Raphaël G. Git Repositories - youtubedl/blob - 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
   9 class AdultSwimIE(InfoExtractor):
  10     _VALID_URL = r'https?://video\.adultswim\.com/(?P<path>.+?)(?:\.html)?(?:\?.*)?(?:#.*)?$'
  11     _TEST = {
  12         'url': 'http://video.adultswim.com/rick-and-morty/close-rick-counters-of-the-rick-kind.html?x=y#title',
  13         'playlist': [
  14             {
  15                 'md5': '4da359ec73b58df4575cd01a610ba5dc',
  16                 'info_dict': {
  17                     'id': '8a250ba1450996e901453d7f02ca02f5',
  18                     'ext': 'flv',
  19                     'title': 'Rick and Morty Close Rick-Counters of the Rick Kind part 1',
  20                     'description': 'Rick has a run in with some old associates, resulting in a fallout with Morty. You got any chips, broh?',
  21                     'uploader': 'Rick and Morty',
  22                     'thumbnail': 'http://i.cdn.turner.com/asfix/repository/8a250ba13f865824013fc9db8b6b0400/thumbnail_267549017116827057.jpg'
  23                 }
  24             },
  25             {
  26                 'md5': 'ffbdf55af9331c509d95350bd0cc1819',
  27                 'info_dict': {
  28                     'id': '8a250ba1450996e901453d7f4bd102f6',
  29                     'ext': 'flv',
  30                     'title': 'Rick and Morty Close Rick-Counters of the Rick Kind part 2',
  31                     'description': 'Rick has a run in with some old associates, resulting in a fallout with Morty. You got any chips, broh?',
  32                     'uploader': 'Rick and Morty',
  33                     'thumbnail': 'http://i.cdn.turner.com/asfix/repository/8a250ba13f865824013fc9db8b6b0400/thumbnail_267549017116827057.jpg'
  34                 }
  35             },
  36             {
  37                 'md5': 'b92409635540304280b4b6c36bd14a0a',
  38                 'info_dict': {
  39                     'id': '8a250ba1450996e901453d7fa73c02f7',
  40                     'ext': 'flv',
  41                     'title': 'Rick and Morty Close Rick-Counters of the Rick Kind part 3',
  42                     'description': 'Rick has a run in with some old associates, resulting in a fallout with Morty. You got any chips, broh?',
  43                     'uploader': 'Rick and Morty',
  44                     'thumbnail': 'http://i.cdn.turner.com/asfix/repository/8a250ba13f865824013fc9db8b6b0400/thumbnail_267549017116827057.jpg'
  45                 }
  46             },
  47             {
  48                 'md5': 'e8818891d60e47b29cd89d7b0278156d',
  49                 'info_dict': {
  50                     'id': '8a250ba1450996e901453d7fc8ba02f8',
  51                     'ext': 'flv',
  52                     'title': 'Rick and Morty Close Rick-Counters of the Rick Kind part 4',
  53                     'description': 'Rick has a run in with some old associates, resulting in a fallout with Morty. You got any chips, broh?',
  54                     'uploader': 'Rick and Morty',
  55                     'thumbnail': 'http://i.cdn.turner.com/asfix/repository/8a250ba13f865824013fc9db8b6b0400/thumbnail_267549017116827057.jpg'
  56                 }
  57             }
  58         ]
  59     }
  60
  61     _video_extensions = {
  62         '3500': 'flv',
  63         '640': 'mp4',
  64         '150': 'mp4',
  65         'ipad': 'm3u8',
  66         'iphone': 'm3u8'
  67     }
  68     _video_dimensions = {
  69         '3500': (1280, 720),
  70         '640': (480, 270),
  71         '150': (320, 180)
  72     }
  73
  74     def _real_extract(self, url):
  75         mobj = re.match(self._VALID_URL, url)
  76         video_path = mobj.group('path')
  77
  78         webpage = self._download_webpage(url, video_path)
  79         episode_id = self._html_search_regex(
  80             r'<link rel="video_src" href="http://i\.adultswim\.com/adultswim/adultswimtv/tools/swf/viralplayer.swf\?id=([0-9a-f]+?)"\s*/?\s*>',
  81             webpage, 'episode_id')
  82         title = self._og_search_title(webpage)
  83
  84         index_url = 'http://asfix.adultswim.com/asfix-svc/episodeSearch/getEpisodesByIDs?networkName=AS&ids=%s' % episode_id
  85         idoc = self._download_xml(index_url, title, 'Downloading episode index', 'Unable to download episode index')
  86
  87         episode_el = idoc.find('.//episode')
  88         show_title = episode_el.attrib.get('collectionTitle')
  89         episode_title = episode_el.attrib.get('title')
  90         thumbnail = episode_el.attrib.get('thumbnailUrl')
  91         description = episode_el.find('./description').text.strip()
  92
  93         entries = []
  94         segment_els = episode_el.findall('./segments/segment')
  95
  96         for part_num, segment_el in enumerate(segment_els):
  97             segment_id = segment_el.attrib.get('id')
  98             segment_title = '%s %s part %d' % (show_title, episode_title, part_num + 1)
  99             thumbnail = segment_el.attrib.get('thumbnailUrl')
 100             duration = segment_el.attrib.get('duration')
 101
 102             segment_url = 'http://asfix.adultswim.com/asfix-svc/episodeservices/getCvpPlaylist?networkName=AS&id=%s' % segment_id
 103             idoc = self._download_xml(
 104                 segment_url, segment_title,
 105                 'Downloading segment information', 'Unable to download segment information')
 106
 107             formats = []
 108             file_els = idoc.findall('.//files/file')
 109
 110             for file_el in file_els:
 111                 bitrate = file_el.attrib.get('bitrate')
 112                 type = file_el.attrib.get('type')
 113                 width, height = self._video_dimensions.get(bitrate, (None, None))
 114                 formats.append({
 115                     'format_id': '%s-%s' % (bitrate, type),
 116                     'url': file_el.text,
 117                     'ext': self._video_extensions.get(bitrate, 'mp4'),
 118                     # The bitrate may not be a number (for example: 'iphone')
 119                     'tbr': int(bitrate) if bitrate.isdigit() else None,
 120                     'height': height,
 121                     'width': width
 122                 })
 123
 124             self._sort_formats(formats)
 125
 126             entries.append({
 127                 'id': segment_id,
 128                 'title': segment_title,
 129                 'formats': formats,
 130                 'uploader': show_title,
 131                 'thumbnail': thumbnail,
 132                 'duration': duration,
 133                 'description': description
 134             })
 135
 136         return {
 137             '_type': 'playlist',
 138             'id': episode_id,
 139             'display_id': video_path,
 140             'entries': entries,
 141             'title': '%s %s' % (show_title, episode_title),
 142             'description': description,
 143             'thumbnail': thumbnail
 144         }