telegram_bots/feeds/modules/tapas.py

import json
import os
import re
from tempfile import TemporaryDirectory

import sentry_sdk
from django.db import models
from djconfig import config
from telebot import TeleBot
from telebot.types import InputMediaPhoto, InputMediaVideo
from bs4 import BeautifulSoup
import requests
from PIL import Image

from feeds.models import FeedModuleConfig


class TapasFeedModuleConfig(FeedModuleConfig):
    display_name = models.CharField(max_length=256)

    MODULE_NAME = 'Tapas.io comic'

    def execute(self, bot: TeleBot, chat_id, last_id):
        config._reload_maybe()

        if last_id is None:
            last_id = 0

        series_page = requests.get(
            f'https://tapas.io/series/{self.display_name}',
            headers={'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:67.0) Gecko/20100101 Firefox/67.0'},
        ).text
        ssoup = BeautifulSoup(series_page, 'html.parser')
        # episode_list = json.loads(re.search(r'episodeList : (\[.*\]),', series_page).group(1))
        for episode in ssoup.select('a[data-ga-category=Episode]'):
            eid = int(episode['data-id'])
            title = episode.select_one('.info__title').text.strip()
            if eid <= last_id:
                continue
            # if not episode['free']:
            #     continue

            caption = f'{title}\nhttps://tapas.io/episode/{eid}'

            esoup = BeautifulSoup(requests.get(
                f'https://tapas.io/episode/{eid}',
                headers={'User-Agent': 'Mozilla/5.0 (X11; Linux x86_64; rv:67.0) Gecko/20100101 Firefox/67.0'},
            ).text, 'html.parser')
            imgs = []
            for img in esoup.select(f'.js-episode-article .content__img'):
                imgs.append(img['data-src'])

            if len(imgs) == 1:
                bot.send_photo(chat_id, imgs[0], caption=caption)
            if len(imgs) > 1:
                with TemporaryDirectory() as d:
                    ims = []
                    for i, img in enumerate(imgs):
                        img_path = os.path.join(d, f'{i}.jpg')
                        with open(img_path, 'wb') as f:
                            f.write(requests.get(img).content)
                        ims.append(Image.open(img_path))
                    widths, heights = zip(*(i.size for i in ims))
                    max_width = max(widths)
                    total_height = sum(heights)
                    new_im = Image.new('RGB', (max_width, total_height))
                    offset = 0
                    for im in ims:
                        new_im.paste(im, (0, offset))
                        offset += im.size[1]
                    new_im_path = os.path.join(d, 'combined.jpg')
                    new_im.save(new_im_path)
                    with open(new_im_path, 'rb') as f:
                        bot.send_document(chat_id, f, caption=caption)
            yield eid