from __future__ import unicode_literals
+import re
+
from .common import InfoExtractor
from ..utils import (
ExtractorError,
int_or_none,
+ float_or_none,
+ unescapeHTML,
)
}, {
'url': 'wistia:sh7fpupwlt',
'only_matching': True,
+ }, {
+ # with hls video
+ 'url': 'wistia:807fafadvk',
+ 'only_matching': True,
}]
+ @staticmethod
+ def _extract_url(webpage):
+ match = re.search(
+ r'<(?:meta[^>]+?content|iframe[^>]+?src)=(["\'])(?P<url>(?:https?:)?//(?:fast\.)?wistia\.net/embed/iframe/.+?)\1', webpage)
+ if match:
+ return unescapeHTML(match.group('url'))
+
+ match = re.search(r'(?:id=["\']wistia_|data-wistia-?id=["\']|Wistia\.embed\(["\'])(?P<id>[^"\']+)', webpage)
+ if match:
+ return 'wistia:%s' % match.group('id')
+
+ match = re.search(
+ r'''(?sx)
+ <script[^>]+src=(["'])(?:https?:)?//fast\.wistia\.com/assets/external/E-v1\.js\1[^>]*>.*?
+ <div[^>]+class=(["']).*?\bwistia_async_(?P<id>[a-z0-9]+)\b.*?\2
+ ''', webpage)
+ if match:
+ return 'wistia:%s' % match.group('id')
+
def _real_extract(self, url):
video_id = self._match_id(url)
formats = []
thumbnails = []
for a in data['assets']:
+ aurl = a.get('url')
+ if not aurl:
+ continue
astatus = a.get('status')
atype = a.get('type')
- if (astatus is not None and astatus != 2) or atype == 'preview':
+ if (astatus is not None and astatus != 2) or atype in ('preview', 'storyboard'):
continue
elif atype in ('still', 'still_image'):
thumbnails.append({
- 'url': a['url'],
- 'resolution': '%dx%d' % (a['width'], a['height']),
+ 'url': aurl,
+ 'width': int_or_none(a.get('width')),
+ 'height': int_or_none(a.get('height')),
})
else:
+ aext = a.get('ext')
+ is_m3u8 = a.get('container') == 'm3u8' or aext == 'm3u8'
formats.append({
'format_id': atype,
- 'url': a['url'],
+ 'url': aurl,
'tbr': int_or_none(a.get('bitrate')),
'vbr': int_or_none(a.get('opt_vbitrate')),
'width': int_or_none(a.get('width')),
'filesize': int_or_none(a.get('size')),
'vcodec': a.get('codec'),
'container': a.get('container'),
- 'ext': a.get('ext'),
+ 'ext': 'mp4' if is_m3u8 else aext,
+ 'protocol': 'm3u8' if is_m3u8 else None,
'preference': 1 if atype == 'original' else None,
})
'description': data.get('seoDescription'),
'formats': formats,
'thumbnails': thumbnails,
- 'duration': int_or_none(data.get('duration')),
+ 'duration': float_or_none(data.get('duration')),
'timestamp': int_or_none(data.get('createdAt')),
}