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

   1 # -*- coding: utf-8 -*-
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     compat_urlparse,
   7     ExtractorError,
   8 )
   9
  10
  11 class GoshgayIE(InfoExtractor):
  12     _VALID_URL = r'^(?:https?://)www.goshgay.com/video(?P<id>\d+?)($|/)'
  13     _TEST = {
  14         'url': 'http://www.goshgay.com/video4116282',
  15         'md5': '268b9f3c3229105c57859e166dd72b03',
  16         'info_dict': {
  17             'id': '4116282',
  18             'ext': 'flv',
  19             'title': 'md5:089833a4790b5e103285a07337f245bf',
  20             'thumbnail': 're:http://.*\.jpg',
  21             'age_limit': 18,
  22         }
  23     }
  24
  25     def _real_extract(self, url):
  26         video_id = self._match_id(url)
  27
  28         webpage = self._download_webpage(url, video_id)
  29         title = self._og_search_title(webpage)
  30         thumbnail = self._og_search_thumbnail(webpage)
  31         family_friendly = self._html_search_meta(
  32             'isFamilyFriendly', webpage, default='false')
  33         config_url = self._search_regex(
  34             r"'config'\s*:\s*'([^']+)'", webpage, 'config URL')
  35
  36         config = self._download_xml(
  37             config_url, video_id, 'Downloading player config XML')
  38
  39         if config is None:
  40             raise ExtractorError('Missing config XML')
  41         if config.tag != 'config':
  42             raise ExtractorError('Missing config attribute')
  43         fns = config.findall('file')
  44         if len(fns) < 1:
  45             raise ExtractorError('Missing media URI')
  46         video_url = fns[0].text
  47
  48         url_comp = compat_urlparse.urlparse(url)
  49         ref = "%s://%s%s" % (url_comp[0], url_comp[1], url_comp[2])
  50
  51         return {
  52             'id': video_id,
  53             'url': video_url,
  54             'title': title,
  55             'thumbnail': thumbnail,
  56             'http_referer': ref,
  57             'age_limit': 0 if family_friendly == 'true' else 18,
  58         }