[soundcloud:pagedplaylist] Improve (closes #19086)

[youtube-dl] / youtube_dl / extractor / soundcloud.py

diff --git a/youtube_dl/extractor/soundcloud.py b/youtube_dl/extractor/soundcloud.py
index 030840fd8d8e92e38367e2af93cfe2b636a5f7be..5536e78514436735c8e4101996a963c6ecadc4ed 100644 (file)
--- a/youtube_dl/extractor/soundcloud.py
+++ b/youtube_dl/extractor/soundcloud.py
@@ -18,6 +18,7 @@ from ..utils import (
     int_or_none,
     unified_strdate,
     update_url_query,
+    url_or_none,
 )
 
 
@@ -395,18 +396,23 @@ class SoundcloudPagedPlaylistBaseIE(SoundcloudPlaylistBaseIE):
             # Empty collection may be returned, in this case we proceed
             # straight to next_href
 
-            def resolve_permalink_url(candidates):
+            def resolve_entry(candidates):
                 for cand in candidates:
-                    if isinstance(cand, dict):
-                        permalink_url = cand.get('permalink_url')
-                        entry_id = self._extract_id(cand)
-                        if permalink_url and permalink_url.startswith('http'):
-                            return permalink_url, entry_id
+                    if not isinstance(cand, dict):
+                        continue
+                    permalink_url = url_or_none(cand.get('permalink_url'))
+                    if not permalink_url:
+                        continue
+                    return self.url_result(
+                        permalink_url,
+                        ie=SoundcloudIE.ie_key() if SoundcloudIE.suitable(permalink_url) else None,
+                        video_id=self._extract_id(cand),
+                        video_title=cand.get('title'))
 
             for e in collection:
-                permalink_url, entry_id = resolve_permalink_url((e, e.get('track'), e.get('playlist')))
-                if permalink_url:
-                    entries.append(self.url_result(permalink_url, video_id=entry_id))
+                entry = resolve_entry((e, e.get('track'), e.get('playlist')))
+                if entry:
+                    entries.append(entry)
 
             next_href = response.get('next_href')
             if not next_href:
@@ -438,40 +444,47 @@ class SoundcloudUserIE(SoundcloudPagedPlaylistBaseIE):
                     '''
     IE_NAME = 'soundcloud:user'
     _TESTS = [{
-        'url': 'https://soundcloud.com/the-akashic-chronicler',
+        'url': 'https://soundcloud.com/soft-cell-official',
         'info_dict': {
-            'id': '114582580',
-            'title': 'The Akashic Chronicler (All)',
+            'id': '207965082',
+            'title': 'Soft Cell (All)',
         },
-        'playlist_mincount': 74,
+        'playlist_mincount': 28,
     }, {
-        'url': 'https://soundcloud.com/the-akashic-chronicler/tracks',
+        'url': 'https://soundcloud.com/soft-cell-official/tracks',
         'info_dict': {
-            'id': '114582580',
-            'title': 'The Akashic Chronicler (Tracks)',
+            'id': '207965082',
+            'title': 'Soft Cell (Tracks)',
         },
-        'playlist_mincount': 37,
+        'playlist_mincount': 27,
     }, {
-        'url': 'https://soundcloud.com/the-akashic-chronicler/sets',
+        'url': 'https://soundcloud.com/soft-cell-official/albums',
+        'info_dict': {
+            'id': '207965082',
+            'title': 'Soft Cell (Albums)',
+        },
+        'playlist_mincount': 1,
+    }, {
+        'url': 'https://soundcloud.com/jcv246/sets',
         'info_dict': {
-            'id': '114582580',
-            'title': 'The Akashic Chronicler (Playlists)',
+            'id': '12982173',
+            'title': 'Jordi / cv (Playlists)',
         },
         'playlist_mincount': 2,
     }, {
-        'url': 'https://soundcloud.com/the-akashic-chronicler/reposts',
+        'url': 'https://soundcloud.com/jcv246/reposts',
         'info_dict': {
-            'id': '114582580',
-            'title': 'The Akashic Chronicler (Reposts)',
+            'id': '12982173',
+            'title': 'Jordi / cv (Reposts)',
         },
-        'playlist_mincount': 7,
+        'playlist_mincount': 6,
     }, {
-        'url': 'https://soundcloud.com/the-akashic-chronicler/likes',
+        'url': 'https://soundcloud.com/clalberg/likes',
         'info_dict': {
-            'id': '114582580',
-            'title': 'The Akashic Chronicler (Likes)',
+            'id': '11817582',
+            'title': 'clalberg (Likes)',
         },
-        'playlist_mincount': 321,
+        'playlist_mincount': 5,
     }, {
         'url': 'https://soundcloud.com/grynpyret/spotlight',
         'info_dict': {
@@ -479,9 +492,6 @@ class SoundcloudUserIE(SoundcloudPagedPlaylistBaseIE):
             'title': 'Grynpyret (Spotlight)',
         },
         'playlist_mincount': 1,
-    }, {
-        'url': 'https://soundcloud.com/soft-cell-official/albums',
-        'only_matching': True,
     }]
 
     _BASE_URL_MAP = {