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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import calendar
   5 import re
   6 import time
   7
   8 from .amp import AMPIE
   9 from .common import InfoExtractor
  10 from .youtube import YoutubeIE
  11 from ..compat import compat_urlparse
  12
  13
  14 class AbcNewsVideoIE(AMPIE):
  15     IE_NAME = 'abcnews:video'
  16     _VALID_URL = r'''(?x)
  17                     https?://
  18                         (?:
  19                             abcnews\.go\.com/
  20                             (?:
  21                                 [^/]+/video/(?P<display_id>[0-9a-z-]+)-|
  22                                 video/embed\?.*?\bid=
  23                             )|
  24                             fivethirtyeight\.abcnews\.go\.com/video/embed/\d+/
  25                         )
  26                         (?P<id>\d+)
  27                     '''
  28
  29     _TESTS = [{
  30         'url': 'http://abcnews.go.com/ThisWeek/video/week-exclusive-irans-foreign-minister-zarif-20411932',
  31         'info_dict': {
  32             'id': '20411932',
  33             'ext': 'mp4',
  34             'display_id': 'week-exclusive-irans-foreign-minister-zarif',
  35             'title': '\'This Week\' Exclusive: Iran\'s Foreign Minister Zarif',
  36             'description': 'George Stephanopoulos goes one-on-one with Iranian Foreign Minister Dr. Javad Zarif.',
  37             'duration': 180,
  38             'thumbnail': r're:^https?://.*\.jpg$',
  39         },
  40         'params': {
  41             # m3u8 download
  42             'skip_download': True,
  43         },
  44     }, {
  45         'url': 'http://abcnews.go.com/video/embed?id=46979033',
  46         'only_matching': True,
  47     }, {
  48         'url': 'http://abcnews.go.com/2020/video/2020-husband-stands-teacher-jail-student-affairs-26119478',
  49         'only_matching': True,
  50     }]
  51
  52     def _real_extract(self, url):
  53         mobj = re.match(self._VALID_URL, url)
  54         display_id = mobj.group('display_id')
  55         video_id = mobj.group('id')
  56         info_dict = self._extract_feed_info(
  57             'http://abcnews.go.com/video/itemfeed?id=%s' % video_id)
  58         info_dict.update({
  59             'id': video_id,
  60             'display_id': display_id,
  61         })
  62         return info_dict
  63
  64
  65 class AbcNewsIE(InfoExtractor):
  66     IE_NAME = 'abcnews'
  67     _VALID_URL = r'https?://abcnews\.go\.com/(?:[^/]+/)+(?P<display_id>[0-9a-z-]+)/story\?id=(?P<id>\d+)'
  68
  69     _TESTS = [{
  70         'url': 'http://abcnews.go.com/Blotter/News/dramatic-video-rare-death-job-america/story?id=10498713#.UIhwosWHLjY',
  71         'info_dict': {
  72             'id': '10505354',
  73             'ext': 'flv',
  74             'display_id': 'dramatic-video-rare-death-job-america',
  75             'title': 'Occupational Hazards',
  76             'description': 'Nightline investigates the dangers that lurk at various jobs.',
  77             'thumbnail': r're:^https?://.*\.jpg$',
  78             'upload_date': '20100428',
  79             'timestamp': 1272412800,
  80         },
  81         'add_ie': ['AbcNewsVideo'],
  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             'display_id': 'justin-timberlake-performs-stop-feeling-eurovision-2016',
  88             'title': 'Justin Timberlake Drops Hints For Secret Single',
  89             'description': 'Lara Spencer reports the buzziest stories of the day in "GMA" Pop News.',
  90             'upload_date': '20160515',
  91             'timestamp': 1463329500,
  92         },
  93         'params': {
  94             # m3u8 download
  95             'skip_download': True,
  96             # The embedded YouTube video is blocked due to copyright issues
  97             'playlist_items': '1',
  98         },
  99         'add_ie': ['AbcNewsVideo'],
 100     }, {
 101         'url': 'http://abcnews.go.com/Technology/exclusive-apple-ceo-tim-cook-iphone-cracking-software/story?id=37173343',
 102         'only_matching': True,
 103     }]
 104
 105     def _real_extract(self, url):
 106         mobj = re.match(self._VALID_URL, url)
 107         display_id = mobj.group('display_id')
 108         video_id = mobj.group('id')
 109
 110         webpage = self._download_webpage(url, video_id)
 111         video_url = self._search_regex(
 112             r'window\.abcnvideo\.url\s*=\s*"([^"]+)"', webpage, 'video URL')
 113         full_video_url = compat_urlparse.urljoin(url, video_url)
 114
 115         youtube_url = YoutubeIE._extract_url(webpage)
 116
 117         timestamp = None
 118         date_str = self._html_search_regex(
 119             r'<span[^>]+class="timestamp">([^<]+)</span>',
 120             webpage, 'timestamp', fatal=False)
 121         if date_str:
 122             tz_offset = 0
 123             if date_str.endswith(' ET'):  # Eastern Time
 124                 tz_offset = -5
 125                 date_str = date_str[:-3]
 126             date_formats = ['%b. %d, %Y', '%b %d, %Y, %I:%M %p']
 127             for date_format in date_formats:
 128                 try:
 129                     timestamp = calendar.timegm(time.strptime(date_str.strip(), date_format))
 130                 except ValueError:
 131                     continue
 132             if timestamp is not None:
 133                 timestamp -= tz_offset * 3600
 134
 135         entry = {
 136             '_type': 'url_transparent',
 137             'ie_key': AbcNewsVideoIE.ie_key(),
 138             'url': full_video_url,
 139             'id': video_id,
 140             'display_id': display_id,
 141             'timestamp': timestamp,
 142         }
 143
 144         if youtube_url:
 145             entries = [entry, self.url_result(youtube_url, ie=YoutubeIE.ie_key())]
 146             return self.playlist_result(entries)
 147
 148         return entry