[youtube] modify regex to get chapters from description (closes #24819)

ytdl-org · Apr 19, 2020 · 78d7146 · 78d7146
1 parent 00eb865
commit 78d7146
Show file tree

Hide file tree

Showing 2 changed files with 36 additions and 1 deletion.
diff --git a/test/test_youtube_chapters.py b/test/test_youtube_chapters.py
@@ -15,6 +15,41 @@
 class TestYoutubeChapters(unittest.TestCase):
 
     _TEST_CASES = [
+        (
+            # https://www.youtube.com/watch?v=gBRKnvK1JUE
+            # pattern: 00:00 - 09:24 <title>
+            '''Here is Nucleus's 1979 album Out Of The Long Dark: https://www.youtube.com/watch?v=GX4Eh1DPb-E<br /><br />And here is their 1971 live album: https://www.youtube.com/watch?v=cpbM75B8qaE<br /><br /><a href="#" onclick="yt.www.watch.player.seekTo(00*60+00);return false;">00:00</a> - <a href="#" onclick="yt.www.watch.player.seekTo(09*60+24);return false;">09:24</a>  roots<br /><a href="#" onclick="yt.www.watch.player.seekTo(09*60+24);return false;">09:24</a> - <a href="#" onclick="yt.www.watch.player.seekTo(14*60+19);return false;">14:19</a>  images<br /><a href="#" onclick="yt.www.watch.player.seekTo(14*60+19);return false;">14:19</a> - <a href="#" onclick="yt.www.watch.player.seekTo(18*60+20);return false;">18:20</a>  caliban<br /><a href="#" onclick="yt.www.watch.player.seekTo(18*60+20);return false;">18:20</a> - <a href="#" onclick="yt.www.watch.player.seekTo(21*60+42);return false;">21:42</a>  whapatiti<br /><a href="#" onclick="yt.www.watch.player.seekTo(21*60+42);return false;">21:42</a> - <a href="#" onclick="yt.www.watch.player.seekTo(26*60+18);return false;">26:18</a>  capricorn<br /><a href="#" onclick="yt.www.watch.player.seekTo(26*60+18);return false;">26:18</a> - <a href="#" onclick="yt.www.watch.player.seekTo(29*60+42);return false;">29:42</a>  odokamona<br /><a href="#" onclick="yt.www.watch.player.seekTo(29*60+42);return false;">29:42</a> - <a href="#" onclick="yt.www.watch.player.seekTo(37*60+26);return false;">37:26</a>  southern roots and celebration<br /><br />Bass Guitar – Roger Sutton<br />Design – Keith Davis (3)<br />Drums – Clive Thacker<br />Engineer – Roger Wake<br />Guitar – Jocelyn Pitchen<br />Percussion – Aureo de Souza<br />Piano, Electric Piano – Dave MacRae<br />Producer – Fritz Fryer<br />Tenor Saxophone, Soprano Saxophone, Flute, Flute [Bamboo] – Brian Smith<br />Trumpet – Ian Carr<br />Vocals – Joy Yates<br />Written-By – Brian Smith (tracks: B1 to B3), Dave MacRae (tracks: B4), Ian Carr (tracks: A)   ''',
+            2246,
+            [{
+                'start_time': 0,
+                'end_time': 564,
+                'title': 'roots',
+            }, {
+                'start_time': 564,
+                'end_time': 859,
+                'title': 'images',
+            }, {
+                'start_time': 859,
+                'end_time': 1100,
+                'title': 'caliban',
+            }, {
+                'start_time': 1100,
+                'end_time': 1302,
+                'title': 'whapatiti',
+            }, {
+                'start_time': 1302,
+                'end_time': 1578,
+                'title': 'capricorn',
+            }, {
+                'start_time': 1578,
+                'end_time': 1782,
+                'title': 'odokamona',
+            }, {
+                'start_time': 1782,
+                'end_time': 2246,
+                'title': 'southern roots and celebration',
+            }]
+        ),
         (
             # https://www.youtube.com/watch?v=A22oy8dFjqc
             # pattern: 00:00 - <title>

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
@@ -1621,7 +1621,7 @@ def _extract_chapters(description, duration):
         if not description:
             return None
         chapter_lines = re.findall(
-            r'(?:^|<br\s*/>)([^<]*<a[^>]+onclick=["\']yt\.www\.watch\.player\.seekTo[^>]+>(\d{1,2}:\d{1,2}(?::\d{1,2})?)</a>[^>]*)(?=$|<br\s*/>)',
+            r'(?:^|<br\s*/>)([^<]*<a[^>]+onclick=["\']yt\.www\.watch\.player\.seekTo[^>]+>(\d{1,2}:\d{1,2}(?::\d{1,2})?)</a>(?:[^<]*<a.*?)?[^>]*)(?=$|<br\s*/>)',
             description)
         if not chapter_lines:
             return None