Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/fivemin.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from ..compat import (
   5     compat_urllib_parse,
   6     compat_parse_qs,
   7     compat_urllib_parse_urlparse,
   8     compat_urlparse,
   9 )
  10 from ..utils import (
  11     ExtractorError,
  12     parse_duration,
  13     replace_extension,
  14 )
  15
  16
  17 class FiveMinIE(InfoExtractor):
  18     IE_NAME = '5min'
  19     _VALID_URL = r'''(?x)
  20         (?:https?://[^/]*?5min\.com/Scripts/PlayerSeed\.js\?(?:.*?&)?playList=|
  21             https?://(?:(?:massively|www)\.)?joystiq\.com/video/|
  22             5min:)
  23         (?P<id>\d+)
  24         '''
  25
  26     _TESTS = [
  27         {
  28             # From http://www.engadget.com/2013/11/15/ipad-mini-retina-display-review/
  29             'url': 'http://pshared.5min.com/Scripts/PlayerSeed.js?sid=281&width=560&height=345&playList=518013791',
  30             'md5': '4f7b0b79bf1a470e5004f7112385941d',
  31             'info_dict': {
  32                 'id': '518013791',
  33                 'ext': 'mp4',
  34                 'title': 'iPad Mini with Retina Display Review',
  35                 'duration': 177,
  36             },
  37         },
  38         {
  39             # From http://on.aol.com/video/how-to-make-a-next-level-fruit-salad-518086247
  40             'url': '5min:518086247',
  41             'md5': 'e539a9dd682c288ef5a498898009f69e',
  42             'info_dict': {
  43                 'id': '518086247',
  44                 'ext': 'mp4',
  45                 'title': 'How to Make a Next-Level Fruit Salad',
  46                 'duration': 184,
  47             },
  48         },
  49     ]
  50     _ERRORS = {
  51         'ErrorVideoNotExist': 'We\'re sorry, but the video you are trying to watch does not exist.',
  52         'ErrorVideoNoLongerAvailable': 'We\'re sorry, but the video you are trying to watch is no longer available.',
  53         'ErrorVideoRejected': 'We\'re sorry, but the video you are trying to watch has been removed.',
  54         'ErrorVideoUserNotGeo': 'We\'re sorry, but the video you are trying to watch cannot be viewed from your current location.',
  55         'ErrorVideoLibraryRestriction': 'We\'re sorry, but the video you are trying to watch is currently unavailable for viewing at this domain.',
  56         'ErrorExposurePermission': 'We\'re sorry, but the video you are trying to watch is currently unavailable for viewing at this domain.',
  57     }
  58     _QUALITIES = {
  59         1: {
  60             'width': 640,
  61             'height': 360,
  62         },
  63         2: {
  64             'width': 854,
  65             'height': 480,
  66         },
  67         4: {
  68             'width': 1280,
  69             'height': 720,
  70         },
  71         8: {
  72             'width': 1920,
  73             'height': 1080,
  74         },
  75         16: {
  76             'width': 640,
  77             'height': 360,
  78         },
  79         32: {
  80             'width': 854,
  81             'height': 480,
  82         },
  83         64: {
  84             'width': 1280,
  85             'height': 720,
  86         },
  87         128: {
  88             'width': 640,
  89             'height': 360,
  90         },
  91     }
  92
  93     def _real_extract(self, url):
  94         video_id = self._match_id(url)
  95         embed_url = 'https://embed.5min.com/playerseed/?playList=%s' % video_id
  96         embed_page = self._download_webpage(embed_url, video_id,
  97                                             'Downloading embed page')
  98         sid = self._search_regex(r'sid=(\d+)', embed_page, 'sid')
  99         query = compat_urllib_parse.urlencode({
 100             'func': 'GetResults',
 101             'playlist': video_id,
 102             'sid': sid,
 103             'isPlayerSeed': 'true',
 104             'url': embed_url,
 105         })
 106         response = self._download_json(
 107             'https://syn.5min.com/handlers/SenseHandler.ashx?' + query,
 108             video_id)
 109         if not response['success']:
 110             raise ExtractorError(
 111                 '%s said: %s' % (
 112                     self.IE_NAME,
 113                     self._ERRORS.get(response['errorMessage'], response['errorMessage'])),
 114                 expected=True)
 115         info = response['binding'][0]
 116
 117         formats = []
 118         parsed_video_url = compat_urllib_parse_urlparse(compat_parse_qs(
 119             compat_urllib_parse_urlparse(info['EmbededURL']).query)['videoUrl'][0])
 120         for rendition in info['Renditions']:
 121             if rendition['RenditionType'] == 'm3u8':
 122                 formats.extend(self._extract_m3u8_formats(rendition['Url'], video_id, m3u8_id='hls'))
 123             elif rendition['RenditionType'] == 'aac':
 124                 continue
 125             else:
 126                 rendition_url = compat_urlparse.urlunparse(parsed_video_url._replace(path=replace_extension(parsed_video_url.path.replace('//', '/%s/' % rendition['ID']), rendition['RenditionType'])))
 127                 quality = self._QUALITIES.get(rendition['ID'], {})
 128                 formats.append({
 129                     'format_id': '%s-%d' % (rendition['RenditionType'], rendition['ID']),
 130                     'url': rendition_url,
 131                     'width': quality.get('width'),
 132                     'height': quality.get('height'),
 133                 })
 134         self._sort_formats(formats)
 135
 136         return {
 137             'id': video_id,
 138             'title': info['Title'],
 139             'thumbnail': info.get('ThumbURL'),
 140             'duration': parse_duration(info.get('Duration')),
 141             'formats': formats,
 142         }