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 from .screenwavemedia import ScreenwaveMediaIE
  10
  11
  12 class CinemassacreIE(InfoExtractor):
  13     _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>[^?#/]+)'
  14     _TESTS = [
  15         {
  16             'url': 'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/',
  17             'md5': 'fde81fbafaee331785f58cd6c0d46190',
  18             'info_dict': {
  19                 'id': 'Cinemassacre-19911',
  20                 'ext': 'mp4',
  21                 'upload_date': '20121110',
  22                 'title': '“Angry Video Game Nerd: The Movie” – Trailer',
  23                 'description': 'md5:fb87405fcb42a331742a0dce2708560b',
  24             },
  25         },
  26         {
  27             'url': 'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940',
  28             'md5': 'd72f10cd39eac4215048f62ab477a511',
  29             'info_dict': {
  30                 'id': 'Cinemassacre-521be8ef82b16',
  31                 'ext': 'mp4',
  32                 'upload_date': '20131002',
  33                 'title': 'The Mummy’s Hand (1940)',
  34             },
  35         },
  36         {
  37             # blip.tv embedded video
  38             'url': 'http://cinemassacre.com/2006/12/07/chronologically-confused-about-bad-movie-and-video-game-sequel-titles/',
  39             'md5': 'ca9b3c8dd5a66f9375daeb5135f5a3de',
  40             'info_dict': {
  41                 'id': '4065369',
  42                 'ext': 'flv',
  43                 'title': 'AVGN: Chronologically Confused about Bad Movie and Video Game Sequel Titles',
  44                 'upload_date': '20061207',
  45                 'uploader': 'cinemassacre',
  46                 'uploader_id': '250778',
  47                 'timestamp': 1283233867,
  48                 'description': 'md5:0a108c78d130676b207d0f6d029ecffd',
  49             }
  50         },
  51         {
  52             # Youtube embedded video
  53             'url': 'http://cinemassacre.com/2006/09/01/mckids/',
  54             'md5': '6eb30961fa795fedc750eac4881ad2e1',
  55             'info_dict': {
  56                 'id': 'FnxsNhuikpo',
  57                 'ext': 'mp4',
  58                 'upload_date': '20060901',
  59                 'uploader': 'Cinemassacre Extras',
  60                 'description': 'md5:de9b751efa9e45fbaafd9c8a1123ed53',
  61                 'uploader_id': 'Cinemassacre',
  62                 'title': 'AVGN: McKids',
  63             }
  64         },
  65         {
  66             'url': 'http://cinemassacre.com/2015/05/25/mario-kart-64-nintendo-64-james-mike-mondays/',
  67             'md5': '1376908e49572389e7b06251a53cdd08',
  68             'info_dict': {
  69                 'id': 'Cinemassacre-555779690c440',
  70                 'ext': 'mp4',
  71                 'description': 'Let’s Play Mario Kart 64 !! Mario Kart 64 is a classic go-kart racing game released for the Nintendo 64 (N64). Today James & Mike do 4 player Battle Mode with Kyle and Bootsy!',
  72                 'title': 'Mario Kart 64 (Nintendo 64) James & Mike Mondays',
  73                 'upload_date': '20150525',
  74             }
  75         }
  76     ]
  77
  78     def _real_extract(self, url):
  79         mobj = re.match(self._VALID_URL, url)
  80         display_id = mobj.group('display_id')
  81         video_date = mobj.group('date_y') + mobj.group('date_m') + mobj.group('date_d')
  82
  83         webpage = self._download_webpage(url, display_id)
  84
  85         playerdata_url = self._search_regex(
  86             [
  87                 ScreenwaveMediaIE.EMBED_PATTERN,
  88                 r'<iframe[^>]+src="(?P<url>(?:https?:)?//(?:[^.]+\.)?youtube\.com/.+?)"',
  89             ],
  90             webpage, 'player data URL', default=None, group='url')
  91         if not playerdata_url:
  92             playerdata_url = BlipTVIE._extract_url(webpage)
  93         if not playerdata_url:
  94             raise ExtractorError('Unable to find player data')
  95
  96         video_title = self._html_search_regex(
  97             r'<title>(?P<title>.+?)\|', webpage, 'title')
  98         video_description = self._html_search_regex(
  99             r'<div class="entry-content">(?P<description>.+?)</div>',
 100             webpage, 'description', flags=re.DOTALL, fatal=False)
 101         video_thumbnail = self._og_search_thumbnail(webpage)
 102
 103         return {
 104             '_type': 'url_transparent',
 105             'display_id': display_id,
 106             'title': video_title,
 107             'description': video_description,
 108             'upload_date': video_date,
 109             'thumbnail': video_thumbnail,
 110             'url': playerdata_url,
 111         }