Telegram Relation Graph

crawl.py

# requirements: telethon, networkx, python-pysocks[asyncio], pygraphviz, tqdm

from telethon.client import TelegramClient
from telethon.tl.types import User as TgUser, Channel as TgChannel, Chat as TgChat
import telethon
import asyncio
import networkx as nx
from tqdm import tqdm
import pickle

client = TelegramClient(
    api_id=114514,
    api_hash='xxxxxxxxxxxxxxxxxxxxxxxxxxxxx',
    session='relation-bot',
    proxy=('socks5', 'localhost', 1080)
).start()


def format_entity_name(entity):
    if isinstance(entity, TgUser):
        first_name = entity.first_name or ''
        last_name = entity.last_name or ''
        return (first_name + ' ' + last_name).strip()
    elif isinstance(entity, TgChat) or isinstance(entity, TgChannel):
        return entity.title
    else:
        raise ValueError(f'Unknown entity {entity}')


async def start():
    g = nx.Graph()
    for dialog in tqdm(await client.get_dialogs()):
        g.add_node(dialog.entity.id, _type='Chat', _name=format_entity_name(dialog.entity))
        try:
            async for person in client.iter_participants(dialog):
                if dialog.entity.id != person.id:
                    g.add_node(person.id, _type='Person', _name=format_entity_name(person))
                    g.add_edge(dialog.entity.id, person.id)
        except telethon.errors.rpcerrorlist.ChatAdminRequiredError:
            pass

    print(g.degree)
    with open('relation.pickle', 'wb') as fp:
        pickle.dump(g, fp)


loop = asyncio.get_event_loop()
loop.run_until_complete(start())

vis.py

import networkx
import pickle
import graphistry
import pandas as pd

if __name__ == '__main__':
    with open('relation.pickle', 'rb') as fp:
        g: networkx.Graph = pickle.load(fp)
    for n, deg in list(g.degree):
        if deg >= 2000:
            g.remove_node(n)
    for n, deg in list(g.degree):
        if deg <= 1:
            g.remove_node(n)
    labels = []
    for n in g.nodes:
        labels.append(g.nodes[n]['_name'])

    reverse_edges = [(e[1], e[0]) for e in g.edges]
    all_edges = list(g.edges) + reverse_edges

    edges = pd.DataFrame(all_edges, columns=['chat', 'user'])
    nodes = pd.DataFrame({'id': g.nodes, 'name': labels})

    graphistry.register(api=3, protocol="https", server="hub.graphistry.com", username="<your-username>", password="<your-password>")

    gra = graphistry.nodes(nodes, 'id').edges(edges).bind(source='chat', destination='user', point_title='name')
    gra.plot()