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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     qualities,
   7     unified_strdate,
   8     clean_html,
   9 )
  10
  11
  12 class UltimediaIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?ultimedia\.com/default/index/video[^/]+/id/(?P<id>[\d+a-z]+)'
  14     _TESTS = [{
  15         # news
  16         'url': 'https://www.ultimedia.com/default/index/videogeneric/id/s8uk0r',
  17         'md5': '276a0e49de58c7e85d32b057837952a2',
  18         'info_dict': {
  19             'id': 's8uk0r',
  20             'ext': 'mp4',
  21             'title': 'Loi sur la fin de vie: le texte prévoit un renforcement des directives anticipées',
  22             'description': 'md5:3e5c8fd65791487333dda5db8aed32af',
  23             'thumbnail': 're:^https?://.*\.jpg',
  24             'upload_date': '20150317',
  25         },
  26     }, {
  27         # music
  28         'url': 'https://www.ultimedia.com/default/index/videomusic/id/xvpfp8',
  29         'md5': '2ea3513813cf230605c7e2ffe7eca61c',
  30         'info_dict': {
  31             'id': 'xvpfp8',
  32             'ext': 'mp4',
  33             'title': "Two - C'est la vie (Clip)",
  34             'description': 'Two',
  35             'thumbnail': 're:^https?://.*\.jpg',
  36             'upload_date': '20150224',
  37         },
  38     }]
  39
  40     def _real_extract(self, url):
  41         video_id = self._match_id(url)
  42
  43         webpage = self._download_webpage(url, video_id)
  44
  45         deliver_url = self._search_regex(
  46             r'<iframe[^>]+src="(https?://(?:www\.)?ultimedia\.com/deliver/[^"]+)"',
  47             webpage, 'deliver URL')
  48
  49         deliver_page = self._download_webpage(
  50             deliver_url, video_id, 'Downloading iframe page')
  51
  52         player = self._parse_json(
  53             self._search_regex(
  54                 r"jwplayer\('player(?:_temp)?'\)\.setup\(({.+?})\)\.on", deliver_page, 'player'),
  55             video_id)
  56
  57         quality = qualities(['flash', 'html5'])
  58
  59         formats = [{
  60             'url': mode['config']['file'],
  61             'format_id': mode.get('type'),
  62             'quality': quality(mode.get('type')),
  63         } for mode in player['modes']]
  64         self._sort_formats(formats)
  65
  66         thumbnail = player.get('image')
  67
  68         title = clean_html((
  69             self._html_search_regex(
  70                 r'(?s)<div\s+id="catArticle">.+?</div>(.+?)</h1>',
  71                 webpage, 'title', default=None)
  72             or self._search_regex(
  73                 r"var\s+nameVideo\s*=\s*'([^']+)'",
  74                 deliver_page, 'title')))
  75
  76         description = clean_html(self._html_search_regex(
  77             r'(?s)<span>Description</span>(.+?)</p>', webpage,
  78             'description', fatal=False))
  79
  80         upload_date = unified_strdate(self._search_regex(
  81             r'Ajouté le\s*<span>([^<]+)', webpage,
  82             'upload date', fatal=False))
  83
  84         return {
  85             'id': video_id,
  86             'title': title,
  87             'description': description,
  88             'thumbnail': thumbnail,
  89             'upload_date': upload_date,
  90             'formats': formats,
  91         }