[twitter] add 'highlights' extractor (#7826)

docs/configuration.rst

@@ -5422,14 +5422,16 @@ Description
     when processing a user profile.
 
     Possible values are
-    ``"info"``,
-    ``"avatar"``,
-    ``"background"``,
-    ``"timeline"``,
-    ``"tweets"``,
-    ``"media"``,
-    ``"replies"``,
-    ``"likes"``.
+
+    * ``"info"``
+    * ``"avatar"``
+    * ``"background"``
+    * ``"timeline"``
+    * ``"tweets"``
+    * ``"media"``
+    * ``"replies"``
+    * ``"highlights"``
+    * ``"likes"``
 
     It is possible to use ``"all"`` instead of listing all values separately.
 

docs/supportedsites.md

@@ -994,7 +994,7 @@ Consider all listed sites to potentially be NSFW.
 <tr id="twitter" title="twitter">
     <td>Twitter</td>
     <td>https://x.com/</td>
-    <td>Avatars, Backgrounds, Bookmarks, Communities, Events, Followers, Followed Users, Hashtags, individual Images, User Profile Information, Likes, Lists, List Members, Media Timelines, Quotes, Search Results, Timelines, Tweets, User Profiles</td>
+    <td>Avatars, Backgrounds, Bookmarks, Communities, Events, Followers, Followed Users, Hashtags, Highlights, individual Images, User Profile Information, Likes, Lists, List Members, Media Timelines, Quotes, Search Results, Timelines, Tweets, User Profiles</td>
     <td>Supported</td>
 </tr>
 <tr id="unsplash" title="unsplash">

gallery_dl/extractor/twitter.py

@@ -649,18 +649,19 @@ class TwitterUserExtractor(Dispatch, TwitterExtractor):
     def items(self):
         user, user_id = self.groups
         if user_id is not None:
-            user = "id:" + user_id
+            user = f"id:{user_id}"
 
         base = f"{self.root}/{user}/"
         return self._dispatch_extractors((
-            (TwitterInfoExtractor      , base + "info"),
-            (TwitterAvatarExtractor    , base + "photo"),
-            (TwitterBackgroundExtractor, base + "header_photo"),
-            (TwitterTimelineExtractor  , base + "timeline"),
-            (TwitterTweetsExtractor    , base + "tweets"),
-            (TwitterMediaExtractor     , base + "media"),
-            (TwitterRepliesExtractor   , base + "with_replies"),
-            (TwitterLikesExtractor     , base + "likes"),
+            (TwitterInfoExtractor      , f"{base}info"),
+            (TwitterAvatarExtractor    , f"{base}photo"),
+            (TwitterBackgroundExtractor, f"{base}header_photo"),
+            (TwitterTimelineExtractor  , f"{base}timeline"),
+            (TwitterTweetsExtractor    , f"{base}tweets"),
+            (TwitterMediaExtractor     , f"{base}media"),
+            (TwitterRepliesExtractor   , f"{base}with_replies"),
+            (TwitterHighlightsExtractor, f"{base}highlights"),
+            (TwitterLikesExtractor     , f"{base}likes"),
         ), ("timeline",))
 
 
@@ -781,6 +782,16 @@ class TwitterRepliesExtractor(TwitterExtractor):
         return self.api.user_tweets_and_replies(self.user)
 
 
+class TwitterHighlightsExtractor(TwitterExtractor):
+    """Extractor for Tweets from a user's highlights timeline"""
+    subcategory = "highlights"
+    pattern = BASE_PATTERN + r"/(?!search)([^/?#]+)/highlights(?!\w)"
+    example = "https://x.com/USER/highlights"
+
+    def tweets(self):
+        return self.api.user_highlights(self.user)
+
+
 class TwitterMediaExtractor(TwitterExtractor):
     """Extractor for Tweets from a user's Media timeline"""
     subcategory = "media"
@@ -1354,6 +1365,20 @@ class TwitterAPI():
         return self._pagination_tweets(
             endpoint, variables, field_toggles=field_toggles)
 
+    def user_highlights(self, screen_name):
+        endpoint = "/graphql/gmHw9geMTncZ7jeLLUUNOw/UserHighlightsTweets"
+        variables = {
+            "userId": self._user_id_by_screen_name(screen_name),
+            "count": 100,
+            "includePromotedContent": False,
+            "withVoice": True,
+        }
+        field_toggles = {
+            "withArticlePlainText": False,
+        }
+        return self._pagination_tweets(
+            endpoint, variables, field_toggles=field_toggles)
+
     def user_media(self, screen_name):
         endpoint = "/graphql/jCRhbOzdgOHp6u9H4g2tEg/UserMedia"
         variables = {

test/results/twitter.py

@@ -772,4 +772,9 @@ The Washington Post writes, "Three weeks after the toxic train derailment in Ohi
     "#class"   : twitter.TwitterImageExtractor,
 },
 
+{
+    "#url"     : "https://x.com/tetsuoai/highlights",
+    "#class"   : twitter.TwitterHighlightsExtractor,
+},
+
 )