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         },
  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             # Youtube embedded video
  37             'url': 'http://cinemassacre.com/2006/12/07/chronologically-confused-about-bad-movie-and-video-game-sequel-titles/',
  38             'md5': 'df4cf8a1dcedaec79a73d96d83b99023',
  39             'info_dict': {
  40                 'id': 'OEVzPCY2T-g',
  41                 'ext': 'mp4',
  42                 'title': 'AVGN: Chronologically Confused about Bad Movie and Video Game Sequel Titles',
  43                 'upload_date': '20061207',
  44                 'uploader': 'Cinemassacre',
  45                 'uploader_id': 'JamesNintendoNerd',
  46                 'description': 'md5:784734696c2b8b7f4b8625cc799e07f6',
  47             }
  48         },
  49         {
  50             # Youtube embedded video
  51             'url': 'http://cinemassacre.com/2006/09/01/mckids/',
  52             'md5': '6eb30961fa795fedc750eac4881ad2e1',
  53             'info_dict': {
  54                 'id': 'FnxsNhuikpo',
  55                 'ext': 'mp4',
  56                 'upload_date': '20060901',
  57                 'uploader': 'Cinemassacre Extras',
  58                 'description': 'md5:de9b751efa9e45fbaafd9c8a1123ed53',
  59                 'uploader_id': 'Cinemassacre',
  60                 'title': 'AVGN: McKids',
  61             }
  62         },
  63         {
  64             'url': 'http://cinemassacre.com/2015/05/25/mario-kart-64-nintendo-64-james-mike-mondays/',
  65             'md5': '1376908e49572389e7b06251a53cdd08',
  66             'info_dict': {
  67                 'id': 'Cinemassacre-555779690c440',
  68                 'ext': 'mp4',
  69                 '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!',
  70                 'title': 'Mario Kart 64 (Nintendo 64) James & Mike Mondays',
  71                 'upload_date': '20150525',
  72             }
  73         }
  74     ]
  75
  76     def _real_extract(self, url):
  77         mobj = re.match(self._VALID_URL, url)
  78         display_id = mobj.group('display_id')
  79         video_date = mobj.group('date_y') + mobj.group('date_m') + mobj.group('date_d')
  80
  81         webpage = self._download_webpage(url, display_id)
  82
  83         playerdata_url = self._search_regex(
  84             [
  85                 ScreenwaveMediaIE.EMBED_PATTERN,
  86                 r'<iframe[^>]+src="(?P<url>(?:https?:)?//(?:[^.]+\.)?youtube\.com/.+?)"',
  87             ],
  88             webpage, 'player data URL', default=None, group='url')
  89         if not playerdata_url:
  90             raise ExtractorError('Unable to find player data')
  91
  92         video_title = self._html_search_regex(
  93             r'<title>(?P<title>.+?)\|', webpage, 'title')
  94         video_description = self._html_search_regex(
  95             r'<div class="entry-content">(?P<description>.+?)</div>',
  96             webpage, 'description', flags=re.DOTALL, fatal=False)
  97         video_thumbnail = self._og_search_thumbnail(webpage)
  98
  99         return {
 100             '_type': 'url_transparent',
 101             'display_id': display_id,
 102             'title': video_title,
 103             'description': video_description,
 104             'upload_date': video_date,
 105             'thumbnail': video_thumbnail,
 106             'url': playerdata_url,
 107         }