markov-quotes.py

# Dependencies: pip install markovify python-telegram-bot

import logging
import json

import telegram
import markovify
from telegram.ext import CommandHandler, Updater, Filters

TOKEN='.........'

logging.basicConfig(format='%(asctime)s - %(name)s - %(levelname)s - %(message)s', level=logging.INFO)

msg_agg = json.load(open('messages.json'))

models = {}
all_messages = []
for username, messages in msg_agg.items():
    models[username] = markovify.NewlineText('\n'.join(messages))
    all_messages.extend(messages)
all_model = markovify.NewlineText('\n'.join(all_messages), 2)


updater = Updater(token=TOKEN)
dispatcher = updater.dispatcher

bot = telegram.Bot(token=TOKEN)

def chat_name(update):
    if update.message.chat.type == 'private':
        return 'private'
    else:
        return update.message.chat.title

def start(bot, update):
    logging.info('[%s] [%s] /start', chat_name(update), update.message.from_user.name)
    bot.send_message(chat_id=update.message.chat_id, text="Hi there. Type /quote <username>")

start_handler = CommandHandler('start', start)
dispatcher.add_handler(start_handler)

def quote(bot, update, args):
    logging.info('[%s] [%s] %s', chat_name(update), update.message.from_user.name, update.message.text)
    username = None

    if args:
        username = args[0]
        username = username.strip()
        username = username.replace('@', '')

    if username:
        if not username in models:
            bot.send_message(chat_id=update.message.chat_id, text='unknown user')
            return
        message = models[username].make_sentence(tries=100)
    else:
        message = all_model.make_sentence(tries=100)

    if not message:
        message = 'Что-то ничего не приходит в голову'

    logging.info('[%s] Sending %s', chat_name(update), message)
    bot.send_message(chat_id=update.message.chat_id, text=message)

quote_handler = CommandHandler('quote', quote, pass_args=True)
dispatcher.add_handler(quote_handler)

updater.start_polling()



################################
# update.py
from telethon import TelegramClient
client = TelegramClient('session_name', 'app id', 'key')
history = client.get_message_history(rt.entity, limit=10000)

msg_agg = {}
for msg in history:
    text = msg.message or ''
    text = text.strip()
    if not text:
        continue
    user_history = msg_agg.setdefault(msg.sender.username, [])
    user_history.append(msg.message)

print(sorted(((len(msgs), username) for username, msgs in msg_agg.items()), reverse=True))

with open('messages.json', 'w') as f:
    json.dump(msg_agg, f, indent=True, ensure_ascii=False)