[extractor/patreon] Fix and improve extractors (#4398)

* Add workaround for 403s - Fixes https://github.com/yt-dlp/yt-dlp/issues/3631
* Support m3u8 post file videos - Fixes https://github.com/yt-dlp/yt-dlp/issues/2277
* Raise useful error messages - Fixes https://github.com/yt-dlp/yt-dlp/issues/2914
* `--write-comments` support

Authored by: coletdjnz, pukkandan

yt_dlp/extractor/_extractors.py

@@ -1244,7 +1244,7 @@
 from .parlview import ParlviewIE
 from .patreon import (
     PatreonIE,
-    PatreonUserIE
+    PatreonCampaignIE
 )
 from .pbs import PBSIE
 from .pearvideo import PearVideoIE

yt_dlp/extractor/patreon.py

@@ -1,4 +1,5 @@
 import itertools
+from urllib.error import HTTPError
 
 from .common import InfoExtractor
 from .vimeo import VimeoIE
@@ -7,17 +8,45 @@
 from ..utils import (
     clean_html,
     determine_ext,
+    ExtractorError,
     int_or_none,
     KNOWN_EXTENSIONS,
     mimetype2ext,
     parse_iso8601,
     str_or_none,
+    traverse_obj,
     try_get,
     url_or_none,
 )
 
 
-class PatreonIE(InfoExtractor):
+class PatreonBaseIE(InfoExtractor):
+    USER_AGENT = 'Patreon/7.6.28 (Android; Android 11; Scale/2.10)'
+
+    def _call_api(self, ep, item_id, query=None, headers=None, fatal=True, note=None):
+        if headers is None:
+            headers = {}
+        if 'User-Agent' not in headers:
+            headers['User-Agent'] = self.USER_AGENT
+        if query:
+            query.update({'json-api-version': 1.0})
+
+        try:
+            return self._download_json(
+                f'https://www.patreon.com/api/{ep}',
+                item_id, note='Downloading API JSON' if not note else note,
+                query=query, fatal=fatal, headers=headers)
+        except ExtractorError as e:
+            if not isinstance(e.cause, HTTPError) or mimetype2ext(e.cause.headers.get('Content-Type')) != 'json':
+                raise
+            err_json = self._parse_json(self._webpage_read_content(e.cause, None, item_id), item_id, fatal=False)
+            err_message = traverse_obj(err_json, ('errors', ..., 'detail'), get_all=False)
+            if err_message:
+                raise ExtractorError(f'Patreon said: {err_message}', expected=True)
+            raise
+
+
+class PatreonIE(PatreonBaseIE):
     _VALID_URL = r'https?://(?:www\.)?patreon\.com/(?:creation\?hid=|posts/(?:[\w-]+-)?)(?P<id>\d+)'
     _TESTS = [{
         'url': 'http://www.patreon.com/creation?hid=743933',
@@ -26,12 +55,18 @@ class PatreonIE(InfoExtractor):
             'id': '743933',
             'ext': 'mp3',
             'title': 'Episode 166: David Smalley of Dogma Debate',
-            'description': 'md5:713b08b772cd6271b9f3906683cfacdf',
+            'description': 'md5:34d207dd29aa90e24f1b3f58841b81c7',
             'uploader': 'Cognitive Dissonance Podcast',
             'thumbnail': 're:^https?://.*$',
             'timestamp': 1406473987,
             'upload_date': '20140727',
             'uploader_id': '87145',
+            'like_count': int,
+            'comment_count': int,
+            'uploader_url': 'https://www.patreon.com/dissonancepod',
+            'channel_id': '80642',
+            'channel_url': 'https://www.patreon.com/dissonancepod',
+            'channel_follower_count': int,
         },
     }, {
         'url': 'http://www.patreon.com/creation?hid=754133',
@@ -42,6 +77,9 @@ class PatreonIE(InfoExtractor):
             'title': 'CD 167 Extra',
             'uploader': 'Cognitive Dissonance Podcast',
             'thumbnail': 're:^https?://.*$',
+            'like_count': int,
+            'comment_count': int,
+            'uploader_url': 'https://www.patreon.com/dissonancepod',
         },
         'skip': 'Patron-only content',
     }, {
@@ -53,8 +91,23 @@ class PatreonIE(InfoExtractor):
             'uploader': 'TraciJHines',
             'thumbnail': 're:^https?://.*$',
             'upload_date': '20150211',
-            'description': 'md5:c5a706b1f687817a3de09db1eb93acd4',
+            'description': 'md5:8af6425f50bd46fbf29f3db0fc3a8364',
             'uploader_id': 'TraciJHines',
+            'categories': ['Entertainment'],
+            'duration': 282,
+            'view_count': int,
+            'tags': 'count:39',
+            'age_limit': 0,
+            'channel': 'TraciJHines',
+            'channel_url': 'https://www.youtube.com/channel/UCGLim4T2loE5rwCMdpCIPVg',
+            'live_status': 'not_live',
+            'like_count': int,
+            'channel_id': 'UCGLim4T2loE5rwCMdpCIPVg',
+            'availability': 'public',
+            'channel_follower_count': int,
+            'playable_in_embed': True,
+            'uploader_url': 'http://www.youtube.com/user/TraciJHines',
+            'comment_count': int,
         },
         'params': {
             'noplaylist': True,
@@ -80,38 +133,40 @@ class PatreonIE(InfoExtractor):
             'uploader_id': '14936315',
         },
         'skip': 'Patron-only content'
-    }]
+    }, {
-
+        # m3u8 video (https://github.com/yt-dlp/yt-dlp/issues/2277)
-    # Currently Patreon exposes download URL via hidden CSS, so login is not
+        'url': 'https://www.patreon.com/posts/video-sketchbook-32452882',
-    # needed. Keeping this commented for when this inevitably changes.
+        'info_dict': {
-    '''
+            'id': '32452882',
-    def _perform_login(self, username, password):
+            'ext': 'mp4',
-        login_form = {
+            'comment_count': int,
-            'redirectUrl': 'http://www.patreon.com/',
+            'uploader_id': '4301314',
-            'email': username,
+            'like_count': int,
-            'password': password,
+            'timestamp': 1576696962,
+            'upload_date': '20191218',
+            'thumbnail': r're:^https?://.*$',
+            'uploader_url': 'https://www.patreon.com/loish',
+            'description': 'md5:e2693e97ee299c8ece47ffdb67e7d9d2',
+            'title': 'VIDEO // sketchbook flipthrough',
+            'uploader': 'Loish ',
+            'tags': ['sketchbook', 'video'],
+            'channel_id': '1641751',
+            'channel_url': 'https://www.patreon.com/loish',
+            'channel_follower_count': int,
         }
-
+    }]
-        request = sanitized_Request(
-            'https://www.patreon.com/processLogin',
-            compat_urllib_parse_urlencode(login_form).encode('utf-8')
-        )
-        login_page = self._download_webpage(request, None, note='Logging in')
-
-        if re.search(r'onLoginFailed', login_page):
-            raise ExtractorError('Unable to login, incorrect username and/or password', expected=True)
-
-    '''
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
-        post = self._download_json(
+        post = self._call_api(
-            'https://www.patreon.com/api/posts/' + video_id, video_id, query={
+            f'posts/{video_id}', video_id, query={
                 'fields[media]': 'download_url,mimetype,size_bytes',
-                'fields[post]': 'comment_count,content,embed,image,like_count,post_file,published_at,title',
+                'fields[post]': 'comment_count,content,embed,image,like_count,post_file,published_at,title,current_user_can_view',
                 'fields[user]': 'full_name,url',
+                'fields[post_tag]': 'value',
+                'fields[campaign]': 'url,name,patron_count',
                 'json-api-use-default-includes': 'false',
-                'include': 'media,user',
+                'include': 'media,user,user_defined_tags,campaign',
             })
         attributes = post['data']['attributes']
         title = attributes['title'].strip()
@@ -125,6 +180,9 @@ def _real_extract(self, url):
             'like_count': int_or_none(attributes.get('like_count')),
             'comment_count': int_or_none(attributes.get('comment_count')),
         }
+        can_view_post = traverse_obj(attributes, 'current_user_can_view')
+        if can_view_post and info['comment_count']:
+            info['__post_extractor'] = self.extract_comments(video_id)
 
         for i in post.get('included', []):
             i_type = i.get('type')
@@ -133,11 +191,12 @@ def _real_extract(self, url):
                 download_url = media_attributes.get('download_url')
                 ext = mimetype2ext(media_attributes.get('mimetype'))
                 if download_url and ext in KNOWN_EXTENSIONS:
-                    info.update({
+                    return {
+                        **info,
                         'ext': ext,
                         'filesize': int_or_none(media_attributes.get('size_bytes')),
                         'url': download_url,
-                    })
+                    }
             elif i_type == 'user':
                 user_attributes = i.get('attributes')
                 if user_attributes:
@@ -147,67 +206,72 @@ def _real_extract(self, url):
                         'uploader_url': user_attributes.get('url'),
                     })
 
-        if not info.get('url'):
+            elif i_type == 'post_tag':
+                info.setdefault('tags', []).append(traverse_obj(i, ('attributes', 'value')))
+
+            elif i_type == 'campaign':
+                info.update({
+                    'channel': traverse_obj(i, ('attributes', 'title')),
+                    'channel_id': str_or_none(i.get('id')),
+                    'channel_url': traverse_obj(i, ('attributes', 'url')),
+                    'channel_follower_count': int_or_none(traverse_obj(i, ('attributes', 'patron_count'))),
+                })
+
         # handle Vimeo embeds
         if try_get(attributes, lambda x: x['embed']['provider']) == 'Vimeo':
             embed_html = try_get(attributes, lambda x: x['embed']['html'])
             v_url = url_or_none(compat_urllib_parse_unquote(
                 self._search_regex(r'(https(?:%3A%2F%2F|://)player\.vimeo\.com.+app_id(?:=|%3D)+\d+)', embed_html, 'vimeo url', fatal=False)))
             if v_url:
-                    info.update({
+                return {
+                    **info,
                     '_type': 'url_transparent',
                     'url': VimeoIE._smuggle_referrer(v_url, 'https://patreon.com'),
                     'ie_key': 'Vimeo',
-                    })
+                }
 
-        if not info.get('url'):
         embed_url = try_get(attributes, lambda x: x['embed']['url'])
         if embed_url:
-                info.update({
+            return {
+                **info,
                 '_type': 'url',
                 'url': embed_url,
-                })
+            }
 
-        if not info.get('url'):
+        post_file = traverse_obj(attributes, 'post_file')
-            post_file = attributes['post_file']
+        if post_file:
-            ext = determine_ext(post_file.get('name'))
+            name = post_file.get('name')
+            ext = determine_ext(name)
             if ext in KNOWN_EXTENSIONS:
-                info.update({
+                return {
+                    **info,
                     'ext': ext,
                     'url': post_file['url'],
-                })
+                }
+            elif name == 'video':
+                formats, subtitles = self._extract_m3u8_formats_and_subtitles(post_file['url'], video_id)
+                return {
+                    **info,
+                    'formats': formats,
+                    'subtitles': subtitles,
+                }
 
+        if can_view_post is False:
+            self.raise_no_formats('You do not have access to this post', video_id=video_id, expected=True)
+        else:
+            self.raise_no_formats('No supported media found in this post', video_id=video_id, expected=True)
         return info
 
-
+    def _get_comments(self, post_id):
-class PatreonUserIE(InfoExtractor):
-
-    _VALID_URL = r'https?://(?:www\.)?patreon\.com/(?!rss)(?P<id>[-\w]+)'
-
-    _TESTS = [{
-        'url': 'https://www.patreon.com/dissonancepod/',
-        'info_dict': {
-            'title': 'dissonancepod',
-        },
-        'playlist_mincount': 68,
-        'expected_warnings': 'Post not viewable by current user! Skipping!',
-    }, {
-        'url': 'https://www.patreon.com/dissonancepod/posts',
-        'only_matching': True
-    }, ]
-
-    @classmethod
-    def suitable(cls, url):
-        return False if PatreonIE.suitable(url) else super(PatreonUserIE, cls).suitable(url)
-
-    def _entries(self, campaign_id, user_id):
         cursor = None
+        count = 0
         params = {
-            'fields[campaign]': 'show_audio_post_download_links,name,url',
+            'page[count]': 50,
-            'fields[post]': 'current_user_can_view,embed,image,is_paid,post_file,published_at,patreon_url,url,post_type,thumbnail_url,title',
+            'include': 'parent.commenter.campaign,parent.post.user,parent.post.campaign.creator,parent.replies.parent,parent.replies.commenter.campaign,parent.replies.post.user,parent.replies.post.campaign.creator,commenter.campaign,post.user,post.campaign.creator,replies.parent,replies.commenter.campaign,replies.post.user,replies.post.campaign.creator,on_behalf_of_campaign',
-            'filter[campaign_id]': campaign_id,
+            'fields[comment]': 'body,created,is_by_creator',
-            'filter[is_draft]': 'false',
+            'fields[user]': 'image_url,full_name,url',
-            'sort': '-published_at',
+            'filter[flair]': 'image_tiny_url,name',
+            'sort': '-created',
             'json-api-version': 1.0,
             'json-api-use-default-includes': 'false',
         }
@@ -215,19 +279,149 @@ def _entries(self, campaign_id, user_id):
         for page in itertools.count(1):
 
             params.update({'page[cursor]': cursor} if cursor else {})
-            posts_json = self._download_json('https://www.patreon.com/api/posts', user_id, note='Downloading posts page %d' % page, query=params, headers={'Cookie': '.'})
+            response = self._call_api(
+                f'posts/{post_id}/comments', post_id, query=params, note='Downloading comments page %d' % page)
 
-            cursor = try_get(posts_json, lambda x: x['meta']['pagination']['cursors']['next'])
+            cursor = None
+            for comment in traverse_obj(response, (('data', ('included', lambda _, v: v['type'] == 'comment')), ...), default=[]):
+                count += 1
+                comment_id = comment.get('id')
+                attributes = comment.get('attributes') or {}
+                if comment_id is None:
+                    continue
+                author_id = traverse_obj(comment, ('relationships', 'commenter', 'data', 'id'))
+                author_info = traverse_obj(
+                    response, ('included', lambda _, v: v['id'] == author_id and v['type'] == 'user', 'attributes'),
+                    get_all=False, expected_type=dict, default={})
 
+                yield {
+                    'id': comment_id,
+                    'text': attributes.get('body'),
+                    'timestamp': parse_iso8601(attributes.get('created')),
+                    'parent': traverse_obj(comment, ('relationships', 'parent', 'data', 'id'), default='root'),
+                    'author_is_uploader': attributes.get('is_by_creator'),
+                    'author_id': author_id,
+                    'author': author_info.get('full_name'),
+                    'author_thumbnail': author_info.get('image_url'),
+                }
+
+            if count < traverse_obj(response, ('meta', 'count')):
+                cursor = traverse_obj(response, ('data', -1, 'id'))
+
+            if cursor is None:
+                break
+
+
+class PatreonCampaignIE(PatreonBaseIE):
+
+    _VALID_URL = r'https?://(?:www\.)?patreon\.com/(?!rss)(?:(?:m/(?P<campaign_id>\d+))|(?P<vanity>[-\w]+))'
+    _TESTS = [{
+        'url': 'https://www.patreon.com/dissonancepod/',
+        'info_dict': {
+            'title': 'Cognitive Dissonance Podcast',
+            'channel_url': 'https://www.patreon.com/dissonancepod',
+            'id': '80642',
+            'description': 'md5:eb2fa8b83da7ab887adeac34da6b7af7',
+            'channel_id': '80642',
+            'channel': 'Cognitive Dissonance Podcast',
+            'age_limit': 0,
+            'channel_follower_count': int,
+            'uploader_id': '87145',
+            'uploader_url': 'https://www.patreon.com/dissonancepod',
+            'uploader': 'Cognitive Dissonance Podcast',
+            'thumbnail': r're:^https?://.*$',
+        },
+        'playlist_mincount': 68,
+    }, {
+        'url': 'https://www.patreon.com/m/4767637/posts',
+        'info_dict': {
+            'title': 'Not Just Bikes',
+            'channel_follower_count': int,
+            'id': '4767637',
+            'channel_id': '4767637',
+            'channel_url': 'https://www.patreon.com/notjustbikes',
+            'description': 'md5:595c6e7dca76ae615b1d38c298a287a1',
+            'age_limit': 0,
+            'channel': 'Not Just Bikes',
+            'uploader_url': 'https://www.patreon.com/notjustbikes',
+            'uploader': 'Not Just Bikes',
+            'uploader_id': '37306634',
+            'thumbnail': r're:^https?://.*$',
+        },
+        'playlist_mincount': 71
+    }, {
+        'url': 'https://www.patreon.com/dissonancepod/posts',
+        'only_matching': True
+    }, {
+        'url': 'https://www.patreon.com/m/5932659',
+        'only_matching': True
+    }]
+
+    @classmethod
+    def suitable(cls, url):
+        return False if PatreonIE.suitable(url) else super(PatreonCampaignIE, cls).suitable(url)
+
+    def _entries(self, campaign_id):
+        cursor = None
+        params = {
+            'fields[post]': 'patreon_url,url',
+            'filter[campaign_id]': campaign_id,
+            'filter[is_draft]': 'false',
+            'sort': '-published_at',
+            'json-api-use-default-includes': 'false',
+        }
+
+        for page in itertools.count(1):
+
+            params.update({'page[cursor]': cursor} if cursor else {})
+            posts_json = self._call_api('posts', campaign_id, query=params, note='Downloading posts page %d' % page)
+
+            cursor = traverse_obj(posts_json, ('meta', 'pagination', 'cursors', 'next'))
             for post in posts_json.get('data') or []:
-                yield self.url_result(url_or_none(try_get(post, lambda x: x['attributes']['patreon_url'])), 'Patreon')
+                yield self.url_result(url_or_none(traverse_obj(post, ('attributes', 'patreon_url'))), 'Patreon')
 
             if cursor is None:
                 break
 
     def _real_extract(self, url):
 
-        user_id = self._match_id(url)
+        campaign_id, vanity = self._match_valid_url(url).group('campaign_id', 'vanity')
-        webpage = self._download_webpage(url, user_id, headers={'Cookie': '.'})
+        if campaign_id is None:
+            webpage = self._download_webpage(url, vanity, headers={'User-Agent': self.USER_AGENT})
             campaign_id = self._search_regex(r'https://www.patreon.com/api/campaigns/(\d+)/?', webpage, 'Campaign ID')
-        return self.playlist_result(self._entries(campaign_id, user_id), playlist_title=user_id)
+
+        params = {
+            'json-api-use-default-includes': 'false',
+            'fields[user]': 'full_name,url',
+            'fields[campaign]': 'name,summary,url,patron_count,creation_count,is_nsfw,avatar_photo_url',
+            'include': 'creator'
+        }
+
+        campaign_response = self._call_api(
+            f'campaigns/{campaign_id}', campaign_id,
+            note='Downloading campaign info', fatal=False,
+            query=params) or {}
+
+        campaign_info = campaign_response.get('data') or {}
+        channel_name = traverse_obj(campaign_info, ('attributes', 'name'))
+        user_info = traverse_obj(
+            campaign_response, ('included', lambda _, v: v['type'] == 'user'),
+            default={}, expected_type=dict, get_all=False)
+
+        return {
+            '_type': 'playlist',
+            'id': campaign_id,
+            'title': channel_name,
+            'entries': self._entries(campaign_id),
+            'description': clean_html(traverse_obj(campaign_info, ('attributes', 'summary'))),
+            'channel_url': traverse_obj(campaign_info, ('attributes', 'url')),
+            'channel_follower_count': int_or_none(traverse_obj(campaign_info, ('attributes', 'patron_count'))),
+            'channel_id': campaign_id,
+            'channel': channel_name,
+            'uploader_url': traverse_obj(user_info, ('attributes', 'url')),
+            'uploader_id': str_or_none(user_info.get('id')),
+            'uploader': traverse_obj(user_info, ('attributes', 'full_name')),
+            'playlist_count': traverse_obj(campaign_info, ('attributes', 'creation_count')),
+            'age_limit': 18 if traverse_obj(campaign_info, ('attributes', 'is_nsfw')) else 0,
+            'thumbnail': url_or_none(traverse_obj(campaign_info, ('attributes', 'avatar_photo_url'))),
+        }