2013-10-05 00:37:11 -07:00
|
|
|
|
# encoding: utf-8
|
2014-02-24 05:44:29 -08:00
|
|
|
|
from __future__ import unicode_literals
|
2013-10-05 00:37:11 -07:00
|
|
|
|
import re
|
|
|
|
|
|
|
|
|
|
from .common import InfoExtractor
|
|
|
|
|
from ..utils import (
|
|
|
|
|
ExtractorError,
|
|
|
|
|
)
|
|
|
|
|
|
2013-10-11 11:36:50 -07:00
|
|
|
|
|
2013-10-05 00:37:11 -07:00
|
|
|
|
class CinemassacreIE(InfoExtractor):
|
2014-03-24 05:42:20 -07:00
|
|
|
|
_VALID_URL = r'http://(?:www\.)?cinemassacre\.com/(?P<date_Y>[0-9]{4})/(?P<date_m>[0-9]{2})/(?P<date_d>[0-9]{2})/(?P<display_id>[^?#/]+)'
|
2014-02-24 05:44:29 -08:00
|
|
|
|
_TESTS = [
|
|
|
|
|
{
|
|
|
|
|
'url': 'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/',
|
|
|
|
|
'file': '19911.mp4',
|
2014-05-22 01:33:30 -07:00
|
|
|
|
'md5': 'fde81fbafaee331785f58cd6c0d46190',
|
2014-02-24 05:44:29 -08:00
|
|
|
|
'info_dict': {
|
|
|
|
|
'upload_date': '20121110',
|
|
|
|
|
'title': '“Angry Video Game Nerd: The Movie” – Trailer',
|
|
|
|
|
'description': 'md5:fb87405fcb42a331742a0dce2708560b',
|
|
|
|
|
},
|
2013-10-05 04:36:13 -07:00
|
|
|
|
},
|
2014-02-24 05:44:29 -08:00
|
|
|
|
{
|
|
|
|
|
'url': 'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940',
|
|
|
|
|
'file': '521be8ef82b16.mp4',
|
2014-05-22 01:33:30 -07:00
|
|
|
|
'md5': 'd72f10cd39eac4215048f62ab477a511',
|
2014-02-24 05:44:29 -08:00
|
|
|
|
'info_dict': {
|
|
|
|
|
'upload_date': '20131002',
|
|
|
|
|
'title': 'The Mummy’s Hand (1940)',
|
|
|
|
|
},
|
|
|
|
|
}
|
|
|
|
|
]
|
2013-10-05 00:37:11 -07:00
|
|
|
|
|
2013-10-11 11:36:50 -07:00
|
|
|
|
def _real_extract(self, url):
|
2013-10-05 00:37:11 -07:00
|
|
|
|
mobj = re.match(self._VALID_URL, url)
|
2014-03-24 05:42:20 -07:00
|
|
|
|
display_id = mobj.group('display_id')
|
2013-10-05 00:37:11 -07:00
|
|
|
|
|
2014-03-24 05:42:20 -07:00
|
|
|
|
webpage = self._download_webpage(url, display_id)
|
2013-10-05 00:37:11 -07:00
|
|
|
|
video_date = mobj.group('date_Y') + mobj.group('date_m') + mobj.group('date_d')
|
2013-10-31 17:24:23 -07:00
|
|
|
|
mobj = re.search(r'src="(?P<embed_url>http://player\.screenwavemedia\.com/play/[a-zA-Z]+\.php\?id=(?:Cinemassacre-)?(?P<video_id>.+?))"', webpage)
|
2013-10-05 04:36:13 -07:00
|
|
|
|
if not mobj:
|
2014-02-24 05:44:29 -08:00
|
|
|
|
raise ExtractorError('Can\'t extract embed url and video id')
|
|
|
|
|
playerdata_url = mobj.group('embed_url')
|
|
|
|
|
video_id = mobj.group('video_id')
|
2013-10-05 04:36:13 -07:00
|
|
|
|
|
2014-03-24 05:42:20 -07:00
|
|
|
|
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>',
|
2014-02-24 05:44:29 -08:00
|
|
|
|
webpage, 'description', flags=re.DOTALL, fatal=False)
|
2013-10-11 11:36:50 -07:00
|
|
|
|
|
2013-10-05 00:37:11 -07:00
|
|
|
|
playerdata = self._download_webpage(playerdata_url, video_id)
|
2014-05-22 01:33:30 -07:00
|
|
|
|
video_thumbnail = self._search_regex(r'image: \'(?P<thumbnail>[^\']+)\'', playerdata, 'thumbnail', fatal=False)
|
|
|
|
|
sd_url = self._search_regex(r'file: \'([^\']+)\', label: \'SD\'', playerdata, 'sd_file')
|
|
|
|
|
videolist_url = self._search_regex(r'file: \'([^\']+\.smil)\'}', playerdata, 'videolist_url')
|
|
|
|
|
|
|
|
|
|
videolist = self._download_webpage(videolist_url, video_id)
|
|
|
|
|
formats = []
|
|
|
|
|
baseurl = sd_url[:sd_url.rfind('/')+1]
|
|
|
|
|
for match in re.finditer('<video src="mp4:(?P<file>[^"]+_(?P<format_id>[^"]+)\.[^"]+)" system-bitrate="(?P<br>\d+)"(?: width="(?P<width>\d+)" height="(?P<height>\d+)")?/>', videolist):
|
|
|
|
|
format = {
|
|
|
|
|
'url': baseurl + match.group('file'),
|
|
|
|
|
'format_id': match.group('format_id')
|
|
|
|
|
}
|
|
|
|
|
if match.group('width'):
|
|
|
|
|
format.update({
|
|
|
|
|
'tbr': int(match.group('br')) // 1000,
|
|
|
|
|
'width': int(match.group('width')),
|
|
|
|
|
'height': int(match.group('height'))
|
|
|
|
|
})
|
|
|
|
|
else:
|
|
|
|
|
format.update({
|
|
|
|
|
'abr': int(match.group('br')) // 1000,
|
|
|
|
|
'vcodec': 'none'
|
|
|
|
|
})
|
|
|
|
|
formats.append(format)
|
2014-03-24 05:42:20 -07:00
|
|
|
|
self._sort_formats(formats)
|
2013-10-05 00:37:11 -07:00
|
|
|
|
|
2013-10-23 11:21:25 -07:00
|
|
|
|
return {
|
2013-10-11 11:36:50 -07:00
|
|
|
|
'id': video_id,
|
|
|
|
|
'title': video_title,
|
|
|
|
|
'formats': formats,
|
|
|
|
|
'description': video_description,
|
|
|
|
|
'upload_date': video_date,
|
|
|
|
|
'thumbnail': video_thumbnail,
|
|
|
|
|
}
|