[qqmusic] Do not capture braced text from the middle of the string
[youtube-dl] / youtube_dl / extractor / qqmusic.py
index 8a724ab5117ae8e4b05f427b47b0562f7a037f71..21cf6f2d6f0d31544aebb4807ab4b0cd16946529 100644 (file)
@@ -16,6 +16,7 @@ from ..compat import compat_urllib_request
 
 class QQMusicIE(InfoExtractor):
     IE_NAME = 'qqmusic'
+    IE_DESC = 'QQ音乐'
     _VALID_URL = r'http://y.qq.com/#type=song&mid=(?P<id>[0-9A-Za-z]+)'
     _TESTS = [{
         'url': 'http://y.qq.com/#type=song&mid=004295Et37taLD',
@@ -27,7 +28,7 @@ class QQMusicIE(InfoExtractor):
             'upload_date': '20141227',
             'creator': '林俊杰',
             'description': 'md5:d327722d0361576fde558f1ac68a7065',
-            'thumbnail': 'http://i.gtimg.cn/music/photo/mid_album_500/7/p/001IV22P1RDX7p.jpg',
+            'thumbnail': 're:^https?://.*\.jpg$',
         }
     }, {
         'note': 'There is no mp3-320 version of this song.',
@@ -40,7 +41,7 @@ class QQMusicIE(InfoExtractor):
             'upload_date': '20050626',
             'creator': '李季美',
             'description': 'md5:46857d5ed62bc4ba84607a805dccf437',
-            'thumbnail': 'http://i.gtimg.cn/music/photo/mid_album_500/r/Q/0042owYj46IxrQ.jpg',
+            'thumbnail': 're:^https?://.*\.jpg$',
         }
     }]
 
@@ -111,15 +112,27 @@ class QQMusicIE(InfoExtractor):
         self._check_formats(formats, mid)
         self._sort_formats(formats)
 
-        return {
+        actual_lrc_lyrics = ''.join(
+            line + '\n' for line in re.findall(
+                r'(?m)^(\[[0-9]{2}:[0-9]{2}\.[0-9]{2,}\][^\n]*|\[[^\]]*\])', lrc_content))
+
+        info_dict = {
             'id': mid,
             'formats': formats,
             'title': song_name,
             'upload_date': publish_time,
             'creator': singer,
             'description': lrc_content,
-            'thumbnail': thumbnail_url,
+            'thumbnail': thumbnail_url
         }
+        if actual_lrc_lyrics:
+            info_dict['subtitles'] = {
+                'origin': [{
+                    'ext': 'lrc',
+                    'data': actual_lrc_lyrics,
+                }]
+            }
+        return info_dict
 
 
 class QQPlaylistBaseIE(InfoExtractor):
@@ -142,6 +155,7 @@ class QQPlaylistBaseIE(InfoExtractor):
 
 class QQMusicSingerIE(QQPlaylistBaseIE):
     IE_NAME = 'qqmusic:singer'
+    IE_DESC = 'QQ音乐 - 歌手'
     _VALID_URL = r'http://y.qq.com/#type=singer&mid=(?P<id>[0-9A-Za-z]+)'
     _TEST = {
         'url': 'http://y.qq.com/#type=singer&mid=001BLpXF2DyJe2',
@@ -186,6 +200,7 @@ class QQMusicSingerIE(QQPlaylistBaseIE):
 
 class QQMusicAlbumIE(QQPlaylistBaseIE):
     IE_NAME = 'qqmusic:album'
+    IE_DESC = 'QQ音乐 - 专辑'
     _VALID_URL = r'http://y.qq.com/#type=album&mid=(?P<id>[0-9A-Za-z]+)'
 
     _TESTS = [{
@@ -228,6 +243,7 @@ class QQMusicAlbumIE(QQPlaylistBaseIE):
 
 class QQMusicToplistIE(QQPlaylistBaseIE):
     IE_NAME = 'qqmusic:toplist'
+    IE_DESC = 'QQ音乐 - 排行榜'
     _VALID_URL = r'http://y\.qq\.com/#type=toplist&p=(?P<id>(top|global)_[0-9]+)'
 
     _TESTS = [{
@@ -281,6 +297,7 @@ class QQMusicToplistIE(QQPlaylistBaseIE):
 
 class QQMusicPlaylistIE(QQPlaylistBaseIE):
     IE_NAME = 'qqmusic:playlist'
+    IE_DESC = 'QQ音乐 - 歌单'
     _VALID_URL = r'http://y\.qq\.com/#type=taoge&id=(?P<id>[0-9]+)'
 
     _TEST = {