Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/foxnews.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     parse_iso8601,
   6     int_or_none,
   7 )
   8
   9
  10 class FoxNewsIE(InfoExtractor):
  11     _VALID_URL = r'https?://video\.foxnews\.com/v/(?:video-embed\.html\?video_id=)?(?P<id>\d+)'
  12     _TESTS = [
  13         {
  14             'url': 'http://video.foxnews.com/v/3937480/frozen-in-time/#sp=show-clips',
  15             'md5': '32aaded6ba3ef0d1c04e238d01031e5e',
  16             'info_dict': {
  17                 'id': '3937480',
  18                 'ext': 'flv',
  19                 'title': 'Frozen in Time',
  20                 'description': 'Doctors baffled by 16-year-old girl that is the size of a toddler',
  21                 'duration': 265,
  22                 'timestamp': 1304411491,
  23                 'upload_date': '20110503',
  24                 'thumbnail': 're:^https?://.*\.jpg$',
  25             },
  26         },
  27         {
  28             'url': 'http://video.foxnews.com/v/3922535568001/rep-luis-gutierrez-on-if-obamas-immigration-plan-is-legal/#sp=show-clips',
  29             'md5': '5846c64a1ea05ec78175421b8323e2df',
  30             'info_dict': {
  31                 'id': '3922535568001',
  32                 'ext': 'mp4',
  33                 'title': "Rep. Luis Gutierrez on if Obama's immigration plan is legal",
  34                 'description': "Congressman discusses the president's executive action",
  35                 'duration': 292,
  36                 'timestamp': 1417662047,
  37                 'upload_date': '20141204',
  38                 'thumbnail': 're:^https?://.*\.jpg$',
  39             },
  40         },
  41         {
  42             'url': 'http://video.foxnews.com/v/video-embed.html?video_id=3937480&d=video.foxnews.com',
  43             'only_matching': True,
  44         },
  45     ]
  46
  47     def _real_extract(self, url):
  48         video_id = self._match_id(url)
  49
  50         video = self._download_json(
  51             'http://video.foxnews.com/v/feed/video/%s.js?template=fox' % video_id, video_id)
  52
  53         item = video['channel']['item']
  54         title = item['title']
  55         description = item['description']
  56         timestamp = parse_iso8601(item['dc-date'])
  57
  58         media_group = item['media-group']
  59         duration = None
  60         formats = []
  61         for media in media_group['media-content']:
  62             attributes = media['@attributes']
  63             video_url = attributes['url']
  64             if video_url.endswith('.f4m'):
  65                 formats.extend(self._extract_f4m_formats(video_url + '?hdcore=3.4.0&plugin=aasp-3.4.0.132.124', video_id))
  66             elif video_url.endswith('.m3u8'):
  67                 formats.extend(self._extract_m3u8_formats(video_url, video_id, 'flv'))
  68             elif not video_url.endswith('.smil'):
  69                 duration = int_or_none(attributes.get('duration'))
  70                 formats.append({
  71                     'url': video_url,
  72                     'format_id': media['media-category']['@attributes']['label'],
  73                     'preference': 1,
  74                     'vbr': int_or_none(attributes.get('bitrate')),
  75                     'filesize': int_or_none(attributes.get('fileSize'))
  76                 })
  77         self._sort_formats(formats)
  78
  79         media_thumbnail = media_group['media-thumbnail']['@attributes']
  80         thumbnails = [{
  81             'url': media_thumbnail['url'],
  82             'width': int_or_none(media_thumbnail.get('width')),
  83             'height': int_or_none(media_thumbnail.get('height')),
  84         }] if media_thumbnail else []
  85
  86         return {
  87             'id': video_id,
  88             'title': title,
  89             'description': description,
  90             'duration': duration,
  91             'timestamp': timestamp,
  92             'formats': formats,
  93             'thumbnails': thumbnails,
  94         }