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 .screenwavemedia import ScreenwaveMediaIE
   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             'params': {
  25                 # m3u8 download
  26                 'skip_download': True,
  27             },
  28         },
  29         {
  30             'url': 'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940',
  31             'md5': 'd72f10cd39eac4215048f62ab477a511',
  32             'info_dict': {
  33                 'id': 'Cinemassacre-521be8ef82b16',
  34                 'ext': 'mp4',
  35                 'upload_date': '20131002',
  36                 'title': 'The Mummy’s Hand (1940)',
  37             },
  38             'params': {
  39                 # m3u8 download
  40                 'skip_download': True,
  41             },
  42         },
  43         {
  44             # Youtube embedded video
  45             'url': 'http://cinemassacre.com/2006/12/07/chronologically-confused-about-bad-movie-and-video-game-sequel-titles/',
  46             'md5': 'ec9838a5520ef5409b3e4e42fcb0a3b9',
  47             'info_dict': {
  48                 'id': 'OEVzPCY2T-g',
  49                 'ext': 'webm',
  50                 'title': 'AVGN: Chronologically Confused about Bad Movie and Video Game Sequel Titles',
  51                 'upload_date': '20061207',
  52                 'uploader': 'Cinemassacre',
  53                 'uploader_id': 'JamesNintendoNerd',
  54                 'description': 'md5:784734696c2b8b7f4b8625cc799e07f6',
  55             }
  56         },
  57         {
  58             # Youtube embedded video
  59             'url': 'http://cinemassacre.com/2006/09/01/mckids/',
  60             'md5': '7393c4e0f54602ad110c793eb7a6513a',
  61             'info_dict': {
  62                 'id': 'FnxsNhuikpo',
  63                 'ext': 'webm',
  64                 'upload_date': '20060901',
  65                 'uploader': 'Cinemassacre Extra',
  66                 'description': 'md5:de9b751efa9e45fbaafd9c8a1123ed53',
  67                 'uploader_id': 'Cinemassacre',
  68                 'title': 'AVGN: McKids',
  69             }
  70         },
  71         {
  72             'url': 'http://cinemassacre.com/2015/05/25/mario-kart-64-nintendo-64-james-mike-mondays/',
  73             'md5': '1376908e49572389e7b06251a53cdd08',
  74             'info_dict': {
  75                 'id': 'Cinemassacre-555779690c440',
  76                 'ext': 'mp4',
  77                 '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!',
  78                 'title': 'Mario Kart 64 (Nintendo 64) James & Mike Mondays',
  79                 'upload_date': '20150525',
  80             },
  81             'params': {
  82                 # m3u8 download
  83                 'skip_download': True,
  84             },
  85         }
  86     ]
  87
  88     def _real_extract(self, url):
  89         mobj = re.match(self._VALID_URL, url)
  90         display_id = mobj.group('display_id')
  91         video_date = mobj.group('date_y') + mobj.group('date_m') + mobj.group('date_d')
  92
  93         webpage = self._download_webpage(url, display_id)
  94
  95         playerdata_url = self._search_regex(
  96             [
  97                 ScreenwaveMediaIE.EMBED_PATTERN,
  98                 r'<iframe[^>]+src="(?P<url>(?:https?:)?//(?:[^.]+\.)?youtube\.com/.+?)"',
  99             ],
 100             webpage, 'player data URL', default=None, group='url')
 101         if not playerdata_url:
 102             raise ExtractorError('Unable to find player data')
 103
 104         video_title = self._html_search_regex(
 105             r'<title>(?P<title>.+?)\|', webpage, 'title')
 106         video_description = self._html_search_regex(
 107             r'<div class="entry-content">(?P<description>.+?)</div>',
 108             webpage, 'description', flags=re.DOTALL, fatal=False)
 109         video_thumbnail = self._og_search_thumbnail(webpage)
 110
 111         return {
 112             '_type': 'url_transparent',
 113             'display_id': display_id,
 114             'title': video_title,
 115             'description': video_description,
 116             'upload_date': video_date,
 117             'thumbnail': video_thumbnail,
 118             'url': playerdata_url,
 119         }