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

   1 import re
   2 import json
   3
   4 from .common import InfoExtractor
   5 from ..utils import unified_strdate, determine_ext
   6
   7
   8 class RoxwelIE(InfoExtractor):
   9     _VALID_URL = r'https?://www\.roxwel\.com/player/(?P<filename>.+?)(\.|\?|$)'
  10
  11     _TEST = {
  12         u'url': u'http://www.roxwel.com/player/passionpittakeawalklive.html',
  13         u'file': u'passionpittakeawalklive.flv',
  14         u'md5': u'd9dea8360a1e7d485d2206db7fe13035',
  15         u'info_dict': {
  16             u'title': u'Take A Walk (live)',
  17             u'uploader': u'Passion Pit',
  18             u'description': u'Passion Pit performs "Take A Walk\" live at The Backyard in Austin, Texas. ',
  19         },
  20         u'skip': u'Requires rtmpdump',
  21     }
  22
  23     def _real_extract(self, url):
  24         mobj = re.match(self._VALID_URL, url)
  25         filename = mobj.group('filename')
  26         info_url = 'http://www.roxwel.com/api/videos/%s' % filename
  27         info_page = self._download_webpage(info_url, filename,
  28                                            u'Downloading video info')
  29
  30         self.report_extraction(filename)
  31         info = json.loads(info_page)
  32         rtmp_rates = sorted([int(r.replace('flv_', '')) for r in info['media_rates'] if r.startswith('flv_')])
  33         best_rate = rtmp_rates[-1]
  34         url_page_url = 'http://roxwel.com/pl_one_time.php?filename=%s&quality=%s' % (filename, best_rate)
  35         rtmp_url = self._download_webpage(url_page_url, filename, u'Downloading video url')
  36         ext = determine_ext(rtmp_url)
  37         if ext == 'f4v':
  38             rtmp_url = rtmp_url.replace(filename, 'mp4:%s' % filename)
  39
  40         return {'id': filename,
  41                 'title': info['title'],
  42                 'url': rtmp_url,
  43                 'ext': 'flv',
  44                 'description': info['description'],
  45                 'thumbnail': info.get('player_image_url') or info.get('image_url_large'),
  46                 'uploader': info['artist'],
  47                 'uploader_id': info['artistname'],
  48                 'upload_date': unified_strdate(info['dbdate']),
  49                 }