]>
jfr.im git - yt-dlp.git/blob - yt_dlp/extractor/channel9.py
1 from __future__
import unicode_literals
5 from .common
import InfoExtractor
15 class Channel9IE(InfoExtractor
):
18 _VALID_URL
= r
'https?://(?:www\.)?(?:channel9\.msdn\.com|s\.ch9\.ms)/(?P<contentpath>.+?)(?P<rss>/RSS)?/?(?:[?#&]|$)'
21 'url': 'http://channel9.msdn.com/Events/TechEd/Australia/2013/KOS002',
22 'md5': '32083d4eaf1946db6d454313f44510ca',
24 'id': '6c413323-383a-49dc-88f9-a22800cab024',
26 'title': 'Developer Kick-Off Session: Stuff We Love',
27 'description': 'md5:b80bf9355a503c193aff7ec6cd5a7731',
29 'thumbnail': r
're:https?://.*\.jpg',
30 'timestamp': 1377717420,
31 'upload_date': '20130828',
32 'session_code': 'KOS002',
33 'session_room': 'Arena 1A',
34 'session_speakers': 'count:5',
37 'url': 'http://channel9.msdn.com/posts/Self-service-BI-with-Power-BI-nuclear-testing',
38 'md5': 'dcf983ee6acd2088e7188c3cf79b46bc',
40 'id': 'fe8e435f-bb93-4e01-8e97-a28c01887024',
42 'title': 'Self-service BI with Power BI - nuclear testing',
43 'description': 'md5:2d17fec927fc91e9e17783b3ecc88f54',
45 'thumbnail': r
're:https?://.*\.jpg',
46 'timestamp': 1386381991,
47 'upload_date': '20131207',
48 'authors': ['Mike Wilmot'],
51 # low quality mp4 is best
52 'url': 'https://channel9.msdn.com/Events/CPP/CppCon-2015/Ranges-for-the-Standard-Library',
54 'id': '33ad69d2-6a4e-4172-83a1-a523013dec76',
56 'title': 'Ranges for the Standard Library',
57 'description': 'md5:9895e0a9fd80822d2f01c454b8f4a372',
59 'thumbnail': r
're:https?://.*\.jpg',
60 'upload_date': '20150930',
61 'timestamp': 1443640735,
64 'skip_download': True,
67 'url': 'https://channel9.msdn.com/Events/DEVintersection/DEVintersection-2016/RSS',
69 'id': 'Events/DEVintersection/DEVintersection-2016',
70 'title': 'DEVintersection 2016 Orlando Sessions',
72 'playlist_mincount': 14,
74 'url': 'https://channel9.msdn.com/Niners/Splendid22/Queue/76acff796e8f411184b008028e0d492b/RSS',
75 'only_matching': True,
77 'url': 'https://channel9.msdn.com/Events/Speakers/scott-hanselman/RSS?UrlSafeName=scott-hanselman',
78 'only_matching': True,
81 _RSS_URL
= 'http://channel9.msdn.com/%s/RSS'
84 def _extract_urls(webpage
):
86 r
'<iframe[^>]+src=["\'](https?
://channel9\
.msdn\
.com
/(?
:[^
/]+/)+)player
\b',
89 def _extract_list(self, video_id, rss_url=None):
91 rss_url = self._RSS_URL % video_id
92 rss = self._download_xml(rss_url, video_id, 'Downloading RSS
')
93 entries = [self.url_result(session_url.text, 'Channel9
')
94 for session_url in rss.findall('./channel
/item
/link
')]
95 title_text = rss.find('./channel
/title
').text
96 return self.playlist_result(entries, video_id, title_text)
98 def _real_extract(self, url):
99 content_path, rss = self._match_valid_url(url).groups()
102 return self._extract_list(content_path, url)
104 webpage = self._download_webpage(
105 url, content_path, 'Downloading web page
')
107 episode_data = self._search_regex(
108 r"data-episode='([^
']+)'", webpage, 'episode data', default=None)
110 episode_data = self._parse_json(unescapeHTML(
111 episode_data), content_path)
112 content_id = episode_data['contentId']
113 is_session = '/Sessions(' in episode_data['api']
114 content_url = 'https://channel9.msdn.com/odata' + episode_data['api'] + '?$select=Captions,CommentCount,MediaLengthInSeconds,PublishedDate,Rating,RatingCount,Title,VideoMP4High,VideoMP4Low,VideoMP4Medium,VideoPlayerPreviewImage,VideoWMV,VideoWMVHQ,Views,'
116 content_url += 'Code,Description,Room,Slides,Speakers,ZipFile&$expand=Speakers'
118 content_url += 'Authors,Body&$expand=Authors'
119 content_data = self._download_json(content_url, content_id)
120 title = content_data['Title']
125 'wmv-low', 'mp4-low',
126 'wmv-mid', 'mp4-mid',
127 'wmv-high', 'mp4-high',
130 quality_key = qualities(QUALITIES)
132 def quality(quality_id, format_url):
133 return (len(QUALITIES) if '_Source.' in format_url
134 else quality_key(quality_id))
142 'Low Quality WMV': 'wmv-low',
143 'Low Quality MP4': 'mp4-low',
144 'Mid Quality WMV': 'wmv-mid',
145 'Mid Quality MP4': 'mp4-mid',
146 'High Quality WMV': 'wmv-high',
147 'High Quality MP4': 'mp4-high',
150 formats_select = self._search_regex(
151 r'(?s)<select[^>]+name=["\']format
[^
>]+>(.+?
)</select
', webpage,
152 'formats select
', default=None)
154 for mobj in re.finditer(
155 r'<option
\b[^
>]+\bvalue
=(["\'])(?P<url>(?:(?!\1).)+)\1[^>]*>\s*(?P<format>[^<]+?)\s*<',
157 format_url = mobj.group('url')
158 if format_url in urls:
161 format_id = mobj.group('format')
162 quality_id = SITE_QUALITIES.get(format_id, format_id)
165 'format_id': quality_id,
166 'quality': quality(quality_id, format_url),
167 'vcodec': 'none' if quality_id == 'mp3' else None,
171 'VideoMP4Low': 'mp4-low',
172 'VideoWMV': 'wmv-mid',
173 'VideoMP4Medium': 'mp4-mid',
174 'VideoMP4High': 'mp4-high',
175 'VideoWMVHQ': 'wmv-hq',
178 for format_id, q in API_QUALITIES.items():
179 q_url = content_data.get(format_id)
180 if not q_url or q_url in urls:
186 'quality': quality(q, q_url),
189 slides = content_data.get('Slides')
190 zip_file = content_data.get('ZipFile')
192 if not formats and not slides and not zip_file:
193 self.raise_no_formats(
194 'None of recording, slides or zip are available for %s' % content_path)
195 self._sort_formats(formats)
198 for caption in content_data.get('Captions', []):
199 caption_url = caption.get('Url')
202 subtitles.setdefault(caption.get('Language', 'en'), []).append({
210 'description': clean_html(content_data.get('Description') or content_data.get('Body')),
211 'thumbnail': content_data.get('VideoPlayerPreviewImage'),
212 'duration': int_or_none(content_data.get('MediaLengthInSeconds')),
213 'timestamp': parse_iso8601(content_data.get('PublishedDate')),
214 'avg_rating': int_or_none(content_data.get('Rating')),
215 'rating_count': int_or_none(content_data.get('RatingCount')),
216 'view_count': int_or_none(content_data.get('Views')),
217 'comment_count': int_or_none(content_data.get('CommentCount')),
218 'subtitles': subtitles,
222 for s in content_data.get('Speakers', []):
223 speaker_name = s.get('FullName')
226 speakers.append(speaker_name)
229 'session_code': content_data.get('Code'),
230 'session_room': content_data.get('Room'),
231 'session_speakers': speakers,
235 for a in content_data.get('Authors', []):
236 author_name = a.get('DisplayName')
239 authors.append(author_name)
240 common['authors'] = authors
246 d.update({'title': title + '-Slides', 'url': slides})
251 d.update({'title': title + '-Zip', 'url': zip_file})
256 d.update({'title': title, 'formats': formats})
258 return self.playlist_result(contents)
260 return self._extract_list(content_path)