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

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     compat_str,
   8     compat_urllib_parse,
   9 )
  10
  11
  12 class FiveMinIE(InfoExtractor):
  13     IE_NAME = '5min'
  14     _VALID_URL = r'''(?x)
  15         (?:https?://[^/]*?5min\.com/Scripts/PlayerSeed\.js\?(.*?&)?playList=|
  16             5min:)
  17         (?P<id>\d+)
  18         '''
  19
  20     _TESTS = [
  21         {
  22             # From http://www.engadget.com/2013/11/15/ipad-mini-retina-display-review/
  23             'url': 'http://pshared.5min.com/Scripts/PlayerSeed.js?sid=281&width=560&height=345&playList=518013791',
  24             'md5': '4f7b0b79bf1a470e5004f7112385941d',
  25             'info_dict': {
  26                 'id': '518013791',
  27                 'ext': 'mp4',
  28                 'title': 'iPad Mini with Retina Display Review',
  29             },
  30         },
  31         {
  32             # From http://on.aol.com/video/how-to-make-a-next-level-fruit-salad-518086247
  33             'url': '5min:518086247',
  34             'md5': 'e539a9dd682c288ef5a498898009f69e',
  35             'info_dict': {
  36                 'id': '518086247',
  37                 'ext': 'mp4',
  38                 'title': 'How to Make a Next-Level Fruit Salad',
  39             },
  40         },
  41     ]
  42
  43     @classmethod
  44     def _build_result(cls, video_id):
  45         return cls.url_result('5min:%s' % video_id, cls.ie_key())
  46
  47     def _real_extract(self, url):
  48         mobj = re.match(self._VALID_URL, url)
  49         video_id = mobj.group('id')
  50         embed_url = 'https://embed.5min.com/playerseed/?playList=%s' % video_id
  51         embed_page = self._download_webpage(embed_url, video_id,
  52             'Downloading embed page')
  53         sid = self._search_regex(r'sid=(\d+)', embed_page, 'sid')
  54         query = compat_urllib_parse.urlencode({
  55             'func': 'GetResults',
  56             'playlist': video_id,
  57             'sid': sid,
  58             'isPlayerSeed': 'true',
  59             'url': embed_url,
  60         })
  61         info = self._download_json(
  62             'https://syn.5min.com/handlers/SenseHandler.ashx?' + query,
  63             video_id)['binding'][0]
  64
  65         second_id = compat_str(int(video_id[:-2]) + 1)
  66         formats = []
  67         for quality, height in [(1, 320), (2, 480), (4, 720), (8, 1080)]:
  68             if any(r['ID'] == quality for r in info['Renditions']):
  69                 formats.append({
  70                     'format_id': compat_str(quality),
  71                     'url': 'http://avideos.5min.com/%s/%s/%s_%s.mp4' % (second_id[-3:], second_id, video_id, quality),
  72                     'height': height,
  73                 })
  74
  75         return {
  76             'id': video_id,
  77             'title': info['Title'],
  78             'formats': formats,
  79         }