yt_dlp/extractor/abcnews.py

   1 from .amp import AMPIE
   2 from .common import InfoExtractor
   3 from ..utils import (
   4     parse_duration,
   5     parse_iso8601,
   6     try_get,
   7 )
   8
   9
  10 class AbcNewsVideoIE(AMPIE):
  11     IE_NAME = 'abcnews:video'
  12     _VALID_URL = r'''(?x)
  13                     https?://
  14                         (?:
  15                             abcnews\.go\.com/
  16                             (?:
  17                                 (?:[^/]+/)*video/(?P<display_id>[0-9a-z-]+)-|
  18                                 video/(?:embed|itemfeed)\?.*?\bid=
  19                             )|
  20                             fivethirtyeight\.abcnews\.go\.com/video/embed/\d+/
  21                         )
  22                         (?P<id>\d+)
  23                     '''
  24
  25     _TESTS = [{
  26         'url': 'http://abcnews.go.com/ThisWeek/video/week-exclusive-irans-foreign-minister-zarif-20411932',
  27         'info_dict': {
  28             'id': '20411932',
  29             'ext': 'mp4',
  30             'display_id': 'week-exclusive-irans-foreign-minister-zarif',
  31             'title': '\'This Week\' Exclusive: Iran\'s Foreign Minister Zarif',
  32             'description': 'George Stephanopoulos goes one-on-one with Iranian Foreign Minister Dr. Javad Zarif.',
  33             'duration': 180,
  34             'thumbnail': r're:^https?://.*\.jpg$',
  35             'timestamp': 1380454200,
  36             'upload_date': '20130929',
  37         },
  38         'params': {
  39             # m3u8 download
  40             'skip_download': True,
  41         },
  42     }, {
  43         'url': 'http://abcnews.go.com/video/embed?id=46979033',
  44         'only_matching': True,
  45     }, {
  46         'url': 'http://abcnews.go.com/2020/video/2020-husband-stands-teacher-jail-student-affairs-26119478',
  47         'only_matching': True,
  48     }, {
  49         'url': 'http://abcnews.go.com/video/itemfeed?id=46979033',
  50         'only_matching': True,
  51     }, {
  52         'url': 'https://abcnews.go.com/GMA/News/video/history-christmas-story-67894761',
  53         'only_matching': True,
  54     }]
  55
  56     def _real_extract(self, url):
  57         mobj = self._match_valid_url(url)
  58         display_id = mobj.group('display_id')
  59         video_id = mobj.group('id')
  60         info_dict = self._extract_feed_info(
  61             f'http://abcnews.go.com/video/itemfeed?id={video_id}')
  62         info_dict.update({
  63             'id': video_id,
  64             'display_id': display_id,
  65         })
  66         return info_dict
  67
  68
  69 class AbcNewsIE(InfoExtractor):
  70     IE_NAME = 'abcnews'
  71     _VALID_URL = r'https?://abcnews\.go\.com/(?:[^/]+/)+(?P<display_id>[0-9a-z-]+)/story\?id=(?P<id>\d+)'
  72
  73     _TESTS = [{
  74         # Youtube Embeds
  75         'url': 'https://abcnews.go.com/Entertainment/peter-billingsley-child-actor-christmas-story-hollywood-power/story?id=51286501',
  76         'info_dict': {
  77             'id': '51286501',
  78             'title': "Peter Billingsley: From child actor in 'A Christmas Story' to Hollywood power player",
  79             'description': 'Billingsley went from a child actor to Hollywood power player.',
  80         },
  81         'playlist_count': 5,
  82     }, {
  83         'url': 'http://abcnews.go.com/Entertainment/justin-timberlake-performs-stop-feeling-eurovision-2016/story?id=39125818',
  84         'info_dict': {
  85             'id': '38897857',
  86             'ext': 'mp4',
  87             'title': 'Justin Timberlake Drops Hints For Secret Single',
  88             'description': 'Lara Spencer reports the buzziest stories of the day in "GMA" Pop News.',
  89             'upload_date': '20160505',
  90             'timestamp': 1462442280,
  91         },
  92         'params': {
  93             # m3u8 download
  94             'skip_download': True,
  95             # The embedded YouTube video is blocked due to copyright issues
  96             'playlist_items': '1',
  97         },
  98         'add_ie': ['AbcNewsVideo'],
  99     }, {
 100         'url': 'http://abcnews.go.com/Technology/exclusive-apple-ceo-tim-cook-iphone-cracking-software/story?id=37173343',
 101         'only_matching': True,
 102     }, {
 103         # inline.type == 'video'
 104         'url': 'http://abcnews.go.com/Technology/exclusive-apple-ceo-tim-cook-iphone-cracking-software/story?id=37173343',
 105         'only_matching': True,
 106     }]
 107
 108     def _real_extract(self, url):
 109         story_id = self._match_id(url)
 110         webpage = self._download_webpage(url, story_id)
 111         story = self._parse_json(self._search_regex(
 112             r"window\['__abcnews__'\]\s*=\s*({.+?});",
 113             webpage, 'data'), story_id)['page']['content']['story']['everscroll'][0]
 114         article_contents = story.get('articleContents') or {}
 115
 116         def entries():
 117             featured_video = story.get('featuredVideo') or {}
 118             feed = try_get(featured_video, lambda x: x['video']['feed'])
 119             if feed:
 120                 yield {
 121                     '_type': 'url',
 122                     'id': featured_video.get('id'),
 123                     'title': featured_video.get('name'),
 124                     'url': feed,
 125                     'thumbnail': featured_video.get('images'),
 126                     'description': featured_video.get('description'),
 127                     'timestamp': parse_iso8601(featured_video.get('uploadDate')),
 128                     'duration': parse_duration(featured_video.get('duration')),
 129                     'ie_key': AbcNewsVideoIE.ie_key(),
 130                 }
 131
 132             for inline in (article_contents.get('inlines') or []):
 133                 inline_type = inline.get('type')
 134                 if inline_type == 'iframe':
 135                     iframe_url = try_get(inline, lambda x: x['attrs']['src'])
 136                     if iframe_url:
 137                         yield self.url_result(iframe_url)
 138                 elif inline_type == 'video':
 139                     video_id = inline.get('id')
 140                     if video_id:
 141                         yield {
 142                             '_type': 'url',
 143                             'id': video_id,
 144                             'url': 'http://abcnews.go.com/video/embed?id=' + video_id,
 145                             'thumbnail': inline.get('imgSrc') or inline.get('imgDefault'),
 146                             'description': inline.get('description'),
 147                             'duration': parse_duration(inline.get('duration')),
 148                             'ie_key': AbcNewsVideoIE.ie_key(),
 149                         }
 150
 151         return self.playlist_result(
 152             entries(), story_id, article_contents.get('headline'),
 153             article_contents.get('subHead'))