Raphaël G. Git Repositories - youtubedl/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     ExtractorError,
  10 )
  11
  12
  13 class FiveMinIE(InfoExtractor):
  14     IE_NAME = '5min'
  15     _VALID_URL = r'''(?x)
  16         (?:https?://[^/]*?5min\.com/Scripts/PlayerSeed\.js\?(.*?&)?playList=|
  17             5min:)
  18         (?P<id>\d+)
  19         '''
  20
  21     _TESTS = [
  22         {
  23             # From http://www.engadget.com/2013/11/15/ipad-mini-retina-display-review/
  24             'url': 'http://pshared.5min.com/Scripts/PlayerSeed.js?sid=281&width=560&height=345&playList=518013791',
  25             'md5': '4f7b0b79bf1a470e5004f7112385941d',
  26             'info_dict': {
  27                 'id': '518013791',
  28                 'ext': 'mp4',
  29                 'title': 'iPad Mini with Retina Display Review',
  30             },
  31         },
  32         {
  33             # From http://on.aol.com/video/how-to-make-a-next-level-fruit-salad-518086247
  34             'url': '5min:518086247',
  35             'md5': 'e539a9dd682c288ef5a498898009f69e',
  36             'info_dict': {
  37                 'id': '518086247',
  38                 'ext': 'mp4',
  39                 'title': 'How to Make a Next-Level Fruit Salad',
  40             },
  41         },
  42     ]
  43
  44     @classmethod
  45     def _build_result(cls, video_id):
  46         return cls.url_result('5min:%s' % video_id, cls.ie_key())
  47
  48     def _real_extract(self, url):
  49         mobj = re.match(self._VALID_URL, url)
  50         video_id = mobj.group('id')
  51         embed_url = 'https://embed.5min.com/playerseed/?playList=%s' % video_id
  52         embed_page = self._download_webpage(embed_url, video_id,
  53                                             'Downloading embed page')
  54         sid = self._search_regex(r'sid=(\d+)', embed_page, 'sid')
  55         query = compat_urllib_parse.urlencode({
  56             'func': 'GetResults',
  57             'playlist': video_id,
  58             'sid': sid,
  59             'isPlayerSeed': 'true',
  60             'url': embed_url,
  61         })
  62         response = self._download_json(
  63             'https://syn.5min.com/handlers/SenseHandler.ashx?' + query,
  64             video_id)
  65         if not response['success']:
  66             err_msg = response['errorMessage']
  67             if err_msg == 'ErrorVideoUserNotGeo':
  68                 msg = 'Video not available from your location'
  69             else:
  70                 msg = 'Aol said: %s' % err_msg
  71             raise ExtractorError(msg, expected=True, video_id=video_id)
  72         info = response['binding'][0]
  73
  74         second_id = compat_str(int(video_id[:-2]) + 1)
  75         formats = []
  76         for quality, height in [(1, 320), (2, 480), (4, 720), (8, 1080)]:
  77             if any(r['ID'] == quality for r in info['Renditions']):
  78                 formats.append({
  79                     'format_id': compat_str(quality),
  80                     'url': 'http://avideos.5min.com/%s/%s/%s_%s.mp4' % (second_id[-3:], second_id, video_id, quality),
  81                     'height': height,
  82                 })
  83
  84         return {
  85             'id': video_id,
  86             'title': info['Title'],
  87             'formats': formats,
  88         }