Datasets:
Sub-tasks:
dialogue-generation
Languages:
Russian
Multilinguality:
monolingual
Size Categories:
10K<n<100K
Language Creators:
found
Annotations Creators:
no-annotation
Source Datasets:
original
Tags:
File size: 3,830 Bytes
27615f9 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 |
import pandas as pd
import numpy as np
from tqdm import tqdm, trange
import json
from time import sleep
import api2ch
import html2text
import re
import os
API = api2ch.DvachApi('b')
if os.path.exists("parsed_threads.json"):
with open("parsed_threads.json", "r") as file:
PARSED_TREADS = set(json.load(file)["ids"])
else:
PARSED_TREADS = set()
REPLY_RE = re.compile(">>[0-9]+")
HTML_STRIPPER = html2text.HTML2Text()
HTML_STRIPPER.ignore_links = True
HTML_STRIPPER.ignore_images = True
HTML_STRIPPER.ignore_emphasis = True
def SaveParsed():
with open("parsed_threads.json", "w") as file:
json.dump({"ids": list(PARSED_TREADS)}, file)
def ThreadGood(thread):
return thread.reply_count > 20
def AppendDialogues(id, dialogues):
with open("dialogues.jsonl", "a") as file:
for dialogue in dialogues:
json.dump({"id": id, "dialogue": dialogue}, file, ensure_ascii=False)
file.write("\n")
def FilterAndPrepareText(text):
text = HTML_STRIPPER.handle(text)
text = text.lower()
text = text.replace("\n", " ")
text = text.replace("(op)", "")
if text.find("бамп") != -1:
return
text = re.sub(' +', ' ', text)
text = re.sub('^ ', '', text)
text = re.sub(' $', '', text)
return text
class Post:
def __init__(self, idx: int, text: str):
self.idx = idx
self.parents = []
while True:
reply_match = REPLY_RE.match(text)
if reply_match is not None:
parent_id = int(reply_match.group(0)[2:])
span = reply_match.span()
self.parents.append(parent_id)
text = text[:span[0]] + text[span[1]:]
text = re.sub('^ ', '', text)
text = re.sub(' $', '', text)
else:
break
self.text = text
self.children = []
self.dialogue = [self.text]
self.appeared = False
def __repr__(self):
return {"idx": self.idx, "parents": self.parents, "text": self.text}.__repr__()
def build_random_dialogue(self, context):
self.appeared = True
if len(self.dialogue) != 1:
return self.dialogue
if len(self.parents) == 0:
return self.dialogue
chosen_parent = self.parents[0]
if chosen_parent in context.keys():
self.dialogue.extend(context[chosen_parent].build_random_dialogue(context))
return self.dialogue
def BuildDialogues(thread):
posts = {}
for post in thread:
idx = post.num
text = FilterAndPrepareText(post.comment)
if text is not None:
posts[idx] = Post(idx, text)
for _, post in reversed(posts.items()):
if not post.appeared:
post.build_random_dialogue(posts)
return [post.dialogue for post in list(posts.values()) if len(post.dialogue) > 1]
def main():
print("Started collecting...")
while True:
while True:
try:
board = API.get_board()
break
except:
print(f"Failed to fetch board. Sleeping for 60s")
sleep(60)
print("Got board")
for thread in board:
if thread.num not in PARSED_TREADS and ThreadGood(thread):
PARSED_TREADS.add(thread.num)
try:
thread = API.get_thread(thread.num)
except:
continue
dialogues = BuildDialogues(thread)
AppendDialogues(thread[0].num, dialogues)
print("Parsed")
SaveParsed()
print("Saved. Sleeping for 10m")
sleep(600)
if __name__ == "__main__":
main()
|