'description': 'md5:a950cc4285c43e44d763d036710cd9cd',
'timestamp': 1368213670,
'upload_date': '20130510',
- 'uploader_id': 1589608506001,
+ 'uploader_id': '1589608506001',
}
},
{
'uploader': 'Oracle',
'timestamp': 1344975024,
'upload_date': '20120814',
- 'uploader_id': 1460825906,
+ 'uploader_id': '1460825906',
},
},
{
'uploader': 'Mashable',
'timestamp': 1382041798,
'upload_date': '20131017',
- 'uploader_id': 1130468786001,
+ 'uploader_id': '1130468786001',
},
},
{
'description': 'UCI MTB World Cup 2014: Fort William, UK - Downhill Finals',
'timestamp': 1409122195,
'upload_date': '20140827',
- 'uploader_id': 710858724001,
+ 'uploader_id': '710858724001',
},
},
{
playlist_title=playlist_info['mediaCollectionDTO']['displayName'])
def _extract_video_info(self, video_info):
+ video_id = compat_str(video_info['id'])
+ publisher_id = video_info.get('publisherId')
info = {
- 'id': compat_str(video_info['id']),
+ 'id': video_id,
'title': video_info['displayName'].strip(),
'description': video_info.get('shortDescription'),
'thumbnail': video_info.get('videoStillURL') or video_info.get('thumbnailURL'),
'uploader': video_info.get('publisherName'),
- 'uploader_id': video_info.get('publisherId'),
+ 'uploader_id': compat_str(publisher_id) if publisher_id else None,
'duration': float_or_none(video_info.get('length'), 1000),
'timestamp': int_or_none(video_info.get('creationDate'), 1000),
}
url_comp = compat_urllib_parse_urlparse(url)
if url_comp.path.endswith('.m3u8'):
formats.extend(
- self._extract_m3u8_formats(url, info['id'], 'mp4'))
+ self._extract_m3u8_formats(
+ url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
continue
elif 'akamaihd.net' in url_comp.netloc:
# This type of renditions are served through
ext = 'flv'
if ext is None:
ext = determine_ext(url)
- tbr = int_or_none(rend.get('encodingRate'), 1000),
+ tbr = int_or_none(rend.get('encodingRate'), 1000)
a_format = {
'format_id': 'http%s' % ('-%s' % tbr if tbr else ''),
'url': url,
a_format.update({
'format_id': 'hls%s' % ('-%s' % tbr if tbr else ''),
'ext': 'mp4',
- 'protocol': 'm3u8',
+ 'protocol': 'm3u8_native',
})
formats.append(a_format)
return ad_info
if 'url' not in info and not info.get('formats'):
- raise ExtractorError('Unable to extract video url for %s' % info['id'])
+ raise ExtractorError('Unable to extract video url for %s' % video_id)
return info
# non numeric ref: prefixed video id
'url': 'http://players.brightcove.net/710858724001/default_default/index.html?videoId=ref:event-stream-356',
'only_matching': True,
+ }, {
+ # unavailable video without message but with error_code
+ 'url': 'http://players.brightcove.net/1305187701/c832abfb-641b-44eb-9da0-2fe76786505f_default/index.html?videoId=4377407326001',
+ 'only_matching': True,
}]
@staticmethod
})
except ExtractorError as e:
if isinstance(e.cause, compat_HTTPError) and e.cause.code == 403:
- json_data = self._parse_json(e.cause.read().decode(), video_id)
- raise ExtractorError(json_data[0]['message'], expected=True)
+ json_data = self._parse_json(e.cause.read().decode(), video_id)[0]
+ raise ExtractorError(
+ json_data.get('message') or json_data['error_code'], expected=True)
raise
- title = json_data['name']
+ title = json_data['name'].strip()
formats = []
for source in json_data.get('sources', []):
if not src:
continue
formats.extend(self._extract_m3u8_formats(
- src, video_id, 'mp4', m3u8_id='hls', fatal=False))
+ src, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
elif source_type == 'application/dash+xml':
if not src:
continue
formats.append(f)
self._sort_formats(formats)
- description = json_data.get('description')
- thumbnail = json_data.get('thumbnail')
- timestamp = parse_iso8601(json_data.get('published_at'))
- duration = float_or_none(json_data.get('duration'), 1000)
- tags = json_data.get('tags', [])
+ subtitles = {}
+ for text_track in json_data.get('text_tracks', []):
+ if text_track.get('src'):
+ subtitles.setdefault(text_track.get('srclang'), []).append({
+ 'url': text_track['src'],
+ })
return {
'id': video_id,
'title': title,
- 'description': description,
- 'thumbnail': thumbnail,
- 'duration': duration,
- 'timestamp': timestamp,
+ 'description': json_data.get('description'),
+ 'thumbnail': json_data.get('thumbnail') or json_data.get('poster'),
+ 'duration': float_or_none(json_data.get('duration'), 1000),
+ 'timestamp': parse_iso8601(json_data.get('published_at')),
'uploader_id': account_id,
'formats': formats,
- 'tags': tags,
+ 'subtitles': subtitles,
+ 'tags': json_data.get('tags', []),
}