]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/cctv.py
3 from .common
import InfoExtractor
4 from ..compat
import compat_str
12 class CCTVIE(InfoExtractor
):
14 _VALID_URL
= r
'https?://(?:(?:[^/]+)\.(?:cntv|cctv)\.(?:com|cn)|(?:www\.)?ncpa-classic\.com)/(?:[^/]+/)*?(?P<id>[^/?#&]+?)(?:/index)?(?:\.s?html|[?#&]|$)'
16 # fo.addVariable("videoCenterId","id")
17 'url': 'http://sports.cntv.cn/2016/02/12/ARTIaBRxv4rTT1yWf1frW2wi160212.shtml',
18 'md5': 'd61ec00a493e09da810bf406a078f691',
20 'id': '5ecdbeab623f4973b40ff25f18b174e8',
22 'title': '[NBA]二少联手砍下46分 雷霆主场击败鹈鹕(快讯)',
23 'description': 'md5:7e14a5328dc5eb3d1cd6afbbe0574e95',
25 'uploader': 'songjunjie',
26 'timestamp': 1455279956,
27 'upload_date': '20160212',
31 'url': 'http://tv.cctv.com/2016/02/05/VIDEUS7apq3lKrHG9Dncm03B160205.shtml',
33 'id': 'efc5d49e5b3b4ab2b34f3a502b73d3ae',
35 'title': '[赛车]“车王”舒马赫恢复情况成谜(快讯)',
36 'description': '2月4日,蒙特泽莫罗透露了关于“车王”舒马赫恢复情况,但情况是否属实遭到了质疑。',
39 'timestamp': 1454677291,
40 'upload_date': '20160205',
43 'skip_download': True,
47 'url': 'http://english.cntv.cn/special/four_comprehensives/index.shtml',
49 'id': '4bb9bb4db7a6471ba85fdeda5af0381e',
51 'title': 'NHnews008 ANNUAL POLITICAL SEASON',
52 'description': 'Four Comprehensives',
54 'uploader': 'zhangyunlei',
55 'timestamp': 1425385521,
56 'upload_date': '20150303',
59 'skip_download': True,
63 'url': 'http://cctv.cntv.cn/lm/tvseries_russian/yilugesanghua/index.shtml',
65 'id': 'b15f009ff45c43968b9af583fc2e04b2',
67 'title': 'Путь,усыпанный космеями Серия 1',
68 'description': 'Путь, усыпанный космеями',
71 'timestamp': 1477479241,
72 'upload_date': '20161026',
75 'skip_download': True,
78 # var initMyAray = 'id'
79 'url': 'http://www.ncpa-classic.com/2013/05/22/VIDE1369219508996867.shtml',
81 'id': 'a194cfa7f18c426b823d876668325946',
83 'title': '小泽征尔音乐塾 音乐梦想无国界',
85 'timestamp': 1369248264,
86 'upload_date': '20130522',
89 'skip_download': True,
93 'url': 'http://www.ncpa-classic.com/clt/more/416/index.shtml',
95 'id': 'a8606119a4884588a79d81c02abecc16',
97 'title': '来自维也纳的新年贺礼',
98 'description': 'md5:f13764ae8dd484e84dd4b39d5bcba2a7',
101 'timestamp': 1482942419,
102 'upload_date': '20161228',
105 'skip_download': True,
107 'expected_warnings': ['Failed to download m3u8 information'],
109 'url': 'http://ent.cntv.cn/2016/01/18/ARTIjprSSJH8DryTVr5Bx8Wb160118.shtml',
110 'only_matching': True,
112 'url': 'http://tv.cntv.cn/video/C39296/e0210d949f113ddfb38d31f00a4e5c44',
113 'only_matching': True,
115 'url': 'http://english.cntv.cn/2016/09/03/VIDEhnkB5y9AgHyIEVphCEz1160903.shtml',
116 'only_matching': True,
118 'url': 'http://tv.cctv.com/2016/09/07/VIDE5C1FnlX5bUywlrjhxXOV160907.shtml',
119 'only_matching': True,
121 'url': 'http://tv.cntv.cn/video/C39296/95cfac44cabd3ddc4a9438780a4e5c44',
122 'only_matching': True,
125 def _real_extract(self
, url
):
126 video_id
= self
._match
_id
(url
)
127 webpage
= self
._download
_webpage
(url
, video_id
)
129 video_id
= self
._search
_regex
(
130 [r
'var\s+guid\s*=\s*["\']([\da
-fA
-F
]+)',
131 r'videoCenterId
["\']\s*,\s*["\']([\da
-fA
-F
]+)',
132 r'changePlayer\s
*\
(\s
*["\']([\da-fA-F]+)',
133 r'load[Vv]ideo\s*\(\s*["\']([\da
-fA
-F
]+)',
134 r'var\s
+initMyAray\s
*=\s
*["\']([\da-fA-F]+)',
135 r'var\s+ids\s*=\s*\[["\']([\da
-fA
-F
]+)'],
138 data = self._download_json(
139 'http
://vdn
.apps
.cntv
.cn
/api
/getHttpVideoInfo
.do
', video_id,
148 title = data['title
']
152 video = data.get('video
')
153 if isinstance(video, dict):
154 for quality, chapters_key in enumerate(('lowChapters
', 'chapters
')):
156 video, lambda x: x[chapters_key][0]['url
'], compat_str)
166 hls_url = try_get(data, lambda x: x['hls_url
'], compat_str)
168 hls_url = re.sub(r'maxbr
=\d
+&?
', '', hls_url)
169 formats.extend(self._extract_m3u8_formats(
170 hls_url, video_id, 'mp4
', entry_protocol='m3u8_native
',
171 m3u8_id='hls
', fatal=False))
173 self._sort_formats(formats)
175 uploader = data.get('editer_name
')
176 description = self._html_search_meta(
177 'description
', webpage, default=None)
178 timestamp = unified_timestamp(data.get('f_pgmtime
'))
179 duration = float_or_none(try_get(video, lambda x: x['totalLength
']))
184 'description
': description,
185 'uploader
': uploader,
186 'timestamp
': timestamp,
187 'duration
': duration,