represents a group on a host or network.' }), ('it:logon', ('guid', {}), { 'doc': 'A GUID that represents an individual logon/logoff event.' }), ('it:hosturl', ('comp', {'fields': (('host', 'it:host'), ('url', 'inet:url'))}), { 'doc': 'A url hosted on or served by a host or system.', }), ('it:sec:cve', ('str', {'lower': True, 'regex': r'(?i)^CVE-[0-9]{4}-[0-9]{4,}$'}), { 'doc': 'A vulnerability as designated by a Common Vulnerabilities and Exposures (CVE) number.', 'ex': 'cve-2012-0158' }), ('it:sec:cwe', ('str', {'regex': r'^CWE-[0-9]{1,8}$'}), { 'doc': 'NIST NVD Common Weaknesses Enumeration Specification', 'ex': 'CWE-120', }), ('it:mitre:attack:status', ('str', {'enums': 'current,deprecated,withdrawn'}), { 'doc': 'A Mitre ATT&CK element status.', 'ex': 'current', }), ('it:mitre:attack:group', ('str', {'regex': r'^G[0-9]{4}$'}), { 'doc': 'A Mitre ATT&CK Group ID.', 'ex': 'G0100', }), ('it:mitre:attack:tactic', ('str', {'regex': r'^TA[0-9]{4}$'}), { 'doc': 'A Mitre ATT&CK Tactic ID.', 'ex': 'TA0040', }), ('it:mitre:attack:technique', ('str', {'regex': r'^T[0-9]{4}(.[0-9]{3})?$'}), { 'doc': 'A Mitre ATT&CK Technique ID.', 'ex': 'T1548', }), ('it:mitre:attack:mitigation', ('str', {'regex': r'^M[0-9]{4}$'}), { 'doc': 'A Mitre ATT&CK Mitigation ID.', 'ex': 'M1036', }), ('it:mitre:attack:software', ('str', {'regex': r'^S[0-9]{4}$'}), { 'doc': 'A Mitre ATT&CK Software ID.', 'ex': 'S0154', }), ('it:dev:str', ('str', {}), { 'doc': 'A developer-selected string.' }), ('it:dev:pipe', ('str', {}), { 'doc': 'A string representing a named pipe.', }), ('it:dev:mutex', ('str', {}), { 'doc': 'A string representing a mutex.', }), ('it:dev:int', ('int', {}), { 'doc': 'A developer selected integer constant.', }), ('it:dev:regkey', ('str', {}), { 'doc': 'A Windows registry key.', 'ex': 'HKEY_LOCAL_MACHINE\\SOFTWARE\\Microsoft\\Windows\\CurrentVersion\\Run', }), ('it:dev:regval', ('guid', {}), { 'doc': 'A Windows registry key/value pair.', }), ('it:prod:soft', ('guid', {}), { 'doc': 'A arbitrary, unversioned software product.', }), ('it:adid', ('str', {'lower': True, 'strip': True}), { 'doc': 'An advertising identification string.'}), ('it:os:windows:sid', ('str', {'regex': r'^S-1-[0-59]-\d{2}-\d{8,10}-\d{8,10}-\d{8,10}-[1-9]\d{3}$'}), { 'doc': 'A Microsoft Windows Security Identifier.', 'ex': 'S-1-5-21-1220945662-1202665555-839525555-5555', }), ('it:os:ios:idfa', ('it:adid', {}), { 'doc': 'An iOS advertising identification string.'}), ('it:os:android:aaid', ('it:adid', {}), { 'doc': 'An android advertising identification string.'}), ('it:os:android:perm', ('str', {}), { 'doc': 'An android permission string.'}), ('it:os:android:intent', ('str', {}), { 'doc': 'An android intent string.'}), ('it:os:android:reqperm', ('comp', {'fields': ( ('app', 'it:prod:soft'), ('perm', 'it:os:android:perm'))}), { 'doc': 'The given software requests the android permission.'}), ('it:os:android:ilisten', ('comp', {'fields': ( ('app', 'it:prod:soft'), ('intent', 'it:os:android:intent'))}), { 'doc': 'The given software listens for an android intent.'}), ('it:os:android:ibroadcast', ('comp', {'fields': ( ('app', 'it:prod:soft'), ('intent', 'it:os:android:intent') )}), { 'doc': 'The given software broadcasts the given Android intent.'}), ('it:prod:softver', ('guid', {}), { 'doc': 'A specific version of a software product.'}), ('it:prod:softfile', ('comp', {'fields': ( ('soft', 'it:prod:softver'), ('file', 'file:bytes'))}), { 'doc': 'A file is distributed by a specific software version.'}), ('it:prod:softlib', ('comp', {'fields': ( ('soft', 'it:prod:softver'), ('lib', 'it:prod:softver'))}), { 'doc': 'A software version contains a library software version.'}), ('it:prod:softos', ('comp', {'fields': ( ('soft', 'it:prod:softver'), ('os', 'it:prod:softver'))}), { 'doc': 'The software version is known to be compatible with the given os software version.'}), ('it:hostsoft', ('comp', {'fields': (('host', 'it:host'), ('softver', 'it:prod:softver'))}), { 'doc': 'A version of a software product which is present on a given host.', }), ('it:av:sig', ('comp', {'fields': (('soft', 'it:prod:soft'), ('name', ('str', {'lower': True})))}), { 'doc': 'A signature name within the namespace of an antivirus engine name.' }), ('it:av:filehit', ('comp', {'fields': (('file', 'file:bytes'), ('sig', 'it:av:sig'))}), { 'doc': 'A file that triggered an alert on a specific antivirus signature.', }), ('it:av:prochit', ('guid', {}), { 'doc': 'An instance of a process triggering an alert on a specific antivirus signature.' }), ('it:auth:passwdhash', ('guid', {}), { 'doc': 'An instance of a password hash.', }), ('it:exec:proc', ('guid', {}), { 'doc': 'A process executing on a host. May be an actual (e.g., endpoint) or virtual (e.g., malware sandbox) host.', }), ('it:exec:thread', ('guid', {}), { 'doc': 'A thread executing in a process.', }), ('it:exec:loadlib', ('guid', {}), { 'doc': 'A library load event in a process.', }), ('it:exec:mmap', ('guid', {}), { 'doc': 'A memory mapped segment located in a process.', }), ('it:cmd', ('str', {'strip': True}), { 'doc': 'A unique command-line string.', 'ex': 'foo.exe --dostuff bar', }), ('it:exec:mutex', ('guid', {}), { 'doc': 'A mutex created by a process at runtime.', }), ('it:exec:pipe', ('guid', {}), { 'doc': 'A named pipe created by a process at runtime.', }), ('it:exec:url', ('guid', {}), { 'doc': 'An instance of a host requesting a URL.', }), ('it:exec:bind', ('guid', {}), { 'doc': 'An instance of a host binding a listening port.', }), ('it:fs:file', ('guid', {}), { 'doc': 'A file on a host.' }), ('it:exec:file:add', ('guid', {}), { 'doc': 'An instance of a host adding a file to a filesystem.', }), ('it:exec:file:del', ('guid', {}), { 'doc': 'An instance of a host deleting a file from a filesystem.', }), ('it:exec:file:read', ('guid', {}), { 'doc': 'An instance of a host reading a file from a filesystem.', }), ('it:exec:file:write', ('guid', {}), { 'doc': 'An instance of a host writing a file to a filesystem.', }), ('it:exec:reg:get', ('guid', {}), { 'doc': 'An instance of a host getting a registry key.', }), ('it:exec:reg:set', ('guid', {}), { 'doc': 'An instance of a host creating or setting a registry key.', }), ('it:exec:reg:del', ('guid', {}), { 'doc': 'An instance of a host deleting a registry key.', }), ('it:app:yara:rule', ('guid', {}), { 'doc': 'A YARA rule unique identifier.', }), ('it:app:yara:match', ('comp', {'fields': (('rule', 'it:app:yara:rule'), ('file', 'file:bytes'))}), { 'doc': 'A YARA rule match to a file.', }), ('it:app:yara:procmatch', ('guid', {}), { 'doc': 'An instance of a YARA rule match to a process.', }), ('it:app:snort:rule', ('guid', {}), { 'doc': 'A snort rule unique identifier.', }), ('it:app:snort:hit', ('guid', {}), { 'doc': 'An instance of a snort rule hit.', }), ('it:reveng:function', ('guid', {}), { 'doc': 'A function inside an executable.', }), ('it:reveng:filefunc', ('comp', {'fields': (('file', 'file:bytes'), ('function', 'it:reveng:function'))}), { 'doc': 'An instance of a function in an executable.', }), ('it:reveng:funcstr', ('comp', {'fields': (('function', 'it:reveng:function'), ('string', 'str'))}), { 'deprecated': True, 'doc': 'A reference to a string inside a function.', }), ('it:reveng:impfunc', ('str', {'lower': 1}), { 'doc': 'A function from an imported library.', }), ), 'interfaces': ( ('it:host:activity', { 'props': ( ('exe', ('file:bytes', {}), { 'doc': 'The executable file which caused the activity.'}), ('proc', ('it:exec:proc', {}), { 'doc': 'The host process which caused the activity.'}), ('thread', ('it:exec:thread', {}), { 'doc': 'The host thread which caused the activity.'}), ('host', ('it:host', {}), { 'doc': 'The host on which the activity occurred.'}), ('time', ('time', {}), { 'doc': 'The time that the activity started.'}), ), }), ), 'forms': ( ('it:hostname', {}, ()), ('it:host', {}, ( ('name', ('it:hostname', {}), { 'doc': 'The name of the host or system.', }), ('desc', ('str', {}), { 'doc': 'A free-form description of the host.', }), ('domain', ('it:domain', {}), { 'doc': 'The authentication domain that the host is a member of.', }), ('ipv4', ('inet:ipv4', {}), { 'doc': 'The last known ipv4 address for the host.' }), ('latlong', ('geo:latlong', {}), { 'doc': 'The last known location for the host.' }), ('place', ('geo:place', {}), { 'doc': 'The place where the host resides.', }), ('loc', ('loc', {}), { 'doc': 'The geo-political location string for the node.', }), ('os', ('it:prod:softver', {}), { 'doc': 'The operating system of the host.' }), ('manu', ('str', {}), { 'doc': 'The manufacturer of the host.', }), ('model', ('str', {}), { 'doc': 'The product model of the host.', }), ('serial', ('str', {}), { 'doc': 'The serial number of the host.', }), ('operator', ('ps:contact', {}), { 'doc': 'The operator of the host.', }), ('org', ('ou:org', {}), { 'doc': 'The org that operates the given host.', }), )), ('it:log:event', {}, ( ('mesg', ('str', {}), { 'doc': 'The log messsage text.', }), ('severity', ('int', {'enums': loglevels}), { 'doc': 'A log level integer that increases with severity.', }), ('data', ('data', {}), { 'doc': 'A raw JSON record of the log event.', }), )), ('it:domain', {}, ( ('name', ('str', {'lower': True, 'strip': True, 'onespace': True}), { 'doc': 'The name of the domain.', }), ('desc', ('str', {}), { 'doc': 'A brief description of the domain.', }), ('org', ('ou:org', {}), { 'doc': 'The org that operates the given domain.', }), )), ('it:network', {}, ( ('name', ('str', {'lower': True, 'strip': True, 'onespace': True}), { 'doc': 'The name of the network.', }), ('desc', ('str', {}), { 'doc': 'A brief description of the network.', }), ('org', ('ou:org', {}), { 'doc': 'The org that owns/operates the network.', }), ('net4', ('inet:net4', {}), { 'doc': 'The optional contiguous IPv4 address range of this network.', }), ('net6', ('inet:net6', {}), { 'doc': 'The optional contiguous IPv6 address range of this network.', }), )), ('it:account', {}, ( ('user', ('inet:user', {}), { 'doc': 'The username associated with the account', }), ('contact', ('ps:contact', {}), { 'doc': 'Additional contact information associated with this account.', }), ('host', ('it:host', {}), { 'doc': 'The host where the account is registered.', }), ('domain', ('it:domain', {}), { 'doc': 'The authentication domain where the account is registered.', }), ('posix:uid', ('int', {}), { 'doc': 'The user ID of the account.', 'ex': '1001', }), ('posix:gid', ('int', {}), { 'doc': 'The primary group ID of the account.', 'ex': '1001', }), ('posix:gecos', ('int', {}), { 'doc': 'The GECOS field for the POSIX account.', }), ('posix:home', ('file:path', {}), { 'doc': "The path to the POSIX account's home directory.", 'ex': '/home/visi', }), ('posix:shell', ('file:path', {}), { 'doc': "The path to the POSIX account's default shell.", 'ex': '/bin/bash', }), ('windows:sid', ('it:os:windows:sid', {}), { 'doc': 'The Microsoft Windows Security Identifier of the account.', }), ('groups', ('array', {'type': 'it:group'}), { 'doc': 'An array of groups that the account is a member of.', }), )), ('it:group', {}, ( ('name', ('str', {'lower': True, 'strip': True, 'onespace': True}), { 'doc': 'The name of the group.', }), ('desc', ('str', {}), { 'doc': 'A brief description of the group.', }), ('host', ('it:host', {}), { 'doc': 'The host where the group is registered.', }), ('domain', ('it:domain', {}), { 'doc': 'The authentication domain where the group is registered.', }), ('groups', ('array', {'type': 'it:group'}), { 'doc': 'Groups that are a member of this group.', }), ('posix:gid', ('int', {}), { 'doc': 'The primary group ID of the account.', 'ex': '1001', }), ('windows:sid', ('it:os:windows:sid', {}), { 'doc': 'The Microsoft Windows Security Identifier of the group.', }), )), ('it:logon', {}, ( ('time', ('time', {}), { 'doc': 'The time the logon occured.', }), ('success', ('bool', {}), { 'doc': 'Set to false to indicate an unsuccessful logon attempt.', }), ('logoff:time', ('time', {}), { 'doc': 'The time the logon session ended.', }), ('host', ('it:host', {}), { 'doc': 'The host that the account logged in to.', }), ('account', ('it:account', {}), { 'doc': 'The account that logged in.', }), ('creds', ('auth:creds', {}), { 'doc': 'The credentials that were used for the logon.', }), ('duration', ('duration', {}), { 'doc': 'The duration of the logon session.', }), ('client:host', ('it:host', {}), { 'doc': 'The host where the logon originated.', }), ('client:ipv4', ('inet:ipv4', {}), { 'doc': 'The IPv4 where the logon originated.', }), ('client:ipv6', ('inet:ipv6', {}), { 'doc': 'The IPv6 where the logon originated.', }), )), ('it:hosturl', {}, ( ('host', ('it:host', {}), { 'ro': True, 'doc': 'Host serving a url.', }), ('url', ('inet:url', {}), { 'ro': True, 'doc': 'URL available on the host.', }), )), ('it:dev:str', {}, ( ('norm', ('str', {'lower': True}), { 'doc': 'Lower case normalized version of the it:dev:str.', }), )), ('it:sec:cve', {}, ( ('desc', ('str', {}), { 'doc': 'A free-form description of the CVE vulnerability.', 'disp': {'hint': 'text'}, }), ('url', ('inet:url', {}), { 'doc': 'A URL linking this CVE to a full description.', }), ('references', ('array', {'type': 'inet:url', 'uniq': True}), { 'doc': 'An array of URLs that document the CVE ID.', }), )), ('it:sec:cpe', {}, ( ('part', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The "part" field from the CPE 2.3 string.'}), ('vendor', ('ou:name', {}), { 'ro': True, 'doc': 'The "vendor" field from the CPE 2.3 string.'}), ('product', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The "product" field from the CPE 2.3 string.'}), ('version', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The "version" field from the CPE 2.3 string.'}), ('update', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The "update" field from the CPE 2.3 string.'}), ('edition', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The "edition" field from the CPE 2.3 string.'}), ('language', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The "language" field from the CPE 2.3 string.'}), ('sw_edition', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The "sw_edition" field from the CPE 2.3 string.'}), ('target_sw', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The "target_sw" field from the CPE 2.3 string.'}), ('target_hw', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The "target_hw" field from the CPE 2.3 string.'}), ('other', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The "other" field from the CPE 2.3 string.'}), )), ('it:sec:cwe', {}, ( ('name', ('str', {}), { 'doc': 'The CWE description field.', 'ex': 'Buffer Copy without Checking Size of Input (Classic Buffer Overflow)', }), ('desc', ('str', {}), { 'doc': 'The CWE description field.', 'disp': {'hint': 'text'}, }), ('url', ('inet:url', {}), { 'doc': 'A URL linking this CWE to a full description.', }), ('parents', ('array', {'type': 'it:sec:cwe', 'uniq': True, 'sorted': True, 'split': ','}), { 'doc': 'An array of ChildOf CWE Relationships.' }), )), ('it:mitre:attack:group', {}, ( ('org', ('ou:org', {}), { 'doc': 'Used to map an ATT&CK group to a synapse ou:org.', }), ('name', ('ou:name', {}), { 'doc': 'The primary name for the ATT&CK group.', }), ('names', ('array', {'type': 'ou:name', 'uniq': True, 'sorted': True}), { 'doc': 'An array of alternate names for the ATT&CK group.', }), ('desc', ('str', {}), { 'doc': 'A description of the ATT&CK group.', 'disp': {'hint': 'text'}, }), ('url', ('inet:url', {}), { 'doc': 'The URL that documents the ATT&CK group.', }), ('tag', ('syn:tag', {}), { 'doc': 'The synapse tag used to annotate nodes included in this ATT&CK group ID.', 'ex': 'cno.mitre.g0100', }), ('references', ('array', {'type': 'inet:url', 'uniq': True}), { 'doc': 'An array of URLs that document the ATT&CK group.', }), ('techniques', ('array', {'type': 'it:mitre:attack:technique', 'uniq': True, 'sorted': True, 'split': ','}), { 'doc': 'An array of ATT&CK technique IDs used by the group.', }), ('software', ('array', {'type': 'it:mitre:attack:software', 'uniq': True, 'sorted': True, 'split': ','}), { 'doc': 'An array of ATT&CK software IDs used by the group.', }), )), ('it:mitre:attack:tactic', {}, ( ('name', ('str', {'strip': True}), { 'doc': 'The primary name for the ATT&CK tactic.', }), ('desc', ('str', {}), { 'doc': 'A description of the ATT&CK tactic.', 'disp': {'hint': 'text'}, }), ('url', ('inet:url', {}), { 'doc': 'The URL that documents the ATT&CK tactic.', }), ('tag', ('syn:tag', {}), { 'doc': 'The synapse tag used to annotate nodes included in this ATT&CK tactic.', 'ex': 'cno.mitre.ta0100', }), ('references', ('array', {'type': 'inet:url', 'uniq': True}), { 'doc': 'An array of URLs that document the ATT&CK tactic.', }), )), ('it:mitre:attack:technique', {}, ( ('name', ('str', {'strip': True}), { 'doc': 'The primary name for the ATT&CK technique.', }), ('status', ('it:mitre:attack:status', {}), { 'doc': 'The status of this ATT&CK technique.', }), ('isnow', ('it:mitre:attack:technique', {}), { 'doc': 'If deprecated, this field may contain the current value for the technique.', }), ('desc', ('str', {'strip': True}), { 'doc': 'A description of the ATT&CK technique.', 'disp': {'hint': 'text'}, }), ('url', ('inet:url', {}), { 'doc': 'The URL that documents the ATT&CK technique.', }), ('tag', ('syn:tag', {}), { 'doc': 'The synapse tag used to annotate nodes included in this ATT&CK technique.', 'ex': 'cno.mitre.t0100', }), ('references', ('array', {'type': 'inet:url', 'uniq': True}), { 'doc': 'An array of URLs that document the ATT&CK technique.', }), ('parent', ('it:mitre:attack:technique', {}), { 'doc': 'The parent ATT&CK technique on this sub-technique.', }), ('tactics', ('array', {'type': 'it:mitre:attack:tactic', 'uniq': True, 'sorted': True, 'split': ','}), { 'doc': 'An array of ATT&CK tactics that include this technique.', }), )), ('it:mitre:attack:software', {}, ( ('software', ('it:prod:soft', {}), { 'doc': 'Used to map an ATT&CK software to a synapse it:prod:soft.', }), ('name', ('str', {'strip': True}), { 'doc': 'The primary name for the ATT&CK software.', }), ('names', ('array', {'type': 'str', 'uniq': True, 'sorted': True}), { 'doc': 'Associated names for the ATT&CK software.', }), ('desc', ('str', {'strip': True}), { 'doc': 'A description of the ATT&CK software.', 'disp': {'hint': 'text'}, }), ('url', ('inet:url', {}), { 'doc': 'The URL that documents the ATT&CK software.', }), ('tag', ('syn:tag', {}), { 'doc': 'The synapse tag used to annotate nodes included in this ATT&CK software.', 'ex': 'cno.mitre.s0100', }), ('references', ('array', {'type': 'inet:url', 'uniq': True}), { 'doc': 'An array of URLs that document the ATT&CK software.', }), ('techniques', ('array', {'type': 'it:mitre:attack:technique', 'uniq': True, 'sorted': True, 'split': ','}), { 'doc': 'An array of techniques used by the software.', }), )), ('it:mitre:attack:mitigation', {}, ( # TODO map to an eventual risk:mitigation ('name', ('str', {'strip': True}), { 'doc': 'The primary name for the ATT&CK mitigation.', }), ('desc', ('str', {'strip': True}), { 'doc': 'A description of the ATT&CK mitigation.', 'disp': {'hint': 'text'}, }), ('url', ('inet:url', {}), { 'doc': 'The URL that documents the ATT&CK mitigation.', }), ('tag', ('syn:tag', {}), { 'doc': 'The synapse tag used to annotate nodes included in this ATT&CK mitigation.', 'ex': 'cno.mitre.m0100', }), ('references', ('array', {'type': 'inet:url', 'uniq': True}), { 'doc': 'An array of URLs that document the ATT&CK mitigation.', }), ('addresses', ('array', {'type': 'it:mitre:attack:technique', 'uniq': True, 'sorted': True, 'split': ','}), { 'doc': 'An array of ATT&CK technique IDs addressed by the mitigation.', }), )), ('it:dev:int', {}, ()), ('it:dev:pipe', {}, ()), ('it:dev:mutex', {}, ()), ('it:dev:regkey', {}, ()), ('it:dev:regval', {}, ( ('key', ('it:dev:regkey', {}), { 'doc': 'The Windows registry key.', }), ('str', ('it:dev:str', {}), { 'doc': 'The value of the registry key, if the value is a string.', }), ('int', ('it:dev:int', {}), { 'doc': 'The value of the registry key, if the value is an integer.', }), ('bytes', ('file:bytes', {}), { 'doc': 'The file representing the value of the registry key, if the value is binary data.', }), )), ('it:prod:soft', {}, ( ('name', ('str', {'lower': True, 'strip': True}), { 'doc': 'Name of the software.', }), ('names', ('array', {'type': 'it:dev:str', 'uniq': True, 'sorted': True}), { 'doc': 'Observed/variant names for this software.', }), ('desc', ('str', {}), { 'doc': 'A description of the software.', 'disp': {'hint': 'text'}, }), ('desc:short', ('str', {'lower': True}), { 'doc': 'A short description of the software.', }), ('cpe', ('it:sec:cpe', {}), { 'doc': 'The NIST CPE 2.3 string specifying this software.', }), ('author', ('ps:contact', {}), { 'doc': 'The contact information of the org or person who authored the software.', }), ('author:org', ('ou:org', {}), { 'deprecated': True, 'doc': 'Organization which authored the software.', }), ('author:acct', ('inet:web:acct', {}), { 'deprecated': True, 'doc': 'Web account of the software author.', }), ('author:email', ('inet:email', {}), { 'deprecated': True, 'doc': 'Email address of the sofware author.', }), ('author:person', ('ps:person', {}), { 'deprecated': True, 'doc': 'Person who authored the software.', }), ('url', ('inet:url', {}), { 'doc': 'URL relevant for the software.', }), ('isos', ('bool', {}), { 'doc': 'Set to True if the software is an operating system.'}), ('islib', ('bool', {}), { 'doc': 'Set to True if the software is a library.'}), )), ('it:adid', {}, ()), ('it:os:ios:idfa', {}, ()), ('it:os:android:aaid', {}, ()), ('it:os:android:perm', {}, ()), ('it:os:android:intent', {}, ()), ('it:os:android:reqperm', {}, ( ('app', ('it:prod:softver', {}), {'ro': True, 'doc': 'The android app which requests the permission.'}), ('perm', ('it:os:android:perm', {}), {'ro': True, 'doc': 'The android permission requested by the app.'}), )), ('it:prod:softos', {}, ( ('soft', ('it:prod:softver', {}), {'ro': True, 'doc': 'The software which can run on the operating system.'}), ('os', ('it:prod:softver', {}), {'ro': True, 'doc': 'The operating system which the software can run on.'}), )), ('it:os:android:ilisten', {}, ( ('app', ('it:prod:softver', {}), {'ro': True, 'doc': 'The app software which listens for the android intent.'}), ('intent', ('it:os:android:intent', {}), {'ro': True, 'doc': 'The android intent which is listened for by the app.'}), )), ('it:os:android:ibroadcast', {}, ( ('app', ('it:prod:softver', {}), {'ro': True, 'doc': 'The app software which broadcasts the android intent.'}), ('intent', ('it:os:android:intent', {}), {'ro': True, 'doc': 'The android intent which is broadcast by the app.'}), )), ('it:prod:softver', {}, ( ('software', ('it:prod:soft', {}), { 'doc': 'Software associated with this version instance.', }), ('software:name', ('str', {'lower': True, 'strip': True}), { 'doc': 'The name of the software at a particular version.', }), ('names', ('array', {'type': 'it:dev:str', 'uniq': True, 'sorted': True}), { 'doc': 'Observed/variant names for this software version.', }), ('cpe', ('it:sec:cpe', {}), { 'doc': 'The NIST CPE 2.3 string specifying this software version', }), ('cves', ('array', {'type': 'it:sec:cve', 'uniq': True, 'sorted': True}), { 'doc': 'A list of CVEs that apply to this software version.', }), ('vers', ('it:dev:str', {}), { 'doc': 'Version string associated with this version instance.', }), ('vers:norm', ('str', {'lower': True}), { 'doc': 'Normalized version of the version string.', }), ('arch', ('it:dev:str', {}), { 'doc': 'Software architecture.', }), ('released', ('time', {}), { 'doc': 'Timestamp for when this version of the software was released.', }), ('semver', ('it:semver', {}), { 'doc': 'System normalized semantic version number.', }), ('semver:major', ('int', {}), { 'doc': 'Version major number.', }), ('semver:minor', ('int', {}), { 'doc': 'Version minor number.', }), ('semver:patch', ('int', {}), { 'doc': 'Version patch number.', }), ('semver:pre', ('str', {}), { 'doc': 'Semver prerelease string.', }), ('semver:build', ('str', {}), { 'doc': 'Semver build string.', }), ('url', ('inet:url', {}), { 'doc': 'URL where a specific version of the software is available from.', }), )), ('it:prod:softlib', {}, ( ('soft', ('it:prod:softver', {}), {'ro': True, 'doc': 'The software version that contains the library.'}), ('lib', ('it:prod:softver', {}), {'ro': True, 'doc': 'The library software version.'}), )), ('it:prod:softfile', {}, ( ('soft', ('it:prod:softver', {}), {'ro': True, 'doc': 'The software which distributes the file.'}), ('file', ('file:bytes', {}), {'ro': True, 'doc': 'The file distributed by the software.'}), ('path', ('file:path', {}), { 'doc': 'The default installation path of the file.'}), )), ('it:hostsoft', {}, ( ('host', ('it:host', {}), {'ro': True, 'doc': 'Host with the software.'}), ('softver', ('it:prod:softver', {}), {'ro': True, 'doc': 'Software on the host.'}) )), ('it:av:sig', {}, ( ('soft', ('it:prod:soft', {}), { 'ro': True, 'doc': 'The anti-virus product which contains the signature.', }), ('name', ('str', {'lower': True}), { 'ro': True, 'doc': 'The signature name.' }), ('desc', ('str', {}), { 'doc': 'A free-form description of the signature.', 'disp': {'hint': 'text'}, }), ('url', ('inet:url', {}), { 'doc': 'A reference URL for information about the signature.', }) )), ('it:av:filehit', {}, ( ('file', ('file:bytes', {}), { 'ro': True, 'doc': 'The file that triggered the signature hit.', }), ('sig', ('it:av:sig', {}), { 'ro': True, 'doc': 'The signature that the file triggered on.' }), ('sig:name', ('str', {'lower': True}), { 'ro': True, 'doc': 'The signature name.', }), ('sig:soft', ('it:prod:soft', {}), { 'ro': True, 'doc': 'The anti-virus product which contains the signature.', }), )), ('it:av:prochit', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The file that triggered the signature hit.', }), ('sig', ('it:av:sig', {}), { 'doc': 'The signature that the file triggered on.' }), ('time', ('time', {}), { 'doc': 'The time that the AV engine detected the signature.' }), )), ('it:auth:passwdhash', {}, ( ('salt', ('hex', {}), { 'doc': 'The (optional) hex encoded salt value used to calculate the password hash.', }), ('hash:md5', ('hash:md5', {}), { 'doc': 'The MD5 password hash value.', }), ('hash:sha1', ('hash:sha1', {}), { 'doc': 'The SHA1 password hash value.', }), ('hash:sha256', ('hash:sha256', {}), { 'doc': 'The SHA256 password hash value.', }), ('hash:sha512', ('hash:sha512', {}), { 'doc': 'The SHA512 password hash value.', }), ('hash:lm', ('hash:lm', {}), { 'doc': 'The LM password hash value.', }), ('hash:ntlm', ('hash:ntlm', {}), { 'doc': 'The NTLM password hash value.', }), ('passwd', ('inet:passwd', {}), { 'doc': 'The (optional) clear text password for this password hash.', }), )), ('it:cmd', {}, ()), ('it:exec:proc', {}, ( ('host', ('it:host', {}), { 'doc': 'The host that executed the process. May be an actual or a virtual / notional host.', }), ('exe', ('file:bytes', {}), { 'doc': 'The file considered the "main" executable for the process. For example, rundll32.exe may be considered the "main" executable for DLLs loaded by that program.', }), ('cmd', ('it:cmd', {}), { 'doc': 'The command string used to launch the process, including any command line parameters.', 'disp': {'hint': 'text'}, }), ('pid', ('int', {}), { 'doc': 'The process ID.', }), ('time', ('time', {}), { 'doc': 'The start time for the process.', }), ('exited', ('time', {}), { 'doc': 'The time the process exited.', }), ('exitcode', ('int', {}), { 'doc': 'The exit code for the process.', }), ('user', ('inet:user', {}), { 'doc': 'The user name of the process owner.', }), ('path', ('file:path', {}), { 'doc': 'The path to the executable of the process.', }), ('src:exe', ('file:path', {}), { 'doc': 'The path to the executable which started the process.', }), ('src:proc', ('it:exec:proc', {}), { 'doc': 'The process which created the process.' }), ('killedby', ('it:exec:proc', {}), { 'doc': 'The process which killed this process.', }), )), ('it:exec:thread', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The process which contains the thread.', }), ('created', ('time', {}), { 'doc': 'The time the thread was created.', }), ('exited', ('time', {}), { 'doc': 'The time the thread exited.', }), ('exitcode', ('int', {}), { 'doc': 'The exit code or return value for the thread.', }), ('src:proc', ('it:exec:proc', {}), { 'doc': 'An external process which created the thread.', }), ('src:thread', ('it:exec:thread', {}), { 'doc': 'The thread which created this thread.', }), )), ('it:exec:loadlib', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The process where the library was loaded.', }), ('va', ('int', {}), { 'doc': 'The base memory address where the library was loaded in the process.', }), ('loaded', ('time', {}), { 'doc': 'The time the library was loaded.', }), ('unloaded', ('time', {}), { 'doc': 'The time the library was unloaded.', }), ('path', ('file:path', {}), { 'doc': 'The path that the library was loaded from.', }), ('file', ('file:bytes', {}), { 'doc': 'The library file that was loaded.', }), )), ('it:exec:mmap', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The process where the memory was mapped.', }), ('va', ('int', {}), { 'doc': 'The base memory address where the map was created in the process.', }), ('size', ('int', {}), { 'doc': 'The size of the memory map in bytes.', }), ('perms:read', ('bool', {}), { 'doc': 'True if the mmap is mapped with read permissions.', }), ('perms:write', ('bool', {}), { 'doc': 'True if the mmap is mapped with write permissions.', }), ('perms:execute', ('bool', {}), { 'doc': 'True if the mmap is mapped with execute permissions.', }), ('created', ('time', {}), { 'doc': 'The time the memory map was created.', }), ('deleted', ('time', {}), { 'doc': 'The time the memory map was deleted.', }), ('path', ('file:path', {}), { 'doc': 'The file path if the mmap is a mapped view of a file.', }), ('hash:sha256', ('hash:sha256', {}), { 'doc': 'A SHA256 hash of the memory map. Bytes may optionally be present in the axon.', }), )), ('it:exec:mutex', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that created the mutex.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that created the mutex. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that created the mutex. May or may not be the same :exe specified in :proc, if present.', }), ('time', ('time', {}), { 'doc': 'The time the mutex was created.', }), ('name', ('it:dev:mutex', {}), { 'doc': 'The mutex string.', }), )), ('it:exec:pipe', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that created the named pipe.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that created the named pipe. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that created the named pipe. May or may not be the same :exe specified in :proc, if present.', }), ('time', ('time', {}), { 'doc': 'The time the named pipe was created.', }), ('name', ('it:dev:pipe', {}), { 'doc': 'The named pipe string.', }), )), ('it:exec:url', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that requested the URL.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that requested the URL. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that requested the URL. May or may not be the same :exe specified in :proc, if present.', }), ('time', ('time', {}), { 'doc': 'The time the URL was requested.', }), ('url', ('inet:url', {}), { 'doc': 'The URL that was requested.', }), ('client', ('inet:client', {}), { 'doc': 'The address of the client during the URL retrieval.' }), ('client:ipv4', ('inet:ipv4', {}), { 'doc': 'The IPv4 of the client during the URL retrieval..' }), ('client:ipv6', ('inet:ipv6', {}), { 'doc': 'The IPv6 of the client during the URL retrieval..' }), ('client:port', ('inet:port', {}), { 'doc': 'The client port during the URL retrieval..' }), )), ('it:exec:bind', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that bound the listening port.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that bound the listening port. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that bound the listening port. May or may not be the same :exe specified in :proc, if present.', }), ('time', ('time', {}), { 'doc': 'The time the port was bound.', }), ('server', ('inet:server', {}), { 'doc': 'The inet:addr of the server when binding the port.' }), ('server:ipv4', ('inet:ipv4', {}), { 'doc': 'The IPv4 address specified to bind().' }), ('server:ipv6', ('inet:ipv6', {}), { 'doc': 'The IPv6 address specified to bind().' }), ('server:port', ('inet:port', {}), { 'doc': 'The bound (listening) TCP port.' }), )), ('it:fs:file', {}, ( ('host', ('it:host', {}), { 'doc': 'The host containing the file.', }), ('path', ('file:path', {}), { 'doc': 'The path for the file.', }), ('path:dir', ('file:path', {}), { 'ro': True, 'doc': 'The parent directory of the file path (parsed from :path).', }), ('path:ext', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The file extension of the file name (parsed from :path).', }), ('path:base', ('file:base', {}), { 'ro': True, 'doc': 'The final component of the file path (parsed from :path).', }), ('file', ('file:bytes', {}), { 'doc': 'The file on the host.', }), ('ctime', ('time', {}), { 'doc': 'The file creation time.', }), ('mtime', ('time', {}), { 'doc': 'The file modification time.', }), ('atime', ('time', {}), { 'doc': 'The file access time.', }), ('user', ('inet:user', {}), { 'doc': 'The owner of the file.', }), ('group', ('inet:user', {}), { 'doc': 'The group owner of the file.', }), )), ('it:exec:file:add', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that created the new file.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that created the new file. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that created the new file. May or may not be the same :exe specified in :proc, if present.'}), ('time', ('time', {}), { 'doc': 'The time the file was created.', }), ('path', ('file:path', {}), { 'doc': 'The path where the file was created.', }), ('path:dir', ('file:path', {}), { 'ro': True, 'doc': 'The parent directory of the file path (parsed from :path).', }), ('path:ext', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The file extension of the file name (parsed from :path).', }), ('path:base', ('file:base', {}), { 'ro': True, 'doc': 'The final component of the file path (parsed from :path).', }), ('file', ('file:bytes', {}), { 'doc': 'The file that was created.', }), )), ('it:exec:file:del', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that deleted the file.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that deleted the file. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that deleted the file. May or may not be the same :exe specified in :proc, if present.'}), ('time', ('time', {}), { 'doc': 'The time the file was deleted.', }), ('path', ('file:path', {}), { 'doc': 'The path where the file was deleted.', }), ('path:dir', ('file:path', {}), { 'ro': True, 'doc': 'The parent directory of the file path (parsed from :path).', }), ('path:ext', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The file extension of the file name (parsed from :path).', }), ('path:base', ('file:base', {}), { 'ro': True, 'doc': 'The final component of the file path (parsed from :path).', }), ('file', ('file:bytes', {}), { 'doc': 'The file that was deleted.', }), )), ('it:exec:file:read', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that read the file.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that read the file. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that read the file. May or may not be the same :exe specified in :proc, if present.'}), ('time', ('time', {}), { 'doc': 'The time the file was read.', }), ('path', ('file:path', {}), { 'doc': 'The path where the file was read.', }), ('path:dir', ('file:path', {}), { 'ro': True, 'doc': 'The parent directory of the file path (parsed from :path).', }), ('path:ext', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The file extension of the file name (parsed from :path).', }), ('path:base', ('file:base', {}), { 'ro': True, 'doc': 'The final component of the file path (parsed from :path).', }), ('file', ('file:bytes', {}), { 'doc': 'The file that was read.', }), )), ('it:exec:file:write', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that wrote to / modified the existing file.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that wrote to the file. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that wrote to the file. May or may not be the same :exe specified in :proc, if present.'}), ('time', ('time', {}), { 'doc': 'The time the file was written to/modified.', }), ('path', ('file:path', {}), { 'doc': 'The path where the file was written to/modified.', }), ('path:dir', ('file:path', {}), { 'ro': True, 'doc': 'The parent directory of the file path (parsed from :path).', }), ('path:ext', ('str', {'lower': True, 'strip': True}), { 'ro': True, 'doc': 'The file extension of the file name (parsed from :path).', }), ('path:base', ('file:base', {}), { 'ro': True, 'doc': 'The final component of the file path (parsed from :path).', }), ('file', ('file:bytes', {}), { 'doc': 'The file that was modified.', }), )), ('it:exec:reg:get', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that read the registry.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that read the registry. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that read the registry. May or may not be the same :exe referenced in :proc, if present.', }), ('time', ('time', {}), { 'doc': 'The time the registry was read.', }), ('reg', ('it:dev:regval', {}), { 'doc': 'The registry key or value that was read.', }), )), ('it:exec:reg:set', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that wrote to the registry.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that wrote to the registry. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that wrote to the registry. May or may not be the same :exe referenced in :proc, if present.', }), ('time', ('time', {}), { 'doc': 'The time the registry was written to.', }), ('reg', ('it:dev:regval', {}), { 'doc': 'The registry key or value that was written to.', }), )), ('it:exec:reg:del', {}, ( ('proc', ('it:exec:proc', {}), { 'doc': 'The main process executing code that deleted data from the registry.', }), ('host', ('it:host', {}), { 'doc': 'The host running the process that deleted data from the registry. Typically the same host referenced in :proc, if present.', }), ('exe', ('file:bytes', {}), { 'doc': 'The specific file containing code that deleted data from the registry. May or may not be the same :exe referenced in :proc, if present.', }), ('time', ('time', {}), { 'doc': 'The time the data from the registry was deleted.', }), ('reg', ('it:dev:regval', {}), { 'doc': 'The registry key or value that was deleted.', }), )), ('it:app:snort:rule', {}, ( ('text', ('str', {}), { 'doc': 'The snort rule text.', 'disp': {'hint': 'text'}, }), ('name', ('str', {}), { 'doc': 'The name of the snort rule.'}), ('version', ('it:semver', {}), { 'doc': 'The current version of the rule.'}), )), ('it:app:snort:hit', {}, ( ('rule', ('it:app:snort:rule', {}), { 'doc': 'The snort rule that matched the file.'}), ('flow', ('inet:flow', {}), { 'doc': 'The inet:flow that matched the snort rule.'}), ('src', ('inet:addr', {}), { 'doc': 'The source address of flow that caused the hit.'}), ('src:ipv4', ('inet:ipv4', {}), { 'doc': 'The source IPv4 address of the flow that caused the hit.'}), ('src:ipv6', ('inet:ipv6', {}), { 'doc': 'The source IPv6 address of the flow that caused the hit.'}), ('src:port', ('inet:port', {}), { 'doc': 'The source port of the flow that caused the hit.'}), ('dst', ('inet:addr', {}), { 'doc': 'The destination address of the trigger.'}), ('dst:ipv4', ('inet:ipv4', {}), { 'doc': 'The destination IPv4 address of the flow that caused the hit.'}), ('dst:ipv6', ('inet:ipv6', {}), { 'doc': 'The destination IPv4 address of the flow that caused the hit.'}), ('dst:port', ('inet:port', {}), { 'doc': 'The destination port of the flow that caused the hit.'}), ('time', ('time', {}), { 'doc': 'The time of the network flow that caused the hit.'}), ('sensor', ('it:host', {}), { 'doc': 'The sensor host node that produced the hit.'}), ('version', ('it:semver', {}), { 'doc': 'The version of the rule at the time of match.'}), )), ('it:app:yara:rule', {}, ( ('text', ('str', {}), { 'doc': 'The YARA rule text.', 'disp': {'hint': 'text'}, }), ('name', ('str', {}), { 'doc': 'The name of the YARA rule.'}), ('author', ('ps:contact', {}), { 'doc': 'Contact info for the author of the YARA rule.'}), ('version', ('it:semver', {}), { 'doc': 'The current version of the rule.'}), ('enabled', ('bool', {}), { 'doc': 'The rule enabled status to be used for YARA evaluation engines.'}), )), ('it:app:yara:match', {}, ( ('rule', ('it:app:yara:rule', {}), { 'ro': True, 'doc': 'The YARA rule that matched the file.'}), ('file', ('file:bytes', {}), { 'ro': True, 'doc': 'The file that matched the YARA rule.'}), ('version', ('it:semver', {}), { 'doc': 'The most recent version of the rule evaluated as a match.'}), )), ('it:app:yara:procmatch', {}, ( ('rule', ('it:app:yara:rule', {}), { 'doc': 'The YARA rule that matched the file.'}), ('proc', ('it:exec:proc', {}), { 'doc': 'The process that matched the YARA rule.'}), ('time', ('time', {}), { 'doc': 'The time that the YARA engine matched the process to the rule.'}), ('version', ('it:semver', {}), { 'doc': 'The most recent version of the rule evaluated as a match.'}), )), ('it:reveng:function', {}, ( ('name', ('str', {}), { 'doc': 'The name of the function.'}), ('description', ('str', {}), { 'doc': 'Notes concerning the function.'}), ('impcalls', ('array', {'type': 'it:reveng:impfunc'}), { 'doc': 'Calls to imported library functions within the scope of the function.', }), ('strings', ('array', {'type': 'it:dev:str', 'uniq': True}), { 'doc': 'An array of strings referenced within the function.', }), )), ('it:reveng:filefunc', {}, ( ('function', ('it:reveng:function', {}), { 'ro': True, 'doc': 'The guid matching the function.'}), ('file', ('file:bytes', {}), { 'ro': True, 'doc': 'The file that contains the function.'}), ('va', ('int', {}), { 'doc': 'The virtual address of the first codeblock of the function.'}), ('rank', ('int', {}), { 'doc': 'The function rank score used to evaluate if it exhibits interesting behavior.'}), ('complexity', ('int', {}), { 'doc': 'The complexity of the function.'}), ('funccalls', ('array', {'type': 'it:reveng:filefunc'}), { 'doc': 'Other function calls within the scope of the function.', }), )), ('it:reveng:funcstr', {}, ( ('function', ('it:reveng:function', {}), { 'ro': True, 'doc': 'The guid matching the function.'}), ('string', ('str', {}), { 'ro': True, 'doc': 'The string that the function references.'}), )), ('it:reveng:impfunc', {}, ()), ), } name = 'it' return ((name, modl), )
1718
import mxnet as mx def slice_symbol_to_seq_symobls(net, seq_len, axis=1, squeeze_axis=True): net = mx.sym.SliceChannel(data=net, num_outputs=seq_len, axis=axis, squeeze_axis=squeeze_axis) hidden_all = [] for seq_index in range(seq_len): hidden_all.append(net[seq_index]) net = hidden_all return net
1776
import logging from ariadne import MutationType, convert_kwargs_to_snake_case from config import clients, messages, queue mutation = MutationType() @mutation.field("createMessage") @convert_kwargs_to_snake_case async def resolve_create_message(obj, info, content, client_id): try: message = {"content": content, "client_id": client_id} messages.append(message) await queue.put(message) return {"success": True, "message": message} except Exception as error: return {"success": False, "errors": [str(error)]} @mutation.field("createClient") @convert_kwargs_to_snake_case async def resolve_create_client(obj, info, client_id): try: logging.info(f"Client id: {client_id}") if not clients.get(client_id): client = {"client_id": client_id} clients[client_id] = client return {"success": True, "client": client} return {"success": False, "errors": ["Client is taken"]} except Exception as error: return {"success": False, "errors": [str(error)]}
1779
from typing import Dict, List, Any from ..df.types import Definition from ..df.base import op from ..util.data import traverse_get MAPPING = Definition(name="mapping", primitive="map") MAPPING_TRAVERSE = Definition(name="mapping_traverse", primitive="List[str]") MAPPING_KEY = Definition(name="key", primitive="str") MAPPING_VALUE = Definition(name="value", primitive="generic") @op( name="dffml.mapping.extract", inputs={"mapping": MAPPING, "traverse": MAPPING_TRAVERSE}, outputs={"value": MAPPING_VALUE}, ) def mapping_extract_value(mapping: Dict[str, Any], traverse: List[str]): """ Extracts value from a given mapping. Parameters ---------- mapping : dict The mapping to extract the value from. traverse : list[str] A list of keys to traverse through the mapping dictionary and extract the values. Returns ------- dict A dictionary containing the value of the keys. Examples -------- >>> import asyncio >>> from dffml import * >>> >>> dataflow = DataFlow.auto(mapping_extract_value, GetSingle) >>> >>> dataflow.seed.append( ... Input( ... value=[mapping_extract_value.op.outputs["value"].name], ... definition=GetSingle.op.inputs["spec"], ... ) ... ) >>> inputs = [ ... Input( ... value={"key1": {"key2": 42}}, ... definition=mapping_extract_value.op.inputs["mapping"], ... ), ... Input( ... value=["key1", "key2"], ... definition=mapping_extract_value.op.inputs["traverse"], ... ), ... ] >>> >>> async def main(): ... async for ctx, result in MemoryOrchestrator.run(dataflow, inputs): ... print(result) >>> >>> asyncio.run(main()) {'value': 42} """ return {"value": traverse_get(mapping, *traverse)} @op( name="dffml.mapping.create", inputs={"key": MAPPING_KEY, "value": MAPPING_VALUE}, outputs={"mapping": MAPPING}, ) def create_mapping(key: str, value: Any): """ Creates a mapping of a given key and value. Parameters ---------- key : str The key for the mapping. value : Any The value for the mapping. Returns ------- dict A dictionary containing the mapping created. Examples -------- >>> import asyncio >>> from dffml import * >>> >>> dataflow = DataFlow.auto(create_mapping, GetSingle) >>> dataflow.seed.append( ... Input( ... value=[create_mapping.op.outputs["mapping"].name], ... definition=GetSingle.op.inputs["spec"], ... ) ... ) >>> inputs = [ ... Input( ... value="key1", definition=create_mapping.op.inputs["key"], ... ), ... Input( ... value=42, definition=create_mapping.op.inputs["value"], ... ), ... ] >>> >>> async def main(): ... async for ctx, result in MemoryOrchestrator.run(dataflow, inputs): ... print(result) >>> >>> asyncio.run(main()) {'mapping': {'key1': 42}} """ return {"mapping": {key: value}}
1789
import unittest import base class Test(base.BaseScriptTest, unittest.TestCase): command_line = "./scripts/maf_extract_ranges_indexed.py ./test_data/maf_tests/mm8_chr7_tiny.maf -c -m 5 -p mm8." input_stdin = base.TestFile(filename="./test_data/maf_tests/dcking_ghp074.bed") output_stdout = base.TestFile(filename="./test_data/maf_tests/dcking_ghp074.maf")
1793
import svhn2mnist import usps import syn2gtrsb import syndig2svhn def Generator(source, target, pixelda=False): if source == 'usps' or target == 'usps': return usps.Feature() elif source == 'svhn': return svhn2mnist.Feature() elif source == 'synth': return syn2gtrsb.Feature() def Classifier(source, target): if source == 'usps' or target == 'usps': return usps.Predictor() if source == 'svhn': return svhn2mnist.Predictor() if source == 'synth': return syn2gtrsb.Predictor()
1800
from datetime import datetime from kubernetes import client from kubernetes.client.rest import ApiException import os import time import yaml from tests import config as conf import tests.utils as ut def remove_clusterrole_binding(shipper_name, crb_name): # remove clusterrolebind k8s_client = client.RbacAuthorizationV1Api() try: k8s_client.delete_cluster_role_binding(crb_name) print(f"\nsuccessfully deleted: {crb_name}") except Exception as e: print(f"\n{shipper_name} cluster role binding deletion has failed, please manually delete {crb_name}:") print(f"kubectl delete clusterrolebinding {crb_name}") def filebeat_teardown(namespace): # remove clusterrolebind # TODO: find a solution for sharing the name both here and in the kube object crb_name = f"filebeat-cluster-role-binding-{namespace}" remove_clusterrole_binding("filebeat", crb_name) def fluent_bit_teardown(namespace): # remove clusterrolebind # TODO: find a solution for sharing the name both here and in the kube object crb_name = f"fluent-bit-clusterrole-binding-{namespace}" remove_clusterrole_binding("fluent-bit", crb_name) def add_elastic_cluster(namespace): print("\nDeploying ElasticSearch\n") add_deployment_dir(namespace, conf.ELASTIC_CONF_DIR) def add_filebeat_cluster(namespace): print("\nDeploying FileBeat\n") add_deployment_dir(namespace, conf.FILEBEAT_CONF_DIR) def add_fluent_bit_cluster(namespace): print("\nDeploying Fluent-bit\n") add_deployment_dir(namespace, conf.FLUENT_BIT_CONF_DIR) def add_kibana_cluster(namespace): print("\nDeploying Kibana\n") add_deployment_dir(namespace, conf.KIBANA_CONF_DIR) def add_logstash_cluster(namespace): print("\nDeploying LogStash\n") add_deployment_dir(namespace, conf.LOGSTASH_CONF_DIR) def add_deployment_dir(namespace, dir_path, delete=False): with open(os.path.join(dir_path, 'dep_order.txt')) as f: dep_order = f.readline() dep_lst = [x.strip() for x in dep_order.split(',')] print(dep_lst) phrases_to_replace = ["(?<!_)NAMESPACE", "REP_ES_USER", "REP_ES_PASS"] values_for_replacement = [namespace, conf.ES_USER_LOCAL, conf.ES_PASS_LOCAL] for filename in dep_lst: # replace all phrases with the actual values if exists modified_file_path, is_change = ut.duplicate_file_and_replace_phrases( dir_path, filename, f"{namespace}_{filename}", phrases_to_replace, values_for_replacement ) print(f"applying file: {filename}") with open(modified_file_path) as f: dep = yaml.safe_load(f) if modified_file_path != os.path.join(dir_path, filename) and is_change: # remove modified file ut.delete_file(modified_file_path) name = dep["metadata"]["name"] if dep['kind'] == 'StatefulSet': k8s_client = client.AppsV1Api() if not delete: k8s_client.create_namespaced_stateful_set(body=dep, namespace=namespace) else: k8s_client.delete_namespaced_stateful_set(name=name, namespace=namespace) elif dep['kind'] == 'DaemonSet': k8s_client = client.AppsV1Api() k8s_client.create_namespaced_daemon_set(body=dep, namespace=namespace) elif dep['kind'] == 'Deployment': k8s_client = client.AppsV1Api() k8s_client.create_namespaced_deployment(body=dep, namespace=namespace) elif dep['kind'] == 'Service': try: k8s_client = client.CoreV1Api() k8s_client.create_namespaced_service(body=dep, namespace=namespace) except ApiException as e: if e.status == 409: print(f"Service exists: {dep['metadata']['name']}") continue raise e elif dep['kind'] == 'PodDisruptionBudget': k8s_client = client.PolicyV1beta1Api() k8s_client.create_namespaced_pod_disruption_budget(body=dep, namespace=namespace) elif dep["kind"] == 'Role': k8s_client = client.RbacAuthorizationV1Api() k8s_client.create_namespaced_role(body=dep, namespace=namespace) elif dep["kind"] == 'ClusterRole': try: k8s_client = client.RbacAuthorizationV1Api() k8s_client.create_cluster_role(body=dep) except ApiException as e: if e.status == 409: print(f"cluster role already exists") continue raise e elif dep["kind"] == 'RoleBinding': k8s_client = client.RbacAuthorizationV1Api() dep["subjects"][0]["namespace"] = namespace k8s_client.create_namespaced_role_binding(body=dep, namespace=namespace) elif dep["kind"] == 'ClusterRoleBinding': k8s_client = client.RbacAuthorizationV1Api() try: k8s_client.create_cluster_role_binding(body=dep) except ApiException as e: if e.status == 409: print(f"cluster role binding already exists") continue raise e elif dep["kind"] == 'ConfigMap': k8s_client = client.CoreV1Api() k8s_client.create_namespaced_config_map(body=dep, namespace=namespace) elif dep["kind"] == 'ServiceAccount': k8s_client = client.CoreV1Api() k8s_client.create_namespaced_service_account(body=dep, namespace=namespace) print("\nDone\n") def remove_deployment_dir(namespace, dir_path): with open(os.path.join(dir_path, 'dep_order.txt')) as f: dep_order = f.readline() dep_lst = [x.strip() for x in dep_order.split(',')] print(dep_lst) for filename in dep_lst: print(f"deleting {filename}") with open(os.path.join(dir_path, filename)) as f: dep = yaml.safe_load(f) name = dep["metadata"]["name"] if dep['kind'] == 'StatefulSet': k8s_client = client.AppsV1Api() k8s_client.delete_namespaced_stateful_set(name=name, namespace=namespace) elif dep['kind'] == 'DaemonSet': k8s_client = client.AppsV1Api() k8s_client.delete_namespaced_daemon_set(name=name, namespace=namespace) elif dep['kind'] == 'Deployment': k8s_client = client.AppsV1Api() k8s_client.delete_namespaced_deployment(name=name, namespace=namespace) elif dep['kind'] == 'Service': k8s_client = client.CoreV1Api() k8s_client.delete_namespaced_service(name=name, namespace=namespace, grace_period_seconds=0) delete_func = k8s_client.delete_namespaced_service list_func = k8s_client.list_namespaced_service wait_for_namespaced_deletion(name, namespace, delete_func, list_func) elif dep['kind'] == 'PodDisruptionBudget': k8s_client = client.PolicyV1beta1Api() k8s_client.delete_namespaced_pod_disruption_budget(name=name, namespace=namespace) elif dep["kind"] == 'Role': k8s_client = client.RbacAuthorizationV1Api() k8s_client.delete_namespaced_role(name=name, namespace=namespace) elif dep["kind"] == 'RoleBinding': k8s_client = client.RbacAuthorizationV1Api() k8s_client.delete_namespaced_role_binding(name=name, namespace=namespace) elif dep["kind"] == 'ClusterRoleBinding': k8s_client = client.RbacAuthorizationV1Api() k8s_client.delete_cluster_role_binding(name=name) elif dep["kind"] == 'ConfigMap': k8s_client = client.CoreV1Api() k8s_client.delete_namespaced_config_map(name=name, namespace=namespace) elif dep["kind"] == 'ServiceAccount': k8s_client = client.CoreV1Api() k8s_client.delete_namespaced_service_account(name=name, namespace=namespace) print("\nDone\n") def wait_for_namespaced_deletion(name, namespace, deletion_func, list_func, timeout=15): deleted = False orig_timeout = timeout while not deleted: # find by name and delete requested item for item in list_func(namespace).items: if item.metadata.name == name: if timeout < 0: raise TimeoutError(f"{orig_timeout} was not enough for deleting item:\n{item}\n") deletion_func(name=name, namespace=namespace) print(f"service {name} was not deleted, retrying") time.sleep(1) timeout -= 1 # validate item was deleted for item in list_func(namespace).items: deleted = True if item.metadata.name == name: deleted = False return deleted def wait_for_daemonset_to_be_ready(name, namespace, timeout=None): wait_for_to_be_ready("daemonset", name, namespace, timeout=timeout) def resolve_read_status_func(obj_name): if obj_name == "daemonset": return client.AppsV1Api().read_namespaced_daemon_set_status else: raise ValueError(f"resolve_read_status_func: {obj_name} is not a valid value") def wait_for_to_be_ready(obj_name, name, namespace, timeout=None): start = datetime.now() while True: read_func = resolve_read_status_func(obj_name) resp = read_func(name=name, namespace=namespace) total_sleep_time = (datetime.now()-start).total_seconds() number_ready = resp.status.number_ready updated_number_scheduled = resp.status.updated_number_scheduled if number_ready and updated_number_scheduled and number_ready == updated_number_scheduled: print("Total time waiting for {3} {0} [size: {1}]: {2} sec".format(name, number_ready, total_sleep_time, obj_name)) break print("{0}/{1} pods ready {2} sec ".format(number_ready, updated_number_scheduled, total_sleep_time), end="\r") time.sleep(1) if timeout and total_sleep_time > timeout: raise Exception(f"Timeout waiting for {obj_name} to be ready")
1810
import json import inspect import hashlib from _plotly_utils.utils import PlotlyJSONEncoder from dash.long_callback.managers import BaseLongCallbackManager class CeleryLongCallbackManager(BaseLongCallbackManager): def __init__(self, celery_app, cache_by=None, expire=None): """ Long callback manager that runs callback logic on a celery task queue, and stores results using a celery result backend. :param celery_app: A celery.Celery application instance that must be configured with a result backend. See the celery documentation for information on configuration options. :param cache_by: A list of zero-argument functions. When provided, caching is enabled and the return values of these functions are combined with the callback function's input arguments and source code to generate cache keys. :param expire: If provided, a cache entry will be removed when it has not been accessed for ``expire`` seconds. If not provided, the lifetime of cache entries is determined by the default behavior of the celery result backend. """ try: import celery # pylint: disable=import-outside-toplevel,import-error from celery.backends.base import ( # pylint: disable=import-outside-toplevel,import-error DisabledBackend, ) except ImportError as missing_imports: raise ImportError( """\ CeleryLongCallbackManager requires extra dependencies which can be installed doing $ pip install "dash[celery]"\n""" ) from missing_imports if not isinstance(celery_app, celery.Celery): raise ValueError("First argument must be a celery.Celery object") if isinstance(celery_app.backend, DisabledBackend): raise ValueError("Celery instance must be configured with a result backend") super().__init__(cache_by) self.handle = celery_app self.expire = expire def terminate_job(self, job): if job is None: return self.handle.control.terminate(job) def terminate_unhealthy_job(self, job): task = self.get_task(job) if task and task.status in ("FAILURE", "REVOKED"): return self.terminate_job(job) return False def job_running(self, job): future = self.get_task(job) return future and future.status in ( "PENDING", "RECEIVED", "STARTED", "RETRY", "PROGRESS", ) def make_job_fn(self, fn, progress, args_deps): return _make_job_fn(fn, self.handle, progress, args_deps) def get_task(self, job): if job: return self.handle.AsyncResult(job) return None def clear_cache_entry(self, key): self.handle.backend.delete(key) def call_job_fn(self, key, job_fn, args): task = job_fn.delay(key, self._make_progress_key(key), args) return task.task_id def get_progress(self, key): progress_key = self._make_progress_key(key) progress_data = self.handle.backend.get(progress_key) if progress_data: return json.loads(progress_data) return None def result_ready(self, key): return self.handle.backend.get(key) is not None def get_result(self, key, job): # Get result value result = self.handle.backend.get(key) if result is None: return None result = json.loads(result) # Clear result if not caching if self.cache_by is None: self.clear_cache_entry(key) else: if self.expire: # Set/update expiration time self.handle.backend.expire(key, self.expire) self.clear_cache_entry(self._make_progress_key(key)) self.terminate_job(job) return result def _make_job_fn(fn, celery_app, progress, args_deps): cache = celery_app.backend # Hash function source and module to create a unique (but stable) celery task name fn_source = inspect.getsource(fn) fn_str = fn_source fn_hash = hashlib.sha1(fn_str.encode("utf-8")).hexdigest() @celery_app.task(name=f"long_callback_{fn_hash}") def job_fn(result_key, progress_key, user_callback_args, fn=fn): def _set_progress(progress_value): cache.set(progress_key, json.dumps(progress_value, cls=PlotlyJSONEncoder)) maybe_progress = [_set_progress] if progress else [] if isinstance(args_deps, dict): user_callback_output = fn(*maybe_progress, **user_callback_args) elif isinstance(args_deps, (list, tuple)): user_callback_output = fn(*maybe_progress, *user_callback_args) else: user_callback_output = fn(*maybe_progress, user_callback_args) cache.set(result_key, json.dumps(user_callback_output, cls=PlotlyJSONEncoder)) return job_fn
1838
import os import os.path as osp import numpy as np from joblib import Parallel, delayed from tensorflow.keras.utils import get_file from tqdm import tqdm from spektral.data import Dataset, Graph from spektral.utils import label_to_one_hot, sparse from spektral.utils.io import load_csv, load_sdf ATOM_TYPES = [1, 6, 7, 8, 9] BOND_TYPES = [1, 2, 3, 4] class QM9(Dataset): """ The QM9 chemical data set of small molecules. In this dataset, nodes represent atoms and edges represent chemical bonds. There are 5 possible atom types (H, C, N, O, F) and 4 bond types (single, double, triple, aromatic). Node features represent the chemical properties of each atom and include: - The atomic number, one-hot encoded; - The atom's position in the X, Y, and Z dimensions; - The atomic charge; - The mass difference from the monoisotope; The edge features represent the type of chemical bond between two atoms, one-hot encoded. Each graph has an 19-dimensional label for regression. **Arguments** - `amount`: int, load this many molecules instead of the full dataset (useful for debugging). - `n_jobs`: number of CPU cores to use for reading the data (-1, to use all available cores). """ url = "https://deepchemdata.s3-us-west-1.amazonaws.com/datasets/gdb9.tar.gz" def __init__(self, amount=None, n_jobs=1, **kwargs): self.amount = amount self.n_jobs = n_jobs super().__init__(**kwargs) def download(self): get_file( "qm9.tar.gz", self.url, extract=True, cache_dir=self.path, cache_subdir=self.path, ) os.remove(osp.join(self.path, "qm9.tar.gz")) def read(self): print("Loading QM9 dataset.") sdf_file = osp.join(self.path, "gdb9.sdf") data = load_sdf(sdf_file, amount=self.amount) # Internal SDF format def read_mol(mol): x = np.array([atom_to_feature(atom) for atom in mol["atoms"]]) a, e = mol_to_adj(mol) return x, a, e data = Parallel(n_jobs=self.n_jobs)( delayed(read_mol)(mol) for mol in tqdm(data, ncols=80) ) x_list, a_list, e_list = list(zip(*data)) # Load labels labels_file = osp.join(self.path, "gdb9.sdf.csv") labels = load_csv(labels_file) labels = labels.set_index("mol_id").values if self.amount is not None: labels = labels[: self.amount] return [ Graph(x=x, a=a, e=e, y=y) for x, a, e, y in zip(x_list, a_list, e_list, labels) ] def atom_to_feature(atom): atomic_num = label_to_one_hot(atom["atomic_num"], ATOM_TYPES) coords = atom["coords"] charge = atom["charge"] iso = atom["iso"] return np.concatenate((atomic_num, coords, [charge, iso]), -1) def mol_to_adj(mol): row, col, edge_features = [], [], [] for bond in mol["bonds"]: start, end = bond["start_atom"], bond["end_atom"] row += [start, end] col += [end, start] edge_features += [bond["type"]] * 2 a, e = sparse.edge_index_to_matrix( edge_index=np.array((row, col)).T, edge_weight=np.ones_like(row), edge_features=label_to_one_hot(edge_features, BOND_TYPES), ) return a, e
1851
from copy import copy try: # Python 2 only: from StringIO import StringIO # create a variant that can serve as a context manager class StringIO(StringIO): def __enter__(self): return self def __exit__(self, exception_type, exception_value, traceback): self.close() except ImportError: from io import StringIO try: # python 3.5+ from typing import Dict, Any from yamlable import Y except ImportError: pass import pytest from yaml import dump, load from yamlable import YamlAble, yaml_info def test_yamlable_incomplete_description(): """ Tests that if __yaml_tag_suffix__ is not provided a YamlAble subclass cannot be declared """ with pytest.raises(NotImplementedError) as err_info: class Foo(YamlAble): # __yaml_tag_suffix__ = 'foo' def __to_yaml_dict__(self): # type: (...) -> Dict[str, Any] return copy(vars(self)) @classmethod def __from_yaml_dict__(cls, # type: Type[Y] dct, # type: Dict[str, Any] yaml_tag # type: str ): # type: (...) -> Y return Foo(**dct) # instantiate f = Foo() # dump f.dumps_yaml() assert "does not seem to have a non-None '__yaml_tag_suffix__' field" in str(err_info.value) def test_yamlable(): """ Tests that YamlAble works correctly """ @yaml_info(yaml_tag_ns='yaml.tests') class Foo(YamlAble): # __yaml_tag_suffix__ = 'foo' not needed: we used @yaml_info def __init__(self, a, b): self.a = a self.b = b def __eq__(self, other): return vars(self) == vars(other) def __to_yaml_dict__(self): # type: (...) -> Dict[str, Any] return copy(vars(self)) @classmethod def __from_yaml_dict__(cls, # type: Type[Y] dct, # type: Dict[str, Any] yaml_tag # type: str ): # type: (...) -> Y return Foo(**dct) # instantiate f = Foo(1, 'hello') # note: # dump y = f.dumps_yaml(default_flow_style=False) assert y == """!yamlable/yaml.tests.Foo a: 1 b: hello """ # dump io class MemorizingStringIO(StringIO): """ A StringIO object that memorizes its buffer when it is closed (as opposed to the standard StringIO) """ def close(self): self.value = self.getvalue() # super(StringIO, self).close() # this does not work with python 2 old-style classes (StringIO is one) StringIO.close(self) s = MemorizingStringIO() f.dump_yaml(s, default_flow_style=False) assert s.value == y # dump pyyaml assert dump(f, default_flow_style=False) == y # load assert f == Foo.loads_yaml(y) # load io assert f == Foo.load_yaml(StringIO(y)) # load pyyaml assert f == load(y) def test_yamlable_legacy_method_names(): """ Tests that YamlAbleMixIn works correctly """ global enc global dec enc, dec = False, False @yaml_info(yaml_tag_ns='yaml.tests') class FooLegacy(YamlAble): # __yaml_tag_suffix__ = 'foo' not needed: we used @yaml_info def __init__(self, a, b): self.a = a self.b = b def __eq__(self, other): return vars(self) == vars(other) def to_yaml_dict(self): # type: (...) -> Dict[str, Any] global enc enc = True return copy(vars(self)) @classmethod def from_yaml_dict(cls, # type: Type[Y] dct, # type: Dict[str, Any] yaml_tag # type: str ): # type: (...) -> Y global dec dec = True return FooLegacy(**dct) # instantiate f = FooLegacy(1, 'hello') # dump y = f.dumps_yaml(default_flow_style=False) assert y == """!yamlable/yaml.tests.FooLegacy a: 1 b: hello """ # dump io class MemorizingStringIO(StringIO): """ A StringIO object that memorizes its buffer when it is closed (as opposed to the standard StringIO) """ def close(self): self.value = self.getvalue() # super(StringIO, self).close() # this does not work with python 2 old-style classes (StringIO is one) StringIO.close(self) s = MemorizingStringIO() f.dump_yaml(s, default_flow_style=False) assert s.value == y # dump pyyaml assert dump(f, default_flow_style=False) == y # load assert f == FooLegacy.loads_yaml(y) # load io assert f == FooLegacy.load_yaml(StringIO(y)) # load pyyaml assert f == load(y) assert enc assert dec # TODO override so that tag is not supported, to check error message def test_yamlable_not_supported(): @yaml_info(yaml_tag_ns='yaml.tests') class Foo_Err(YamlAble): # __yaml_tag_suffix__ = 'foo' not needed: we used @yaml_info def __init__(self, a, b): self.a = a self.b = b def __eq__(self, other): return vars(self) == vars(other) def __to_yaml_dict__(self): # type: (...) -> Dict[str, Any] return copy(vars(self)) @classmethod def __from_yaml_dict__(cls, # type: Type[Y] dct, # type: Dict[str, Any] yaml_tag # type: str ): # type: (...) -> Y return Foo_Err(**dct) @classmethod def is_yaml_tag_supported(cls, yaml_tag # type: str ): # type: (...) -> bool # ALWAYS return false return False with pytest.raises(TypeError) as err_info: Foo_Err.loads_yaml("!yamlable/yaml.tests.Foo_Err {a: 1, b: hello}\n") assert "No YamlAble subclass found able to decode object" in str(err_info.value) def test_yamlable_default_impl(): """ tests that the default implementation works """ @yaml_info(yaml_tag_ns='yaml.tests') class Foo_Default(YamlAble): def __init__(self, a, b): self.a = a self.b = b f = Foo_Default(1, 'hello') s = """!yamlable/yaml.tests.Foo_Default a: 1 b: hello """ assert dump(f, default_flow_style=False) == s assert dump(load(dump(load(s))), default_flow_style=False) == s def test_help_yaml_info(): @yaml_info("com.example.MyFoo") class Foo(YamlAble): pass assert Foo.__yaml_tag_suffix__ == "com.example.MyFoo" @yaml_info(yaml_tag_ns="com.example") class Foo(YamlAble): pass assert Foo.__yaml_tag_suffix__ == "com.example.Foo" assert Foo().dumps_yaml() == """!yamlable/com.example.Foo {} """ def test_abstract_parent_error(): """This tests that we can define an abstract parent class with the YamlAble behaviour and inherit it""" class AbstractFooE(YamlAble): pass class FooError(AbstractFooE): """ This class inherits from the parent without redefining a yaml tag """ def __init__(self, a, b): self.a = a self.b = b def __eq__(self, other): return vars(self) == vars(other) # instantiate e = FooError(1, 'hello') # dump with pytest.raises(NotImplementedError): e.dumps_yaml() def test_abstract_parent(): """This tests that we can define an abstract parent class with the YamlAble behaviour and inherit it""" class AbstractFooV(YamlAble): pass @yaml_info(yaml_tag_ns='yaml.tests') class FooValid(AbstractFooV): def __init__(self, a, b): self.a = a self.b = b def __eq__(self, other): return vars(self) == vars(other) # instantiate f = FooValid(1, 'hello') # note: # dump y = f.dumps_yaml(default_flow_style=False) assert y == """!yamlable/yaml.tests.FooValid a: 1 b: hello """ # dump io class MemorizingStringIO(StringIO): """ A StringIO object that memorizes its buffer when it is closed (as opposed to the standard StringIO) """ def close(self): self.value = self.getvalue() # super(StringIO, self).close() # this does not work with python 2 old-style classes (StringIO is one) StringIO.close(self) s = MemorizingStringIO() f.dump_yaml(s, default_flow_style=False) assert s.value == y # dump pyyaml assert dump(f, default_flow_style=False) == y # load assert f == FooValid.loads_yaml(y) # load io assert f == FooValid.load_yaml(StringIO(y)) # load pyyaml assert f == load(y)
1859
import scene class MyScene(scene.Scene): def setup(self): self.label_node = scene.LabelNode('A', position=(100,400), parent=self) self.start_flag = False def update(self): if self.start_flag: x,y = self.label_node.position if x < 340: self.label_node.position = (x+2, y) else: self.start_flag = False def touch_ended(self, touch): self.start_flag = True scene.run(MyScene())
1918
import os import sys DIR_OF_THIS_SCRIPT = os.path.abspath( os.path.dirname( __file__ ) ) def Settings( **kwargs ): return { 'interpreter_path': sys.executable, 'sys_path': [ os.path.join( DIR_OF_THIS_SCRIPT, 'third_party' ) ] }
1925
import numpy as np import xml.etree.ElementTree as ET class Geom(object): def __init__(self, geom): self.xml = geom self.params = [] def get_params(self): return self.params.copy() def set_params(self, new_params): self.params = new_params def update_point(self, p, new_params): pass def update_xml(self): pass def update(self, new_params): self.set_params(new_params) self.update_xml() def get_smallest_z(self): pass def get_param_limits(self): pass def get_param_names(self): pass def get_volume(self): pass class Sphere(Geom): min_radius = .05 max_radius = .4 def __init__(self, geom): self.xml = geom self.params = [float(self.xml.get('size'))] # radius self.center = np.array([float(x) for x in self.xml.get('pos').split()]) def update_point(self, p, new_params): return ((p - self.center) * new_params[0] / self.params[0]) + self.center def update_xml(self): self.xml.set('size', str(self.params[0])) def get_smallest_z(self): return self.center[2] - self.params[0] def get_param_limits(self): return [[self.min_radius], [self.max_radius]] def get_param_names(self): return ['radius'] def get_volume(self): return 4./3. * np.pi * self.params[0] ** 3 class Capsule(Geom): min_length = 0.175 max_length = 0.8 min_radius = 0.035 max_radius = 0.085 def __init__(self, geom): self.xml = geom fromto = [float(x) for x in self.xml.get('fromto').split()] self.p1 = np.array(fromto[:3]) self.p2 = np.array(fromto[3:]) length = np.sqrt(np.sum((self.p2 - self.p1) ** 2)) radius = float(self.xml.get('size')) self.params = [length, radius] self.axis = (self.p2 - self.p1) / length def update_point(self, p, new_params): lfac = p.dot(self.axis) * self.axis rfac = p - lfac return p + lfac * (-1.0 + new_params[0] / self.params[0])# + rfac * (new_params[1] / self.params[1]) def update_xml(self): self.xml.set('fromto', ' '.join([str(x) for x in np.concatenate([self.p1, self.p2])])) self.xml.set('size', str(self.params[1])) # radius def set_params(self, new_params): p1 = self.update_point(self.p1, new_params) p2 = self.update_point(self.p2, new_params) # update only after computing p1, p2 self.p1 = p1 self.p2 = p2 super().set_params(new_params) def get_smallest_z(self): return min(self.p1[2], self.p2[2]) - self.params[1] def get_param_limits(self): return [[self.min_length, self.min_radius], [self.max_length, self.max_radius]] def get_param_names(self): return ['length','radius'] def get_volume(self): return 4./3. * np.pi * self.params[1]**3 + self.params[0] * np.pi * self.params[1]**2 class Body: geoms = {'sphere': Sphere, 'capsule': Capsule} # dictionary of legal geometry types def __init__(self, body, worldbody=False): self.xml = body self.worldbody = worldbody geom_xml = body.find('geom') # assume only one geometry per body self.geom = self.geoms[geom_xml.get('type')](geom_xml) self.joints = [j for j in body.findall('joint') if 'ignore' not in j.get('name')] self.parts = [Body(b) for b in body.findall('body')] pos = [b.get('pos') for b in body.findall('body')] self.part_positions = [np.array([float(x) for x in p.split()]) for p in pos] pos = [j.get('pos') for j in self.joints] self.joint_positions = [np.array([float(x) for x in p.split()]) for p in pos] self.n = len(self.geom.get_params()) self.n_all_params = len(self.get_params()) self.zmin = float(self.xml.get("pos").split()[2]) - self.get_height() def get_height(self): max_height = -self.geom.get_smallest_z() for body, pos in zip(self.parts, self.part_positions): max_height = max(max_height, body.get_height() - pos[2]) return max_height def update_initial_position(self): pos = self.xml.get("pos").split() pos[2] = str(self.get_height() + self.zmin) self.xml.set("pos", ' '.join(pos)) def update_xml(self): for body, pos in zip(self.parts, self.part_positions): body.xml.set('pos', ' '.join([str(x) for x in pos])) for joint, pos in zip(self.joints, self.joint_positions): joint.set('pos', ' '.join([str(x) for x in pos])) def set_body_positions(self, new_params): for i, pos in enumerate(self.part_positions): self.part_positions[i] = self.geom.update_point(pos, new_params) for i, pos in enumerate(self.joint_positions): self.joint_positions[i] = self.geom.update_point(pos, new_params) def update(self, new_params): self.set_body_positions(new_params) self.geom.update(new_params) self.update_xml() def get_params(self): params = self.geom.get_params() for body in self.parts: params += body.get_params() return params def get_param_limits(self): limits = self.geom.get_param_limits() for body in self.parts: body_limits = body.get_param_limits() limits[0] += body_limits[0] limits[1] += body_limits[1] return limits def get_param_names(self): name = self.xml.get('name') param_names = [name + '-' + p for p in self.geom.get_param_names()] for body in self.parts: param_names += body.get_param_names() return param_names def update_params(self, new_params): if self.worldbody: assert len(new_params) == self.n_all_params, "Wrong number of parameters" self.update(new_params[:self.n]) remaining_params = new_params[self.n:] for body in self.parts: remaining_params = body.update_params(remaining_params) if self.worldbody: self.update_initial_position() else: return remaining_params def get_body_names(self): names = [self.xml.get('name')] for body in self.parts: names += body.get_names() return names def get_joints(self): joints = {} for body,pos in zip(self.parts, self.part_positions): for j in body.joints: joints[j.get('name')] = (self.xml.get('name'), body.xml.get('name'), self.geom, body.geom, pos) joints.update(body.get_joints()) return joints def get_volumes(self): volumes = {} if len(self.joints) > 0: for j in self.joints: v1 = self.geom.get_volume() v2 = sum([b.geom.get_volume() for b in self.parts]) volumes[j.get('name')] = np.array((v1, v2)) for body in self.parts: volumes.update(body.get_volumes()) return volumes class MuJoCoXmlRobot: def __init__(self, model_xml): self.model_xml = model_xml self.tree = ET.parse(self.model_xml) worldbody = self.tree.getroot().find('worldbody') self.body = Body(worldbody.find('body'), worldbody=True) def get_params(self): return self.body.get_params() def get_param_limits(self): return self.body.get_param_limits() def get_param_names(self): return self.body.get_param_names() def get_height(self): return self.body.get_height() def get_joints(self): return self.body.get_joints() def get_volumes(self): return self.body.get_volumes() def update(self, params, xml_file=None): if xml_file is None: xml_file = self.model_xml self.body.update_params(list(params)) self.tree.write(xml_file) if __name__ == '__main__': robot = MuJoCoXmlRobot('mujoco_assets/hopper.xml') params = list(1.0 * np.array(robot.get_params())) robot.update(params, 'mujoco_assets/hopper_test.xml') assert robot.get_params() == params #assert robot.get_height() == 1.31 print(robot.get_param_limits()) print(robot.get_param_names()) robot = MuJoCoXmlRobot('mujoco_assets/walker2d.xml') params = [.4,.04,.5,.05,.55,.055,.6,.06,.5,.05,.55,.055,.6,.06] robot.update(params, 'mujoco_assets/walker2d_test.xml') assert robot.get_params() == params assert robot.get_height() == 1.31 print(robot.get_param_limits()) print(robot.get_param_names()) robot = MuJoCoXmlRobot('mujoco_assets/ant.xml') params = [.2, .2,.06,.2,.06,.4,.06, .2,.06,.2,.06,.4,.06, .2,.06,.2,.06,.4,.06, .2,.06,.2,.06,.4,.06] robot.update(params, 'mujoco_assets/ant_test.xml') assert robot.get_params() == params assert robot.get_height() == .2 print(robot.get_param_limits()) print(robot.get_param_names()) robot = MuJoCoXmlRobot('mujoco_assets/humanoid.xml') params = list(.8 * np.array(robot.get_params())) robot.update(params, 'mujoco_assets/humanoid_test.xml') assert robot.get_params() == params print(robot.get_height()) #assert robot.get_height() == .6085 print(robot.get_param_limits()) print(robot.get_param_names()) import gym, roboschool env = gym.make("RoboschoolHopper-v1") env.unwrapped.model_xml = 'mujoco_assets/hopper_test.xml' env.reset() #env.render() import os from scipy.misc import imsave import subprocess as sp outdir = 'xml_vid' os.makedirs(outdir, exist_ok=True) i = 0 for _ in range(10): env.reset() for _ in range(100): env.step(env.action_space.sample()) rgb = env.render('rgb_array') imsave(os.path.join(outdir, '{:05d}.png'.format(i)), rgb) i+=1 sp.call(['ffmpeg', '-r', '60', '-f', 'image2', '-i', os.path.join(outdir, '%05d.png'), '-vcodec', 'libx264', '-pix_fmt', 'yuv420p', os.path.join(outdir, 'out.mp4')]) env.close()
1934
from django.conf import settings from django.conf.urls.static import static from django.contrib import admin from django.urls import path, include, re_path from django.views.generic import TemplateView urlpatterns = [ path('api-auth/', include('rest_framework.urls')), path('rest-auth/', include('rest_auth.urls')), path('rest-auth/registration/', include('rest_auth.registration.urls')), path('admin/', admin.site.urls), path('api/', include('core.api.urls')), ] if settings.DEBUG: urlpatterns += static(settings.MEDIA_URL, document_root=settings.MEDIA_ROOT) if not settings.DEBUG: urlpatterns += [re_path(r'^.*', TemplateView.as_view(template_name='index.html'))]
1946
import h5py import numpy as np np.set_printoptions(threshold=np.nan) from shutil import copyfile copyfile("dummy_lutnet.h5", "pretrained_bin.h5") # create pretrained.h5 using datastructure from dummy.h5 bl = h5py.File("baseline_pruned.h5", 'r') #dummy = h5py.File("dummy.h5", 'r') pretrained = h5py.File("pretrained_bin.h5", 'r+') # dense layer 1 bl_w1 = bl["model_weights"]["binary_dense_1"]["binary_dense_1"]["Variable_1:0"] bl_pruning_mask = bl["model_weights"]["binary_dense_1"]["binary_dense_1"]["pruning_mask:0"] bl_gamma = bl["model_weights"]["binary_dense_1"]["binary_dense_1"]["Variable:0"] zero_fill = np.zeros(np.shape(np.array(bl_w1))) pret_w1 = pretrained["model_weights"]["binary_dense_1"]["binary_dense_1"]["Variable_1:0"] pret_pruning_mask = pretrained["model_weights"]["binary_dense_1"]["binary_dense_1"]["pruning_mask:0"] p_gamma = pretrained["model_weights"]["binary_dense_1"]["binary_dense_1"]["Variable:0"] pret_w1[...] = np.array(bl_w1) p_gamma[...] = np.array(bl_gamma) pret_pruning_mask[...] = np.array(bl_pruning_mask) print(np.sum(np.array(bl_pruning_mask)), np.prod(np.shape(np.array(bl_pruning_mask)))) # dense layer 2 bl_w1 = bl["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_1:0"] bl_rand_map_0 = bl["model_weights"]["binary_dense_2"]["binary_dense_2"]["rand_map_0:0"] bl_pruning_mask = bl["model_weights"]["binary_dense_2"]["binary_dense_2"]["pruning_mask:0"] bl_gamma = bl["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable:0"] bl_means = bl["model_weights"]["residual_sign_1"]["residual_sign_1"]["means:0"] pret_rand_map_0 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["rand_map_0:0"] pret_rand_map_1 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["rand_map_1:0"] pret_rand_map_2 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["rand_map_2:0"] pret_pruning_mask = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["pruning_mask:0"] p_gamma = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable:0"] pret_means = pretrained["model_weights"]["residual_sign_1"]["residual_sign_1"]["means:0"] pret_c1 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_1:0"] pret_c2 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_2:0"] pret_c3 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_3:0"] pret_c4 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_4:0"] pret_c5 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_5:0"] pret_c6 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_6:0"] pret_c7 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_7:0"] pret_c8 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_8:0"] pret_c9 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_9:0"] pret_c10= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_10:0"] pret_c11= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_11:0"] pret_c12= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_12:0"] pret_c13= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_13:0"] pret_c14= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_14:0"] pret_c15= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_15:0"] pret_c16= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_16:0"] pret_c17= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_17:0"] pret_c18= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_18:0"] pret_c19= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_19:0"] pret_c20= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_20:0"] pret_c21= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_21:0"] pret_c22= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_22:0"] pret_c23= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_23:0"] pret_c24= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_24:0"] pret_c25= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_25:0"] pret_c26= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_26:0"] pret_c27= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_27:0"] pret_c28= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_28:0"] pret_c29= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_29:0"] pret_c30= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_30:0"] pret_c31= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_31:0"] pret_c32= pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_32:0"] pret_w1 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["Variable_33:0"] pret_rand_map_exp_0 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["rand_map_exp_0:0"] pret_rand_map_exp_1 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["rand_map_exp_1:0"] pret_rand_map_exp_2 = pretrained["model_weights"]["binary_dense_2"]["binary_dense_2"]["rand_map_exp_2:0"] weight_shape = np.shape(bl_w1) tile_shape = np.shape(pret_c1) zero_fill = np.zeros(tile_shape) one_fill = np.ones(tile_shape) neg_one_fill = -np.ones(tile_shape) # randomisation and pruning recovery bl_w1_unroll = np.array(bl_w1) bl_w1 = np.array(bl_w1) rand_map_0 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_0) rand_map_1 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_1) rand_map_2 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_2) pruning_mask = np.array(bl_pruning_mask).astype(bool) init_mask = np.logical_not(pruning_mask[rand_map_0]) pruning_mask_recover = np.logical_and(pruning_mask, init_mask)[np.argsort(rand_map_0)] pruning_mask = np.logical_or(pruning_mask, pruning_mask_recover) init_mask = np.reshape(init_mask, tile_shape) # expand randomisation map across tiles rand_map_0_expand = np.tile(rand_map_0,[weight_shape[0]/tile_shape[0]]) rand_map_1_expand = np.tile(rand_map_1,[weight_shape[0]/tile_shape[0]]) rand_map_2_expand = np.tile(rand_map_2,[weight_shape[0]/tile_shape[0]]) for i in range(weight_shape[0]): rand_map_0_expand[i] = rand_map_0_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_0_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) rand_map_1_expand[i] = rand_map_1_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_1_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) rand_map_2_expand[i] = rand_map_2_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_2_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) bl_w1_rand_0 = bl_w1_unroll[rand_map_0_expand] bl_w1_rand_0 = np.reshape(bl_w1_rand_0, weight_shape) w1 = bl_w1 # connect1 only c1 = one_fill c2 = neg_one_fill c3 = one_fill c4 = neg_one_fill c5 = one_fill c6 = neg_one_fill c7 = one_fill c8 = neg_one_fill c9 = one_fill c10 = neg_one_fill c11 = one_fill c12 = neg_one_fill c13 = one_fill c14 = neg_one_fill c15 = one_fill c16 = neg_one_fill c17 = neg_one_fill c18 = one_fill c19 = neg_one_fill c20 = one_fill c21 = neg_one_fill c22 = one_fill c23 = neg_one_fill c24 = one_fill c25 = neg_one_fill c26 = one_fill c27 = neg_one_fill c28 = one_fill c29 = neg_one_fill c30 = one_fill c31 = neg_one_fill c32 = one_fill pret_w1 [...] = w1 pret_c1 [...] = c1 pret_c2 [...] = c2 pret_c3 [...] = c3 pret_c4 [...] = c4 pret_c5 [...] = c5 pret_c6 [...] = c6 pret_c7 [...] = c7 pret_c8 [...] = c8 pret_c9 [...] = c9 pret_c10[...] = c10 pret_c11[...] = c11 pret_c12[...] = c12 pret_c13[...] = c13 pret_c14[...] = c14 pret_c15[...] = c15 pret_c16[...] = c16 pret_c17[...] = c17 pret_c18[...] = c18 pret_c19[...] = c19 pret_c20[...] = c20 pret_c21[...] = c21 pret_c22[...] = c22 pret_c23[...] = c23 pret_c24[...] = c24 pret_c25[...] = c25 pret_c26[...] = c26 pret_c27[...] = c27 pret_c28[...] = c28 pret_c29[...] = c29 pret_c30[...] = c30 pret_c31[...] = c31 pret_c32[...] = c32 pret_rand_map_0[...] = np.reshape(rand_map_0, (-1,1)).astype(float) pret_rand_map_1[...] = np.reshape(rand_map_1, (-1,1)).astype(float) pret_rand_map_2[...] = np.reshape(rand_map_2, (-1,1)).astype(float) p_gamma[...] = np.array(bl_gamma) pret_means[...] = np.array(bl_means) pret_pruning_mask[...] = np.array(bl_pruning_mask) rand_map_0_expand = np.reshape(rand_map_0_expand, [-1,1]).astype(float) pret_rand_map_exp_0[...] = rand_map_0_expand rand_map_1_expand = np.reshape(rand_map_1_expand, [-1,1]).astype(float) pret_rand_map_exp_1[...] = rand_map_1_expand rand_map_2_expand = np.reshape(rand_map_2_expand, [-1,1]).astype(float) pret_rand_map_exp_2[...] = rand_map_2_expand print(np.sum(np.array(bl_pruning_mask)), np.prod(np.shape(np.array(bl_pruning_mask)))) # dense layer 3 bl_w1 = bl["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_1:0"] bl_rand_map_0 = bl["model_weights"]["binary_dense_3"]["binary_dense_3"]["rand_map_0:0"] bl_pruning_mask = bl["model_weights"]["binary_dense_3"]["binary_dense_3"]["pruning_mask:0"] bl_gamma = bl["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable:0"] bl_means = bl["model_weights"]["residual_sign_2"]["residual_sign_2"]["means:0"] pret_rand_map_0 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["rand_map_0:0"] pret_rand_map_1 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["rand_map_1:0"] pret_rand_map_2 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["rand_map_2:0"] pret_pruning_mask = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["pruning_mask:0"] p_gamma = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable:0"] pret_means = pretrained["model_weights"]["residual_sign_2"]["residual_sign_2"]["means:0"] pret_c1 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_1:0"] pret_c2 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_2:0"] pret_c3 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_3:0"] pret_c4 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_4:0"] pret_c5 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_5:0"] pret_c6 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_6:0"] pret_c7 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_7:0"] pret_c8 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_8:0"] pret_c9 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_9:0"] pret_c10= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_10:0"] pret_c11= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_11:0"] pret_c12= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_12:0"] pret_c13= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_13:0"] pret_c14= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_14:0"] pret_c15= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_15:0"] pret_c16= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_16:0"] pret_c17= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_17:0"] pret_c18= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_18:0"] pret_c19= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_19:0"] pret_c20= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_20:0"] pret_c21= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_21:0"] pret_c22= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_22:0"] pret_c23= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_23:0"] pret_c24= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_24:0"] pret_c25= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_25:0"] pret_c26= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_26:0"] pret_c27= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_27:0"] pret_c28= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_28:0"] pret_c29= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_29:0"] pret_c30= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_30:0"] pret_c31= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_31:0"] pret_c32= pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_32:0"] pret_w1 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["Variable_33:0"] pret_rand_map_exp_0 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["rand_map_exp_0:0"] pret_rand_map_exp_1 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["rand_map_exp_1:0"] pret_rand_map_exp_2 = pretrained["model_weights"]["binary_dense_3"]["binary_dense_3"]["rand_map_exp_2:0"] weight_shape = np.shape(bl_w1) tile_shape = np.shape(pret_c1) zero_fill = np.zeros(tile_shape) one_fill = np.ones(tile_shape) neg_one_fill = -np.ones(tile_shape) # randomisation and pruning recovery bl_w1_unroll = np.array(bl_w1) bl_w1 = np.array(bl_w1) rand_map_0 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_0) rand_map_1 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_1) rand_map_2 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_2) pruning_mask = np.array(bl_pruning_mask).astype(bool) init_mask = np.logical_not(pruning_mask[rand_map_0]) pruning_mask_recover = np.logical_and(pruning_mask, init_mask)[np.argsort(rand_map_0)] pruning_mask = np.logical_or(pruning_mask, pruning_mask_recover) init_mask = np.reshape(init_mask, tile_shape) # expand randomisation map across tiles rand_map_0_expand = np.tile(rand_map_0,[weight_shape[0]/tile_shape[0]]) rand_map_1_expand = np.tile(rand_map_1,[weight_shape[0]/tile_shape[0]]) rand_map_2_expand = np.tile(rand_map_2,[weight_shape[0]/tile_shape[0]]) for i in range(weight_shape[0]): rand_map_0_expand[i] = rand_map_0_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_0_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) rand_map_1_expand[i] = rand_map_1_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_1_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) rand_map_2_expand[i] = rand_map_2_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_2_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) bl_w1_rand_0 = bl_w1_unroll[rand_map_0_expand] bl_w1_rand_0 = np.reshape(bl_w1_rand_0, weight_shape) w1 = bl_w1 # connect1 only c1 = one_fill c2 = neg_one_fill c3 = one_fill c4 = neg_one_fill c5 = one_fill c6 = neg_one_fill c7 = one_fill c8 = neg_one_fill c9 = one_fill c10 = neg_one_fill c11 = one_fill c12 = neg_one_fill c13 = one_fill c14 = neg_one_fill c15 = one_fill c16 = neg_one_fill c17 = neg_one_fill c18 = one_fill c19 = neg_one_fill c20 = one_fill c21 = neg_one_fill c22 = one_fill c23 = neg_one_fill c24 = one_fill c25 = neg_one_fill c26 = one_fill c27 = neg_one_fill c28 = one_fill c29 = neg_one_fill c30 = one_fill c31 = neg_one_fill c32 = one_fill pret_w1 [...] = w1 pret_c1 [...] = c1 pret_c2 [...] = c2 pret_c3 [...] = c3 pret_c4 [...] = c4 pret_c5 [...] = c5 pret_c6 [...] = c6 pret_c7 [...] = c7 pret_c8 [...] = c8 pret_c9 [...] = c9 pret_c10[...] = c10 pret_c11[...] = c11 pret_c12[...] = c12 pret_c13[...] = c13 pret_c14[...] = c14 pret_c15[...] = c15 pret_c16[...] = c16 pret_c17[...] = c17 pret_c18[...] = c18 pret_c19[...] = c19 pret_c20[...] = c20 pret_c21[...] = c21 pret_c22[...] = c22 pret_c23[...] = c23 pret_c24[...] = c24 pret_c25[...] = c25 pret_c26[...] = c26 pret_c27[...] = c27 pret_c28[...] = c28 pret_c29[...] = c29 pret_c30[...] = c30 pret_c31[...] = c31 pret_c32[...] = c32 pret_rand_map_0[...] = np.reshape(rand_map_0, (-1,1)).astype(float) pret_rand_map_1[...] = np.reshape(rand_map_1, (-1,1)).astype(float) pret_rand_map_2[...] = np.reshape(rand_map_2, (-1,1)).astype(float) p_gamma[...] = np.array(bl_gamma) pret_means[...] = np.array(bl_means) pret_pruning_mask[...] = np.array(bl_pruning_mask) rand_map_0_expand = np.reshape(rand_map_0_expand, [-1,1]).astype(float) pret_rand_map_exp_0[...] = rand_map_0_expand rand_map_1_expand = np.reshape(rand_map_1_expand, [-1,1]).astype(float) pret_rand_map_exp_1[...] = rand_map_1_expand rand_map_2_expand = np.reshape(rand_map_2_expand, [-1,1]).astype(float) pret_rand_map_exp_2[...] = rand_map_2_expand print(np.sum(np.array(bl_pruning_mask)), np.prod(np.shape(np.array(bl_pruning_mask)))) # dense layer 4 bl_w1 = bl["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_1:0"] bl_rand_map_0 = bl["model_weights"]["binary_dense_4"]["binary_dense_4"]["rand_map_0:0"] bl_pruning_mask = bl["model_weights"]["binary_dense_4"]["binary_dense_4"]["pruning_mask:0"] bl_gamma = bl["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable:0"] bl_means = bl["model_weights"]["residual_sign_3"]["residual_sign_3"]["means:0"] pret_rand_map_0 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["rand_map_0:0"] pret_rand_map_1 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["rand_map_1:0"] pret_rand_map_2 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["rand_map_2:0"] pret_pruning_mask = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["pruning_mask:0"] p_gamma = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable:0"] pret_means = pretrained["model_weights"]["residual_sign_3"]["residual_sign_3"]["means:0"] pret_c1 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_1:0"] pret_c2 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_2:0"] pret_c3 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_3:0"] pret_c4 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_4:0"] pret_c5 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_5:0"] pret_c6 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_6:0"] pret_c7 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_7:0"] pret_c8 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_8:0"] pret_c9 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_9:0"] pret_c10= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_10:0"] pret_c11= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_11:0"] pret_c12= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_12:0"] pret_c13= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_13:0"] pret_c14= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_14:0"] pret_c15= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_15:0"] pret_c16= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_16:0"] pret_c17= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_17:0"] pret_c18= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_18:0"] pret_c19= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_19:0"] pret_c20= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_20:0"] pret_c21= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_21:0"] pret_c22= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_22:0"] pret_c23= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_23:0"] pret_c24= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_24:0"] pret_c25= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_25:0"] pret_c26= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_26:0"] pret_c27= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_27:0"] pret_c28= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_28:0"] pret_c29= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_29:0"] pret_c30= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_30:0"] pret_c31= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_31:0"] pret_c32= pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_32:0"] pret_w1 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["Variable_33:0"] pret_rand_map_exp_0 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["rand_map_exp_0:0"] pret_rand_map_exp_1 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["rand_map_exp_1:0"] pret_rand_map_exp_2 = pretrained["model_weights"]["binary_dense_4"]["binary_dense_4"]["rand_map_exp_2:0"] weight_shape = np.shape(bl_w1) tile_shape = np.shape(pret_c1) zero_fill = np.zeros(tile_shape) one_fill = np.ones(tile_shape) neg_one_fill = -np.ones(tile_shape) # randomisation and pruning recovery bl_w1_unroll = np.array(bl_w1) bl_w1 = np.array(bl_w1) rand_map_0 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_0) rand_map_1 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_1) rand_map_2 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_2) pruning_mask = np.array(bl_pruning_mask).astype(bool) init_mask = np.logical_not(pruning_mask[rand_map_0]) pruning_mask_recover = np.logical_and(pruning_mask, init_mask)[np.argsort(rand_map_0)] pruning_mask = np.logical_or(pruning_mask, pruning_mask_recover) init_mask = np.reshape(init_mask, tile_shape) # expand randomisation map across tiles rand_map_0_expand = np.tile(rand_map_0,[weight_shape[0]/tile_shape[0]]) rand_map_1_expand = np.tile(rand_map_1,[weight_shape[0]/tile_shape[0]]) rand_map_2_expand = np.tile(rand_map_2,[weight_shape[0]/tile_shape[0]]) for i in range(weight_shape[0]): rand_map_0_expand[i] = rand_map_0_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_0_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) rand_map_1_expand[i] = rand_map_1_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_1_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) rand_map_2_expand[i] = rand_map_2_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_2_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) bl_w1_rand_0 = bl_w1_unroll[rand_map_0_expand] bl_w1_rand_0 = np.reshape(bl_w1_rand_0, weight_shape) w1 = bl_w1 # connect1 only c1 = one_fill c2 = neg_one_fill c3 = one_fill c4 = neg_one_fill c5 = one_fill c6 = neg_one_fill c7 = one_fill c8 = neg_one_fill c9 = one_fill c10 = neg_one_fill c11 = one_fill c12 = neg_one_fill c13 = one_fill c14 = neg_one_fill c15 = one_fill c16 = neg_one_fill c17 = neg_one_fill c18 = one_fill c19 = neg_one_fill c20 = one_fill c21 = neg_one_fill c22 = one_fill c23 = neg_one_fill c24 = one_fill c25 = neg_one_fill c26 = one_fill c27 = neg_one_fill c28 = one_fill c29 = neg_one_fill c30 = one_fill c31 = neg_one_fill c32 = one_fill pret_w1 [...] = w1 pret_c1 [...] = c1 pret_c2 [...] = c2 pret_c3 [...] = c3 pret_c4 [...] = c4 pret_c5 [...] = c5 pret_c6 [...] = c6 pret_c7 [...] = c7 pret_c8 [...] = c8 pret_c9 [...] = c9 pret_c10[...] = c10 pret_c11[...] = c11 pret_c12[...] = c12 pret_c13[...] = c13 pret_c14[...] = c14 pret_c15[...] = c15 pret_c16[...] = c16 pret_c17[...] = c17 pret_c18[...] = c18 pret_c19[...] = c19 pret_c20[...] = c20 pret_c21[...] = c21 pret_c22[...] = c22 pret_c23[...] = c23 pret_c24[...] = c24 pret_c25[...] = c25 pret_c26[...] = c26 pret_c27[...] = c27 pret_c28[...] = c28 pret_c29[...] = c29 pret_c30[...] = c30 pret_c31[...] = c31 pret_c32[...] = c32 pret_rand_map_0[...] = np.reshape(rand_map_0, (-1,1)).astype(float) pret_rand_map_1[...] = np.reshape(rand_map_1, (-1,1)).astype(float) pret_rand_map_2[...] = np.reshape(rand_map_2, (-1,1)).astype(float) p_gamma[...] = np.array(bl_gamma) pret_means[...] = np.array(bl_means) pret_pruning_mask[...] = np.array(bl_pruning_mask) rand_map_0_expand = np.reshape(rand_map_0_expand, [-1,1]).astype(float) pret_rand_map_exp_0[...] = rand_map_0_expand rand_map_1_expand = np.reshape(rand_map_1_expand, [-1,1]).astype(float) pret_rand_map_exp_1[...] = rand_map_1_expand rand_map_2_expand = np.reshape(rand_map_2_expand, [-1,1]).astype(float) pret_rand_map_exp_2[...] = rand_map_2_expand print(np.sum(np.array(bl_pruning_mask)), np.prod(np.shape(np.array(bl_pruning_mask)))) # dense layer 5 bl_w1 = bl["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_1:0"] bl_rand_map_0 = bl["model_weights"]["binary_dense_5"]["binary_dense_5"]["rand_map_0:0"] bl_pruning_mask = bl["model_weights"]["binary_dense_5"]["binary_dense_5"]["pruning_mask:0"] bl_gamma = bl["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable:0"] bl_means = bl["model_weights"]["residual_sign_4"]["residual_sign_4"]["means:0"] pret_rand_map_0 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["rand_map_0:0"] pret_rand_map_1 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["rand_map_1:0"] pret_rand_map_2 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["rand_map_2:0"] pret_pruning_mask = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["pruning_mask:0"] p_gamma = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable:0"] pret_means = pretrained["model_weights"]["residual_sign_4"]["residual_sign_4"]["means:0"] pret_c1 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_1:0"] pret_c2 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_2:0"] pret_c3 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_3:0"] pret_c4 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_4:0"] pret_c5 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_5:0"] pret_c6 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_6:0"] pret_c7 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_7:0"] pret_c8 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_8:0"] pret_c9 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_9:0"] pret_c10= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_10:0"] pret_c11= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_11:0"] pret_c12= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_12:0"] pret_c13= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_13:0"] pret_c14= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_14:0"] pret_c15= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_15:0"] pret_c16= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_16:0"] pret_c17= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_17:0"] pret_c18= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_18:0"] pret_c19= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_19:0"] pret_c20= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_20:0"] pret_c21= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_21:0"] pret_c22= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_22:0"] pret_c23= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_23:0"] pret_c24= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_24:0"] pret_c25= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_25:0"] pret_c26= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_26:0"] pret_c27= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_27:0"] pret_c28= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_28:0"] pret_c29= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_29:0"] pret_c30= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_30:0"] pret_c31= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_31:0"] pret_c32= pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_32:0"] pret_w1 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["Variable_33:0"] pret_rand_map_exp_0 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["rand_map_exp_0:0"] pret_rand_map_exp_1 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["rand_map_exp_1:0"] pret_rand_map_exp_2 = pretrained["model_weights"]["binary_dense_5"]["binary_dense_5"]["rand_map_exp_2:0"] weight_shape = np.shape(bl_w1) tile_shape = np.shape(pret_c1) zero_fill = np.zeros(tile_shape) one_fill = np.ones(tile_shape) neg_one_fill = -np.ones(tile_shape) # randomisation and pruning recovery bl_w1_unroll = np.array(bl_w1) bl_w1 = np.array(bl_w1) rand_map_0 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_0) rand_map_1 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_1) rand_map_2 = np.arange(tile_shape[0]) np.random.shuffle(rand_map_2) pruning_mask = np.array(bl_pruning_mask).astype(bool) init_mask = np.logical_not(pruning_mask[rand_map_0]) pruning_mask_recover = np.logical_and(pruning_mask, init_mask)[np.argsort(rand_map_0)] pruning_mask = np.logical_or(pruning_mask, pruning_mask_recover) init_mask = np.reshape(init_mask, tile_shape) # expand randomisation map across tiles rand_map_0_expand = np.tile(rand_map_0,[weight_shape[0]/tile_shape[0]]) rand_map_1_expand = np.tile(rand_map_1,[weight_shape[0]/tile_shape[0]]) rand_map_2_expand = np.tile(rand_map_2,[weight_shape[0]/tile_shape[0]]) for i in range(weight_shape[0]): rand_map_0_expand[i] = rand_map_0_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_0_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) rand_map_1_expand[i] = rand_map_1_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_1_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) rand_map_2_expand[i] = rand_map_2_expand[i] + (tile_shape[0]*(weight_shape[0]/tile_shape[0]-1)) * (rand_map_2_expand[i]/tile_shape[0]) + tile_shape[0]*(i%weight_shape[0]/tile_shape[0]) bl_w1_rand_0 = bl_w1_unroll[rand_map_0_expand] bl_w1_rand_0 = np.reshape(bl_w1_rand_0, weight_shape) w1 = bl_w1 # connect1 only c1 = one_fill c2 = neg_one_fill c3 = one_fill c4 = neg_one_fill c5 = one_fill c6 = neg_one_fill c7 = one_fill c8 = neg_one_fill c9 = one_fill c10 = neg_one_fill c11 = one_fill c12 = neg_one_fill c13 = one_fill c14 = neg_one_fill c15 = one_fill c16 = neg_one_fill c17 = neg_one_fill c18 = one_fill c19 = neg_one_fill c20 = one_fill c21 = neg_one_fill c22 = one_fill c23 = neg_one_fill c24 = one_fill c25 = neg_one_fill c26 = one_fill c27 = neg_one_fill c28 = one_fill c29 = neg_one_fill c30 = one_fill c31 = neg_one_fill c32 = one_fill pret_w1 [...] = w1 pret_c1 [...] = c1 pret_c2 [...] = c2 pret_c3 [...] = c3 pret_c4 [...] = c4 pret_c5 [...] = c5 pret_c6 [...] = c6 pret_c7 [...] = c7 pret_c8 [...] = c8 pret_c9 [...] = c9 pret_c10[...] = c10 pret_c11[...] = c11 pret_c12[...] = c12 pret_c13[...] = c13 pret_c14[...] = c14 pret_c15[...] = c15 pret_c16[...] = c16 pret_c17[...] = c17 pret_c18[...] = c18 pret_c19[...] = c19 pret_c20[...] = c20 pret_c21[...] = c21 pret_c22[...] = c22 pret_c23[...] = c23 pret_c24[...] = c24 pret_c25[...] = c25 pret_c26[...] = c26 pret_c27[...] = c27 pret_c28[...] = c28 pret_c29[...] = c29 pret_c30[...] = c30 pret_c31[...] = c31 pret_c32[...] = c32 pret_rand_map_0[...] = np.reshape(rand_map_0, (-1,1)).astype(float) pret_rand_map_1[...] = np.reshape(rand_map_1, (-1,1)).astype(float) pret_rand_map_2[...] = np.reshape(rand_map_2, (-1,1)).astype(float) p_gamma[...] = np.array(bl_gamma) pret_means[...] = np.array(bl_means) pret_pruning_mask[...] = np.array(bl_pruning_mask) rand_map_0_expand = np.reshape(rand_map_0_expand, [-1,1]).astype(float) pret_rand_map_exp_0[...] = rand_map_0_expand rand_map_1_expand = np.reshape(rand_map_1_expand, [-1,1]).astype(float) pret_rand_map_exp_1[...] = rand_map_1_expand rand_map_2_expand = np.reshape(rand_map_2_expand, [-1,1]).astype(float) pret_rand_map_exp_2[...] = rand_map_2_expand print(np.sum(np.array(bl_pruning_mask)), np.prod(np.shape(np.array(bl_pruning_mask)))) # bn 1 bl_beta = bl["model_weights"]["batch_normalization_1"]["batch_normalization_1"]["beta:0"] bl_gamma = bl["model_weights"]["batch_normalization_1"]["batch_normalization_1"]["gamma:0"] bl_moving_mean = bl["model_weights"]["batch_normalization_1"]["batch_normalization_1"]["moving_mean:0"] bl_moving_variance = bl["model_weights"]["batch_normalization_1"]["batch_normalization_1"]["moving_variance:0"] p_beta = pretrained["model_weights"]["batch_normalization_1"]["batch_normalization_1"]["beta:0"] p_gamma = pretrained["model_weights"]["batch_normalization_1"]["batch_normalization_1"]["gamma:0"] p_moving_mean = pretrained["model_weights"]["batch_normalization_1"]["batch_normalization_1"]["moving_mean:0"] p_moving_variance = pretrained["model_weights"]["batch_normalization_1"]["batch_normalization_1"]["moving_variance:0"] p_beta[...] = np.array(bl_beta) p_gamma[...] = np.array(bl_gamma) p_moving_mean[...] = np.array(bl_moving_mean) p_moving_variance[...] = np.array(bl_moving_variance) # bn 2 bl_beta = bl["model_weights"]["batch_normalization_2"]["batch_normalization_2"]["beta:0"] bl_gamma = bl["model_weights"]["batch_normalization_2"]["batch_normalization_2"]["gamma:0"] bl_moving_mean = bl["model_weights"]["batch_normalization_2"]["batch_normalization_2"]["moving_mean:0"] bl_moving_variance = bl["model_weights"]["batch_normalization_2"]["batch_normalization_2"]["moving_variance:0"] p_beta = pretrained["model_weights"]["batch_normalization_2"]["batch_normalization_2"]["beta:0"] p_gamma = pretrained["model_weights"]["batch_normalization_2"]["batch_normalization_2"]["gamma:0"] p_moving_mean = pretrained["model_weights"]["batch_normalization_2"]["batch_normalization_2"]["moving_mean:0"] p_moving_variance = pretrained["model_weights"]["batch_normalization_2"]["batch_normalization_2"]["moving_variance:0"] p_beta[...] = np.array(bl_beta) p_gamma[...] = np.array(bl_gamma) p_moving_mean[...] = np.array(bl_moving_mean) p_moving_variance[...] = np.array(bl_moving_variance) # bn 3 bl_beta = bl["model_weights"]["batch_normalization_3"]["batch_normalization_3"]["beta:0"] bl_gamma = bl["model_weights"]["batch_normalization_3"]["batch_normalization_3"]["gamma:0"] bl_moving_mean = bl["model_weights"]["batch_normalization_3"]["batch_normalization_3"]["moving_mean:0"] bl_moving_variance = bl["model_weights"]["batch_normalization_3"]["batch_normalization_3"]["moving_variance:0"] p_beta = pretrained["model_weights"]["batch_normalization_3"]["batch_normalization_3"]["beta:0"] p_gamma = pretrained["model_weights"]["batch_normalization_3"]["batch_normalization_3"]["gamma:0"] p_moving_mean = pretrained["model_weights"]["batch_normalization_3"]["batch_normalization_3"]["moving_mean:0"] p_moving_variance = pretrained["model_weights"]["batch_normalization_3"]["batch_normalization_3"]["moving_variance:0"] p_beta[...] = np.array(bl_beta) p_gamma[...] = np.array(bl_gamma) p_moving_mean[...] = np.array(bl_moving_mean) p_moving_variance[...] = np.array(bl_moving_variance) # bn 4 bl_beta = bl["model_weights"]["batch_normalization_4"]["batch_normalization_4"]["beta:0"] bl_gamma = bl["model_weights"]["batch_normalization_4"]["batch_normalization_4"]["gamma:0"] bl_moving_mean = bl["model_weights"]["batch_normalization_4"]["batch_normalization_4"]["moving_mean:0"] bl_moving_variance = bl["model_weights"]["batch_normalization_4"]["batch_normalization_4"]["moving_variance:0"] p_beta = pretrained["model_weights"]["batch_normalization_4"]["batch_normalization_4"]["beta:0"] p_gamma = pretrained["model_weights"]["batch_normalization_4"]["batch_normalization_4"]["gamma:0"] p_moving_mean = pretrained["model_weights"]["batch_normalization_4"]["batch_normalization_4"]["moving_mean:0"] p_moving_variance = pretrained["model_weights"]["batch_normalization_4"]["batch_normalization_4"]["moving_variance:0"] p_beta[...] = np.array(bl_beta) p_gamma[...] = np.array(bl_gamma) p_moving_mean[...] = np.array(bl_moving_mean) p_moving_variance[...] = np.array(bl_moving_variance) # bn 5 bl_beta = bl["model_weights"]["batch_normalization_5"]["batch_normalization_5"]["beta:0"] bl_gamma = bl["model_weights"]["batch_normalization_5"]["batch_normalization_5"]["gamma:0"] bl_moving_mean = bl["model_weights"]["batch_normalization_5"]["batch_normalization_5"]["moving_mean:0"] bl_moving_variance = bl["model_weights"]["batch_normalization_5"]["batch_normalization_5"]["moving_variance:0"] p_beta = pretrained["model_weights"]["batch_normalization_5"]["batch_normalization_5"]["beta:0"] p_gamma = pretrained["model_weights"]["batch_normalization_5"]["batch_normalization_5"]["gamma:0"] p_moving_mean = pretrained["model_weights"]["batch_normalization_5"]["batch_normalization_5"]["moving_mean:0"] p_moving_variance = pretrained["model_weights"]["batch_normalization_5"]["batch_normalization_5"]["moving_variance:0"] p_beta[...] = np.array(bl_beta) p_gamma[...] = np.array(bl_gamma) p_moving_mean[...] = np.array(bl_moving_mean) p_moving_variance[...] = np.array(bl_moving_variance) pretrained.close()
1953
from Ifc.ClassRegistry import ifc_class, ifc_abstract_class, ifc_fallback_class @ifc_abstract_class class IfcEntity: """ Generic IFC entity, only for subclassing from it """ def __init__(self, rtype, args): """ rtype: Resource type args: Arguments in *reverse* order, so you can just args.pop() from it """ self.rtype = rtype def __str__(self): return self.rtype def __json__(self): return {'rtype': self.rtype} @ifc_fallback_class class IfcGenericEntity(IfcEntity): """ Generic IFC entity: type and args """ def __init__(self, rtype, args): IfcEntity.__init__(self, rtype, args) self.args = args self.args.reverse() def __str__(self): return "Gen<{sup}>{a}".format( sup=IfcEntity.__str__(self), a=self.args) @ifc_class class IfcScalarValue(IfcEntity): def __init__(self, rtype, args): IfcEntity.__init__(self, rtype, args) self.value = args.pop() def __str__(self): return str(self.value) @ifc_class class BOOLEAN(IfcScalarValue): pass @ifc_class class REAL(IfcScalarValue): pass @ifc_class class BINARY(IfcScalarValue): pass @ifc_class class INTEGER(IfcScalarValue): pass @ifc_class class NUMBER(IfcScalarValue): pass @ifc_class class STRING(IfcScalarValue): pass @ifc_class class LOGICAL(IfcScalarValue): pass class Omitted: """ Marked with '*' it states that some supertype had defined that attribute, but in the subtype it is a derived (calculated) value, so it no longer makes sense to explicitely assign value to it. """ # TODO: Haven't tried if it can be handled 'just as expected' def __init__(self): pass def __str__(self): return "<omitted>" def __json__(self): return None # class-level, enough to reference, no need to create multiple instances (doesn't hurt though) omitted = Omitted() class Reference: """ Refers to another entity by its index """ def __init__(self, index): self.index = index def __str__(self): return "<#{idx}>".format(idx=self.index) def __json__(self): return {'ref': self.index} class EnumValue: """ Item from some set of enumerated values. """ def __init__(self, value): self.value = value def __str__(self): return "<.{val}.>".format(val=self.value) def __json__(self): return self.value @ifc_class class STEPHeader(IfcEntity): def __init__(self): IfcEntity.__init__(self, "STEPHeader", []) self.fields = {} def add(self, e): self.fields[e.rtype] = e def __str__(self): return "STEPHeader({f})".format(f=", ".join(map(lambda f: "{n}: {v}".format(n=f[0], v=str(f[1])), self.fields.iteritems()))) # vim: set sw=4 ts=4 et:
1995
from feemodel.app.transient import TransientOnline from feemodel.app.pools import PoolsOnlineEstimator from feemodel.app.predict import Prediction from feemodel.app.simonline import SimOnline __all__ = [ 'TransientOnline', 'PoolsOnlineEstimator', 'Prediction', 'SimOnline' ]
2040
from typing import Dict, List from simulator.services.resources.directory import Directory from simulator.services.services import Services class Atlas(Directory): def __init__(self, services: Services, name: str, parent: str, create: bool = False) -> None: super().__init__(services, name, parent, create) if create: metadata: Dict[str, any] = { "next_index": 0, } self._save_metadata(metadata) def append(self, obj: any) -> None: self.save(str(self._get_next_index()), obj) self._increment_index() def load_all(self, max_els: int = float("inf")) -> List[any]: ret: List[any] = [] idx: int = 0 while idx < max_els: obj: any = self.load(str(idx)) if obj: ret.append(obj) idx += 1 else: break return ret def _get_next_index(self) -> int: metadata: Dict[str, any] = self._get_metadata() return metadata["next_index"] def _increment_index(self) -> None: metadata: Dict[str, any] = self._get_metadata() metadata["next_index"] += 1 self._save_metadata(metadata) def _save_metadata(self, metadata: Dict[str, any]) -> None: super().save("metadata", metadata) def _get_metadata(self) -> Dict[str, any]: return super().load("metadata")
2063
from common.make_tx import make_swap_tx from sol.handle_simple import handle_unknown_detect_transfers def handle_metaplex(exporter, txinfo): transfers_in, transfers_out, _ = txinfo.transfers_net if len(transfers_in) == 1 and len(transfers_out) == 1: sent_amount, sent_currency, _, _ = transfers_out[0] received_amount, received_currency, _, _ = transfers_in[0] row = make_swap_tx(txinfo, sent_amount, sent_currency, received_amount, received_currency) exporter.ingest_row(row) else: handle_unknown_detect_transfers(exporter, txinfo) def is_nft_mint(txinfo): log_instructions = txinfo.log_instructions transfers_in, transfers_out, _ = txinfo.transfers_net if "MintTo" in log_instructions and len(transfers_out) == 1 and len(transfers_in) == 0: return True elif ("MintTo" in log_instructions and len(transfers_out) == 1 and len(transfers_in) == 1 and transfers_in[0][0] == 1): return True else: return False def handle_nft_mint(exporter, txinfo): transfers_in, transfers_out, transfers_unknown = txinfo.transfers_net if len(transfers_in) == 1 and len(transfers_out) == 1: sent_amount, sent_currency, _, _ = transfers_out[0] received_amount, received_currency, _, _ = transfers_in[0] row = make_swap_tx(txinfo, sent_amount, sent_currency, received_amount, received_currency) exporter.ingest_row(row) return handle_unknown_detect_transfers(exporter, txinfo)
2086
from flask import Flask from flask_appconfig import HerokuConfig def create_sample_app(): app = Flask('testapp') HerokuConfig(app) return app def test_herokupostgres(monkeypatch): monkeypatch.setenv('HEROKU_POSTGRESQL_ORANGE_URL', 'heroku-db-uri') app = create_sample_app() assert app.config['SQLALCHEMY_DATABASE_URI'] == 'heroku-db-uri'
2101
import numpy as np from keras import backend as K import os import sys K.set_image_dim_ordering('tf') def patch_path(path): return os.path.join(os.path.dirname(__file__), path) def main(): sys.path.append(patch_path('..')) data_dir_path = patch_path('very_large_data') model_dir_path = patch_path('models/UCF-101') from keras_video_classifier.library.convolutional import CnnVideoClassifier from keras_video_classifier.library.utility.ucf.UCF101_loader import load_ucf, scan_ucf_with_labels config_file_path = CnnVideoClassifier.get_config_file_path(model_dir_path) weight_file_path = CnnVideoClassifier.get_weight_file_path(model_dir_path) np.random.seed(42) load_ucf(data_dir_path) predictor = CnnVideoClassifier() predictor.load_model(config_file_path, weight_file_path) videos = scan_ucf_with_labels(data_dir_path, [label for (label, label_index) in predictor.labels.items()]) video_file_path_list = np.array([file_path for file_path in videos.keys()]) np.random.shuffle(video_file_path_list) for video_file_path in video_file_path_list: label = videos[video_file_path] predicted_label = predictor.predict(video_file_path) print('predicted: ' + predicted_label + ' actual: ' + label) if __name__ == '__main__': main()
2116
import warnings from collections import OrderedDict from distutils.version import LooseVersion from functools import partial from inspect import isclass from typing import Callable, Optional, Dict, Union import numpy as np import torch import tqdm from torch import Tensor, nn from torch.nn import functional as F from adv_lib.distances.lp_norms import l0_distances, l1_distances, l2_distances, linf_distances from adv_lib.utils import ForwardCounter, BackwardCounter, predict_inputs def generate_random_targets(labels: Tensor, num_classes: int) -> Tensor: """ Generates one random target in (num_classes - 1) possibilities for each label that is different from the original label. Parameters ---------- labels: Tensor Original labels. Generated targets will be different from labels. num_classes: int Number of classes to generate the random targets from. Returns ------- targets: Tensor Random target for each label. Has the same shape as labels. """ random = torch.rand(len(labels), num_classes, device=labels.device, dtype=torch.float) random.scatter_(1, labels.unsqueeze(-1), 0) return random.argmax(1) def get_all_targets(labels: Tensor, num_classes: int): """ Generates all possible targets that are different from the original labels. Parameters ---------- labels: Tensor Original labels. Generated targets will be different from labels. num_classes: int Number of classes to generate the random targets from. Returns ------- targets: Tensor Random targets for each label. shape: (len(labels), num_classes - 1). """ all_possible_targets = torch.zeros(len(labels), num_classes - 1, dtype=torch.long) all_classes = set(range(num_classes)) for i in range(len(labels)): this_label = labels[i].item() other_labels = list(all_classes.difference({this_label})) all_possible_targets[i] = torch.tensor(other_labels) return all_possible_targets def run_attack(model: nn.Module, inputs: Tensor, labels: Tensor, attack: Callable, targets: Optional[Tensor] = None, batch_size: Optional[int] = None) -> dict: device = next(model.parameters()).device to_device = lambda tensor: tensor.to(device) targeted, adv_labels = False, labels if targets is not None: targeted, adv_labels = True, targets batch_size = batch_size or len(inputs) # run attack only on non already adversarial samples already_adv = [] chunks = [tensor.split(batch_size) for tensor in [inputs, adv_labels]] for (inputs_chunk, label_chunk) in zip(*chunks): batch_chunk_d, label_chunk_d = [to_device(tensor) for tensor in [inputs_chunk, label_chunk]] preds = model(batch_chunk_d).argmax(1) is_adv = (preds == label_chunk_d) if targeted else (preds != label_chunk_d) already_adv.append(is_adv.cpu()) not_adv = ~torch.cat(already_adv, 0) start, end = torch.cuda.Event(enable_timing=True), torch.cuda.Event(enable_timing=True) forward_counter, backward_counter = ForwardCounter(), BackwardCounter() model.register_forward_pre_hook(forward_counter) if LooseVersion(torch.__version__) >= LooseVersion('1.8'): model.register_full_backward_hook(backward_counter) else: model.register_backward_hook(backward_counter) average_forwards, average_backwards = [], [] # number of forward and backward calls per sample advs_chunks = [] chunks = [tensor.split(batch_size) for tensor in [inputs[not_adv], adv_labels[not_adv]]] total_time = 0 for (inputs_chunk, label_chunk) in tqdm.tqdm(zip(*chunks), ncols=80, total=len(chunks[0])): batch_chunk_d, label_chunk_d = [to_device(tensor.clone()) for tensor in [inputs_chunk, label_chunk]] start.record() advs_chunk_d = attack(model, batch_chunk_d, label_chunk_d, targeted=targeted) # performance monitoring end.record() torch.cuda.synchronize() total_time += (start.elapsed_time(end)) / 1000 # times for cuda Events are in milliseconds average_forwards.append(forward_counter.num_samples_called / len(batch_chunk_d)) average_backwards.append(backward_counter.num_samples_called / len(batch_chunk_d)) forward_counter.reset(), backward_counter.reset() advs_chunks.append(advs_chunk_d.cpu()) if isinstance(attack, partial) and (callback := attack.keywords.get('callback')) is not None: callback.reset_windows() adv_inputs = inputs.clone() adv_inputs[not_adv] = torch.cat(advs_chunks, 0) data = { 'inputs': inputs, 'labels': labels, 'targets': adv_labels if targeted else None, 'adv_inputs': adv_inputs, 'time': total_time, 'num_forwards': sum(average_forwards) / len(chunks[0]), 'num_backwards': sum(average_backwards) / len(chunks[0]), } return data _default_metrics = OrderedDict([ ('linf', linf_distances), ('l0', l0_distances), ('l1', l1_distances), ('l2', l2_distances), ]) def compute_attack_metrics(model: nn.Module, attack_data: Dict[str, Union[Tensor, float]], batch_size: Optional[int] = None, metrics: Dict[str, Callable] = _default_metrics) -> Dict[str, Union[Tensor, float]]: inputs, labels, targets, adv_inputs = map(attack_data.get, ['inputs', 'labels', 'targets', 'adv_inputs']) if adv_inputs.min() < 0 or adv_inputs.max() > 1: warnings.warn('Values of produced adversarials are not in the [0, 1] range -> Clipping to [0, 1].') adv_inputs.clamp_(min=0, max=1) device = next(model.parameters()).device to_device = lambda tensor: tensor.to(device) batch_size = batch_size or len(inputs) chunks = [tensor.split(batch_size) for tensor in [inputs, labels, adv_inputs]] all_predictions = [[] for _ in range(6)] distances = {k: [] for k in metrics.keys()} metrics = {k: v().to(device) if (isclass(v.func) if isinstance(v, partial) else False) else v for k, v in metrics.items()} append = lambda list, data: list.append(data.cpu()) for inputs_chunk, labels_chunk, adv_chunk in zip(*chunks): inputs_chunk, adv_chunk = map(to_device, [inputs_chunk, adv_chunk]) clean_preds, adv_preds = [predict_inputs(model, chunk.to(device)) for chunk in [inputs_chunk, adv_chunk]] list(map(append, all_predictions, [*clean_preds, *adv_preds])) for metric, metric_func in metrics.items(): distances[metric].append(metric_func(adv_chunk, inputs_chunk).detach().cpu()) logits, probs, preds, logits_adv, probs_adv, preds_adv = [torch.cat(l) for l in all_predictions] for metric in metrics.keys(): distances[metric] = torch.cat(distances[metric], 0) accuracy_orig = (preds == labels).float().mean().item() if targets is not None: success = (preds_adv == targets) labels = targets else: success = (preds_adv != labels) prob_orig = probs.gather(1, labels.unsqueeze(1)).squeeze(1) prob_adv = probs_adv.gather(1, labels.unsqueeze(1)).squeeze(1) labels_infhot = torch.zeros_like(logits_adv).scatter_(1, labels.unsqueeze(1), float('inf')) real = logits_adv.gather(1, labels.unsqueeze(1)).squeeze(1) other = (logits_adv - labels_infhot).max(1).values diff_vs_max_adv = (real - other) nll = F.cross_entropy(logits, labels, reduction='none') nll_adv = F.cross_entropy(logits_adv, labels, reduction='none') data = { 'time': attack_data['time'], 'num_forwards': attack_data['num_forwards'], 'num_backwards': attack_data['num_backwards'], 'targeted': targets is not None, 'preds': preds, 'adv_preds': preds_adv, 'accuracy_orig': accuracy_orig, 'success': success, 'probs_orig': prob_orig, 'probs_adv': prob_adv, 'logit_diff_adv': diff_vs_max_adv, 'nll': nll, 'nll_adv': nll_adv, 'distances': distances, } return data def print_metrics(metrics: dict) -> None: np.set_printoptions(formatter={'float': '{:0.3f}'.format}, threshold=16, edgeitems=3, linewidth=120) # To print arrays with less precision print('Original accuracy: {:.2%}'.format(metrics['accuracy_orig'])) print('Attack done in: {:.2f}s with {:.4g} forwards and {:.4g} backwards.'.format( metrics['time'], metrics['num_forwards'], metrics['num_backwards'])) success = metrics['success'].numpy() fail = bool(success.mean() != 1) print('Attack success: {:.2%}'.format(success.mean()) + fail * ' - {}'.format(success)) for distance, values in metrics['distances'].items(): data = values.numpy() print('{}: {} - Average: {:.3f} - Median: {:.3f}'.format(distance, data, data.mean(), np.median(data)) + fail * ' | Avg over success: {:.3f}'.format(data[success].mean())) attack_type = 'targets' if metrics['targeted'] else 'correct' print('Logit({} class) - max_Logit(other classes): {} - Average: {:.2f}'.format( attack_type, metrics['logit_diff_adv'].numpy(), metrics['logit_diff_adv'].numpy().mean())) print('NLL of target/pred class: {:.3f}'.format(metrics['nll_adv'].numpy().mean()))
2135
from dataclasses import dataclass from typing import List from greendoge.types.condition_opcodes import ConditionOpcode from greendoge.util.streamable import Streamable, streamable @dataclass(frozen=True) @streamable class ConditionWithArgs(Streamable): """ This structure is used to store parsed CLVM conditions Conditions in CLVM have either format of (opcode, var1) or (opcode, var1, var2) """ opcode: ConditionOpcode vars: List[bytes]
2141
hiddenimports = ['sip', 'PyQt4.QtGui', 'PyQt4._qt'] from PyInstaller.hooks.hookutils import qt4_plugins_binaries def hook(mod): mod.binaries.extend(qt4_plugins_binaries('phonon_backend')) return mod
2209
import hashlib import mimetypes from urllib.parse import unquote from django.conf import settings from django.core.exceptions import ValidationError from django.db import models from django.http import HttpResponseRedirect from django.template.loader import render_to_string from django.urls import reverse from django.utils.functional import cached_property from django.utils.safestring import mark_safe from django.utils.text import slugify from django.utils.translation import ugettext_lazy as _ from django_extensions.db.fields import CreationDateTimeField, ModificationDateTimeField from great_components.mixins import GA360Mixin from modelcluster.contrib.taggit import ClusterTaggableManager from modelcluster.models import ClusterableModel, ParentalKey from taggit.managers import TaggableManager from taggit.models import ItemBase, TagBase, TaggedItemBase from wagtail.admin.edit_handlers import ( FieldPanel, InlinePanel, MultiFieldPanel, ObjectList, PageChooserPanel, StreamFieldPanel, TabbedInterface, ) from wagtail.contrib.redirects.models import Redirect from wagtail.contrib.settings.models import BaseSetting, register_setting from wagtail.core import blocks from wagtail.core.blocks.stream_block import StreamBlockValidationError from wagtail.core.fields import RichTextField, StreamField from wagtail.core.models import Orderable, Page from wagtail.images import get_image_model_string from wagtail.images.edit_handlers import ImageChooserPanel from wagtail.images.models import AbstractImage, AbstractRendition, Image from wagtail.snippets.models import register_snippet from wagtail.utils.decorators import cached_classmethod from wagtailmedia.models import Media from core import blocks as core_blocks, mixins from core.case_study_index import delete_cs_index, update_cs_index from core.constants import BACKLINK_QUERYSTRING_NAME, RICHTEXT_FEATURES__MINIMAL from core.context import get_context_provider from core.utils import PageTopicHelper, get_first_lesson from exportplan.core.data import ( SECTION_SLUGS as EXPORTPLAN_SLUGS, SECTIONS as EXPORTPLAN_URL_MAP, ) # If we make a Redirect appear as a Snippet, we can sync it via Wagtail-Transfer register_snippet(Redirect) class GreatMedia(Media): transcript = models.TextField( verbose_name=_('Transcript'), blank=False, null=True # left null because was an existing field ) subtitles_en = models.TextField( verbose_name=_('English subtitles'), null=True, blank=True, help_text='English-language subtitles for this video, in VTT format', ) admin_form_fields = Media.admin_form_fields + ( 'transcript', 'subtitles_en', ) @property def sources(self): return [ { 'src': self.url, 'type': mimetypes.guess_type(self.filename)[0] or 'application/octet-stream', 'transcript': self.transcript, } ] @property def subtitles(self): output = [] # TO COME: support for more than just English if self.subtitles_en: output.append( { 'srclang': 'en', 'label': 'English', 'url': reverse('core:subtitles-serve', args=[self.id, 'en']), 'default': False, }, ) return output class AbstractObjectHash(models.Model): class Meta: abstract = True content_hash = models.CharField(max_length=1000) @staticmethod def generate_content_hash(field_file): filehash = hashlib.md5() field_file.open() filehash.update(field_file.read()) field_file.close() return filehash.hexdigest() class DocumentHash(AbstractObjectHash): document = models.ForeignKey( 'wagtaildocs.Document', null=True, blank=True, on_delete=models.CASCADE, related_name='+' ) class ImageHash(AbstractObjectHash): image = models.ForeignKey('wagtailimages.Image', null=True, blank=True, on_delete=models.CASCADE, related_name='+') class AltTextImage(AbstractImage): alt_text = models.CharField(max_length=255, blank=True) admin_form_fields = Image.admin_form_fields + ('alt_text',) class Rendition(AbstractRendition): image = models.ForeignKey(AltTextImage, on_delete=models.CASCADE, related_name='renditions') class Meta: unique_together = ('image', 'filter_spec', 'focal_point_key') @property def alt(self): return self.image.alt_text @register_snippet class Tour(ClusterableModel): page = models.OneToOneField('wagtailcore.Page', on_delete=models.CASCADE, related_name='tour') title = models.CharField(max_length=255) body = models.CharField(max_length=255) button_text = models.CharField(max_length=255) panels = [ PageChooserPanel('page'), FieldPanel('title'), FieldPanel('body'), FieldPanel('button_text'), MultiFieldPanel([InlinePanel('steps')], heading='Steps'), ] def __str__(self): return self.page.title class TourStep(Orderable): title = models.CharField(max_length=255) body = models.CharField(max_length=255) position = models.CharField(max_length=255) selector = models.CharField(max_length=255) tour = ParentalKey(Tour, on_delete=models.CASCADE, related_name='steps') panels = [ FieldPanel('title'), FieldPanel('body'), FieldPanel('position'), FieldPanel('selector'), ] @register_snippet class Product(models.Model): name = models.CharField(max_length=255) panels = [ FieldPanel('name'), ] def __str__(self): return self.name @register_snippet class Region(models.Model): name = models.CharField(max_length=100, unique=True) panels = [FieldPanel('name')] class Meta: ordering = ('name',) def __str__(self): return self.name @register_snippet class Country(models.Model): name = models.CharField(max_length=255) slug = models.SlugField(max_length=100, unique=True) region = models.ForeignKey(Region, null=True, blank=True, on_delete=models.SET_NULL) panels = [ FieldPanel('name'), FieldPanel('region'), ] class Meta: verbose_name_plural = 'Countries' ordering = ('name',) def save(self, *args, **kwargs): # Automatically set slug on save, if not already set if not self.slug: self.slug = slugify(self.name) super().save(*args, **kwargs) def __str__(self): return self.name @register_snippet class Tag(models.Model): name = models.CharField(max_length=100, unique=True) panels = [FieldPanel('name')] class Meta: ordering = ('name',) def __str__(self): return self.name @register_snippet class IndustryTag(models.Model): name = models.CharField(max_length=100, unique=True) icon = models.ForeignKey( AltTextImage, null=True, blank=True, on_delete=models.SET_NULL, related_name='+', ) panels = [FieldPanel('name'), ImageChooserPanel('icon')] class Meta: ordering = ('name',) def __str__(self): return self.name class TimeStampedModel(models.Model): """Modified version of django_extensions.db.models.TimeStampedModel Unfortunately, because null=True needed to be added to create and modified fields, inheritance causes issues with field clash. """ created = CreationDateTimeField('created', null=True) modified = ModificationDateTimeField('modified', null=True) def save(self, **kwargs): self.update_modified = kwargs.pop('update_modified', getattr(self, 'update_modified', True)) super().save(**kwargs) class Meta: get_latest_by = 'modified' ordering = ( '-modified', '-created', ) abstract = True # Content models class CMSGenericPage( mixins.EnableTourMixin, mixins.AuthenticatedUserRequired, mixins.WagtailGA360Mixin, GA360Mixin, Page, ): """ Generic page, freely inspired by Codered page """ class Meta: abstract = True # Do not allow this page type to be created in wagtail admin is_creatable = False template_choices = [] ############### # Layout fields ############### template = models.CharField( max_length=255, choices=None, ) ######### # Panels ########## layout_panels = [FieldPanel('template')] settings_panels = [FieldPanel('slug')] + Page.settings_panels def __init__(self, *args, **kwargs): super().__init__(*args, **kwargs) field = self._meta.get_field('template') field.choices = self.template_choices field.required = True @cached_classmethod def get_edit_handler(cls): # NOQA N805 panels = [ ObjectList(cls.content_panels, heading='Content'), ObjectList(cls.layout_panels, heading='Layout'), ObjectList(cls.settings_panels, heading='Settings', classname='settings'), ] return TabbedInterface(panels).bind_to(model=cls) def get_template(self, request, *args, **kwargs): return self.template def get_context(self, request, *args, **kwargs): context = super().get_context(request) self.set_ga360_payload( page_id=self.id, business_unit=settings.GA360_BUSINESS_UNIT, site_section=str(self.url or '/').split('/')[1], ) self.add_ga360_data_to_payload(request) context['ga360'] = self.ga360_payload provider = get_context_provider(request=request, page=self) if provider: context.update(provider.get_context_data(request=request, page=self)) return context class LandingPage(CMSGenericPage): parent_page_types = [ 'domestic.DomesticHomePage', # TODO: once we've restructured, remove this permission 'domestic.GreatDomesticHomePage', ] subpage_types = [ 'core.ListPage', 'core.InterstitialPage', 'domestic.DomesticDashboard', ] template_choices = ( ('learn/landing_page.html', 'Learn'), ('core/generic_page.html', 'Generic'), ) ################ # Content fields ################ description = RichTextField() button = StreamField([('button', core_blocks.ButtonBlock(icon='cog'))], null=True, blank=True) image = models.ForeignKey( get_image_model_string(), null=True, blank=True, on_delete=models.SET_NULL, related_name='+' ) body = StreamField( [ ('section', core_blocks.SectionBlock()), ('title', core_blocks.TitleBlock()), ('text', blocks.RichTextBlock(icon='openquote', helptext='Add a textblock')), ('image', core_blocks.ImageBlock()), ], null=True, blank=True, ) components = StreamField( [ ('route', core_blocks.RouteSectionBlock()), ], null=True, blank=True, ) ######### # Panels ######### content_panels = CMSGenericPage.content_panels + [ FieldPanel('description'), StreamFieldPanel('button'), ImageChooserPanel('image'), StreamFieldPanel('components'), StreamFieldPanel('body'), ] class InterstitialPage(CMSGenericPage): parent_page_types = ['core.LandingPage'] template_choices = (('learn/interstitial.html', 'Learn'),) ################ # Content fields ################ button = StreamField([('button', core_blocks.ButtonBlock(icon='cog'))], null=True, blank=True) ######### # Panels ######### content_panels = CMSGenericPage.content_panels + [ StreamFieldPanel('button'), ] class ListPage(CMSGenericPage): parent_page_types = ['core.LandingPage'] subpage_types = ['core.CuratedListPage'] template_choices = (('learn/automated_list_page.html', 'Learn'),) record_read_progress = models.BooleanField( default=False, help_text='Should we record when a user views a page in this collection?', ) class Meta: verbose_name = 'Automated list page' verbose_name_plural = 'Automated list pages' def get_context(self, request, *args, **kwargs): from core.helpers import get_high_level_completion_progress from domestic.helpers import get_lesson_completion_status context = super().get_context(request) if request.user.is_authenticated: completion_status = get_lesson_completion_status(request.user) context['high_level_completion_progress'] = get_high_level_completion_progress( completion_status=completion_status, ) return context ################ # Content fields ################ description = RichTextField() button_label = models.CharField(max_length=100) ######### # Panels ######### settings_panels = CMSGenericPage.settings_panels + [FieldPanel('record_read_progress')] content_panels = CMSGenericPage.content_panels + [FieldPanel('description'), FieldPanel('button_label')] class CuratedListPage(CMSGenericPage): parent_page_types = ['core.ListPage'] subpage_types = [ 'core.TopicPage', ] template_choices = (('learn/curated_list_page.html', 'Learn'),) ################ # Content fields ################ heading = RichTextField() image = models.ForeignKey( get_image_model_string(), null=True, blank=True, on_delete=models.SET_NULL, related_name='+' ) ######## # Panels ######## content_panels = CMSGenericPage.content_panels + [ FieldPanel('heading'), ImageChooserPanel('image'), ] def get_topics(self, live=True) -> models.QuerySet: qs = TopicPage.objects.live().specific().descendant_of(self) if live: qs = qs.live() return qs @cached_property def count_topics(self): return self.get_topics().count() @cached_property def count_detail_pages(self): count = 0 for topic in self.get_topics(): count += DetailPage.objects.live().descendant_of(topic).count() return count def get_context(self, request, *args, **kwargs): from core.helpers import ( get_high_level_completion_progress, get_module_completion_progress, ) from domestic.helpers import get_lesson_completion_status context = super().get_context(request) # Give the template a simple way to link back to the parent # learning module (ListPage) context['parent_page_url'] = self.get_parent().url if request.user.is_authenticated: # get this once, so we don't waste the network call to get the data twice completion_status = get_lesson_completion_status(request.user) context['module_completion_progress'] = get_module_completion_progress( completion_status=completion_status, module_page=self, ) context['high_level_completion_progress'] = get_high_level_completion_progress( completion_status=completion_status, ) return context def hero_singular_validation(value): if value and len(value) > 1: raise StreamBlockValidationError( non_block_errors=ValidationError('Only one image or video allowed in Hero section', code='invalid'), ) class TopicPage(mixins.AuthenticatedUserRequired, Page): """Structural page to allow for cleaner mapping of lessons (`DetailPage`s) to modules (`CuratedListPage`s). Not intented to be viewed by end users, so will redirect to the parent module if accessed. Also, for the above reason, mixins.WagtailGA360Mixin and GA360Mixin are not used.""" parent_page_types = ['core.CuratedListPage'] subpage_types = [ 'core.DetailPage', 'core.LessonPlaceholderPage', ] # `title` comes from Page superclass and that's all we need here def _redirect_to_parent_module(self): return HttpResponseRedirect(self.get_parent().url) def serve_preview(self, request, mode_name='dummy'): # It doesn't matter what is passed as mode_name - we always redirect return self._redirect_to_parent_module() def serve(self, request): return self._redirect_to_parent_module() class LessonPlaceholderPage(mixins.AuthenticatedUserRequired, Page): """Structural page to allow for configuring and representing very simple to modules (`CuratedListPage`s). Not intented to be viewed by end users, so will redirect to the parent module if accessed. Also, for the above reason, mixins.WagtailGA360Mixin and GA360Mixin are not used.""" parent_page_types = ['core.TopicPage'] subpage_types = [] # No child pages allowed for placeholders # `title` comes from Page superclass and that's all we need here def _redirect_to_parent_module(self): dest = CuratedListPage.objects.ancestor_of(self).first().url return HttpResponseRedirect(dest) def serve_preview(self, request, mode_name='dummy'): # It doesn't matter what is passed as mode_name - we always redirect return self._redirect_to_parent_module() def serve(self, request): return self._redirect_to_parent_module() class DetailPage(CMSGenericPage): estimated_read_duration = models.DurationField(null=True, blank=True) parent_page_types = [ 'core.CuratedListPage', # TEMPORARY: remove after topics refactor migration has run 'core.TopicPage', ] template_choices = (('learn/detail_page.html', 'Learn'),) class Meta: verbose_name = 'Detail page' verbose_name_plural = 'Detail pages' ################ # Content fields ################ hero = StreamField( [ ('Image', core_blocks.ImageBlock(template='core/includes/_hero_image.html')), ('Video', core_blocks.SimpleVideoBlock(template='core/includes/_hero_video.html')), ], null=True, blank=True, validators=[hero_singular_validation], ) objective = StreamField( [ ( 'paragraph', blocks.RichTextBlock(options={'class': 'objectives'}), ), ('ListItem', core_blocks.Item()), ] ) body = StreamField( [ ( 'paragraph', blocks.StructBlock( [('paragraph', blocks.RichTextBlock())], template='core/struct_paragraph_block.html', icon='fa-font', ), ), ( 'video', blocks.StructBlock( [('video', core_blocks.VideoBlock())], template='core/struct_video_block.html', icon='fa-play', ), ), ('case_study', core_blocks.CaseStudyStaticBlock(icon='fa-book')), ( 'Step', core_blocks.StepByStepBlock(icon='cog'), ), ( 'fictional_example', blocks.StructBlock( [('fiction_body', blocks.RichTextBlock(icon='openquote'))], template='learn/fictional_company_example.html', icon='fa-commenting-o', ), ), ( 'ITA_Quote', core_blocks.ITAQuoteBlock(icon='fa-quote-left'), ), ( 'pros_cons', blocks.StructBlock( [ ( 'pros', blocks.StreamBlock( [ ( 'item', core_blocks.Item(icon='fa-arrow-right'), ) ] ), ), ( 'cons', blocks.StreamBlock( [ ( 'item', core_blocks.Item(icon='fa-arrow-right'), ) ] ), ), ], template='learn/pros_and_cons.html', icon='fa-arrow-right', ), ), ('choose_do_not_choose', core_blocks.ChooseDoNotChooseBlock()), ( 'image', core_blocks.ImageBlock( template='core/includes/_image_full_width.html', help_text='Image displayed within a full-page-width block', ), ), ( 'video', core_blocks.SimpleVideoBlock( template='core/includes/_video_full_width.html', help_text='Video displayed within a full-page-width block', ), ), ] ) recap = StreamField( [ ( 'recap_item', blocks.StructBlock( [ ('title', blocks.CharBlock(icon='fa-header')), ( 'item', blocks.StreamBlock( [ ( 'item', core_blocks.Item(), ) ] ), ), ], template='learn/recap.html', icon='fa-commenting-o', ), ) ] ) ######### # Panels ########## content_panels = Page.content_panels + [ StreamFieldPanel('hero'), StreamFieldPanel('objective'), StreamFieldPanel('body'), StreamFieldPanel('recap'), ] def handle_page_view(self, request): if request.user.is_authenticated: # checking if the page should record read progress # checking if the page is already marked as read list_page = ( ListPage.objects.ancestor_of(self) .filter(record_read_progress=True) .exclude(page_views_list__sso_id=request.user.pk, page_views_list__page=self) .first() ) if list_page: PageView.objects.get_or_create( page=self, list_page=list_page, sso_id=request.user.pk, ) def serve(self, request, *args, **kwargs): self.handle_page_view(request) return super().serve(request, **kwargs) @cached_property def topic_title(self): return self.get_parent().title @cached_property def module(self): """Gets the learning module this lesson belongs to""" return CuratedListPage.objects.live().specific().ancestor_of(self).first() @cached_property def _export_plan_url_map(self): """Return a lookup dictionary of URL Slugs->title for all the Export Plan sections we have.""" return {url: values['title'] for url, values in EXPORTPLAN_URL_MAP.items()} def _get_backlink(self, request): """Try to extract a backlink (used for a link to the export plan) from the querystring on the request that brought us to this view. Only accepts backlinks that we KNOW are for the export plan, else ignore it.""" backlink_path = request.GET.get(BACKLINK_QUERYSTRING_NAME, '') if backlink_path is not None: backlink_path = unquote(backlink_path) if len(backlink_path.split('/')) > 2 and ( backlink_path.split('/')[3] in EXPORTPLAN_SLUGS and '://' not in backlink_path ): # The check for '://' will stop us accepting a backlink which # features a full URL as its OWN querystring param (eg a crafted attack # URL), but that's an acceptable limitation here and is very unlikely # to happen. return backlink_path return None # safe default def _get_backlink_title(self, backlink_path): """For a given backlink, see if we can get a title that goes with it. For now, this is limited only to Export Plan pages/links. """ # We have to re-arrange EXPORT_PLAN_SECTION_TITLES_URLS after import # because it features lazily-evaluated URLs that aren't ready when # models are imported if backlink_path and len(backlink_path.split('/')) > 3: _path = backlink_path.split('/')[3] return self._export_plan_url_map.get(_path) def get_context(self, request, *args, **kwargs): context = super().get_context(request) context['refresh_on_market_change'] = True # Prepare backlink to the export plan if we detect one and can validate it _backlink = self._get_backlink(request) if _backlink: context['backlink'] = _backlink context['backlink_title'] = self._get_backlink_title(_backlink) if isinstance(self.get_parent().specific, TopicPage): # In a conditional because a DetailPage currently MAY be used as # a child of another page type... page_topic_helper = PageTopicHelper(self) next_lesson = page_topic_helper.get_next_lesson() context['current_lesson'] = self context['current_module'] = page_topic_helper.module if page_topic_helper: topic_page = page_topic_helper.get_page_topic() if topic_page: context['current_topic'] = topic_page context['page_topic'] = topic_page.title if next_lesson: context['next_lesson'] = next_lesson else: next_module = self.module.get_next_sibling() if not next_module: return context context['next_module'] = next_module.specific context['next_lesson'] = get_first_lesson(next_module) return context class PageView(TimeStampedModel): page = models.ForeignKey(DetailPage, on_delete=models.CASCADE, related_name='page_views') list_page = models.ForeignKey(ListPage, on_delete=models.CASCADE, related_name='page_views_list') sso_id = models.TextField() class Meta: ordering = ['page__pk'] unique_together = ['page', 'sso_id'] # TODO: deprecate and remove class ContentModuleTag(TaggedItemBase): content_object = ParentalKey('core.ContentModule', on_delete=models.CASCADE, related_name='tagged_items') # TODO: deprecate and remove @register_snippet class ContentModule(ClusterableModel): title = models.CharField(max_length=255) content = RichTextField() tags = TaggableManager(through=ContentModuleTag, blank=True) panels = [ FieldPanel('title'), FieldPanel('content'), FieldPanel('tags'), ] def __str__(self): return self.title class PersonalisationHSCodeTag(TagBase): """Custom tag for personalisation. Tag value will be a HS6, HS4 or HS2 code""" # free_tagging = False # DISABLED until tag data only comes via data migration class Meta: verbose_name = 'HS Code tag for personalisation' verbose_name_plural = 'HS Code tags for personalisation' class PersonalisationCountryTag(TagBase): """Custom tag for personalisation. Tag value will be an ISO-2 Country code ('DE') """ free_tagging = False class Meta: verbose_name = 'Country tag for personalisation' verbose_name_plural = 'Country tags for personalisation' class PersonalisationRegionTag(TagBase): """Custom tag for personalisation. Tag value will be a geographical string ('Europe') """ free_tagging = False class Meta: verbose_name = 'Region tag for personalisation' verbose_name_plural = 'Region tags for personalisation' class PersonalisationTradingBlocTag(TagBase): """Custom tag for personalisation. Tag value will be an Trading blocs """ free_tagging = False class Meta: verbose_name = 'Trading bloc tag for personalisation' verbose_name_plural = 'Trading bloc tags for personalisation' # If you're wondering what's going on here: # https://docs.wagtail.io/en/stable/reference/pages/model_recipes.html#custom-tag-models class HSCodeTaggedCaseStudy(ItemBase): tag = models.ForeignKey( PersonalisationHSCodeTag, related_name='hscode_tagged_case_studies', on_delete=models.CASCADE ) content_object = ParentalKey(to='core.CaseStudy', on_delete=models.CASCADE, related_name='hs_code_tagged_items') class CountryTaggedCaseStudy(ItemBase): tag = models.ForeignKey( PersonalisationCountryTag, related_name='country_tagged_case_studies', on_delete=models.CASCADE ) content_object = ParentalKey(to='core.CaseStudy', on_delete=models.CASCADE, related_name='country_tagged_items') class RegionTaggedCaseStudy(ItemBase): tag = models.ForeignKey( PersonalisationRegionTag, related_name='region_tagged_case_studies', on_delete=models.CASCADE ) content_object = ParentalKey(to='core.CaseStudy', on_delete=models.CASCADE, related_name='region_tagged_items') class TradingBlocTaggedCaseStudy(ItemBase): tag = models.ForeignKey( PersonalisationTradingBlocTag, related_name='trading_bloc_tagged_case_studies', on_delete=models.CASCADE ) content_object = ParentalKey( to='core.CaseStudy', on_delete=models.CASCADE, related_name='trading_bloc_tagged_items' ) def _high_level_validation(value, error_messages): TEXT_BLOCK = 'text' # noqa N806 MEDIA_BLOCK = 'media' # noqa N806 QUOTE_BLOCK = 'quote' # noqa N806 # we need to be strict about presence and ordering of these nodes if [node.block_type for node in value if node.block_type != QUOTE_BLOCK] != [MEDIA_BLOCK, TEXT_BLOCK]: error_messages.append( ( 'This block must contain one Media section (with one or ' 'two items in it) and/or a Quote section, then one Text section following it.' ) ) return error_messages def _low_level_validation(value, error_messages): # Check content of media node, which should be present here MEDIA_BLOCK = 'media' # noqa N806 VIDEO_BLOCK = 'video' # noqa N806 for node in value: if node.block_type == MEDIA_BLOCK: subnode_block_types = [subnode.block_type for subnode in node.value] if len(subnode_block_types) == 2: if set(subnode_block_types) == {VIDEO_BLOCK}: # Two videos: not allowed error_messages.append('Only one video may be used in a case study.') elif subnode_block_types[1] == VIDEO_BLOCK: # implicitly, [0] must be an image # video after image: not allowed error_messages.append('The video must come before a still image.') return error_messages def case_study_body_validation(value): """Ensure the case study has exactly both a media node and a text node and that the media node has the following content: * One image, only * One video, only * One video + One image * (video must comes first so that it is displayed first) * Two images """ error_messages = [] if value: error_messages = _high_level_validation(value, error_messages) error_messages = _low_level_validation(value, error_messages) if error_messages: raise StreamBlockValidationError( non_block_errors=ValidationError('; '.join(error_messages), code='invalid'), ) class MagnaPageChooserPanel(PageChooserPanel): show_label = False field_template = 'admin/wagtailadmin/edit_handlers/field_panel_field.html' def render_as_field(self): instance_obj = self.get_chosen_item() context = { 'field': self.bound_field, self.object_type_name: instance_obj, 'is_chosen': bool(instance_obj), # DEPRECATED - passed to templates for backwards compatibility only # Added obj_type on base class method render_as_field 'obj_type': instance_obj.specific.__class__.__name__ if instance_obj else None, } return mark_safe(render_to_string(self.field_template, context)) class CaseStudyRelatedPages(Orderable): case_study = ParentalKey( 'core.CaseStudy', related_name='related_pages', on_delete=models.SET_NULL, null=True, blank=True, ) page = models.ForeignKey( 'wagtailcore.Page', on_delete=models.CASCADE, related_name='+', ) panels = [ MagnaPageChooserPanel('page', [DetailPage, CuratedListPage, TopicPage]), ] class Meta: unique_together = ['case_study', 'page'] @register_snippet class CaseStudy(ClusterableModel): """Dedicated snippet for use as a case study. Supports personalised selection via its tags. The decision about the appropriate Case Study block to show will happen when the page attempts to render the relevant CaseStudyBlock. Note that this is rendered via Wagtail's ModelAdmin, so appears in the sidebar, but we have to keep it registered as a Snippet to be able to transfer it with Wagtail-Transfer """ title = models.CharField( max_length=255, blank=False, verbose_name='Internal case study title', ) # old name company_name summary_context = models.CharField(max_length=255, blank=False, default='How we did it') # old name summary lead_title = models.TextField(blank=False) # Deliberately not rich-text / no formatting body = StreamField( [ ( 'media', blocks.StreamBlock( [ ('video', core_blocks.SimpleVideoBlock(template='core/includes/_case_study_video.html')), ('image', core_blocks.ImageBlock()), ], min_num=1, max_num=2, ), ), ( 'text', blocks.RichTextBlock( features=RICHTEXT_FEATURES__MINIMAL, ), ), ( 'quote', core_blocks.CaseStudyQuoteBlock(), ), ], validators=[case_study_body_validation], help_text=( 'This block must contain one Media section (with one or two items in it) ' 'and/or Quote sections, then one Text section.' ), ) # We are keeping the personalisation-relevant tags in separate # fields to aid lookup and make the UX easier for editors hs_code_tags = ClusterTaggableManager(through='core.HSCodeTaggedCaseStudy', blank=True, verbose_name='HS-code tags') country_code_tags = ClusterTaggableManager( through='core.CountryTaggedCaseStudy', blank=True, verbose_name='Country tags' ) region_code_tags = ClusterTaggableManager( through='core.RegionTaggedCaseStudy', blank=True, verbose_name='Region tags' ) trading_bloc_code_tags = ClusterTaggableManager( through='core.TradingBlocTaggedCaseStudy', blank=True, verbose_name='Trading bloc tags' ) created = CreationDateTimeField('created', null=True) modified = ModificationDateTimeField('modified', null=True) panels = [ MultiFieldPanel( [ FieldPanel('title'), FieldPanel('lead_title'), FieldPanel('summary_context'), StreamFieldPanel('body'), ], heading='Case Study content', ), MultiFieldPanel( [ FieldPanel('hs_code_tags'), FieldPanel('country_code_tags'), FieldPanel('region_code_tags'), FieldPanel('trading_bloc_code_tags'), ], heading='Case Study tags for Personalisation', ), MultiFieldPanel( [ InlinePanel('related_pages', label='Related pages'), ], heading='Related Lesson, Topic & Module, also used for Personalisation', ), ] def __str__(self): display_name = self.title if self.title else self.summary_context return f'{display_name}' def save(self, **kwargs): # When we create a new CS need to call create to obtain an ID for indexing self.update_modified = kwargs.pop('update_modified', getattr(self, 'update_modified', True)) super().save(**kwargs) update_cs_index(self) def delete(self, **kwargs): delete_cs_index(self.id) super().delete(**kwargs) def get_cms_standalone_view_url(self): return reverse('cms_extras:case-study-view', args=[self.id]) class Meta: verbose_name_plural = 'Case studies' get_latest_by = 'modified' ordering = ( '-modified', '-created', ) @register_setting class CaseStudyScoringSettings(BaseSetting): threshold = models.DecimalField( help_text='This is the minimum score which a case study needs to have to be ' 'considered before being presented to users. ', default=10, decimal_places=3, max_digits=5, ) lesson = models.DecimalField( help_text="Score given when user's lesson is tagged in the case study.", default=8, decimal_places=3, max_digits=5, ) topic = models.DecimalField( help_text="Score given when user's lesson's topic is tagged in the case study " 'unless there is also lesson match.', default=4, decimal_places=3, max_digits=5, ) module = models.DecimalField( help_text="Score given when the user's lesson's module is tagged in the case study " 'unless there is also lesson or topic match.', default=2, decimal_places=3, max_digits=5, ) product_hs6 = models.DecimalField( help_text='Score given when any case study HS6 tag matches the complete HS6 code of ' "any of the user's products", default=8, decimal_places=3, max_digits=5, ) product_hs4 = models.DecimalField( help_text="Score given when any case study HS4 tag matches the first 4 digits of any of the user's products " 'unless there is an HS6 match.', default=4, decimal_places=3, max_digits=5, ) product_hs2 = models.DecimalField( help_text="Score given when any case study HS2 tag matches the first 2 digits of any of the user's products " 'unless there is an HS6 or HS4 match.', default=2, decimal_places=3, max_digits=5, ) country_exact = models.DecimalField( help_text="Score given when any case study country tag exactly matches one of the user's export markets.", default=4, decimal_places=3, max_digits=5, ) country_region = models.DecimalField( help_text="Score given when any case study region tag matches the region of any of the user's export markets " 'unless there is an exact country match.', default=2, decimal_places=3, max_digits=5, ) trading_blocs = models.DecimalField( help_text='Score given when any case study trading bloc tag matches the any trading bloc that any of ' "the user's export markets falls into unless there is an exact country or region match.", default=2, decimal_places=3, max_digits=5, ) product_tab = [MultiFieldPanel([FieldPanel('product_hs6'), FieldPanel('product_hs4'), FieldPanel('product_hs2')])] market_tab = [ MultiFieldPanel([FieldPanel('country_exact'), FieldPanel('country_region'), FieldPanel('trading_blocs')]) ] lesson_tab = [MultiFieldPanel([FieldPanel('lesson'), FieldPanel('topic'), FieldPanel('module')])] threshold_tab = [ MultiFieldPanel( [ FieldPanel('threshold'), ] ) ] edit_handler = TabbedInterface( [ ObjectList(product_tab, heading='Product'), ObjectList(market_tab, heading='Market'), ObjectList(lesson_tab, heading='Lesson'), ObjectList(threshold_tab, heading='Threshold'), ] ) class Meta: verbose_name = 'Case Study Scoring'
2253
import pytest import numpy as np from numpy.testing import assert_allclose from keras import backend as K from keras import activations def get_standard_values(): ''' These are just a set of floats used for testing the activation functions, and are useful in multiple tests. ''' return np.array([[0, 0.1, 0.5, 0.9, 1.0]], dtype=K.floatx()) def test_softmax(): ''' Test using a reference implementation of softmax ''' def softmax(values): m = np.max(values) e = np.exp(values - m) return e / np.sum(e) x = K.placeholder(ndim=2) f = K.function([x], [activations.softmax(x)]) test_values = get_standard_values() result = f([test_values])[0] expected = softmax(test_values) assert_allclose(result, expected, rtol=1e-05) def test_time_distributed_softmax(): x = K.placeholder(shape=(1, 1, 5)) f = K.function([x], [activations.softmax(x)]) test_values = get_standard_values() test_values = np.reshape(test_values, (1, 1, np.size(test_values))) f([test_values])[0] def test_softplus(): ''' Test using a reference softplus implementation ''' def softplus(x): return np.log(np.ones_like(x) + np.exp(x)) x = K.placeholder(ndim=2) f = K.function([x], [activations.softplus(x)]) test_values = get_standard_values() result = f([test_values])[0] expected = softplus(test_values) assert_allclose(result, expected, rtol=1e-05) def test_softsign(): ''' Test using a reference softsign implementation ''' def softsign(x): return np.divide(x, np.ones_like(x) + np.absolute(x)) x = K.placeholder(ndim=2) f = K.function([x], [activations.softsign(x)]) test_values = get_standard_values() result = f([test_values])[0] expected = softsign(test_values) assert_allclose(result, expected, rtol=1e-05) def test_sigmoid(): ''' Test using a numerically stable reference sigmoid implementation ''' def ref_sigmoid(x): if x >= 0: return 1 / (1 + np.exp(-x)) else: z = np.exp(x) return z / (1 + z) sigmoid = np.vectorize(ref_sigmoid) x = K.placeholder(ndim=2) f = K.function([x], [activations.sigmoid(x)]) test_values = get_standard_values() result = f([test_values])[0] expected = sigmoid(test_values) assert_allclose(result, expected, rtol=1e-05) def test_hard_sigmoid(): ''' Test using a reference hard sigmoid implementation ''' def ref_hard_sigmoid(x): ''' Reference hard sigmoid with slope and shift values from theano, see https://github.com/Theano/Theano/blob/master/theano/tensor/nnet/sigm.py ''' x = (x * 0.2) + 0.5 z = 0.0 if x <= 0 else (1.0 if x >= 1 else x) return z hard_sigmoid = np.vectorize(ref_hard_sigmoid) x = K.placeholder(ndim=2) f = K.function([x], [activations.hard_sigmoid(x)]) test_values = get_standard_values() result = f([test_values])[0] expected = hard_sigmoid(test_values) assert_allclose(result, expected, rtol=1e-05) def test_relu(): ''' Relu implementation doesn't depend on the value being a theano variable. Testing ints, floats and theano tensors. ''' x = K.placeholder(ndim=2) f = K.function([x], [activations.relu(x)]) test_values = get_standard_values() result = f([test_values])[0] # because no negatives in test values assert_allclose(result, test_values, rtol=1e-05) def test_elu(): x = K.placeholder(ndim=2) f = K.function([x], [activations.elu(x, 0.5)]) test_values = get_standard_values() result = f([test_values])[0] # because no negatives in test values assert_allclose(result, test_values, rtol=1e-05) negative_values = np.array([[-1, -2]], dtype=K.floatx()) result = f([negative_values])[0] true_result = (np.exp(negative_values) - 1) / 2 assert_allclose(result, true_result) def test_tanh(): test_values = get_standard_values() x = K.placeholder(ndim=2) exp = activations.tanh(x) f = K.function([x], [exp]) result = f([test_values])[0] expected = np.tanh(test_values) assert_allclose(result, expected, rtol=1e-05) def test_linear(): ''' This function does no input validation, it just returns the thing that was passed in. ''' xs = [1, 5, True, None, 'foo'] for x in xs: assert(x == activations.linear(x)) if __name__ == '__main__': pytest.main([__file__])
2280
from arcapix.fs.gpfs.policy import PlacementPolicy from arcapix.fs.gpfs.rule import MigrateRule # load placement policy for mmfs1 policy = PlacementPolicy('mmfs1') # create a new migrate rule for 'sata1' r = MigrateRule(source='sata1', threshold=(90, 50)) # add rule to start of the policy policy.rules.insert(r, 0) # save changes policy.save()
2285
from allennlp.common import JsonDict from allennlp.data import DatasetReader, Instance from allennlp.models import Model from allennlp.predictors import Predictor from overrides import overrides @Predictor.register("sentence_classifier") class SentenceClassifierPredictor(Predictor): def predict(self, sentence: str) -> JsonDict: return self.predict_json({"sentence": sentence}) @overrides def _json_to_instance(self, json_dict: JsonDict) -> Instance: sentence = json_dict["sentence"] return self._dataset_reader.text_to_instance(sentence)
2298
from __future__ import print_function from scipy.linalg import block_diag from scipy.stats import norm as ndist from scipy.interpolate import interp1d import collections import numpy as np from numpy import log from numpy.linalg import norm, qr, inv, eig import pandas as pd import regreg.api as rr from .randomization import randomization from ..base import restricted_estimator from ..algorithms.barrier_affine import solve_barrier_affine_py as solver from ..distributions.discrete_family import discrete_family class group_lasso(object): def __init__(self, loglike, groups, weights, ridge_term, randomizer, use_lasso=True, # should lasso solver be used where applicable - defaults to True perturb=None): _check_groups(groups) # make sure groups looks sensible # log likelihood : quadratic loss self.loglike = loglike self.nfeature = self.loglike.shape[0] # ridge parameter self.ridge_term = ridge_term # group lasso penalty (from regreg) # use regular lasso penalty if all groups are size 1 if use_lasso and groups.size == np.unique(groups).size: # need to provide weights an an np.array rather than a dictionary weights_np = np.array([w[1] for w in sorted(weights.items())]) self.penalty = rr.weighted_l1norm(weights=weights_np, lagrange=1.) else: self.penalty = rr.group_lasso(groups, weights=weights, lagrange=1.) # store groups as a class variable since the non-group lasso doesn't self.groups = groups self._initial_omega = perturb # gaussian randomization self.randomizer = randomizer def fit(self, solve_args={'tol': 1.e-12, 'min_its': 50}, perturb=None): # solve the randomized version of group lasso (self.initial_soln, self.initial_subgrad) = self._solve_randomized_problem(perturb=perturb, solve_args=solve_args) # initialize variables active_groups = [] # active group labels active_dirs = {} # dictionary: keys are group labels, values are unit-norm coefficients unpenalized = [] # selected groups with no penalty overall = np.ones(self.nfeature, np.bool) # mask of active features ordered_groups = [] # active group labels sorted by label ordered_opt = [] # gamma's ordered by group labels ordered_vars = [] # indices "ordered" by sorting group labels tol = 1.e-20 _, self.randomizer_prec = self.randomizer.cov_prec # now we are collecting the directions and norms of the active groups for g in sorted(np.unique(self.groups)): # g is group label group_mask = self.groups == g soln = self.initial_soln # do not need to keep setting this if norm(soln[group_mask]) > tol * norm(soln): # is group g appreciably nonzero ordered_groups.append(g) # variables in active group ordered_vars.extend(np.flatnonzero(group_mask)) if self.penalty.weights[g] == 0: unpenalized.append(g) else: active_groups.append(g) active_dirs[g] = soln[group_mask] / norm(soln[group_mask]) ordered_opt.append(norm(soln[group_mask])) else: overall[group_mask] = False self.selection_variable = {'directions': active_dirs, 'active_groups': active_groups} # kind of redundant with keys of active_dirs self._ordered_groups = ordered_groups # exception if no groups are selected if len(self.selection_variable['active_groups']) == 0: return np.sign(soln), soln # otherwise continue as before self.observed_opt_state = np.hstack(ordered_opt) # gammas as array _beta_unpenalized = restricted_estimator(self.loglike, # refit OLS on E overall, solve_args=solve_args) beta_bar = np.zeros(self.nfeature) beta_bar[overall] = _beta_unpenalized # refit OLS beta with zeros self._beta_full = beta_bar X, y = self.loglike.data W = self._W = self.loglike.saturated_loss.hessian(X.dot(beta_bar)) # all 1's for LS opt_linearNoU = np.dot(X.T, X[:, ordered_vars] * W[:, np.newaxis]) for i, var in enumerate(ordered_vars): opt_linearNoU[var, i] += self.ridge_term opt_offset = self.initial_subgrad self.observed_score_state = -opt_linearNoU.dot(_beta_unpenalized) self.observed_score_state[~overall] += self.loglike.smooth_objective(beta_bar, 'grad')[~overall] active_signs = np.sign(self.initial_soln) active = np.flatnonzero(active_signs) self.active = active def compute_Vg(ug): pg = ug.size # figure out size of g'th group if pg > 1: Z = np.column_stack((ug, np.eye(pg, pg - 1))) Q, _ = qr(Z) Vg = Q[:, 1:] # drop the first column else: Vg = np.zeros((1, 0)) # if the group is size one, the orthogonal complement is empty return Vg def compute_Lg(g): pg = active_dirs[g].size Lg = self.penalty.weights[g] * np.eye(pg) return Lg sorted_active_dirs = collections.OrderedDict(sorted(active_dirs.items())) Vs = [compute_Vg(ug) for ug in sorted_active_dirs.values()] V = block_diag(*Vs) # unpack the list Ls = [compute_Lg(g) for g in sorted_active_dirs] L = block_diag(*Ls) # unpack the list XE = X[:, ordered_vars] # changed to ordered_vars Q = XE.T.dot(self._W[:, None] * XE) QI = inv(Q) C = V.T.dot(QI).dot(L).dot(V) self.XE = XE self.Q = Q self.QI = QI self.C = C U = block_diag(*[ug for ug in sorted_active_dirs.values()]).T self.opt_linear = opt_linearNoU.dot(U) self.active_dirs = active_dirs self.opt_offset = opt_offset self.ordered_vars = ordered_vars self.linear_part = -np.eye(self.observed_opt_state.shape[0]) self.offset = np.zeros(self.observed_opt_state.shape[0]) return active_signs, soln def _solve_randomized_problem(self, perturb=None, solve_args={'tol': 1.e-15, 'min_its': 100}): # take a new perturbation if supplied if perturb is not None: self._initial_omega = perturb if self._initial_omega is None: self._initial_omega = self.randomizer.sample() quad = rr.identity_quadratic(self.ridge_term, 0, -self._initial_omega, 0) problem = rr.simple_problem(self.loglike, self.penalty) # if all groups are size 1, set up lasso penalty and run usual lasso solver... (see existing code)... initial_soln = problem.solve(quad, **solve_args) initial_subgrad = -(self.loglike.smooth_objective(initial_soln, 'grad') + quad.objective(initial_soln, 'grad')) return initial_soln, initial_subgrad @staticmethod def gaussian(X, Y, groups, weights, sigma=1., quadratic=None, ridge_term=0., perturb=None, use_lasso=True, # should lasso solver be used when applicable - defaults to True randomizer_scale=None): loglike = rr.glm.gaussian(X, Y, coef=1. / sigma ** 2, quadratic=quadratic) n, p = X.shape mean_diag = np.mean((X ** 2).sum(0)) if ridge_term is None: ridge_term = np.std(Y) * np.sqrt(mean_diag) / np.sqrt(n - 1) if randomizer_scale is None: randomizer_scale = np.sqrt(mean_diag) * 0.5 * np.std(Y) * np.sqrt(n / (n - 1.)) randomizer = randomization.isotropic_gaussian((p,), randomizer_scale) return group_lasso(loglike, groups, weights, ridge_term, randomizer, use_lasso, perturb) def _setup_implied_gaussian(self): _, prec = self.randomizer.cov_prec if np.asarray(prec).shape in [(), (0,)]: cond_precision = self.opt_linear.T.dot(self.opt_linear) * prec cond_cov = inv(cond_precision) logdens_linear = cond_cov.dot(self.opt_linear.T) * prec else: cond_precision = self.opt_linear.T.dot(prec.dot(self.opt_linear)) cond_cov = inv(cond_precision) logdens_linear = cond_cov.dot(self.opt_linear.T).dot(prec) cond_mean = -logdens_linear.dot(self.observed_score_state + self.opt_offset) self.cond_mean = cond_mean self.cond_cov = cond_cov self.cond_precision = cond_precision self.logdens_linear = logdens_linear return cond_mean, cond_cov, cond_precision, logdens_linear def selective_MLE(self, solve_args={'tol': 1.e-12}, level=0.9, useJacobian=True, dispersion=None): """Do selective_MLE for group_lasso Note: this masks the selective_MLE inherited from query because that is not adapted for the group_lasso. Also, assumes you have already run the fit method since this uses results from that method. Parameters ---------- observed_target: from selected_targets target_cov: from selected_targets target_cov_score: from selected_targets init_soln: (opt_state) initial (observed) value of optimization variables cond_mean: conditional mean of optimization variables (model on _setup_implied_gaussian) cond_cov: conditional variance of optimization variables (model on _setup_implied_gaussian) logdens_linear: (model on _setup_implied_gaussian) linear_part: like A_scaling (from lasso) offset: like b_scaling (from lasso) solve_args: passed on to solver level: level of confidence intervals useC: whether to use python or C solver JacobianPieces: (use self.C defined in fitting) """ self._setup_implied_gaussian() # Calculate useful quantities (observed_target, target_cov, target_score_cov, alternatives) = self.selected_targets(dispersion) init_soln = self.observed_opt_state # just the gammas cond_mean = self.cond_mean cond_cov = self.cond_cov logdens_linear = self.logdens_linear linear_part = self.linear_part offset = self.offset if np.asarray(observed_target).shape in [(), (0,)]: raise ValueError('no target specified') observed_target = np.atleast_1d(observed_target) prec_target = inv(target_cov) prec_opt = self.cond_precision score_offset = self.observed_score_state + self.opt_offset # target_lin determines how the conditional mean of optimization variables # vary with target # logdens_linear determines how the argument of the optimization density # depends on the score, not how the mean depends on score, hence the minus sign target_linear = target_score_cov.T.dot(prec_target) target_offset = score_offset - target_linear.dot(observed_target) target_lin = - logdens_linear.dot(target_linear) target_off = cond_mean - target_lin.dot(observed_target) if np.asarray(self.randomizer_prec).shape in [(), (0,)]: _P = target_linear.T.dot(target_offset) * self.randomizer_prec _prec = prec_target + (target_linear.T.dot(target_linear) * self.randomizer_prec) - target_lin.T.dot( prec_opt).dot( target_lin) else: _P = target_linear.T.dot(self.randomizer_prec).dot(target_offset) _prec = prec_target + (target_linear.T.dot(self.randomizer_prec).dot(target_linear)) - target_lin.T.dot( prec_opt).dot(target_lin) C = target_cov.dot(_P - target_lin.T.dot(prec_opt).dot(target_off)) conjugate_arg = prec_opt.dot(cond_mean) val, soln, hess = solve_barrier_affine_jacobian_py(conjugate_arg, prec_opt, init_soln, linear_part, offset, self.C, self.active_dirs, useJacobian, **solve_args) final_estimator = target_cov.dot(_prec).dot(observed_target) \ + target_cov.dot(target_lin.T.dot(prec_opt.dot(cond_mean - soln))) + C unbiased_estimator = target_cov.dot(_prec).dot(observed_target) + target_cov.dot( _P - target_lin.T.dot(prec_opt).dot(target_off)) L = target_lin.T.dot(prec_opt) observed_info_natural = _prec + L.dot(target_lin) - L.dot(hess.dot(L.T)) observed_info_mean = target_cov.dot(observed_info_natural.dot(target_cov)) Z_scores = final_estimator / np.sqrt(np.diag(observed_info_mean)) pvalues = ndist.cdf(Z_scores) pvalues = 2 * np.minimum(pvalues, 1 - pvalues) alpha = 1 - level quantile = ndist.ppf(1 - alpha / 2.) intervals = np.vstack([final_estimator - quantile * np.sqrt(np.diag(observed_info_mean)), final_estimator + quantile * np.sqrt(np.diag(observed_info_mean))]).T log_ref = val + conjugate_arg.T.dot(cond_cov).dot(conjugate_arg) / 2. result = pd.DataFrame({'MLE': final_estimator, 'SE': np.sqrt(np.diag(observed_info_mean)), 'Zvalue': Z_scores, 'pvalue': pvalues, 'lower_confidence': intervals[:, 0], 'upper_confidence': intervals[:, 1], 'unbiased': unbiased_estimator}) return result, observed_info_mean, log_ref def selected_targets(self, dispersion=None, solve_args={'tol': 1.e-12, 'min_its': 50}): X, y = self.loglike.data n, p = X.shape XE = self.XE Q = self.Q observed_target = restricted_estimator(self.loglike, self.ordered_vars, solve_args=solve_args) _score_linear = -XE.T.dot(self._W[:, None] * X).T alternatives = ['twosided'] * len(self.active) if dispersion is None: # use Pearson's X^2 dispersion = ((y - self.loglike.saturated_loss.mean_function( XE.dot(observed_target))) ** 2 / self._W).sum() / (n - XE.shape[1]) cov_target = self.QI * dispersion crosscov_target_score = _score_linear.dot(self.QI).T * dispersion return (observed_target, cov_target, crosscov_target_score, alternatives) class approximate_grid_inference(object): def __init__(self, query, dispersion, solve_args={'tol': 1.e-12}, useIP=True): """ Produce p-values and confidence intervals for targets of model including selected features Parameters ---------- query : `gaussian_query` A Gaussian query which has information to describe implied Gaussian. observed_target : ndarray Observed estimate of target. target_cov : ndarray Estimated covaraince of target. target_score_cov : ndarray Estimated covariance of target and score of randomized query. solve_args : dict, optional Arguments passed to solver. """ self.solve_args = solve_args result, inverse_info = query.selective_MLE(dispersion=dispersion)[:2] self.linear_part = query.linear_part self.offset = query.offset self.logdens_linear = query.logdens_linear self.cond_mean = query.cond_mean self.prec_opt = np.linalg.inv(query.cond_cov) self.cond_cov = query.cond_cov self.C = query.C self.active_dirs = query.active_dirs (observed_target, target_cov, target_score_cov, alternatives) = query.selected_targets(dispersion) self.observed_target = observed_target self.target_score_cov = target_score_cov self.target_cov = target_cov self.init_soln = query.observed_opt_state self.randomizer_prec = query.randomizer_prec self.score_offset = query.observed_score_state + query.opt_offset self.ntarget = ntarget = target_cov.shape[0] _scale = 4 * np.sqrt(np.diag(inverse_info)) if useIP == False: ngrid = 1000 self.stat_grid = np.zeros((ntarget, ngrid)) for j in range(ntarget): self.stat_grid[j, :] = np.linspace(observed_target[j] - 1.5 * _scale[j], observed_target[j] + 1.5 * _scale[j], num=ngrid) else: ngrid = 100 self.stat_grid = np.zeros((ntarget, ngrid)) for j in range(ntarget): self.stat_grid[j, :] = np.linspace(observed_target[j] - 1.5 * _scale[j], observed_target[j] + 1.5 * _scale[j], num=ngrid) self.opt_linear = query.opt_linear self.useIP = useIP def summary(self, alternatives=None, parameter=None, level=0.9): """ Produce p-values and confidence intervals for targets of model including selected features Parameters ---------- alternatives : [str], optional Sequence of strings describing the alternatives, should be values of ['twosided', 'less', 'greater'] parameter : np.array Hypothesized value for parameter -- defaults to 0. level : float Confidence level. """ if parameter is not None: pivots = self._approx_pivots(parameter, alternatives=alternatives) else: pivots = None pvalues = self._approx_pivots(np.zeros_like(self.observed_target), alternatives=alternatives) lower, upper = self._approx_intervals(level=level) result = pd.DataFrame({'target': self.observed_target, 'pvalue': pvalues, 'lower_confidence': lower, 'upper_confidence': upper}) if not np.all(parameter == 0): result.insert(4, 'pivot', pivots) result.insert(5, 'parameter', parameter) return result def log_reference(self, observed_target, target_cov, target_score_cov, grid): """ Approximate the log of the reference density on a grid. """ if np.asarray(observed_target).shape in [(), (0,)]: raise ValueError('no target specified') prec_target = np.linalg.inv(target_cov) target_lin = - self.logdens_linear.dot(target_score_cov.T.dot(prec_target)) ref_hat = [] for k in range(grid.shape[0]): # in the usual D = N + Gamma theta.hat, # target_lin is "something" times Gamma, # where "something" comes from implied Gaussian # cond_mean is "something" times D # Gamma is target_score_cov.T.dot(prec_target) num_opt = self.prec_opt.shape[0] num_con = self.linear_part.shape[0] cond_mean_grid = (target_lin.dot(np.atleast_1d(grid[k] - observed_target)) + self.cond_mean) #direction for decomposing o eta = -self.prec_opt.dot(self.logdens_linear.dot(target_score_cov.T)) implied_mean = np.asscalar(eta.T.dot(cond_mean_grid)) implied_cov = np.asscalar(eta.T.dot(self.cond_cov).dot(eta)) implied_prec = 1./implied_cov _A = self.cond_cov.dot(eta) * implied_prec R = np.identity(num_opt) - _A.dot(eta.T) A = self.linear_part.dot(_A).reshape((-1,)) b = self.offset-self.linear_part.dot(R).dot(self.init_soln) conjugate_arg = implied_mean * implied_prec val, soln, _ = solver(np.asarray([conjugate_arg]), np.reshape(implied_prec, (1,1)), eta.T.dot(self.init_soln), A.reshape((A.shape[0],1)), b, **self.solve_args) gamma_ = _A.dot(soln) + R.dot(self.init_soln) log_jacob = jacobian_grad_hess(gamma_, self.C, self.active_dirs) ref_hat.append(-val - ((conjugate_arg ** 2) * implied_cov)/ 2. + log_jacob[0]) return np.asarray(ref_hat) def _construct_families(self): self._construct_density() self._families = [] for m in range(self.ntarget): p = self.target_score_cov.shape[1] observed_target_uni = (self.observed_target[m]).reshape((1,)) target_cov_uni = (np.diag(self.target_cov)[m]).reshape((1, 1)) target_score_cov_uni = self.target_score_cov[m, :].reshape((1, p)) var_target = 1. / ((self.precs[m])[0, 0]) log_ref = self.log_reference(observed_target_uni, target_cov_uni, target_score_cov_uni, self.stat_grid[m]) if self.useIP == False: logW = (log_ref - 0.5 * (self.stat_grid[m] - self.observed_target[m]) ** 2 / var_target) logW -= logW.max() self._families.append(discrete_family(self.stat_grid[m], np.exp(logW))) else: approx_fn = interp1d(self.stat_grid[m], log_ref, kind='quadratic', bounds_error=False, fill_value='extrapolate') grid = np.linspace(self.stat_grid[m].min(), self.stat_grid[m].max(), 1000) logW = (approx_fn(grid) - 0.5 * (grid - self.observed_target[m]) ** 2 / var_target) logW -= logW.max() self._families.append(discrete_family(grid, np.exp(logW))) def _approx_pivots(self, mean_parameter, alternatives=None): if not hasattr(self, "_families"): self._construct_families() if alternatives is None: alternatives = ['twosided'] * self.ntarget pivot = [] for m in range(self.ntarget): family = self._families[m] var_target = 1. / ((self.precs[m])[0, 0]) mean = self.S[m].dot(mean_parameter[m].reshape((1,))) + self.r[m] _cdf = family.cdf((mean[0] - self.observed_target[m]) / var_target, x=self.observed_target[m]) print("variable completed ", m) if alternatives[m] == 'twosided': pivot.append(2 * min(_cdf, 1 - _cdf)) elif alternatives[m] == 'greater': pivot.append(1 - _cdf) elif alternatives[m] == 'less': pivot.append(_cdf) else: raise ValueError('alternative should be in ["twosided", "less", "greater"]') return pivot def _approx_intervals(self, level=0.9): if not hasattr(self, "_families"): self._construct_families() lower, upper = [], [] for m in range(self.ntarget): # construction of intervals from families follows `selectinf.learning.core` family = self._families[m] observed_target = self.observed_target[m] l, u = family.equal_tailed_interval(observed_target, alpha=1 - level) var_target = 1. / ((self.precs[m])[0, 0]) lower.append(l * var_target + observed_target) upper.append(u * var_target + observed_target) return np.asarray(lower), np.asarray(upper) ### Private method def _construct_density(self): precs = {} S = {} r = {} p = self.target_score_cov.shape[1] for m in range(self.ntarget): observed_target_uni = (self.observed_target[m]).reshape((1,)) target_cov_uni = (np.diag(self.target_cov)[m]).reshape((1, 1)) prec_target = 1. / target_cov_uni target_score_cov_uni = self.target_score_cov[m, :].reshape((1, p)) target_linear = target_score_cov_uni.T.dot(prec_target) target_offset = (self.score_offset - target_linear.dot(observed_target_uni)).reshape( (target_linear.shape[0],)) target_lin = -self.logdens_linear.dot(target_linear) target_off = (self.cond_mean - target_lin.dot(observed_target_uni)).reshape((target_lin.shape[0],)) _prec = prec_target + (target_linear.T.dot(target_linear) * self.randomizer_prec) - target_lin.T.dot( self.prec_opt).dot(target_lin) _P = target_linear.T.dot(target_offset) * self.randomizer_prec _r = (1. / _prec).dot(target_lin.T.dot(self.prec_opt).dot(target_off) - _P) _S = np.linalg.inv(_prec).dot(prec_target) S[m] = _S r[m] = _r precs[m] = _prec self.precs = precs self.S = S self.r = r def solve_barrier_affine_jacobian_py(conjugate_arg, precision, feasible_point, con_linear, con_offset, C, active_dirs, useJacobian=True, step=1, nstep=2000, min_its=500, tol=1.e-12): """ This needs to be updated to actually use the Jacobian information (in self.C) arguments conjugate_arg: \\bar{\\Sigma}^{-1} \bar{\\mu} precision: \\bar{\\Sigma}^{-1} feasible_point: gamma's from fitting con_linear: linear part of affine constraint used for barrier function con_offset: offset part of affine constraint used for barrier function C: V^T Q^{-1} \\Lambda V active_dirs: """ scaling = np.sqrt(np.diag(con_linear.dot(precision).dot(con_linear.T))) if feasible_point is None: feasible_point = 1. / scaling def objective(gs): p1 = -gs.T.dot(conjugate_arg) p2 = gs.T.dot(precision).dot(gs) / 2. if useJacobian: p3 = - jacobian_grad_hess(gs, C, active_dirs)[0] else: p3 = 0 p4 = log(1. + 1. / ((con_offset - con_linear.dot(gs)) / scaling)).sum() return p1 + p2 + p3 + p4 def grad(gs): p1 = -conjugate_arg + precision.dot(gs) p2 = -con_linear.T.dot(1. / (scaling + con_offset - con_linear.dot(gs))) if useJacobian: p3 = - jacobian_grad_hess(gs, C, active_dirs)[1] else: p3 = 0 p4 = 1. / (con_offset - con_linear.dot(gs)) return p1 + p2 + p3 + p4 def barrier_hessian(gs): # contribution of barrier and jacobian to hessian p1 = con_linear.T.dot(np.diag(-1. / ((scaling + con_offset - con_linear.dot(gs)) ** 2.) + 1. / ((con_offset - con_linear.dot(gs)) ** 2.))).dot(con_linear) if useJacobian: p2 = - jacobian_grad_hess(gs, C, active_dirs)[2] else: p2 = 0 return p1 + p2 current = feasible_point current_value = np.inf for itercount in range(nstep): cur_grad = grad(current) # make sure proposal is feasible count = 0 while True: count += 1 proposal = current - step * cur_grad if np.all(con_offset - con_linear.dot(proposal) > 0): break step *= 0.5 if count >= 40: raise ValueError('not finding a feasible point') # make sure proposal is a descent count = 0 while True: count += 1 proposal = current - step * cur_grad proposed_value = objective(proposal) if proposed_value <= current_value: break step *= 0.5 if count >= 20: if not (np.isnan(proposed_value) or np.isnan(current_value)): break else: raise ValueError('value is NaN: %f, %f' % (proposed_value, current_value)) # stop if relative decrease is small if np.fabs(current_value - proposed_value) < tol * np.fabs(current_value) and itercount >= min_its: current = proposal current_value = proposed_value break current = proposal current_value = proposed_value if itercount % 4 == 0: step *= 2 hess = inv(precision + barrier_hessian(current)) return current_value, current, hess # Jacobian calculations def calc_GammaMinus(gamma, active_dirs): """Calculate Gamma^minus (as a function of gamma vector, active directions) """ to_diag = [[g] * (ug.size - 1) for (g, ug) in zip(gamma, active_dirs.values())] return block_diag(*[i for gp in to_diag for i in gp]) def jacobian_grad_hess(gamma, C, active_dirs): """ Calculate the log-Jacobian (scalar), gradient (gamma.size vector) and hessian (gamma.size square matrix) """ if C.shape == (0, 0): # when all groups are size one, C will be an empty array return 0, 0, 0 else: GammaMinus = calc_GammaMinus(gamma, active_dirs) # eigendecomposition #evalues, evectors = eig(GammaMinus + C) # log Jacobian #J = log(evalues).sum() J = np.log(np.linalg.det(GammaMinus + C)) # inverse #GpC_inv = evectors.dot(np.diag(1 / evalues).dot(evectors.T)) GpC_inv = np.linalg.inv(GammaMinus + C) # summing matrix (gamma.size by C.shape[0]) S = block_diag(*[np.ones((1, ug.size - 1)) for ug in active_dirs.values()]) # gradient grad_J = S.dot(GpC_inv.diagonal()) # hessian hess_J = -S.dot(np.multiply(GpC_inv, GpC_inv.T).dot(S.T)) return J, grad_J, hess_J def _check_groups(groups): """Make sure that the user-specific groups are ok There are a number of assumptions that group_lasso makes about how groups are specified. Specifically, we assume that `groups` is a 1-d array_like of integers that are sorted in increasing order, start at 0, and have no gaps (e.g., if there is a group 2 and a group 4, there must also be at least one feature in group 3). This function checks the user-specified group scheme and raises an exception if it finds any problems. Sorting feature groups is potentially tedious for the user and in future we might do this for them. """ # check array_like agroups = np.array(groups) # check dimension if len(agroups.shape) != 1: raise ValueError("Groups are not a 1D array_like") # check sorted if np.any(agroups[:-1] > agroups[1:]) < 0: raise ValueError("Groups are not sorted") # check integers if not np.issubdtype(agroups.dtype, np.integer): raise TypeError("Groups are not integers") # check starts with 0 if not np.amin(agroups) == 0: raise ValueError("First group is not 0") # check for no skipped groups if not np.all(np.diff(np.unique(agroups)) == 1): raise ValueError("Some group is skipped")
2300
import six import json import gzip from exporters.default_retries import retry_long from exporters.writers.base_writer import BaseWriter class ODOWriter(BaseWriter): """ Writes items to a odo destination. https://odo.readthedocs.org/en/latest/ Needed parameters: - schema (object) schema object. - odo_uri (str) ODO valid destination uri. """ requirements = { 'schema': {'type': object, 'required': True}, 'odo_uri': {'type': six.string_types, 'required': True} } def __init__(self, options): super(ODOWriter, self).__init__(options) from flatson import Flatson schema = self.read_option('schema', None) self.odo_uri = self.read_option('odo_uri', None) self.flatson = Flatson(schema) self.logger.info('ODOWriter has been initiated. Writing to: {}'.format(self.odo_uri)) @retry_long def write(self, dump_path, group_key=''): from odo import odo, resource, discover import pandas as pd with gzip.open(dump_path) as f: lines = [json.loads(line.replace('\n', '')) for line in f.readlines()] flattened_lines = (self.flatson.flatten(line) for line in lines) pf = pd.DataFrame(flattened_lines, columns=self.flatson.fieldnames) dshape = discover(pf) odo(pf, resource(self.odo_uri), dshape=dshape)
2312
import time import os import sys import shutil import json import argparse from zipfile import ZipFile from contextlib import contextmanager from datetime import datetime from Tests.private_build.upload_packs_private import download_and_extract_index, update_index_with_priced_packs, \ extract_packs_artifacts from Tests.Marketplace.marketplace_services import init_storage_client from Tests.scripts.utils.log_util import install_logging from Tests.scripts.utils import logging_wrapper as logging MAX_SECONDS_TO_WAIT_FOR_LOCK = 600 LOCK_FILE_PATH = 'lock.txt' @contextmanager def lock_and_unlock_dummy_index(public_storage_bucket, dummy_index_lock_path): try: acquire_dummy_index_lock(public_storage_bucket, dummy_index_lock_path) yield except Exception: logging.exception("Error in dummy index lock context manager.") finally: release_dummy_index_lock(public_storage_bucket, dummy_index_lock_path) def change_pack_price_to_zero(path_to_pack_metadata): with open(path_to_pack_metadata, 'r') as pack_metadata_file: pack_metadata = json.load(pack_metadata_file) pack_metadata['price'] = 0 with open(path_to_pack_metadata, 'w') as pack_metadata_file: json.dump(pack_metadata, pack_metadata_file, indent=4) def change_packs_price_to_zero(public_index_folder_path): paths_to_packs_in_merged_index = [pack_dir.path for pack_dir in os.scandir(public_index_folder_path) if pack_dir.is_dir()] for path_to_pack in paths_to_packs_in_merged_index: path_to_pack_metadata = os.path.join(path_to_pack, 'metadata.json') change_pack_price_to_zero(path_to_pack_metadata) def merge_private_index_into_public_index(public_index_folder_path, private_index_folder_path): packs_in_private_index = [pack_dir.name for pack_dir in os.scandir(private_index_folder_path) if pack_dir.is_dir()] for pack_name in packs_in_private_index: path_to_pack_in_private_index = os.path.join(private_index_folder_path, pack_name) path_to_pack_in_public_index = os.path.join(public_index_folder_path, pack_name) shutil.copy(path_to_pack_in_private_index, path_to_pack_in_public_index) def upload_modified_index(public_index_folder_path, extract_destination_path, public_ci_dummy_index_blob, build_number, private_packs): """Upload updated index zip to cloud storage. Args: public_index_folder_path (str): public index folder full path. extract_destination_path (str): extract folder full path. public_ci_dummy_index_blob (Blob): google cloud storage object that represents the dummy index.zip blob. build_number (str): circleCI build number, used as an index revision. private_packs (list): List of private packs and their price. """ with open(os.path.join(public_index_folder_path, "index.json"), "w+") as index_file: for private_pack in private_packs: private_pack['price'] = 0 index = { 'revision': build_number, 'modified': datetime.utcnow().strftime('%Y-%m-%dT%H:%M:%SZ'), 'packs': private_packs } json.dump(index, index_file, indent=4) index_zip_name = os.path.basename(public_index_folder_path) index_zip_path = shutil.make_archive(base_name=public_index_folder_path, format="zip", root_dir=extract_destination_path, base_dir=index_zip_name) try: public_ci_dummy_index_blob.reload() public_ci_dummy_index_blob.cache_control = "no-cache,max-age=0" # disabling caching for index blob public_ci_dummy_index_blob.upload_from_filename(index_zip_path) logging.success("Finished uploading index.zip to storage.") except Exception: logging.exception("Failed in uploading index. Mismatch in index file generation.") sys.exit(1) finally: shutil.rmtree(public_index_folder_path) def option_handler(): """Validates and parses script arguments. Returns: Namespace: Parsed arguments object. """ parser = argparse.ArgumentParser(description="Store packs in cloud storage.") # disable-secrets-detection-start parser.add_argument('-b', '--public_bucket_name', help="CI public bucket name", required=True) parser.add_argument('-pb', '--private_bucket_name', help="CI private bucket name", required=True) parser.add_argument('-s', '--service_account', help=("Path to gcloud service account, is for circleCI usage. " "For local development use your personal account and " "authenticate using Google Cloud SDK by running: " "`gcloud auth application-default login` and leave this parameter blank. " "For more information go to: " "https://googleapis.dev/python/google-api-core/latest/auth.html"), required=False) parser.add_argument('-n', '--ci_build_number', help="CircleCi build number (will be used as hash revision at index file)", required=True) parser.add_argument('-e', '--extract_public_index_path', help="Full path of folder to extract the public index", required=True) parser.add_argument('-sb', '--storage_base_path', help="Storage base path of the directory to upload to.", required=False) parser.add_argument('-p', '--pack_name', help="Modified pack to upload to gcs.") parser.add_argument('-a', '--artifacts_path', help="The full path of packs artifacts", required=True) parser.add_argument('-ea', '--extract_artifacts_path', help="Full path of folder to extract wanted packs", required=True) parser.add_argument('-di', '--dummy_index_dir_path', help="Full path to the dummy index in the private CI bucket", required=True) # disable-secrets-detection-end return parser.parse_args() def is_dummy_index_locked(public_storage_bucket, dummy_index_lock_path): dummy_index_lock_blob = public_storage_bucket.blob(dummy_index_lock_path) return dummy_index_lock_blob.exists() def lock_dummy_index(public_storage_bucket, dummy_index_lock_path): dummy_index_lock_blob = public_storage_bucket.blob(dummy_index_lock_path) with open(LOCK_FILE_PATH, 'w') as lock_file: lock_file.write('locked') with open(LOCK_FILE_PATH, 'rb') as lock_file: dummy_index_lock_blob.upload_from_file(lock_file) def acquire_dummy_index_lock(public_storage_bucket, dummy_index_lock_path): total_seconds_waited = 0 while is_dummy_index_locked(public_storage_bucket, dummy_index_lock_path): if total_seconds_waited >= MAX_SECONDS_TO_WAIT_FOR_LOCK: logging.critical("Error: Failed too long to acquire lock, exceeded max wait time.") sys.exit(1) if total_seconds_waited % 60 == 0: # Printing a message every minute to keep the machine from dying due to no output logging.info("Waiting to acquire lock.") total_seconds_waited += 10 time.sleep(10) lock_dummy_index(public_storage_bucket, dummy_index_lock_path) def release_dummy_index_lock(public_storage_bucket, dummy_index_lock_path): dummy_index_lock_blob = public_storage_bucket.blob(dummy_index_lock_path) dummy_index_lock_blob.delete() os.remove(LOCK_FILE_PATH) def add_private_packs_from_dummy_index(private_packs, dummy_index_blob): downloaded_dummy_index_path = 'current_dummy_index.zip' extracted_dummy_index_path = 'dummy_index' dummy_index_json_path = os.path.join(extracted_dummy_index_path, 'index', 'index.json') dummy_index_blob.download_to_filename(downloaded_dummy_index_path) os.mkdir(extracted_dummy_index_path) if os.path.exists(downloaded_dummy_index_path): with ZipFile(downloaded_dummy_index_path, 'r') as index_zip: index_zip.extractall(extracted_dummy_index_path) with open(dummy_index_json_path) as index_file: index_json = json.load(index_file) packs_from_dummy_index = index_json.get('packs', []) for pack in private_packs: is_pack_in_dummy_index = any( [pack['id'] == dummy_index_pack['id'] for dummy_index_pack in packs_from_dummy_index]) if not is_pack_in_dummy_index: packs_from_dummy_index.append(pack) os.remove(downloaded_dummy_index_path) shutil.rmtree(extracted_dummy_index_path) return packs_from_dummy_index def main(): install_logging('prepare_public_index_for_private_testing.log', logger=logging) upload_config = option_handler() service_account = upload_config.service_account build_number = upload_config.ci_build_number public_bucket_name = upload_config.public_bucket_name private_bucket_name = upload_config.private_bucket_name storage_base_path = upload_config.storage_base_path extract_public_index_path = upload_config.extract_public_index_path changed_pack = upload_config.pack_name extract_destination_path = upload_config.extract_artifacts_path packs_artifacts_path = upload_config.artifacts_path dummy_index_dir_path = upload_config.dummy_index_dir_path dummy_index_path = os.path.join(dummy_index_dir_path, 'index.zip') dummy_index_lock_path = os.path.join(dummy_index_dir_path, 'lock.txt') storage_client = init_storage_client(service_account) public_storage_bucket = storage_client.bucket(public_bucket_name) private_storage_bucket = storage_client.bucket(private_bucket_name) dummy_index_blob = public_storage_bucket.blob(dummy_index_path) with lock_and_unlock_dummy_index(public_storage_bucket, dummy_index_lock_path): extract_packs_artifacts(packs_artifacts_path, extract_destination_path) public_index_folder_path, public_index_blob, _ = download_and_extract_index(public_storage_bucket, extract_public_index_path, storage_base_path) # In order for the packs to be downloaded successfully, their price has to be 0 change_packs_price_to_zero(public_index_folder_path) private_packs, private_index_path, private_index_blob = update_index_with_priced_packs(private_storage_bucket, extract_destination_path, public_index_folder_path, changed_pack, True, storage_base_path) private_packs = add_private_packs_from_dummy_index(private_packs, dummy_index_blob) upload_modified_index(public_index_folder_path, extract_public_index_path, dummy_index_blob, build_number, private_packs) if __name__ == '__main__': main()
2321
r"""Training and evaluating quantum kernels =========================================== .. meta:: :property="og:description": Kernels and alignment training with Pennylane. :property="og:image": https://pennylane.ai/qml/_images/QEK_thumbnail.png .. related:: tutorial_kernel_based_training Kernel-based training with scikit-learn tutorial_data_reuploading_classifier Classification with data reuploading *Authors: <NAME>, <NAME>, <NAME>, <NAME>, <NAME> and <NAME>. Posted: 24 June 2021* Kernel methods are one of the cornerstones of classical machine learning. Here we are concerned with kernels that can be evaluated on quantum computers, *quantum kernels* for short. In this tutorial you will learn how to evaluate kernels, use them for classification and train them with gradient-based optimization, and all that using the functionality of PennyLane's `kernels module <https://pennylane.readthedocs.io/en/latest/code/qml_kernels.html>`__. The demo is based on Ref. [#Training_QEKs]_, a project from Xanadu's own `QHack <https://qhack.ai/>`__ hackathon. What are kernel methods? ------------------------ To understand what a kernel method does, let's first revisit one of the simplest methods to assign binary labels to datapoints: linear classification. Imagine we want to discern two different classes of points that lie in different corners of the plane. A linear classifier corresponds to drawing a line and assigning different labels to the regions on opposing sides of the line: .. figure:: ../demonstrations/kernels_module/linear_classification.png :align: center :width: 30% We can mathematically formalize this by assigning the label :math:`y` via .. math:: y(\boldsymbol{x}) = \operatorname{sgn}(\langle \boldsymbol{w}, \boldsymbol{x}\rangle + b). The vector :math:`\boldsymbol{w}` points perpendicular to the line and thus determine its slope. The independent term :math:`b` specifies the position on the plane. In this form, linear classification can also be extended to higher dimensional vectors :math:`\boldsymbol{x}`, where a line does not divide the entire space into two regions anymore. Instead one needs a *hyperplane*. It is immediately clear that this method is not very powerful, as datasets that are not separable by a hyperplane can't be classified without error. We can actually sneak around this limitation by performing a neat trick: if we define some map :math:`\phi(\boldsymbol{x})` that *embeds* our datapoints into a larger *feature space* and then perform linear classification there, we could actually realise non-linear classification in our original space! .. figure:: ../demonstrations/kernels_module/embedding_nonlinear_classification.png :align: center :width: 65% If we go back to the expression for our prediction and include the embedding, we get .. math:: y(\boldsymbol{x}) = \operatorname{sgn}(\langle \boldsymbol{w}, \phi(\boldsymbol{x})\rangle + b). We will forgo one tiny step, but it can be shown that for the purpose of optimal classification, we can choose the vector defining the decision boundary as a linear combination of the embedded datapoints :math:`\boldsymbol{w} = \sum_i \alpha_i \phi(\boldsymbol{x}_i)`. Putting this into the formula yields .. math:: y(\boldsymbol{x}) = \operatorname{sgn}\left(\sum_i \alpha_i \langle \phi(\boldsymbol{x}_i), \phi(\boldsymbol{x})\rangle + b\right). This rewriting might not seem useful at first, but notice the above formula only contains inner products between vectors in the embedding space: .. math:: k(\boldsymbol{x}_i, \boldsymbol{x}_j) = \langle \phi(\boldsymbol{x}_i), \phi(\boldsymbol{x}_j)\rangle. We call this function the *kernel*. It provides the advantage that we can often find an explicit formula for the kernel :math:`k` that makes it superfluous to actually perform the (potentially expensive) embedding :math:`\phi`. Consider for example the following embedding and the associated kernel: .. math:: \phi((x_1, x_2)) &= (x_1^2, \sqrt{2} x_1 x_2, x_2^2) \\ k(\boldsymbol{x}, \boldsymbol{y}) &= x_1^2 y_1^2 + 2 x_1 x_2 y_1 y_2 + x_2^2 y_2^2 = \langle \boldsymbol{x}, \boldsymbol{y} \rangle^2. This means by just replacing the regular scalar product in our linear classification with the map :math:`k`, we can actually express much more intricate decision boundaries! This is very important, because in many interesting cases the embedding :math:`\phi` will be much costlier to compute than the kernel :math:`k`. In this demo, we will explore one particular kind of kernel that can be realized on near-term quantum computers, namely *Quantum Embedding Kernels (QEKs)*. These are kernels that arise from embedding data into the space of quantum states. We formalize this by considering a parameterised quantum circuit :math:`U(\boldsymbol{x})` that maps a datapoint :math:`\boldsymbol{x}` to the state .. math:: |\psi(\boldsymbol{x})\rangle = U(\boldsymbol{x}) |0 \rangle. The kernel value is then given by the *overlap* of the associated embedded quantum states .. math:: k(\boldsymbol{x}_i, \boldsymbol{x}_j) = | \langle\psi(\boldsymbol{x}_i)|\psi(\boldsymbol{x}_j)\rangle|^2. """ ############################################################################## # A toy problem # ------------- # In this demo, we will treat a toy problem that showcases the # inner workings of classification with quantum embedding kernels, # training variational embedding kernels and the available functionalities # to do both in PennyLane. We of course need to start with some imports: from pennylane import numpy as np import matplotlib as mpl np.random.seed(1359) ############################################################################## # And we proceed right away to create a dataset to work with, the # ``DoubleCake`` dataset. Firstly, we define two functions to enable us to # generate the data. # The details of these functions are not essential for understanding the demo, # so don't mind them if they are confusing. def _make_circular_data(num_sectors): """Generate datapoints arranged in an even circle.""" center_indices = np.array(range(0, num_sectors)) sector_angle = 2 * np.pi / num_sectors angles = (center_indices + 0.5) * sector_angle x = 0.7 * np.cos(angles) y = 0.7 * np.sin(angles) labels = 2 * np.remainder(np.floor_divide(angles, sector_angle), 2) - 1 return x, y, labels def make_double_cake_data(num_sectors): x1, y1, labels1 = _make_circular_data(num_sectors) x2, y2, labels2 = _make_circular_data(num_sectors) # x and y coordinates of the datapoints x = np.hstack([x1, 0.5 * x2]) y = np.hstack([y1, 0.5 * y2]) # Canonical form of dataset X = np.vstack([x, y]).T labels = np.hstack([labels1, -1 * labels2]) # Canonical form of labels Y = labels.astype(int) return X, Y ############################################################################## # Next, we define a function to help plot the ``DoubleCake`` data: def plot_double_cake_data(X, Y, ax, num_sectors=None): """Plot double cake data and corresponding sectors.""" x, y = X.T cmap = mpl.colors.ListedColormap(["#FF0000", "#0000FF"]) ax.scatter(x, y, c=Y, cmap=cmap, s=25, marker="s") if num_sectors is not None: sector_angle = 360 / num_sectors for i in range(num_sectors): color = ["#FF0000", "#0000FF"][(i % 2)] other_color = ["#FF0000", "#0000FF"][((i + 1) % 2)] ax.add_artist( mpl.patches.Wedge( (0, 0), 1, i * sector_angle, (i + 1) * sector_angle, lw=0, color=color, alpha=0.1, width=0.5, ) ) ax.add_artist( mpl.patches.Wedge( (0, 0), 0.5, i * sector_angle, (i + 1) * sector_angle, lw=0, color=other_color, alpha=0.1, ) ) ax.set_xlim(-1, 1) ax.set_ylim(-1, 1) ax.set_aspect("equal") ax.axis("off") return ax ############################################################################## # Let's now have a look at our dataset. In our example, we will work with # 3 sectors: import matplotlib.pyplot as plt num_sectors = 3 X, Y = make_double_cake_data(num_sectors) ax = plot_double_cake_data(X, Y, plt.gca(), num_sectors=num_sectors) ############################################################################## # Defining a Quantum Embedding Kernel # ----------------------------------- # PennyLane's `kernels module <https://pennylane.readthedocs.io/en/latest/code/qml_kernels.html>`__ # allows for a particularly simple # implementation of Quantum Embedding Kernels. The first ingredient we # need for this is an *ansatz*, which we will construct by repeating a # layer as building block. Let's start by defining this layer: import pennylane as qml def layer(x, params, wires, i0=0, inc=1): """Building block of the embedding ansatz""" i = i0 for j, wire in enumerate(wires): qml.Hadamard(wires=[wire]) qml.RZ(x[i % len(x)], wires=[wire]) i += inc qml.RY(params[0, j], wires=[wire]) qml.broadcast(unitary=qml.CRZ, pattern="ring", wires=wires, parameters=params[1]) ############################################################################## # To construct the ansatz, this layer is repeated multiple times, reusing # the datapoint ``x`` but feeding different variational # parameters ``params`` into each of them. # Together, the datapoint and the variational parameters fully determine # the embedding ansatz :math:`U(\boldsymbol{x})`. # In order to construct the full kernel circuit, we also require its adjoint # :math:`U(\boldsymbol{x})^\dagger`, which we can obtain via ``qml.adjoint``. def ansatz(x, params, wires): """The embedding ansatz""" for j, layer_params in enumerate(params): layer(x, layer_params, wires, i0=j * len(wires)) adjoint_ansatz = qml.adjoint(ansatz) def random_params(num_wires, num_layers): """Generate random variational parameters in the shape for the ansatz.""" return np.random.uniform(0, 2 * np.pi, (num_layers, 2, num_wires), requires_grad=True) ############################################################################## # Together with the ansatz we only need a device to run the quantum circuit on. # For the purpose of this tutorial we will use PennyLane's ``default.qubit`` # device with 5 wires in analytic mode. dev = qml.device("default.qubit", wires=5, shots=None) wires = dev.wires.tolist() ############################################################################## # Let us now define the quantum circuit that realizes the kernel. We will compute # the overlap of the quantum states by first applying the embedding of the first # datapoint and then the adjoint of the embedding of the second datapoint. We # finally extract the probabilities of observing each basis state. @qml.qnode(dev) def kernel_circuit(x1, x2, params): ansatz(x1, params, wires=wires) adjoint_ansatz(x2, params, wires=wires) return qml.probs(wires=wires) ############################################################################## # The kernel function itself is now obtained by looking at the probability # of observing the all-zero state at the end of the kernel circuit -- because # of the ordering in ``qml.probs``, this is the first entry: def kernel(x1, x2, params): return kernel_circuit(x1, x2, params)[0] ############################################################################## # # .. note:: # An alternative way to set up the kernel circuit in PennyLane would be # to use the observable type # `Projector <https://pennylane.readthedocs.io/en/latest/code/api/pennylane.Projector.html>`__. # This is shown in the # `demo on kernel-based training of quantum models <https://pennylane.ai/qml/demos/tutorial_kernel_based_training.html>`__, where you will also find more # background information on the kernel circuit structure itself. # # Before focusing on the kernel values we have to provide values for the # variational parameters. At this point we fix the number of layers in the # ansatz circuit to :math:`6`. init_params = random_params(num_wires=5, num_layers=6) ############################################################################## # Now we can have a look at the kernel value between the first and the # second datapoint: kernel_value = kernel(X[0], X[1], init_params) print(f"The kernel value between the first and second datapoint is {kernel_value:.3f}") ############################################################################## # The mutual kernel values between all elements of the dataset form the # *kernel matrix*. We can inspect it via the ``qml.kernels.square_kernel_matrix`` # method, which makes use of symmetry of the kernel, # :math:`k(\boldsymbol{x}_i,\boldsymbol{x}_j) = k(\boldsymbol{x}_j, \boldsymbol{x}_i)`. # In addition, the option ``assume_normalized_kernel=True`` ensures that we do not # calculate the entries between the same datapoints, as we know them to be 1 # for our noiseless simulation. Overall this means that we compute # :math:`\frac{1}{2}(N^2-N)` kernel values for :math:`N` datapoints. # To include the variational parameters, we construct a ``lambda`` function that # fixes them to the values we sampled above. init_kernel = lambda x1, x2: kernel(x1, x2, init_params) K_init = qml.kernels.square_kernel_matrix(X, init_kernel, assume_normalized_kernel=True) with np.printoptions(precision=3, suppress=True): print(K_init) ############################################################################## # Using the Quantum Embedding Kernel for predictions # -------------------------------------------------- # The quantum kernel alone can not be used to make predictions on a # dataset, becaues it is essentially just a tool to measure the similarity # between two datapoints. To perform an actual prediction we will make use # of scikit-learn's Support Vector Classifier (SVC). from sklearn.svm import SVC ############################################################################## # To construct the SVM, we need to supply ``sklearn.svm.SVC`` with a function # that takes two sets of datapoints and returns the associated kernel matrix. # We can make use of the function ``qml.kernels.kernel_matrix`` that provides # this functionality. It expects the kernel to not have additional parameters # besides the datapoints, which is why we again supply the variational # parameters via the ``lambda`` function from above. # Once we have this, we can let scikit-learn adjust the SVM from our Quantum # Embedding Kernel. # # .. note:: # This step does *not* modify the variational parameters in our circuit # ansatz. What it does is solving a different optimization task for the # :math:`\alpha` and :math:`b` vectors we introduced in the beginning. svm = SVC(kernel=lambda X1, X2: qml.kernels.kernel_matrix(X1, X2, init_kernel)).fit(X, Y) ############################################################################## # To see how well our classifier performs we will measure which percentage # of the dataset it classifies correctly. def accuracy(classifier, X, Y_target): return 1 - np.count_nonzero(classifier.predict(X) - Y_target) / len(Y_target) accuracy_init = accuracy(svm, X, Y) print(f"The accuracy of the kernel with random parameters is {accuracy_init:.3f}") ############################################################################## # We are also interested in seeing what the decision boundaries in this # classification look like. This could help us spotting overfitting issues # visually in more complex data sets. To this end we will introduce a # second helper method. def plot_decision_boundaries(classifier, ax, N_gridpoints=14): _xx, _yy = np.meshgrid(np.linspace(-1, 1, N_gridpoints), np.linspace(-1, 1, N_gridpoints)) _zz = np.zeros_like(_xx) for idx in np.ndindex(*_xx.shape): _zz[idx] = classifier.predict(np.array([_xx[idx], _yy[idx]])[np.newaxis, :]) plot_data = {"_xx": _xx, "_yy": _yy, "_zz": _zz} ax.contourf( _xx, _yy, _zz, cmap=mpl.colors.ListedColormap(["#FF0000", "#0000FF"]), alpha=0.2, levels=[-1, 0, 1], ) plot_double_cake_data(X, Y, ax) return plot_data ############################################################################## # With that done, let's have a look at the decision boundaries for our # initial classifier: init_plot_data = plot_decision_boundaries(svm, plt.gca()) ############################################################################## # We see the outer points in the dataset can be correctly classified, but # we still struggle with the inner circle. But remember we have a circuit # with many free parameters! It is reasonable to believe we can give # values to those variational parameters which improve the overall accuracy # of our SVC. # # Training the Quantum Embedding Kernel # ------------------------------------- # # To be able to train the Quantum Embedding Kernel we need some measure of # how well it fits the dataset in question. Performing an exhaustive # search in parameter space is not a good solution because it is very # resource intensive, and since the accuracy is a discrete quantity we # would not be able to detect small improvements. # # We can, however, resort to a more specialized measure, the # *kernel-target alignment* [#Alignment]_. The kernel-target alignment compares the # similarity predicted by the quantum kernel to the actual labels of the # training data. It is based on *kernel alignment*, a similiarity measure # between two kernels with given kernel matrices :math:`K_1` and # :math:`K_2`: # # .. math:: # \operatorname{KA}(K_1, K_2) = \frac{\operatorname{Tr}(K_1 K_2)}{\sqrt{\operatorname{Tr}(K_1^2)\operatorname{Tr}(K_2^2)}}. # # .. note:: # Seen from a more theoretical side, :math:`\operatorname{KA}` # is nothing else than the cosine of the angle between the kernel # matrices :math:`K_1` and :math:`K_2` if we see them as vectors # in the space of matrices with the Hilbert-Schmidt (or # Frobenius) scalar product # :math:`\langle A, B \rangle = \operatorname{Tr}(A^T B)`. This # reinforces the geometric picture of how this measure relates # to objects, namely two kernels, being aligned in a vector space. # # The training data enters the picture by defining an *ideal* kernel # function that expresses the original labelling in the vector # :math:`\boldsymbol{y}` by assigning to two datapoints the product # of the corresponding labels: # # .. math:: # k_{\boldsymbol{y}}(\boldsymbol{x}_i, \boldsymbol{x}_j) = y_i y_j. # # The assigned kernel is thus :math:`+1` if both datapoints lie in the # same class and :math:`-1` otherwise and its kernel matrix is simply # given by the outer product :math:`\boldsymbol{y}\boldsymbol{y}^T`. # The kernel-target alignment is then defined as the kernel alignment # of the kernel matrix :math:`K` generated by the # quantum kernel and :math:`\boldsymbol{y}\boldsymbol{y}^T`: # # .. math:: # \operatorname{KTA}_{\boldsymbol{y}}(K) # = \frac{\operatorname{Tr}(K \boldsymbol{y}\boldsymbol{y}^T)}{\sqrt{\operatorname{Tr}(K^2)\operatorname{Tr}((\boldsymbol{y}\boldsymbol{y}^T)^2)}} # = \frac{\boldsymbol{y}^T K \boldsymbol{y}}{\sqrt{\operatorname{Tr}(K^2)} N} # # where :math:`N` is the number of elements in :math:`\boldsymbol{y}`, # that is the number of datapoints in the dataset. # # In summary, the kernel-target alignment effectively captures how well # the kernel you chose reproduces the actual similarities of the data. It # does have one drawback, however: having a high kernel-target alignment # is only a necessary but not a sufficient condition for a good # performance of the kernel [#Alignment]_. This means having good alignment is # guaranteed for good performance, but optimal alignment will not always # bring optimal training accuracy with it. # # Let's now come back to the actual implementation. PennyLane's # ``kernels`` module allows you to easily evaluate the kernel # target alignment: kta_init = qml.kernels.target_alignment(X, Y, init_kernel, assume_normalized_kernel=True) print(f"The kernel-target alignment for our dataset and random parameters is {kta_init:.3f}") ############################################################################## # Now let's code up an optimization loop and improve the kernel-target alignment! # # We will make use of regular gradient descent optimization. To speed up # the optimization we will not use the entire training set to compute # :math:`\operatorname{KTA}` but rather # sample smaller subsets of the data at each step, we choose :math:`4` # datapoints at random. Remember that PennyLane's built-in optimizer works # to *minimize* the cost function that is given to it, which is why we # have to multiply the kernel target alignment by :math:`-1` to actually # *maximize* it in the process. # # .. note:: # Currently, the function ``qml.kernels.target_alignment`` is not # differentiable yet, making it unfit for gradient descent optimization. # We therefore first define a differentiable version of this function. def target_alignment( X, Y, kernel, assume_normalized_kernel=False, rescale_class_labels=True, ): """Kernel-target alignment between kernel and labels.""" K = qml.kernels.square_kernel_matrix( X, kernel, assume_normalized_kernel=assume_normalized_kernel, ) if rescale_class_labels: nplus = np.count_nonzero(np.array(Y) == 1) nminus = len(Y) - nplus _Y = np.array([y / nplus if y == 1 else y / nminus for y in Y]) else: _Y = np.array(Y) T = np.outer(_Y, _Y) inner_product = np.sum(K * T) norm = np.sqrt(np.sum(K * K) * np.sum(T * T)) inner_product = inner_product / norm return inner_product params = init_params opt = qml.GradientDescentOptimizer(0.2) for i in range(500): # Choose subset of datapoints to compute the KTA on. subset = np.random.choice(list(range(len(X))), 4) # Define the cost function for optimization cost = lambda _params: -target_alignment( X[subset], Y[subset], lambda x1, x2: kernel(x1, x2, _params), assume_normalized_kernel=True, ) # Optimization step params = opt.step(cost, params) # Report the alignment on the full dataset every 50 steps. if (i + 1) % 50 == 0: current_alignment = target_alignment( X, Y, lambda x1, x2: kernel(x1, x2, params), assume_normalized_kernel=True, ) print(f"Step {i+1} - Alignment = {current_alignment:.3f}") ############################################################################## # We want to assess the impact of training the parameters of the quantum # kernel. Thus, let's build a second support vector classifier with the # trained kernel: # First create a kernel with the trained parameter baked into it. trained_kernel = lambda x1, x2: kernel(x1, x2, params) # Second create a kernel matrix function using the trained kernel. trained_kernel_matrix = lambda X1, X2: qml.kernels.kernel_matrix(X1, X2, trained_kernel) # Note that SVC expects the kernel argument to be a kernel matrix function. svm_trained = SVC(kernel=trained_kernel_matrix).fit(X, Y) ############################################################################## # We expect to see an accuracy improvement vs. the SVM with random # parameters: accuracy_trained = accuracy(svm_trained, X, Y) print(f"The accuracy of a kernel with trained parameters is {accuracy_trained:.3f}") ############################################################################## # We have now achieved perfect classification! 🎆 # # Following on the results that SVM's have proven good generalisation # behavior, it will be interesting to inspect the decision boundaries of # our classifier: trained_plot_data = plot_decision_boundaries(svm_trained, plt.gca()) ############################################################################## # Indeed, we see that now not only every data instance falls within the # correct class, but also that there are no strong artifacts that would make us # distrust the model. In this sense, our approach benefits from both: on # one hand it can adjust itself to the dataset, and on the other hand # is not expected to suffer from bad generalisation. # # References # ---------- # # .. [#Training_QEKs] # # <NAME>, <NAME>, <NAME>, <NAME>, # <NAME>, and <NAME>. # "Training Quantum Embedding Kernels on Near-Term Quantum Computers." # `arXiv:2105.02276 <https://arxiv.org/abs/2105.02276>`__, 2021. # # .. [#Alignment] # # <NAME>, <NAME>, and <NAME>. # "An overview of kernel alignment and its applications." # `Artificial Intelligence Review 43.2: 179-192 <https://link.springer.com/article/10.1007/s10462-012-9369-4>`__, 2015.