Merge 9c32f15b7b into f2a4983df7

yt_dlp/extractor/bandlab.py

@@ -3,26 +3,38 @@ from .common import InfoExtractor
 from ..utils import (
     ExtractorError,
     float_or_none,
+    format_field,
     int_or_none,
     parse_iso8601,
     parse_qs,
     truncate_string,
     url_or_none,
 )
-from ..utils.traversal import traverse_obj
+from ..utils.traversal import traverse_obj, value
 
 
 class BandlabBaseIE(InfoExtractor):
     _API_HEADERS = {
+        'accept': 'application/json',
         'referer': 'https://www.bandlab.com/',
         'x-client-id': 'BandLab-Web',
-        'x-client-version': '10.1.123',
+        'x-client-version': '10.1.124',
     }
 
-    def _parse_revision(self, revision_data):
+    def _call_api(self, endpoint, asset_id, **kwargs):
+        return self._download_json(
+            f'https://www.bandlab.com/api/v1.3/{endpoint}/{asset_id}',
+            asset_id, headers=self._API_HEADERS, **kwargs)
+
+    def _parse_revision(self, revision_data, url=None):
         return {
             'vcodec': 'none',
+            'media_type': 'revision',
+            'extractor_key': BandlabIE.ie_key(),
+            'extractor': BandlabIE.IE_NAME,
             **traverse_obj(revision_data, {
+                'webpage_url': (
+                    'id', ({value(url)}, {format_field(template='https://www.bandlab.com/revision/%s')}), filter, any),
                 'id': (('revisionId', 'id'), {str}, any),
                 'title': ('song', 'name', {str}),
                 'track': ('song', 'name', {str}),
@@ -40,10 +52,15 @@ class BandlabBaseIE(InfoExtractor):
             }),
         }
 
-    def _parse_track(self, track_data):
+    def _parse_track(self, track_data, url=None):
         return {
             'vcodec': 'none',
+            'media_type': 'track',
+            'extractor_key': BandlabIE.ie_key(),
+            'extractor': BandlabIE.IE_NAME,
             **traverse_obj(track_data, {
+                'webpage_url': (
+                    'id', ({value(url)}, {format_field(template='https://www.bandlab.com/post/%s')}), filter, any),
                 'id': (('revisionId', 'id'), {str}, any),
                 'url': ('track', 'sample', 'audioUrl', {url_or_none}),
                 'title': ('track', 'name', {str}),
@@ -60,20 +77,27 @@ class BandlabBaseIE(InfoExtractor):
             }),
         }
 
-    def _parse_video(self, video_data):
-        return traverse_obj(video_data, {
-            'id': ('id', {str}),
-            'url': ('video', 'url', {url_or_none}),
-            'title': ('caption', {lambda x: x.replace('\n', ' ')}, {truncate_string(left=50)}),
-            'description': ('caption', {str}),
-            'thumbnail': ('video', 'picture', 'url', {url_or_none}),
-            'view_count': ('video', 'counters', 'plays', {int_or_none}),
-            'like_count': ('video', 'counters', 'likes', {int_or_none}),
-            'comment_count': ('counters', 'comments', {int_or_none}),
-            'duration': ('video', 'duration', {float_or_none}),
-            'uploader': ('creator', 'name', {str}),
-            'uploader_id': ('creator', 'username', {str}),
-        })
+    def _parse_video(self, video_data, url=None):
+        return {
+            'media_type': 'video',
+            'extractor_key': BandlabIE.ie_key(),
+            'extractor': BandlabIE.IE_NAME,
+            **traverse_obj(video_data, {
+                'id': ('id', {str}),
+                'webpage_url': (
+                    'id', ({value(url)}, {format_field(template='https://www.bandlab.com/post/%s')}), filter, any),
+                'url': ('video', 'url', {url_or_none}),
+                'title': ('caption', {lambda x: x.replace('\n', ' ')}, {truncate_string(left=50)}),
+                'description': ('caption', {str}),
+                'thumbnail': ('video', 'picture', 'url', {url_or_none}),
+                'view_count': ('video', 'counters', 'plays', {int_or_none}),
+                'like_count': ('video', 'counters', 'likes', {int_or_none}),
+                'comment_count': ('counters', 'comments', {int_or_none}),
+                'duration': ('video', 'duration', {float_or_none}),
+                'uploader': ('creator', 'name', {str}),
+                'uploader_id': ('creator', 'username', {str}),
+            }),
+        }
 
 
 class BandlabIE(BandlabBaseIE):
@@ -101,6 +125,7 @@ class BandlabIE(BandlabBaseIE):
             'genres': ['Lofi'],
             'uploader': 'ender milze',
             'comment_count': int,
+            'media_type': 'revision',
         },
     }, {
         # Same track as above but post URL
@@ -122,6 +147,7 @@ class BandlabIE(BandlabBaseIE):
             'genres': ['Lofi'],
             'uploader': 'ender milze',
             'comment_count': int,
+            'media_type': 'revision',
         },
     }, {
         # SharedKey Example
@@ -143,6 +169,7 @@ class BandlabIE(BandlabBaseIE):
             'upload_date': '20240305',
             'uploader': 'Erna Wageneder',
             'title': 'PodcastMaerchen4b',
+            'media_type': 'revision',
         },
     }, {
         # Different Revision selected
@@ -164,6 +191,7 @@ class BandlabIE(BandlabBaseIE):
             'upload_date': '20200430',
             'like_count': int,
             'duration': 279.43,
+            'media_type': 'revision',
         },
     }, {
         # Video
@@ -180,6 +208,7 @@ class BandlabIE(BandlabBaseIE):
             'uploader': 'auraa',
             'like_count': int,
             'description': 'backing vocals',
+            'media_type': 'video',
         },
     }, {
         # Embed Example
@@ -201,6 +230,7 @@ class BandlabIE(BandlabBaseIE):
             'uploader_id': 'microfreaks',
             'title': 'Positronic Meltdown',
             'like_count': int,
+            'media_type': 'revision',
         },
     }, {
         # Track without revisions available
@@ -220,6 +250,7 @@ class BandlabIE(BandlabBaseIE):
             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/users/572a351a-0f3a-4c6a-ac39-1a5defdeeb1c/',
             'timestamp': 1691162128,
             'upload_date': '20230804',
+            'media_type': 'track',
         },
     }, {
         'url': 'https://www.bandlab.com/revision/014de0a4-7d82-ea11-a94c-0003ffd19c0f',
@@ -243,6 +274,7 @@ class BandlabIE(BandlabBaseIE):
             'title': 'Vermilion Pt. 2 (Cover)',
             'track': 'Vermilion Pt. 2 (Cover)',
             'thumbnail': 'https://bandlabimages.azureedge.net/v1.0/songs/62b10750-7aef-4f42-ad08-1af52f577e97/',
+            'media_type': 'revision',
         },
     }]
 
@@ -256,9 +288,8 @@ class BandlabIE(BandlabBaseIE):
 
         revision_data = None
         if not revision_id:
-            post_data = self._download_json(
-                f'https://www.bandlab.com/api/v1.3/posts/{display_id}', display_id,
-                note='Downloading post data', headers=self._API_HEADERS,
+            post_data = self._call_api(
+                'posts', display_id, note='Downloading post data',
                 query=traverse_obj(qs, {'sharedKey': ('sharedKey', 0)}))
 
             revision_id = traverse_obj(post_data, (('revisionId', ('revision', 'id')), {str}, any))
@@ -267,17 +298,16 @@ class BandlabIE(BandlabBaseIE):
             if not revision_data and not revision_id:
                 post_type = post_data.get('type')
                 if post_type == 'Video':
-                    return self._parse_video(post_data)
+                    return self._parse_video(post_data, url=url)
                 if post_type == 'Track':
-                    return self._parse_track(post_data)
+                    return self._parse_track(post_data, url=url)
                 raise ExtractorError('Could not extract data')
 
         if not revision_data:
-            revision_data = self._download_json(
-                f'https://www.bandlab.com/api/v1.3/revisions/{revision_id}', revision_id,
-                note='Downloading revision data', headers=self._API_HEADERS, query={'edit': 'false'})
+            revision_data = self._call_api(
+                'revisions', revision_id, note='Downloading revision data', query={'edit': 'false'})
 
-        return self._parse_revision(revision_data)
+        return self._parse_revision(revision_data, url=url)
 
 
 class BandlabPlaylistIE(BandlabBaseIE):
@@ -379,9 +409,8 @@ class BandlabPlaylistIE(BandlabBaseIE):
             'embed': ['collections', 'albums'],
         }.get(playlist_type)
         for endpoint in endpoints:
-            playlist_data = self._download_json(
-                f'https://www.bandlab.com/api/v1.3/{endpoint}/{playlist_id}', playlist_id,
-                note=f'Downloading {endpoint[:-1]} data', headers=self._API_HEADERS,
+            playlist_data = self._call_api(
+                endpoint, playlist_id, note=f'Downloading {endpoint[:-1]} data',
                 fatal=False, expected_status=404)
             if not playlist_data.get('errorCode'):
                 playlist_type = endpoint