Switch codebase to use sanitized_Request instead of
[youtube-dl] / youtube_dl / extractor / qqmusic.py
index 476432330bc6e0ab875f0e28611ad956ebfe0f41..1ba3bbddf654dbe6a407465df0d1fe89af789106 100644 (file)
@@ -7,15 +7,16 @@ import re
 
 from .common import InfoExtractor
 from ..utils import (
+    sanitized_Request,
     strip_jsonp,
     unescapeHTML,
     clean_html,
 )
-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',
@@ -24,7 +25,7 @@ class QQMusicIE(InfoExtractor):
             'id': '004295Et37taLD',
             'ext': 'mp3',
             'title': '可惜没如果',
-            'upload_date': '20141227',
+            'release_date': '20141227',
             'creator': '林俊杰',
             'description': 'md5:d327722d0361576fde558f1ac68a7065',
             'thumbnail': 're:^https?://.*\.jpg$',
@@ -37,11 +38,26 @@ class QQMusicIE(InfoExtractor):
             'id': '004MsGEo3DdNxV',
             'ext': 'mp3',
             'title': '如果',
-            'upload_date': '20050626',
+            'release_date': '20050626',
             'creator': '李季美',
             'description': 'md5:46857d5ed62bc4ba84607a805dccf437',
             'thumbnail': 're:^https?://.*\.jpg$',
         }
+    }, {
+        'note': 'lyrics not in .lrc format',
+        'url': 'http://y.qq.com/#type=song&mid=001JyApY11tIp6',
+        'info_dict': {
+            'id': '001JyApY11tIp6',
+            'ext': 'mp3',
+            'title': 'Shadows Over Transylvania',
+            'release_date': '19970225',
+            'creator': 'Dark Funeral',
+            'description': 'md5:ed14d5bd7ecec19609108052c25b2c11',
+            'thumbnail': 're:^https?://.*\.jpg$',
+        },
+        'params': {
+            'skip_download': True,
+        },
     }]
 
     _FORMATS = {
@@ -111,15 +127,27 @@ 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,
+            'release_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 +170,7 @@ 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',
@@ -172,7 +201,7 @@ class QQMusicSingerIE(QQPlaylistBaseIE):
         singer_desc = None
 
         if singer_id:
-            req = compat_urllib_request.Request(
+            req = sanitized_Request(
                 'http://s.plcloud.music.qq.com/fcgi-bin/fcg_get_singer_desc.fcg?utf8=1&outCharset=utf-8&format=xml&singerid=%s' % singer_id)
             req.add_header(
                 'Referer', 'http://s.plcloud.music.qq.com/xhr_proxy_utf8.html')
@@ -186,6 +215,7 @@ 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 +258,7 @@ 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 +312,7 @@ 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 = {