list-of-demos / demo_list.py
hysts's picture
hysts HF staff
Update
758ca0a
raw
history blame
4.65 kB
import copy
import datetime
import operator
import pathlib
import pandas as pd
import tqdm.auto
import yaml
from huggingface_hub import HfApi
repo_dir = pathlib.Path(__file__).parent
class DemoList:
COLUMN_INFO = [
['status', 'markdown'],
['hardware', 'markdown'],
['title', 'markdown'],
['arxiv', 'markdown'],
['github', 'markdown'],
['likes', 'number'],
['tags', 'str'],
['last_modified', 'str'],
['created', 'str'],
['sdk', 'markdown'],
['sdk_version', 'str'],
['suggested_hardware', 'str'],
]
def __init__(self):
self.api = HfApi()
self.data = self.load_data()
self.df = self.to_df()
def update_data(self) -> None:
self.data = self.load_data()
self.df = self.to_df()
def load_data(self) -> dict:
with open(repo_dir / 'list.yaml') as f:
data = yaml.safe_load(f)
for url in tqdm.auto.tqdm(list(data)):
space_id = self.get_space_id(url)
space_info = self.api.space_info(repo_id=space_id)
card = space_info.cardData
info = data[url]
info['title'] = card['title']
info['sdk'] = self.colorize_sdk(card['sdk'])
info['sdk_version'] = card.get('sdk_version', '')
info['likes'] = space_info.likes
last_modified = datetime.datetime.strptime(
space_info.lastModified, '%Y-%m-%dT%H:%M:%S.000Z')
info['last_modified'] = last_modified.strftime('%Y/%m/%d %H:%M:%S')
created = datetime.datetime.strptime(info['created'],
'%Y-%m-%d-%H-%M-%S')
info['created'] = created.strftime('%Y/%m/%d %H:%M:%S')
info['status'] = self.colorize_status(space_info.runtime['stage'])
info['suggested_hardware'] = card.get('suggested_hardware', '')
info['hardware'] = self.colorize_hardware(
space_info.runtime['hardware']['current'],
info['suggested_hardware'])
return data
@staticmethod
def get_space_id(url: str) -> str:
return '/'.join(url.split('/')[-2:])
@staticmethod
def create_link(text: str, url: str) -> str:
return f'<a href={url} target="_blank">{text}</a>'
def get_arxiv_link(self, url: str) -> str:
links = sorted(self.data[url].get('arxiv', []))
links = [self.create_link(link.split('/')[-1], link) for link in links]
return '\n'.join(links)
def get_github_link(self, url: str) -> str:
links = sorted(self.data[url].get('github', []))
links = [self.create_link('github', link) for link in links]
return '\n'.join(links)
def get_tag_list(self, url: str) -> str:
tags = sorted(self.data[url].get('tags', []))
return ', '.join(tags)
@property
def column_names(self):
return list(map(operator.itemgetter(0), self.COLUMN_INFO))
@property
def column_datatype(self):
return list(map(operator.itemgetter(1), self.COLUMN_INFO))
def colorize_status(self, status: str) -> str:
if status == 'RUNNING':
color = 'green'
elif status in ['STOPPED', 'PAUSED']:
color = 'orange'
elif status in ['RUNTIME_ERROR', 'BUILD_ERROR']:
color = 'red'
else:
color = ''
return f'<div class="{color}">{status}</div>'
def colorize_sdk(self, sdk: str) -> str:
if sdk == 'gradio':
color = 'orange'
elif sdk == 'docker':
color = 'deepskyblue'
else:
color = ''
return f'<div class="{color}">{sdk}</div>'
def colorize_hardware(self, hardware: str | None,
suggested_hardware: str) -> str:
if hardware is None:
return ''
if 't4' in hardware or 'a10g' in hardware or 'a100' in hardware:
color = 'red'
elif suggested_hardware and 'cpu' not in suggested_hardware:
color = 'deepskyblue'
else:
return hardware
return f'<div class="{color}">{hardware}</div>'
def to_df(self) -> pd.DataFrame:
data = copy.deepcopy(self.data)
for url in list(data):
info = data[url]
info['title'] = self.create_link(info['title'], url)
info['arxiv'] = self.get_arxiv_link(url)
info['github'] = self.get_github_link(url)
info['tags'] = self.get_tag_list(url)
df = pd.DataFrame(data).T.loc[:, self.column_names]
return df