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

   1 # encoding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import ExtractorError
   8 from .bliptv import BlipTVIE
   9
  10
  11 class CinemassacreIE(InfoExtractor):
  12     _VALID_URL = 'https?://(?:www\.)?cinemassacre\.com/(?P<date_y>[0-9]{4})/(?P<date_m>[0-9]{2})/(?P<date_d>[0-9]{2})/(?P<display_id>[^?#/]+)'
  13     _TESTS = [
  14         {
  15             'url': 'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/',
  16             'md5': 'fde81fbafaee331785f58cd6c0d46190',
  17             'info_dict': {
  18                 'id': 'Cinemassacre-19911',
  19                 'ext': 'mp4',
  20                 'upload_date': '20121110',
  21                 'title': '“Angry Video Game Nerd: The Movie” – Trailer',
  22                 'description': 'md5:fb87405fcb42a331742a0dce2708560b',
  23             },
  24         },
  25         {
  26             'url': 'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940',
  27             'md5': 'd72f10cd39eac4215048f62ab477a511',
  28             'info_dict': {
  29                 'id': 'Cinemassacre-521be8ef82b16',
  30                 'ext': 'mp4',
  31                 'upload_date': '20131002',
  32                 'title': 'The Mummy’s Hand (1940)',
  33             },
  34         },
  35         {
  36             # blip.tv embedded video
  37             'url': 'http://cinemassacre.com/2006/12/07/chronologically-confused-about-bad-movie-and-video-game-sequel-titles/',
  38             'md5': 'ca9b3c8dd5a66f9375daeb5135f5a3de',
  39             'info_dict': {
  40                 'id': '4065369',
  41                 'ext': 'flv',
  42                 'title': 'AVGN: Chronologically Confused about Bad Movie and Video Game Sequel Titles',
  43                 'upload_date': '20061207',
  44                 'uploader': 'cinemassacre',
  45                 'uploader_id': '250778',
  46                 'timestamp': 1283233867,
  47                 'description': 'md5:0a108c78d130676b207d0f6d029ecffd',
  48             }
  49         },
  50         {
  51             # Youtube embedded video
  52             'url': 'http://cinemassacre.com/2006/09/01/mckids/',
  53             'md5': '6eb30961fa795fedc750eac4881ad2e1',
  54             'info_dict': {
  55                 'id': 'FnxsNhuikpo',
  56                 'ext': 'mp4',
  57                 'upload_date': '20060901',
  58                 'uploader': 'Cinemassacre Extras',
  59                 'description': 'md5:de9b751efa9e45fbaafd9c8a1123ed53',
  60                 'uploader_id': 'Cinemassacre',
  61                 'title': 'AVGN: McKids',
  62             }
  63         }
  64     ]
  65
  66     def _real_extract(self, url):
  67         mobj = re.match(self._VALID_URL, url)
  68         display_id = mobj.group('display_id')
  69         video_date = mobj.group('date_y') + mobj.group('date_m') + mobj.group('date_d')
  70
  71         webpage = self._download_webpage(url, display_id)
  72
  73         playerdata_url = self._search_regex(
  74             [
  75                 r'src="(http://player\.screenwavemedia\.com/play/[a-zA-Z]+\.php\?[^"]*\bid=.+?)"',
  76                 r'<iframe[^>]+src="((?:https?:)?//(?:[^.]+\.)?youtube\.com/.+?)"',
  77             ],
  78             webpage, 'player data URL', default=None)
  79         if not playerdata_url:
  80             playerdata_url = BlipTVIE._extract_url(webpage)
  81         if not playerdata_url:
  82             raise ExtractorError('Unable to find player data')
  83
  84         video_title = self._html_search_regex(
  85             r'<title>(?P<title>.+?)\|', webpage, 'title')
  86         video_description = self._html_search_regex(
  87             r'<div class="entry-content">(?P<description>.+?)</div>',
  88             webpage, 'description', flags=re.DOTALL, fatal=False)
  89         video_thumbnail = self._og_search_thumbnail(webpage)
  90
  91         return {
  92             '_type': 'url_transparent',
  93             'display_id': display_id,
  94             'title': video_title,
  95             'description': video_description,
  96             'upload_date': video_date,
  97             'thumbnail': video_thumbnail,
  98             'url': playerdata_url,
  99         }