# encoding: utf-8
+from __future__ import unicode_literals
+
import re
from .common import InfoExtractor
-from ..utils import (
- ExtractorError,
-)
+from ..utils import ExtractorError
+from .bliptv import BlipTVIE
class CinemassacreIE(InfoExtractor):
- _VALID_URL = r'(?:http://)?(?:www\.)?(?P<url>cinemassacre\.com/(?P<date_Y>[0-9]{4})/(?P<date_m>[0-9]{2})/(?P<date_d>[0-9]{2})/.+?)(?:[/?].*)?'
- _TESTS = [{
- u'url': u'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/',
- u'file': u'19911.flv',
- u'info_dict': {
- u'upload_date': u'20121110',
- u'title': u'“Angry Video Game Nerd: The Movie” – Trailer',
- u'description': u'md5:fb87405fcb42a331742a0dce2708560b',
- },
- u'params': {
- # rtmp download
- u'skip_download': True,
+ _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>[^?#/]+)'
+ _TESTS = [
+ {
+ 'url': 'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/',
+ 'md5': 'fde81fbafaee331785f58cd6c0d46190',
+ 'info_dict': {
+ 'id': 'Cinemassacre-19911',
+ 'ext': 'mp4',
+ 'upload_date': '20121110',
+ 'title': '“Angry Video Game Nerd: The Movie” – Trailer',
+ 'description': 'md5:fb87405fcb42a331742a0dce2708560b',
+ },
},
- },
- {
- u'url': u'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940',
- u'file': u'521be8ef82b16.flv',
- u'info_dict': {
- u'upload_date': u'20131002',
- u'title': u'The Mummy’s Hand (1940)',
+ {
+ 'url': 'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940',
+ 'md5': 'd72f10cd39eac4215048f62ab477a511',
+ 'info_dict': {
+ 'id': 'Cinemassacre-521be8ef82b16',
+ 'ext': 'mp4',
+ 'upload_date': '20131002',
+ 'title': 'The Mummy’s Hand (1940)',
+ },
},
- u'params': {
- # rtmp download
- u'skip_download': True,
+ {
+ # blip.tv embedded video
+ 'url': 'http://cinemassacre.com/2006/12/07/chronologically-confused-about-bad-movie-and-video-game-sequel-titles/',
+ 'md5': 'ca9b3c8dd5a66f9375daeb5135f5a3de',
+ 'info_dict': {
+ 'id': '4065369',
+ 'ext': 'flv',
+ 'title': 'AVGN: Chronologically Confused about Bad Movie and Video Game Sequel Titles',
+ 'upload_date': '20061207',
+ 'uploader': 'cinemassacre',
+ 'uploader_id': '250778',
+ 'timestamp': 1283233867,
+ 'description': 'md5:0a108c78d130676b207d0f6d029ecffd',
+ }
},
- }]
+ {
+ # Youtube embedded video
+ 'url': 'http://cinemassacre.com/2006/09/01/mckids/',
+ 'md5': '6eb30961fa795fedc750eac4881ad2e1',
+ 'info_dict': {
+ 'id': 'FnxsNhuikpo',
+ 'ext': 'mp4',
+ 'upload_date': '20060901',
+ 'uploader': 'Cinemassacre Extras',
+ 'description': 'md5:de9b751efa9e45fbaafd9c8a1123ed53',
+ 'uploader_id': 'Cinemassacre',
+ 'title': 'AVGN: McKids',
+ }
+ }
+ ]
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
+ display_id = mobj.group('display_id')
+ video_date = mobj.group('date_y') + mobj.group('date_m') + mobj.group('date_d')
- webpage_url = u'http://' + mobj.group('url')
- webpage = self._download_webpage(webpage_url, None) # Don't know video id yet
- video_date = mobj.group('date_Y') + mobj.group('date_m') + mobj.group('date_d')
- mobj = re.search(r'src="(?P<embed_url>http://player\.screenwavemedia\.com/play/(?:embed|player)\.php\?id=(?:Cinemassacre-)?(?P<video_id>.+?))"', webpage)
- if not mobj:
- raise ExtractorError(u'Can\'t extract embed url and video id')
- playerdata_url = mobj.group(u'embed_url')
- video_id = mobj.group(u'video_id')
+ webpage = self._download_webpage(url, display_id)
- video_title = self._html_search_regex(r'<title>(?P<title>.+?)\|',
- webpage, u'title')
- video_description = self._html_search_regex(r'<div class="entry-content">(?P<description>.+?)</div>',
- webpage, u'description', flags=re.DOTALL, fatal=False)
- if len(video_description) == 0:
- video_description = None
+ playerdata_url = self._search_regex(
+ [
+ r'src="(http://player\.screenwavemedia\.com/play/[a-zA-Z]+\.php\?[^"]*\bid=.+?)"',
+ r'<iframe[^>]+src="((?:https?:)?//(?:[^.]+\.)?youtube\.com/.+?)"',
+ ],
+ webpage, 'player data URL', default=None)
+ if not playerdata_url:
+ playerdata_url = BlipTVIE._extract_url(webpage)
+ if not playerdata_url:
+ raise ExtractorError('Unable to find player data')
- playerdata = self._download_webpage(playerdata_url, video_id)
- base_url = self._html_search_regex(r'\'streamer\': \'(?P<base_url>rtmp://.*?)/(?:vod|Cinemassacre)\'',
- playerdata, u'base_url')
- base_url += '/Cinemassacre/'
- # Important: The file names in playerdata are not used by the player and even wrong for some videos
- sd_file = 'Cinemassacre-%s_high.mp4' % video_id
- hd_file = 'Cinemassacre-%s.mp4' % video_id
- video_thumbnail = 'http://image.screenwavemedia.com/Cinemassacre/Cinemassacre-%s_thumb_640x360.jpg' % video_id
-
- formats = [
- {
- 'url': base_url + sd_file,
- 'ext': 'flv',
- 'format': 'sd',
- 'format_id': 'sd',
- },
- {
- 'url': base_url + hd_file,
- 'ext': 'flv',
- 'format': 'hd',
- 'format_id': 'hd',
- },
- ]
+ video_title = self._html_search_regex(
+ r'<title>(?P<title>.+?)\|', webpage, 'title')
+ video_description = self._html_search_regex(
+ r'<div class="entry-content">(?P<description>.+?)</div>',
+ webpage, 'description', flags=re.DOTALL, fatal=False)
+ video_thumbnail = self._og_search_thumbnail(webpage)
- info = {
- 'id': video_id,
+ return {
+ '_type': 'url_transparent',
+ 'display_id': display_id,
'title': video_title,
- 'formats': formats,
'description': video_description,
'upload_date': video_date,
'thumbnail': video_thumbnail,
+ 'url': playerdata_url,
}
- # TODO: Remove when #980 has been merged
- info.update(formats[-1])
- return info
Rapsys Git / youtubedl / blobdiff