add extract url from message

.gitignore

@@ -3,4 +3,4 @@ __pycache__/
 cert/
 data/
 docker-compose.yml
-x.py
+utils/x.py

Dockerfile

@@ -7,7 +7,7 @@ ENV PYTHONUNBUFFERED=1
 
 RUN set -eux; \
 	apt-get update; \
-	apt-get install -y gosu; \
+	apt-get install -y git gosu; \
 	rm -rf /var/lib/apt/lists/*; \
 # verify that the binary works
 	gosu nobody true

common.py

@@ -1,9 +1,8 @@
-import logging
 import os
-import re
 
 try:
-    import uvloop, asyncio
+    import uvloop
+    import asyncio
 
     asyncio.set_event_loop_policy(uvloop.EventLoopPolicy())
 except ImportError:
@@ -12,6 +11,8 @@ except ImportError:
 BOT_TOKEN = os.getenv("BOT_TOKEN")
 ADMIN = [int(i) for i in os.getenv("BOT_ADMIN").split(",")]
 
+PIXIV_REFRESH_TOKEN = os.getenv("PIXIV_REFRESH_TOKEN")
+
 WEBHOOK = os.getenv("WEBHOOK", False)
 if WEBHOOK:
     WEBHOOK_LISTEN = os.getenv("WEBHOOK_LISTEN", "0.0.0.0")
@@ -20,17 +21,3 @@ if WEBHOOK:
     WEBHOOK_KEY = os.getenv("WEBHOOK_KEY", "cert/private.key")
     WEBHOOK_CERT = os.getenv("WEBHOOK_CERT", "cert/cert.pem")
     WEBHOOK_SECRET_TOKEN = os.getenv("WEBHOOK_SECRET_TOKEN")
-
-x_url_regex = re.compile(r"^(?:https?://)(?:www\.|mobile\.|)(?:x|twitter|fixvx|vxtwitter)\.com/(.+)/status/(\d+)")
-x_media_regex = re.compile(r"^(?:https?://)(pbs|video)\.twimg\.com/(.*)")
-x_tco_regex = re.compile(r"(?:https?://)t\.co/.+$", re.M)
-message_url_regex = re.compile(r"\[.+]", re.S)
-
-logging.basicConfig(
-    level=os.getenv("LOG_LEVEL", "WARNING"),
-    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
-)
-
-
-def get_logger(name: str) -> logging.Logger:
-    return logging.getLogger(name)

docker-compose.yml.example

@@ -1,12 +1,13 @@
 services:
   tgxmb: image: yoursfunny/telegram-twitter-media-bot: latest
     restart: always
-    ports:
-      - "8443:8443"
+#    ports:
+#      - "8443:8443"
     environment:
       LOCAL_USER_ID: '1000'
       BOT_TOKEN: ''
       BOT_ADMIN: ''
+      PIXIV_REFRESH_TOKEN: ''
       WEBHOOK: false
       WEBHOOK_LISTEN: '127.0.0.1'
       WEBHOOK_PORT: 8443

main.py

@@ -4,25 +4,30 @@ import html
 from functools import wraps
 from typing import TYPE_CHECKING
 
-from telegram import InlineKeyboardButton, InlineKeyboardMarkup
+from telegram import InlineKeyboardButton, InlineKeyboardMarkup, MessageEntity
 from telegram.constants import ChatAction, ChatType, ParseMode
-from telegram.ext import (ApplicationBuilder, CallbackQueryHandler, CommandHandler, Defaults,
+from telegram.ext import (ApplicationBuilder, CallbackQueryHandler, CommandHandler, ContextTypes, Defaults,
                           InlineQueryHandler, MessageHandler, PicklePersistence, filters)
 
 import common
-from tweet import Telegram
+import utils.regex as regex
+from utils.context import ChatData, CustomContext, EditMessage
+from utils.logger import get_logger
+from utils.net import NetClient
+from utils.pixiv import ProcessPixiv
+from utils.telegram import Telegram
 
 if TYPE_CHECKING:
-    from telegram import Chat, Message, Update
-    from telegram.ext import Application, ContextTypes
+    from telegram import Message, Update
+    from telegram.ext import Application
 
-logger = common.get_logger(__name__)
+logger = get_logger(__name__)
 
 
 def send_action(action):
     def decorator(func):
         @wraps(func)
-        async def command_func(update: Update, context: ContextTypes.DEFAULT_TYPE, *args, **kwargs):
+        async def command_func(update: Update, context: CustomContext, *args, **kwargs):
             await update.effective_chat.send_action(action)
             return await func(update, context, *args, **kwargs)
 
@@ -31,27 +36,39 @@ def send_action(action):
     return decorator
 
 
-async def inline_query(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
+def extract_urls(message: Message) -> set[str]:
+    types = [MessageEntity.URL, MessageEntity.TEXT_LINK]
+    res = message.parse_entities(types)
+    res.update(message.parse_caption_entities(types))
+    res.update({key: key.url for key in res if key.type == MessageEntity.TEXT_LINK})
+    return set(res.values())
+
+
+async def inline_query(update: Update, context: CustomContext) -> None:
     query = update.inline_query.query
     if query == "":
         return
     logger.info(f"Query: {query}")
     async with Telegram(query) as tweet:
-        result = list(tweet.inline_query_generator)
-        await update.inline_query.answer(result)
+        if not tweet:
+            return
+        await update.inline_query.answer(tweet.inline_query_result())
 
 
 @send_action(ChatAction.UPLOAD_PHOTO)
-async def url_media(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
-    url = update.message.text
-    logger.info(f"Receiving url: {url}")
+async def url_media(update: Update, context: CustomContext, url: str) -> None:
     async with Telegram(url) as tweet:
-        media = list(tweet.message_media_generator)
+        if not tweet:
+            return
+        media = tweet.message_media_result()
+        if not media:
+            await update.effective_message.reply_text("No media found or media type is not supported.")
+            return
         message_to_send = await update.effective_message.reply_media_group(
             media,
             caption=tweet.message_text,
             reply_to_message_id=update.message.message_id,
-        ) if not isinstance(tuple, media[0]) else await update.effective_message.reply_animation(
+        ) if not isinstance(media[0], tuple) else await update.effective_message.reply_animation(
             media[0][0],
             caption=tweet.message_text,
             reply_to_message_id=update.message.message_id,
@@ -60,80 +77,94 @@ async def url_media(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
         if not isinstance(message_to_send, tuple):
             message_to_send = (message_to_send,)
         url = tweet.url
-    if context.user_data.get('edit_before_forward', False):
+    if context.chat_data.edit_before_forward:
         message_reply = await update.effective_message.reply_text(
-            "Reply to edit message. [URL]",
-            reply_markup=InlineKeyboardMarkup.from_button(
-                InlineKeyboardButton("↩️ Confirm", callback_data="forward")
+            "Reply to edit message.",
+            reply_markup=InlineKeyboardMarkup.from_column(
+                [InlineKeyboardButton(name, callback_data=f"template|{name}") for name in
+                 context.chat_data.template.keys()] + [InlineKeyboardButton("↩️ Confirm", callback_data="forward")]
             ),
             reply_to_message_id=update.message.message_id,
         )
-        context.user_data['message_reply'] = message_reply
-        context.user_data['message_to_send'] = message_to_send
-        context.user_data['message_url'] = url
+        context.chat_data.edit_message[message_reply.id] = EditMessage(
+            url=url,
+            forward=message_to_send
+        )
         return
-    if 'forward_channel_id' in context.user_data:
+    if context.chat_data.forward_channel_id:
         await forward_message(update, context, message_to_send)
 
 
+async def handel_url_media(update: Update, context: CustomContext) -> None:
+    url = update.message.text
+    logger.info(f"Receiving url: {url}")
+    await url_media(update, context, url)
+
+
 async def forward_message(
         update: Update,
-        context: ContextTypes.DEFAULT_TYPE,
+        context: CustomContext,
         message_to_send: tuple[Message, ...],
 ) -> None:
     try:
         await update.effective_chat.copy_messages(
-            context.user_data['forward_channel_id'],
+            context.chat_data.forward_channel_id,
             [m.id for m in message_to_send]
         )
     except Exception as e:
         await update.effective_message.reply_text(str(e))
 
 
-async def edit_message(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
-    if 'message_reply' not in context.user_data:
-        return
-    if update.message.reply_to_message != context.user_data['message_reply']:
-        return
-    template = context.user_data.get('template', None)
-    message_url = '<a href="{0}">{1}</a>'
-    url = context.user_data['message_url']
-    if template:
-        update_text = template.replace("[]", message_url.format(
-            url,
-            html.escape(update.message.text)
-        ))
-    else:
-        update_text = html.escape(update.message.text)
-        match = common.message_url_regex.search(update_text)
-        if match:
-            match = match.span()
-            update_text = update_text[:match[0]] + message_url.format(
-                url,
-                update_text[match[0] + 1:match[1] - 1]
-            ) + update_text[match[1]:]
-    message_to_send = context.user_data['message_to_send']
-    await message_to_send[0].edit_caption(update_text)
+async def edit_message(update: Update, context: CustomContext) -> bool:
+    if not (reply := update.message.reply_to_message):
+        return False
+    _edit_message = context.chat_data.edit_message.get(reply.id, None)
+    if not _edit_message:
+        return False
+    new_text = '<a href="{0}">{1}</a>'.format(
+        _edit_message.url,
+        html.escape(update.message.text)
+    )
+    update_text = context.chat_data.template[template].replace("[]", new_text) if (
+        template := _edit_message.template) else new_text
+    await _edit_message.forward[0].edit_caption(update_text)
+    return True
 
 
-async def query_forward_message(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
-    message_to_send = context.user_data['message_to_send']
-    await forward_message(update, context, message_to_send)
+async def handle_message(update: Update, context: CustomContext) -> None:
+    if edit_message(update, context):
+        return
+    if not (urls := extract_urls(update.message)):
+        return
+    for url in urls:
+        await url_media(update, context, url)
+
+
+async def query_forward_message(update: Update, context: CustomContext) -> None:
+    _edit_message = context.chat_data.edit_message[update.effective_message.id]
+    await forward_message(update, context, _edit_message.forward)
     await update.callback_query.answer('✅ Forwarded')
     await update.callback_query.delete_message()
-    del context.user_data['message_reply']
-    del context.user_data['message_to_send']
-    del context.user_data['message_url']
+    del _edit_message
+
+
+async def query_template(update: Update, context: CustomContext) -> None:
+    query = update.callback_query
+    await query.answer()
+    name = query.data.split("|")[1]
+    _edit_message = context.chat_data.edit_message[query.message.message_id]
+    _edit_message.template = name
+    await _edit_message.forward[0].edit_caption(context.chat_data.template[name])
 
 
 @send_action(ChatAction.TYPING)
-async def cmd_set_forward_channel(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
+async def cmd_set_forward_channel(update: Update, context: CustomContext) -> None:
     if not context.args:
         await update.effective_message.reply_text("Please provide a channel username or id.")
         return
     channel = context.args[0]
     try:
-        channel: Chat = await context.bot.get_chat(channel)
+        channel = await context.bot.get_chat(channel)
     except Exception as e:
         await update.effective_message.reply_text(str(e))
         return
@@ -145,55 +176,60 @@ async def cmd_set_forward_channel(update: Update, context: ContextTypes.DEFAULT_
     except Exception as e:
         await update.effective_message.reply_text(str(e) + "\nPlease add the bot to the channel and set as admin")
         return
+    user = filter(lambda x: x.user.id == update.effective_user.id, channel_admin)
+    user = next(user, None)
+    if not user:
+        await update.effective_message.reply_text("You are not an admin of the channel.")
+        return
     user_bot = filter(lambda x: x.user.id == context.bot.id, channel_admin)
     user_bot = next(user_bot, None)
     if user_bot.can_post_messages:
-        context.user_data['forward_channel_id'] = channel.id
+        context.chat_data.forward_channel_id = channel.id
         await update.effective_message.reply_text("Add successfully.")
 
 
 @send_action(ChatAction.TYPING)
-async def cmd_remove_forward_channel(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
-    if 'forward_channel_id' in context.user_data:
-        del context.user_data['forward_channel_id']
+async def cmd_remove_forward_channel(update: Update, context: CustomContext) -> None:
+    if context.chat_data.forward_channel_id:
+        context.chat_data.forward_channel_id = None
         await update.effective_message.reply_text("Remove successfully.")
         return
     await update.effective_message.reply_text("No channel to remove.")
 
 
 @send_action(ChatAction.TYPING)
-async def cmd_edit_before_forward(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
-    if context.user_data.get('forward_channel_id', None) is None:
+async def cmd_edit_before_forward(update: Update, context: CustomContext) -> None:
+    if context.chat_data.forward_channel_id is None:
         await update.effective_message.reply_text("Please enable forward channel first.")
         return
-    ebf_status = context.user_data.get('edit_before_forward', False)
-    if ebf_status:
-        context.user_data['edit_before_forward'] = False
-        context.user_data.pop('message_reply', None)
-        context.user_data.pop('message_to_send', None)
-        context.user_data.pop('message_url', None)
+    if context.chat_data.edit_before_forward:
+        context.chat_data.edit_before_forward = False
+        context.chat_data.edit_message.clear()
         await update.effective_message.reply_text("Disable edit before forward.")
         return
-    context.user_data['edit_before_forward'] = True
+    context.chat_data.edit_before_forward = True
     await update.effective_message.reply_text("Enable edit before forward.")
 
 
 @send_action(ChatAction.TYPING)
-async def cmd_set_template(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
+async def cmd_set_template(update: Update, context: CustomContext) -> None:
     reply = update.effective_message.reply_to_message
     if not reply:
         await update.effective_message.reply_text("Please reply to a message to set as template.")
         return
-    if '[]' not in reply.text_html:
+    if '[]' not in (template := reply.text_html):
         await update.effective_message.reply_text("Please reply to a message with [] to set as template.")
         return
-    context.user_data['template'] = reply.text_html
+    if not context.args:
+        await update.effective_message.reply_text("Please provide a name for the template.")
+        return
+    context.chat_data.template[''.join(context.args)] = template
     await update.effective_message.reply_text("Template set.")
 
 
 @send_action(ChatAction.TYPING)
-async def cmd_user_dict(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
-    await update.effective_message.reply_text(str(context.user_data))
+async def cmd_user_dict(update: Update, context: CustomContext) -> None:
+    await update.effective_message.reply_text(html.escape(str(context.chat_data)), disable_web_page_preview=True)
 
 
 async def post_init(application: Application) -> None:
@@ -204,7 +240,9 @@ async def post_init(application: Application) -> None:
     DESCRIPTION = "A bot to fetch tweets from Twitter."
     await application.bot.set_my_description(DESCRIPTION)
     await application.bot.set_my_short_description(DESCRIPTION)
-    Telegram.init_client()
+    NetClient.init_client()
+    if common.PIXIV_REFRESH_TOKEN:
+        await ProcessPixiv.init_client(common.PIXIV_REFRESH_TOKEN)
 
 
 async def post_stop(application: Application) -> None:
@@ -212,7 +250,7 @@ async def post_stop(application: Application) -> None:
 
 
 async def post_shutdown(application: Application) -> None:
-    await Telegram.close_client()
+    await NetClient.close_client()
 
 
 def main():
@@ -222,6 +260,7 @@ def main():
                    .token(common.BOT_TOKEN)
                    .defaults(defaults)
                    .persistence(persistence)
+                   .context_types(ContextTypes(context=CustomContext, chat_data=ChatData))
                    .post_init(post_init)
                    .post_stop(post_stop)
                    .post_shutdown(post_shutdown)
@@ -234,15 +273,17 @@ def main():
     user_filter.add_user_ids(common.ADMIN)
 
     handlers = [
-        InlineQueryHandler(inline_query, common.x_url_regex),
-        MessageHandler(filters.Regex(common.x_url_regex) & filters.ChatType.PRIVATE, url_media),
+        InlineQueryHandler(inline_query),
+        MessageHandler((filters.Regex(regex.x_url) | filters.Regex(regex.pixiv_url)) & filters.ChatType.PRIVATE,
+                       handel_url_media),
         CommandHandler("set_forward_channel", cmd_set_forward_channel),
         CommandHandler("remove_forward_channel", cmd_remove_forward_channel),
         CommandHandler("edit_before_forward", cmd_edit_before_forward),
         CommandHandler("set_template", cmd_set_template),
         MessageHandler(~filters.COMMAND & filters.ChatType.PRIVATE, edit_message),
         CallbackQueryHandler(query_forward_message, pattern="forward"),
-        CommandHandler("bot_dict", cmd_user_dict, filters=user_filter),
+        CallbackQueryHandler(query_template, pattern=r"^template\|"),
+        CommandHandler("bot_dict", cmd_user_dict),
     ]
 
     application.add_handlers(handlers)

requirements.txt

@@ -1,3 +1,4 @@
-python-telegram-bot[webhooks]~=21.3
-httpx[http2]~=0.27.0
-uvloop~=0.19.0; sys_platform != 'win32'
+python-telegram-bot[webhooks]~=21.4
+httpx[http2]~=0.27
+uvloop~=0.19.0; sys_platform != 'win32'
+async-pixiv @ git+https://github.com/TheFunny/async-pixiv@main

tweet.py

@@ -1,282 +0,0 @@
-from __future__ import annotations
-
-import html
-from functools import cached_property
-from typing import Generator, TYPE_CHECKING
-from uuid import uuid4
-
-from httpx import AsyncClient
-from telegram import (InlineQueryResultMpeg4Gif, InlineQueryResultPhoto, InlineQueryResultVideo, InputMediaPhoto,
-                      InputMediaVideo)
-
-from common import get_logger, x_media_regex, x_tco_regex, x_url_regex
-
-if TYPE_CHECKING:
-    from types import TweetInfo, TypeInlineQueryResult, TypeMessageMediaResult
-
-logger = get_logger(__name__)
-
-
-twimg_url = 'https://pbs.twimg.com/'
-vx_api_url = 'https://api.vxtwitter.com/{0}/status/{1}'
-
-message_raw_text = """{url}
-<a href="{author_url}">{author}</a>: {text}
-"""
-
-
-def create_client() -> AsyncClient:
-    return AsyncClient(http2=True)
-
-
-async def close_client(_client: AsyncClient) -> None:
-    await _client.aclose()
-
-
-async def fetch_json(_client: AsyncClient, url: str) -> dict:
-    logger.info(f"Fetching {url}")
-    response = await _client.get(url)
-    assert response.is_success, f"Failed to fetch {url}, status code {response.status_code}"
-    return response.json()
-
-
-class TweetMedia:
-    __slots__ = ('_url', '_thumb', '_type', '__dict__')
-
-    def __init__(self, url: str, thumb: str, media_type: str):
-        self._url: str = url
-        self._thumb: str = thumb
-        self._type: str = media_type
-
-    def __str__(self):
-        return f"TweetMedia(url={self.url}, thumb={self.thumb}, type={self.type})"
-
-    @cached_property
-    def _uri(self) -> str | None:
-        if match := x_media_regex.match(self._url):
-            return match.group(2).removesuffix('.jpg').removesuffix('.png')
-        return None
-
-    @cached_property
-    def url(self) -> str:
-        match self._type:
-            case "image":
-                return f"{twimg_url}{self._uri}?format=jpg&name=4096x4096"
-            case "video":
-                return self._url
-            case "gif":
-                return self._url
-            case _:
-                return self._url
-
-    @cached_property
-    def thumb(self) -> str:
-        match self._type:
-            case "image":
-                return f"{twimg_url}{self._uri}?format=jpg&name=thumb"
-            case "video":
-                return self._thumb
-            case "gif":
-                return self._thumb
-            case _:
-                return self._thumb
-
-    @property
-    def type(self) -> str:
-        return self._type
-
-
-class Tweet:
-    __slots__ = ('_id', '_author', '_author_id', '_text', '_media', '_sensitive', '__dict__')
-
-    def __init__(
-            self,
-            tweet_id: str,
-            author: str,
-            author_id: str,
-            text: str,
-            media: list[TweetMedia],
-            sensitive: bool = False
-    ):
-        self._id: str = tweet_id
-        self._author: str = author
-        self._author_id: str = author_id
-        self._text: str = text
-        self._media: list[TweetMedia] = media
-        self._sensitive: bool = sensitive
-
-    @property
-    def id(self) -> str:
-        return self._id
-
-    @cached_property
-    def url(self) -> str:
-        return f"https://twitter.com/{self._author_id}/status/{self._id}"
-
-    @property
-    def author(self) -> str:
-        return self._author
-
-    @cached_property
-    def author_url(self) -> str:
-        return f"https://twitter.com/{self._author_id}"
-
-    @property
-    def text(self) -> str:
-        return self._text
-
-    @property
-    def media(self) -> list[TweetMedia]:
-        return self._media
-
-    @property
-    def sensitive(self) -> bool:
-        return self._sensitive
-
-
-class ProcessTweet:
-    __slots__ = ('_httpx_client', '_url', '_tweet')
-
-    def __init__(self, httpx_client: AsyncClient, url: str):
-        self._httpx_client: AsyncClient = httpx_client
-        self._url: str = url
-
-    async def __aenter__(self):
-        self._tweet = await self._fetch_tweet()
-        return Tweet(
-            tweet_id=self._tweet["tweetID"],
-            author=self._tweet["user_name"],
-            author_id=self._tweet["user_screen_name"],
-            text=self._tweet_text,
-            media=self._tweet_media,
-            sensitive=self._tweet["possibly_sensitive"]
-        )
-
-    async def __aexit__(self, exc_type, exc_val, exc_tb):
-        pass
-
-    async def _fetch_tweet(self) -> TweetInfo:
-        match = x_url_regex.match(self._url)
-        assert match, f"Invalid URL: {self._url}"
-        auther_id, tweet_id = match.group()
-        return await fetch_json(self._httpx_client, vx_api_url.format(auther_id, tweet_id))
-
-    @property
-    def _tweet_text(self) -> str:
-        match = x_tco_regex.search(self._tweet['text'])
-        return self._tweet['text'][:match.start()].strip(" ") if match else self._tweet['text']
-
-    @property
-    def _tweet_media(self) -> list[TweetMedia]:
-        return [
-            TweetMedia(
-                url=tweet_media['url'],
-                thumb=tweet_media['thumbnail_url'],
-                media_type=tweet_media['type']
-            )
-            for tweet_media in self._tweet['media_extended']
-        ]
-
-
-class TelegramTweet:
-    __slots__ = ('_httpx_client', '_url', '_tweet', '__dict__')
-
-    def __init__(self, httpx_client: AsyncClient, url: str):
-        self._httpx_client: AsyncClient = httpx_client
-        self._url: str = url
-
-    async def __aenter__(self):
-        async with ProcessTweet(self._httpx_client, self._url) as tweet:
-            self._tweet = tweet
-        return self
-
-    async def __aexit__(self, exc_type, exc_val, exc_tb):
-        pass
-
-    @cached_property
-    def url(self) -> str:
-        return self._tweet.url
-
-    @cached_property
-    def message_text(self) -> str:
-        tweet = self._tweet
-        return message_raw_text.format(
-            url=tweet.url,
-            author_url=tweet.author_url,
-            author=html.escape(tweet.author),
-            text=html.escape(tweet.text)
-        )
-
-    @property
-    def inline_query_generator(self) -> Generator[TypeInlineQueryResult, None, None]:
-        tweet = self._tweet
-        for tweet_media in tweet.media:
-            logger.info(str(tweet_media))
-            if tweet_media.type == "image":
-                yield InlineQueryResultPhoto(
-                    id=str(uuid4()),
-                    photo_url=tweet_media.url,
-                    thumbnail_url=tweet_media.thumb,
-                    caption=self.message_text
-                )
-            elif tweet_media.type == "video":
-                yield InlineQueryResultVideo(
-                    id=str(uuid4()),
-                    video_url=tweet_media.url,
-                    mime_type="video/mp4",
-                    thumbnail_url=tweet_media.thumb,
-                    title=tweet.text,
-                    caption=self.message_text
-                )
-            elif tweet_media.type == "gif":
-                yield InlineQueryResultMpeg4Gif(
-                    id=str(uuid4()),
-                    mpeg4_url=tweet_media.url,
-                    thumbnail_url=tweet_media.thumb,
-                    caption=self.message_text
-                )
-
-    @property
-    def message_media_generator(self) -> Generator[TypeMessageMediaResult, None, None]:
-        tweet = self._tweet
-        for tweet_media in tweet.media:
-            logger.info(str(tweet_media))
-            if tweet_media.type == "image":
-                yield InputMediaPhoto(
-                    media=tweet_media.url,
-                    has_spoiler=tweet.sensitive
-                )
-            elif tweet_media.type == "video":
-                yield InputMediaVideo(
-                    media=tweet_media.url,
-                    has_spoiler=tweet.sensitive,
-                    thumbnail=tweet_media.thumb
-                )
-            elif tweet_media.type == "gif":
-                if len(tweet.media) == 1:
-                    yield tweet_media.url, tweet.sensitive
-                yield InputMediaVideo(
-                    media=tweet_media.url,
-                    has_spoiler=tweet.sensitive,
-                    thumbnail=tweet_media.thumb
-                )
-
-
-class Telegram:
-    _httpx_client: AsyncClient
-
-    def __init__(self, url: str):
-        self._url = url
-
-    async def __aenter__(self):
-        if x_url_regex.match(self._url):
-            async with TelegramTweet(self._httpx_client, self._url) as tweet:
-                return tweet
-
-    @classmethod
-    def init_client(cls) -> None:
-        cls._httpx_client = create_client()
-
-    @classmethod
-    async def close_client(cls) -> None:
-        await close_client(cls._httpx_client)

types.py

@@ -1,19 +0,0 @@
-from typing import Type, TypedDict
-
-from telegram import InlineQueryResultMpeg4Gif, InlineQueryResultPhoto, InlineQueryResultVideo, InputMediaPhoto, \
-    InputMediaVideo
-
-TypeInlineQueryResult: Type[
-    InlineQueryResultMpeg4Gif | InlineQueryResultPhoto | InlineQueryResultVideo] = InlineQueryResultMpeg4Gif | InlineQueryResultPhoto | InlineQueryResultVideo
-InputMediaAnimation = tuple[str, bool]
-TypeMessageMediaResult: Type[
-    InputMediaPhoto | InputMediaVideo | InputMediaAnimation] = InputMediaPhoto | InputMediaVideo | InputMediaAnimation
-
-
-class TweetInfo(TypedDict):
-    tweetID: str
-    user_name: str
-    user_screen_name: str
-    text: str
-    media_extended: list[dict]
-    possibly_sensitive: bool

utils/context.py

@@ -0,0 +1,42 @@
+import dataclasses
+from typing import Optional
+
+from telegram import Message
+from telegram.ext import Application, CallbackContext, ExtBot
+
+
+@dataclasses.dataclass(repr=False)
+class EditMessage:
+    url: str
+    forward: tuple[Message, ...]
+    template: str = ""
+
+    def __str__(self):
+        forward = ", ".join(f"Message({f.id})" for f in self.forward)
+        return f"EditMessage(url={self.url}, forward={forward}, template={self.template})"
+
+    __repr__ = __str__
+
+
+class ChatData:
+    def __init__(self):
+        self.forward_channel_id: Optional[int] = None
+        self.edit_before_forward: bool = False
+        self.edit_message: dict[int, EditMessage] = {}
+        self.template: dict[str, str] = {}
+
+    def __str__(self):
+        return f"ChatData(forward_channel_id={self.forward_channel_id}, edit_before_forward={self.edit_before_forward}, " \
+               f"edit_message={self.edit_message}, template={self.template})"
+
+    __repr__ = __str__
+
+
+class CustomContext(CallbackContext[ExtBot, dict, ChatData, dict]):
+    def __init__(
+            self,
+            application: Application,
+            chat_id: Optional[int] = None,
+            user_id: Optional[int] = None
+    ):
+        super().__init__(application=application, chat_id=chat_id, user_id=user_id)

utils/logger.py

@@ -0,0 +1,11 @@
+import logging
+import os
+
+logging.basicConfig(
+    level=os.getenv("LOG_LEVEL", "WARNING"),
+    format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
+)
+
+
+def get_logger(name: str) -> logging.Logger:
+    return logging.getLogger(name)

utils/net.py

@@ -0,0 +1,37 @@
+from __future__ import annotations
+
+from httpx import AsyncClient
+
+
+def create_client() -> AsyncClient:
+    return AsyncClient(http2=True)
+
+
+async def close_client(_client: AsyncClient) -> None:
+    return await _client.aclose()
+
+
+async def fetch_json(_client: AsyncClient, url: str) -> dict:
+    response = await _client.get(url)
+    assert response.is_success, f"Failed to fetch {url}, status code {response.status_code}"
+    return response.json()
+
+
+class NetClient:
+    _httpx_client: AsyncClient
+
+    @classmethod
+    def init_client(cls) -> None:
+        cls._httpx_client = create_client()
+
+    @classmethod
+    async def close_client(cls) -> None:
+        await close_client(cls._httpx_client)
+
+    @classmethod
+    def get_client(cls) -> AsyncClient:
+        return cls._httpx_client
+
+    @classmethod
+    async def fetch_json(cls, url: str) -> dict:
+        return await fetch_json(cls._httpx_client, url)

utils/pixiv.py

@@ -0,0 +1,141 @@
+from __future__ import annotations
+
+from typing import Literal, TYPE_CHECKING
+
+from async_pixiv import PixivClient
+from async_pixiv.error import ApiError
+
+if TYPE_CHECKING:
+    from async_pixiv.model.illust import Illust
+
+
+class PixivMedia:
+    __slots__ = ('_url', '_thumb')
+
+    def __init__(self, url: str, thumb: str):
+        self._url: str = url
+        self._thumb: str = thumb
+
+    def __str__(self):
+        return f"PixivMedia(url={self.url}, thumb={self.thumb}, large={self.large})"
+
+    @property
+    def url(self) -> str:
+        return self._url
+
+    @property
+    def thumb(self) -> str:
+        return self._thumb
+
+    @property
+    def large(self) -> str:
+        url = self._url.replace("img-original", "img-master").removesuffix(".jpg").removesuffix(".png")
+        return url + "_master1200.jpg"
+
+
+class Pixiv:
+    __slots__ = ('_illust',)
+
+    def __init__(self, illust: Illust):
+        self._illust: Illust = illust
+
+    @property
+    def url(self) -> str:
+        return str(self._illust.link).rstrip('/')
+
+    @property
+    def type(self) -> Literal["illust", "manga", "ugoira"]:
+        return self._illust.type.value
+
+    @property
+    def title(self) -> str:
+        return self._illust.title
+
+    @property
+    def author(self) -> str:
+        return self._illust.user.name
+
+    @property
+    def author_url(self) -> str:
+        return str(self._illust.user.link).rstrip('/')
+
+    @property
+    def description(self) -> str:
+        return self._illust.caption
+
+    @property
+    def tags(self) -> list[str]:
+        return [tag.name for tag in self._illust.tags]
+
+    @property
+    def is_multiple_pages(self) -> bool:
+        return self._illust.page_count > 1
+
+    @property
+    def is_nsfw(self) -> bool:
+        return self._illust.is_nsfw
+
+    @property
+    def is_ai(self) -> bool:
+        return self._illust.ai_type.value == 2
+
+    @property
+    def images(self) -> list[PixivMedia]:
+        if self.is_multiple_pages:
+            return [
+                PixivMedia(
+                    url=page.image_urls.original,
+                    thumb=page.image_urls.medium
+                )
+                for page in self._illust.meta_pages
+            ]
+        else:
+            return [
+                PixivMedia(  # should observe if image_url.original is always None for single page
+                    url=self._illust.image_urls.original or self._illust.meta_single_page.original,
+                    thumb=self._illust.image_urls.medium
+                )
+            ]
+
+
+class _ProcessPixiv:
+    _client: PixivClient
+
+    @classmethod
+    async def init_client(cls, token: str) -> None:
+        cls._client = PixivClient()
+        await cls._client.login_with_token(token)
+        cls._token = token
+
+    @classmethod
+    async def close_client(cls) -> None:
+        await cls._client.close()
+
+    @classmethod
+    async def refresh_token(cls):
+        await cls._client.login_with_token(cls._token)
+
+
+class ProcessPixiv(_ProcessPixiv):
+    __slots__ = ('_url', '_illust')
+
+    def __init__(self, url: str):
+        self._url: str = url
+
+    async def __aenter__(self):
+        self._illust = await self._fetch_illust()
+        return Pixiv(self._illust)
+
+    async def __aexit__(self, exc_type, exc_val, exc_tb):
+        pass
+
+    async def _fetch_illust(self) -> Illust:
+        illust_id = self._parse_illust_id()
+        try:
+            return (await self._client.ILLUST.detail(illust_id)).illust
+        except ApiError:
+            await self.refresh_token()
+            return (await self._client.ILLUST.detail(illust_id)).illust  # TODO use retry here
+
+    def _parse_illust_id(self) -> int:
+        return int(self._url.split("/")[-1])

utils/regex.py

@@ -0,0 +1,8 @@
+import re
+
+x_url = re.compile(r"^(?:https?://)?(?:www\.|mobile\.)?(?:x|twitter|fixvx|vxtwitter)\.com/(.+)/status/(\d+)")
+x_media_url = re.compile(r"^(?:https?://)?(pbs|video)\.twimg\.com/(.*)")
+x_tco_url = re.compile(r"(?:https?://)?t\.co/.+$", re.M)
+message_url = re.compile(r"\[.+]", re.S)
+
+pixiv_url = re.compile(r"^(?:https?://)?(?:www\.)?pixiv\.net/(?:en/)?(?:artworks/|i/)(\d+)")

utils/telegram.py

@@ -0,0 +1,197 @@
+from __future__ import annotations
+
+import html
+from functools import cached_property
+from typing import Generator, TYPE_CHECKING
+from uuid import uuid4
+
+from telegram import InlineQueryResultMpeg4Gif, InlineQueryResultPhoto, InlineQueryResultVideo, InputMediaPhoto, \
+    InputMediaVideo
+
+from common import PIXIV_REFRESH_TOKEN
+from .logger import get_logger
+from .pixiv import ProcessPixiv
+from .regex import pixiv_url, x_url
+from .tweet import ProcessTweet
+
+if TYPE_CHECKING:
+    from .types import TypeInlineQueryResult, TypeMessageMediaResult
+
+logger = get_logger(__name__)
+
+message_raw_text_tweet = """{url}
+<a href="{author_url}">{author}</a>: {text}
+"""
+
+message_raw_text_pixiv = """<a href="{url}">{text}</a> / <a href="{author_url}">{author}</a>
+{tags}
+"""
+
+
+class Telegram:
+    def __init__(self, url: str):
+        self._url = url
+
+    async def __aenter__(self):
+        if x_url.match(self._url):
+            async with TelegramTweet(self._url) as tweet:
+                return tweet
+        elif PIXIV_REFRESH_TOKEN and pixiv_url.match(self._url):
+            async with TelegramPixiv(self._url) as pixiv:
+                return pixiv
+        else:
+            return None  # TODO add raise and catch
+
+    async def __aexit__(self, exc_type, exc_val, exc_tb):
+        pass
+
+
+class TelegramTweet:
+    message_raw_text = message_raw_text_tweet
+    __slots__ = ('_url', '_tweet', '__dict__')
+
+    def __init__(self, url: str):
+        self._url: str = url
+
+    async def __aenter__(self):
+        async with ProcessTweet(self._url) as tweet:
+            self._tweet = tweet
+            return self
+
+    async def __aexit__(self, exc_type, exc_val, exc_tb):
+        pass
+
+    @cached_property
+    def url(self) -> str:
+        return self._tweet.url
+
+    @cached_property
+    def message_text(self) -> str:
+        tweet = self._tweet
+        return self.message_raw_text.format(
+            url=tweet.url,
+            author_url=tweet.author_url,
+            author=html.escape(tweet.author),
+            text=html.escape(tweet.text)
+        )
+
+    def inline_query_result(self) -> tuple[TypeInlineQueryResult, ...]:
+        return tuple(self.inline_query_generator())
+
+    def message_media_result(self) -> tuple[TypeMessageMediaResult, ...]:
+        return tuple(self.message_media_generator())
+
+    def inline_query_generator(self) -> Generator[TypeInlineQueryResult, None, None]:
+        tweet = self._tweet
+        for tweet_media in tweet.media:
+            logger.info(str(tweet_media))
+            if tweet_media.type == "image":
+                yield InlineQueryResultPhoto(
+                    id=str(uuid4()),
+                    photo_url=tweet_media.url,
+                    thumbnail_url=tweet_media.thumb,
+                    caption=self.message_text
+                )
+            elif tweet_media.type == "video":
+                yield InlineQueryResultVideo(
+                    id=str(uuid4()),
+                    video_url=tweet_media.url,
+                    mime_type="video/mp4",
+                    thumbnail_url=tweet_media.thumb,
+                    title=tweet.text,
+                    caption=self.message_text
+                )
+            elif tweet_media.type == "gif":
+                yield InlineQueryResultMpeg4Gif(
+                    id=str(uuid4()),
+                    mpeg4_url=tweet_media.url,
+                    thumbnail_url=tweet_media.thumb,
+                    caption=self.message_text
+                )
+
+    def message_media_generator(self) -> Generator[TypeMessageMediaResult, None, None]:
+        tweet = self._tweet
+        for tweet_media in tweet.media:
+            logger.info(str(tweet_media))
+            if tweet_media.type == "image":
+                yield InputMediaPhoto(
+                    media=tweet_media.url,
+                    has_spoiler=tweet.sensitive
+                )
+            elif tweet_media.type == "video":
+                yield InputMediaVideo(
+                    media=tweet_media.url,
+                    has_spoiler=tweet.sensitive,
+                    thumbnail=tweet_media.thumb
+                )
+            elif tweet_media.type == "gif":
+                if len(tweet.media) == 1:
+                    yield tweet_media.url, tweet.sensitive
+                yield InputMediaVideo(
+                    media=tweet_media.url,
+                    has_spoiler=tweet.sensitive,
+                    thumbnail=tweet_media.thumb
+                )
+
+
+class TelegramPixiv:
+    message_raw_text = message_raw_text_pixiv
+    __slots__ = ('_url', '_pixiv')
+
+    def __init__(self, url: str):
+        self._url = url
+
+    async def __aenter__(self):
+        async with ProcessPixiv(self._url) as pixiv:
+            self._pixiv = pixiv
+            return self
+
+    async def __aexit__(self, exc_type, exc_val, exc_tb):
+        pass
+
+    @property
+    def url(self) -> str:
+        return self._pixiv.url
+
+    @property
+    def message_text(self) -> str:
+        pixiv = self._pixiv
+        return self.message_raw_text.format(
+            url=pixiv.url,
+            author_url=pixiv.author_url,
+            author=html.escape(pixiv.author),
+            text=html.escape(pixiv.title),
+            tags=html.escape(" ".join(f"#{name}" for name in pixiv.tags))
+        )
+
+    def inline_query_result(self) -> tuple[TypeInlineQueryResult, ...]:
+        return tuple(self.inline_query_generator())
+
+    def message_media_result(self) -> tuple[TypeMessageMediaResult, ...]:
+        return tuple(self.message_media_generator())
+
+    def inline_query_generator(self) -> Generator[TypeInlineQueryResult, None, None]:
+        pixiv = self._pixiv
+        for media in pixiv.images:
+            logger.info(str(media))
+            if pixiv.type in ("illust", "manga"):
+                yield InlineQueryResultPhoto(
+                    id=str(uuid4()),
+                    photo_url=media.large,
+                    thumbnail_url=media.thumb,
+                    caption=self.message_text
+                )
+            else:
+                yield
+
+    def message_media_generator(self) -> Generator[TypeMessageMediaResult, None, None]:
+        pixiv = self._pixiv
+        for media in pixiv.images:
+            logger.info(str(media))
+            if pixiv.type in ("illust", "manga"):
+                yield InputMediaPhoto(
+                    media=media.large,
+                    has_spoiler=pixiv.is_nsfw
+                )
+            else:
+                yield

utils/tweet.py

@@ -0,0 +1,152 @@
+from __future__ import annotations
+
+from functools import cached_property
+from typing import TYPE_CHECKING
+
+from .net import NetClient
+from .regex import x_media_url, x_tco_url, x_url
+
+if TYPE_CHECKING:
+    from .types import TweetInfo
+
+twimg_url = 'https://pbs.twimg.com/'
+vx_api_url = 'https://api.vxtwitter.com/{0}/status/{1}'
+
+
+class TweetMedia:
+    __slots__ = ('_url', '_thumb', '_type', '__dict__')
+
+    def __init__(self, url: str, thumb: str, media_type: str):
+        self._url: str = url
+        self._thumb: str = thumb
+        self._type: str = media_type
+
+    def __str__(self):
+        return f"TweetMedia(url={self.url}, thumb={self.thumb}, type={self.type})"
+
+    @cached_property
+    def _uri(self) -> str | None:
+        if match := x_media_url.match(self._url):
+            return match.group(2).removesuffix('.jpg').removesuffix('.png')
+        return None
+
+    @cached_property
+    def url(self) -> str:
+        match self._type:
+            case "image":
+                return f"{twimg_url}{self._uri}?format=jpg&name=4096x4096"
+            case "video":
+                return self._url
+            case "gif":
+                return self._url
+            case _:
+                return self._url
+
+    @cached_property
+    def thumb(self) -> str:
+        match self._type:
+            case "image":
+                return f"{twimg_url}{self._uri}?format=jpg&name=thumb"
+            case "video":
+                return self._thumb
+            case "gif":
+                return self._thumb
+            case _:
+                return self._thumb
+
+    @property
+    def type(self) -> str:
+        return self._type
+
+
+class Tweet:
+    __slots__ = ('_id', '_author', '_author_id', '_text', '_media', '_sensitive', '__dict__')
+
+    def __init__(
+            self,
+            tweet_id: str,
+            author: str,
+            author_id: str,
+            text: str,
+            media: list[TweetMedia],
+            sensitive: bool = False
+    ):
+        self._id: str = tweet_id
+        self._author: str = author
+        self._author_id: str = author_id
+        self._text: str = text
+        self._media: list[TweetMedia] = media
+        self._sensitive: bool = sensitive
+
+    @property
+    def id(self) -> str:
+        return self._id
+
+    @cached_property
+    def url(self) -> str:
+        return f"https://twitter.com/{self._author_id}/status/{self._id}"
+
+    @property
+    def author(self) -> str:
+        return self._author
+
+    @cached_property
+    def author_url(self) -> str:
+        return f"https://twitter.com/{self._author_id}"
+
+    @property
+    def text(self) -> str:
+        return self._text
+
+    @property
+    def media(self) -> list[TweetMedia]:
+        return self._media
+
+    @property
+    def sensitive(self) -> bool:
+        return self._sensitive
+
+
+class ProcessTweet:
+    __slots__ = ('_url', '_tweet')
+
+    def __init__(self, url: str):
+        self._url: str = url
+
+    async def __aenter__(self):
+        self._tweet = await self._fetch_tweet()
+        return Tweet(
+            tweet_id=self._tweet["tweetID"],
+            author=self._tweet["user_name"],
+            author_id=self._tweet["user_screen_name"],
+            text=self._tweet_text,
+            media=self._tweet_media,
+            sensitive=self._tweet["possibly_sensitive"]
+        )
+
+    async def __aexit__(self, exc_type, exc_val, exc_tb):
+        pass
+
+    async def _fetch_tweet(self) -> TweetInfo:
+        match = x_url.match(self._url)
+        assert match, f"Invalid URL: {self._url}"
+        auther_id, tweet_id = match.groups()
+        return await NetClient.fetch_json(vx_api_url.format(auther_id, tweet_id))
+
+    @property
+    def _tweet_text(self) -> str:
+        match = x_tco_url.search(self._tweet['text'])
+        return self._tweet['text'][:match.start()].strip(" ") if match else self._tweet['text']
+
+    @property
+    def _tweet_media(self) -> list[TweetMedia]:
+        return [
+            TweetMedia(
+                url=tweet_media['url'],
+                thumb=tweet_media['thumbnail_url'],
+                media_type=tweet_media['type']
+            )
+            for tweet_media in self._tweet['media_extended']
+        ]
+
+

utils/types.py

@@ -0,0 +1,17 @@
+from typing import TypedDict
+
+from telegram import InlineQueryResultMpeg4Gif, InlineQueryResultPhoto, InlineQueryResultVideo, InputMediaPhoto, \
+    InputMediaVideo
+
+TypeInlineQueryResult = InlineQueryResultMpeg4Gif | InlineQueryResultPhoto | InlineQueryResultVideo
+InputMediaAnimation = tuple[str, bool]
+TypeMessageMediaResult = InputMediaPhoto | InputMediaVideo | InputMediaAnimation
+
+
+class TweetInfo(TypedDict):
+    tweetID: str
+    user_name: str
+    user_screen_name: str
+    text: str
+    media_extended: list[dict]
+    possibly_sensitive: bool