fix regex match group

.gitignore

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

main.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 import html
 from functools import wraps
 from typing import TYPE_CHECKING
@@ -8,7 +10,7 @@ from telegram.ext import (ApplicationBuilder, CallbackQueryHandler, CommandHandl
                           InlineQueryHandler, MessageHandler, PicklePersistence, filters)
 
 import common
-from tweet import TGTweet
+from utils.tweet import Telegram
 
 if TYPE_CHECKING:
     from telegram import Chat, Message, Update
@@ -34,7 +36,7 @@ async def inline_query(update: Update, context: ContextTypes.DEFAULT_TYPE) -> No
     if query == "":
         return
     logger.info(f"Query: {query}")
-    async with TGTweet(query) as tweet:
+    async with Telegram(query) as tweet:
         result = list(tweet.inline_query_generator)
         await update.inline_query.answer(result)
 
@@ -43,13 +45,13 @@ async def inline_query(update: Update, context: ContextTypes.DEFAULT_TYPE) -> No
 async def url_media(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None:
     url = update.message.text
     logger.info(f"Receiving url: {url}")
-    async with TGTweet(url) as tweet:
-        media = list(tweet.pm_media_generator)
+    async with Telegram(url) as tweet:
+        media = list(tweet.message_media_generator)
         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 tweet.is_single_gif else await update.effective_message.reply_animation(
+        ) if not isinstance(tuple, media[0]) else await update.effective_message.reply_animation(
             media[0][0],
             caption=tweet.message_text,
             reply_to_message_id=update.message.message_id,
@@ -202,7 +204,7 @@ 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)
-    TGTweet.init_client()
+    Telegram.init_client()
 
 
 async def post_stop(application: Application) -> None:
@@ -210,7 +212,7 @@ async def post_stop(application: Application) -> None:
 
 
 async def post_shutdown(application: Application) -> None:
-    await TGTweet.close_client()
+    await Telegram.close_client()
 
 
 def main():
@@ -224,6 +226,7 @@ def main():
                    .post_stop(post_stop)
                    .post_shutdown(post_shutdown)
                    .concurrent_updates(True)
+                   .http_version('2')
                    .build()
                    )
 

utils/tweet.py

@@ -1,5 +1,9 @@
+from __future__ import annotations
+
 import html
-from typing import Generator
+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,
@@ -7,6 +11,9 @@ from telegram import (InlineQueryResultMpeg4Gif, InlineQueryResultPhoto, InlineQ
 
 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__)
 
 
@@ -18,38 +25,39 @@ message_raw_text = """{url}
 """
 
 
-def create_client() -> 'AsyncClient':
+def create_client() -> AsyncClient:
     return AsyncClient(http2=True)
 
 
-async def close_client(_client: 'AsyncClient') -> None:
+async def close_client(_client: AsyncClient) -> None:
     await _client.aclose()
 
 
-async def fetch_json(_client: 'AsyncClient', url: str) -> dict:
+async def fetch_json(_client: AsyncClient, url: str) -> dict:
     logger.info(f"Fetching {url}")
     response = await _client.get(url)
-    assert response.status_code == response.is_success, f"Failed to fetch {url}, status code {response.status_code}"
+    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"Media[url: {self.url} thumb: {self.thumb} type: {self.type}]"
+        return f"TweetMedia(url={self.url}, thumb={self.thumb}, type={self.type})"
 
-    @property
+    @cached_property
     def _uri(self) -> str | None:
-        match = x_media_regex.match(self._url)
-        if match:
+        if match := x_media_regex.match(self._url):
             return match.group(2).removesuffix('.jpg').removesuffix('.png')
         return None
 
-    @property
+    @cached_property
     def url(self) -> str:
         match self._type:
             case "image":
@@ -61,7 +69,7 @@ class TweetMedia:
             case _:
                 return self._url
 
-    @property
+    @cached_property
     def thumb(self) -> str:
         match self._type:
             case "image":
@@ -79,6 +87,8 @@ class TweetMedia:
 
 
 class Tweet:
+    __slots__ = ('_id', '_author', '_author_id', '_text', '_media', '_sensitive', '__dict__')
+
     def __init__(
             self,
             tweet_id: str,
@@ -99,7 +109,7 @@ class Tweet:
     def id(self) -> str:
         return self._id
 
-    @property
+    @cached_property
     def url(self) -> str:
         return f"https://twitter.com/{self._author_id}/status/{self._id}"
 
@@ -107,7 +117,7 @@ class Tweet:
     def author(self) -> str:
         return self._author
 
-    @property
+    @cached_property
     def author_url(self) -> str:
         return f"https://twitter.com/{self._author_id}"
 
@@ -124,19 +134,144 @@ class Tweet:
         return self._sensitive
 
 
-class TGTweet(Tweet):
+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.groups()
+        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: str = url
-        self._api_param: tuple[str] = self._tweet_id
-        self._is_single_gif: bool = False
-        assert self._api_param
+        self._url = url
 
     async def __aenter__(self):
-        self._tweet: dict = await self._fetch_tweet(self._api_param)
-        super().__init__(*self._init_properties)
-        return self
+        if x_url_regex.match(self._url):
+            async with TelegramTweet(self._httpx_client, self._url) as tweet:
+                return tweet
 
     async def __aexit__(self, exc_type, exc_val, exc_tb):
         pass
@@ -148,107 +283,3 @@ class TGTweet(Tweet):
     @classmethod
     async def close_client(cls) -> None:
         await close_client(cls._httpx_client)
-
-    async def _fetch_tweet(self, api_param: tuple[str]) -> dict:
-        return await fetch_json(self._httpx_client, vx_api_url.format(*api_param))
-
-    @property
-    def _tweet_id(self) -> tuple[str] | None:
-        match = x_url_regex.match(self._url)
-        if match:
-            return match.groups()
-        return None
-
-    @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=x['url'],
-                thumb=x['thumbnail_url'],
-                media_type=x['type']
-            )
-            for x in self._tweet['media_extended']
-        ]
-
-    @property
-    def _init_properties(self) -> tuple:
-        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']
-        return id, author, author_id, text, media, sensitive
-
-    @property
-    def is_single_gif(self) -> bool:
-        return self._is_single_gif
-
-    @property
-    def message_text(self) -> str:
-        return message_raw_text.format(
-            url=self.url,
-            author_url=self.author_url,
-            author=html.escape(self.author),
-            text=html.escape(self.text)
-        )
-
-    @property
-    def inline_query_generator(self) -> Generator[
-        InlineQueryResultPhoto | InlineQueryResultVideo | InlineQueryResultMpeg4Gif, None, None
-    ]:
-        for i, tweet_media in enumerate(self.media):
-            logger.info(str(tweet_media))
-            if tweet_media.type == "image":
-                yield InlineQueryResultPhoto(
-                    id=str(i),
-                    photo_url=tweet_media.url,
-                    thumbnail_url=tweet_media.thumb,
-                    caption=self.message_text
-                )
-            elif tweet_media.type == "video":
-                yield InlineQueryResultVideo(
-                    id=str(i),
-                    video_url=tweet_media.url,
-                    mime_type="video/mp4",
-                    thumbnail_url=tweet_media.thumb,
-                    title=self.text,
-                    caption=self.message_text
-                )
-            elif tweet_media.type == "gif":
-                yield InlineQueryResultMpeg4Gif(
-                    id=str(i),
-                    mpeg4_url=tweet_media.url,
-                    thumbnail_url=tweet_media.thumb,
-                    caption=self.message_text
-                )
-
-    @property
-    def pm_media_generator(self) -> Generator[InputMediaPhoto | InputMediaVideo | tuple[str, bool], None, None]:
-        for tweet_media in self.media:
-            logger.info(str(tweet_media))
-            if tweet_media.type == "image":
-                yield InputMediaPhoto(
-                    media=tweet_media.url,
-                    has_spoiler=self.sensitive
-                )
-            elif tweet_media.type == "video":
-                yield InputMediaVideo(
-                    media=tweet_media.url,
-                    has_spoiler=self.sensitive,
-                    thumbnail=tweet_media.thumb
-                )
-            elif tweet_media.type == "gif":
-                if len(self.media) == 1:
-                    self._is_single_gif = True
-                    yield tweet_media.url, self.sensitive
-                yield InputMediaVideo(
-                    media=tweet_media.url,
-                    has_spoiler=self.sensitive,
-                    thumbnail=tweet_media.thumb
-                )

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