_ Git - youtube-dl/blob - youtube_dl/extractor/foxnews.py

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     parse_iso8601,
   8     int_or_none,
   9 )
  10
  11
  12 class FoxNewsIE(InfoExtractor):
  13     _VALID_URL = r'https?://video\.fox(?:news|business)\.com/v/(?:video-embed\.html\?video_id=)?(?P<id>\d+)'
  14     _TESTS = [
  15         {
  16             'url': 'http://video.foxnews.com/v/3937480/frozen-in-time/#sp=show-clips',
  17             'md5': '32aaded6ba3ef0d1c04e238d01031e5e',
  18             'info_dict': {
  19                 'id': '3937480',
  20                 'ext': 'flv',
  21                 'title': 'Frozen in Time',
  22                 'description': 'Doctors baffled by 16-year-old girl that is the size of a toddler',
  23                 'duration': 265,
  24                 'timestamp': 1304411491,
  25                 'upload_date': '20110503',
  26                 'thumbnail': 're:^https?://.*\.jpg$',
  27             },
  28         },
  29         {
  30             'url': 'http://video.foxnews.com/v/3922535568001/rep-luis-gutierrez-on-if-obamas-immigration-plan-is-legal/#sp=show-clips',
  31             'md5': '5846c64a1ea05ec78175421b8323e2df',
  32             'info_dict': {
  33                 'id': '3922535568001',
  34                 'ext': 'mp4',
  35                 'title': "Rep. Luis Gutierrez on if Obama's immigration plan is legal",
  36                 'description': "Congressman discusses the president's executive action",
  37                 'duration': 292,
  38                 'timestamp': 1417662047,
  39                 'upload_date': '20141204',
  40                 'thumbnail': 're:^https?://.*\.jpg$',
  41             },
  42         },
  43         {
  44             'url': 'http://video.foxnews.com/v/video-embed.html?video_id=3937480&d=video.foxnews.com',
  45             'only_matching': True,
  46         },
  47     ]
  48
  49     def _real_extract(self, url):
  50         video_id = self._match_id(url)
  51
  52         m = re.match(r'^https?://video\.fox(news|business)', url)
  53
  54         video = self._download_json(
  55             'http://video.fox' + m.group(1) + '.com/v/feed/video/%s.js?template=fox' % video_id, video_id)
  56
  57         item = video['channel']['item']
  58         title = item['title']
  59         description = item['description']
  60         timestamp = parse_iso8601(item['dc-date'])
  61
  62         media_group = item['media-group']
  63         duration = None
  64         formats = []
  65         for media in media_group['media-content']:
  66             attributes = media['@attributes']
  67             video_url = attributes['url']
  68             if video_url.endswith('.f4m'):
  69                 formats.extend(self._extract_f4m_formats(video_url + '?hdcore=3.4.0&plugin=aasp-3.4.0.132.124', video_id))
  70             elif video_url.endswith('.m3u8'):
  71                 formats.extend(self._extract_m3u8_formats(video_url, video_id, 'flv'))
  72             elif not video_url.endswith('.smil'):
  73                 duration = int_or_none(attributes.get('duration'))
  74                 formats.append({
  75                     'url': video_url,
  76                     'format_id': media['media-category']['@attributes']['label'],
  77                     'preference': 1,
  78                     'vbr': int_or_none(attributes.get('bitrate')),
  79                     'filesize': int_or_none(attributes.get('fileSize'))
  80                 })
  81         self._sort_formats(formats)
  82
  83         media_thumbnail = media_group['media-thumbnail']['@attributes']
  84         thumbnails = [{
  85             'url': media_thumbnail['url'],
  86             'width': int_or_none(media_thumbnail.get('width')),
  87             'height': int_or_none(media_thumbnail.get('height')),
  88         }] if media_thumbnail else []
  89
  90         return {
  91             'id': video_id,
  92             'title': title,
  93             'description': description,
  94             'duration': duration,
  95             'timestamp': timestamp,
  96             'formats': formats,
  97             'thumbnails': thumbnails,
  98         }