youtube_dl/extractor/adultswim.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7
   8 class AdultSwimIE(InfoExtractor):
   9     _VALID_URL = r'https?://video\.adultswim\.com/(?P<path>.+?)(?:\.html)?(?:\?.*)?(?:#.*)?$'
  10     _TEST = {
  11         'url': 'http://video.adultswim.com/rick-and-morty/close-rick-counters-of-the-rick-kind.html?x=y#title',
  12         'playlist': [
  13             {
  14                 'md5': '4da359ec73b58df4575cd01a610ba5dc',
  15                 'info_dict': {
  16                     'id': '8a250ba1450996e901453d7f02ca02f5',
  17                     'ext': 'flv',
  18                     'title': 'Rick and Morty Close Rick-Counters of the Rick Kind part 1',
  19                     'description': 'Rick has a run in with some old associates, resulting in a fallout with Morty. You got any chips, broh?',
  20                     'uploader': 'Rick and Morty',
  21                     'thumbnail': 'http://i.cdn.turner.com/asfix/repository/8a250ba13f865824013fc9db8b6b0400/thumbnail_267549017116827057.jpg'
  22                 }
  23             },
  24             {
  25                 'md5': 'ffbdf55af9331c509d95350bd0cc1819',
  26                 'info_dict': {
  27                     'id': '8a250ba1450996e901453d7f4bd102f6',
  28                     'ext': 'flv',
  29                     'title': 'Rick and Morty Close Rick-Counters of the Rick Kind part 2',
  30                     'description': 'Rick has a run in with some old associates, resulting in a fallout with Morty. You got any chips, broh?',
  31                     'uploader': 'Rick and Morty',
  32                     'thumbnail': 'http://i.cdn.turner.com/asfix/repository/8a250ba13f865824013fc9db8b6b0400/thumbnail_267549017116827057.jpg'
  33                 }
  34             },
  35             {
  36                 'md5': 'b92409635540304280b4b6c36bd14a0a',
  37                 'info_dict': {
  38                     'id': '8a250ba1450996e901453d7fa73c02f7',
  39                     'ext': 'flv',
  40                     'title': 'Rick and Morty Close Rick-Counters of the Rick Kind part 3',
  41                     'description': 'Rick has a run in with some old associates, resulting in a fallout with Morty. You got any chips, broh?',
  42                     'uploader': 'Rick and Morty',
  43                     'thumbnail': 'http://i.cdn.turner.com/asfix/repository/8a250ba13f865824013fc9db8b6b0400/thumbnail_267549017116827057.jpg'
  44                 }
  45             },
  46             {
  47                 'md5': 'e8818891d60e47b29cd89d7b0278156d',
  48                 'info_dict': {
  49                     'id': '8a250ba1450996e901453d7fc8ba02f8',
  50                     'ext': 'flv',
  51                     'title': 'Rick and Morty Close Rick-Counters of the Rick Kind part 4',
  52                     'description': 'Rick has a run in with some old associates, resulting in a fallout with Morty. You got any chips, broh?',
  53                     'uploader': 'Rick and Morty',
  54                     'thumbnail': 'http://i.cdn.turner.com/asfix/repository/8a250ba13f865824013fc9db8b6b0400/thumbnail_267549017116827057.jpg'
  55                 }
  56             }
  57         ]
  58     }
  59
  60     _video_extensions = {
  61         '3500': 'flv',
  62         '640': 'mp4',
  63         '150': 'mp4',
  64         'ipad': 'm3u8',
  65         'iphone': 'm3u8'
  66     }
  67     _video_dimensions = {
  68         '3500': (1280, 720),
  69         '640': (480, 270),
  70         '150': (320, 180)
  71     }
  72
  73     def _real_extract(self, url):
  74         mobj = re.match(self._VALID_URL, url)
  75         video_path = mobj.group('path')
  76
  77         webpage = self._download_webpage(url, video_path)
  78         episode_id = self._html_search_regex(r'<link rel="video_src" href="http://i\.adultswim\.com/adultswim/adultswimtv/tools/swf/viralplayer.swf\?id=([0-9a-f]+?)"\s*/?\s*>', webpage, 'episode_id')
  79         title = self._og_search_title(webpage)
  80
  81         index_url = 'http://asfix.adultswim.com/asfix-svc/episodeSearch/getEpisodesByIDs?networkName=AS&ids=%s' % episode_id
  82         idoc = self._download_xml(index_url, title, 'Downloading episode index', 'Unable to download episode index')
  83
  84         episode_el = idoc.find('.//episode')
  85         show_title = episode_el.attrib.get('collectionTitle')
  86         episode_title = episode_el.attrib.get('title')
  87         thumbnail = episode_el.attrib.get('thumbnailUrl')
  88         description = episode_el.find('./description').text.strip()
  89
  90         entries = []
  91         segment_els = episode_el.findall('./segments/segment')
  92
  93         for part_num, segment_el in enumerate(segment_els):
  94             segment_id = segment_el.attrib.get('id')
  95             segment_title = '%s %s part %d' % (show_title, episode_title, part_num + 1)
  96             thumbnail = segment_el.attrib.get('thumbnailUrl')
  97             duration = segment_el.attrib.get('duration')
  98
  99             segment_url = 'http://asfix.adultswim.com/asfix-svc/episodeservices/getCvpPlaylist?networkName=AS&id=%s' % segment_id
 100             idoc = self._download_xml(segment_url, segment_title, 'Downloading segment information', 'Unable to download segment information')
 101
 102             formats = []
 103             file_els = idoc.findall('.//files/file')
 104
 105             for file_el in file_els:
 106                 bitrate = file_el.attrib.get('bitrate')
 107                 type = file_el.attrib.get('type')
 108                 width, height = self._video_dimensions.get(bitrate, (None, None))
 109                 formats.append({
 110                     'format_id': '%s-%s' % (bitrate, type),
 111                     'url': file_el.text,
 112                     'ext': self._video_extensions.get(bitrate, 'mp4'),
 113                     # The bitrate may not be a number (for example: 'iphone')
 114                     'tbr': int(bitrate) if bitrate.isdigit() else None,
 115                     'height': height,
 116                     'width': width
 117                 })
 118
 119             self._sort_formats(formats)
 120
 121             entries.append({
 122                 'id': segment_id,
 123                 'title': segment_title,
 124                 'formats': formats,
 125                 'uploader': show_title,
 126                 'thumbnail': thumbnail,
 127                 'duration': duration,
 128                 'description': description
 129             })
 130
 131         return {
 132             '_type': 'playlist',
 133             'id': episode_id,
 134             'display_id': video_path,
 135             'entries': entries,
 136             'title': '%s %s' % (show_title, episode_title),
 137             'description': description,
 138             'thumbnail': thumbnail
 139         }