1
0
Fork 0
mirror of https://gitlab.com/dstftw/youtube-dl.git synced 2020-11-16 09:42:26 +00:00
youtube-dl/youtube_dl/extractor/cinemassacre.py

119 lines
4.5 KiB
Python
Raw Normal View History

2013-10-05 07:37:11 +00:00
# encoding: utf-8
2014-02-24 13:44:29 +00:00
from __future__ import unicode_literals
2013-10-05 07:37:11 +00:00
import re
from .common import InfoExtractor
from ..utils import (
ExtractorError,
int_or_none,
2013-10-05 07:37:11 +00:00
)
2013-10-05 07:37:11 +00:00
class CinemassacreIE(InfoExtractor):
2014-03-24 12:42:20 +00:00
_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 13:44:29 +00:00
_TESTS = [
{
'url': 'http://cinemassacre.com/2012/11/10/avgn-the-movie-trailer/',
'md5': 'fde81fbafaee331785f58cd6c0d46190',
2014-02-24 13:44:29 +00:00
'info_dict': {
'id': '19911',
'ext': 'mp4',
2014-02-24 13:44:29 +00:00
'upload_date': '20121110',
'title': '“Angry Video Game Nerd: The Movie” Trailer',
'description': 'md5:fb87405fcb42a331742a0dce2708560b',
},
},
2014-02-24 13:44:29 +00:00
{
'url': 'http://cinemassacre.com/2013/10/02/the-mummys-hand-1940',
'md5': 'd72f10cd39eac4215048f62ab477a511',
2014-02-24 13:44:29 +00:00
'info_dict': {
'id': '521be8ef82b16',
'ext': 'mp4',
2014-02-24 13:44:29 +00:00
'upload_date': '20131002',
'title': 'The Mummys Hand (1940)',
},
}
]
2013-10-05 07:37:11 +00:00
def _real_extract(self, url):
2013-10-05 07:37:11 +00:00
mobj = re.match(self._VALID_URL, url)
2014-03-24 12:42:20 +00:00
display_id = mobj.group('display_id')
2013-10-05 07:37:11 +00:00
2014-03-24 12:42:20 +00:00
webpage = self._download_webpage(url, display_id)
2013-10-05 07:37:11 +00:00
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/[a-zA-Z]+\.php\?[^"]*\bid=(?P<full_video_id>(?:Cinemassacre-)?(?P<video_id>.+?)))"', webpage)
if not mobj:
2014-02-24 13:44:29 +00:00
raise ExtractorError('Can\'t extract embed url and video id')
playerdata_url = mobj.group('embed_url')
video_id = mobj.group('video_id')
full_video_id = mobj.group('full_video_id')
2014-03-24 12:42:20 +00:00
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 13:44:29 +00:00
webpage, 'description', flags=re.DOTALL, fatal=False)
video_thumbnail = self._og_search_thumbnail(webpage)
playerdata = self._download_webpage(playerdata_url, video_id, 'Downloading player webpage')
vidurl = self._search_regex(
r'\'vidurl\'\s*:\s*"([^\']+)"', playerdata, 'vidurl').replace('\\/', '/')
videolist_url = None
mobj = re.search(r"'videoserver'\s*:\s*'(?P<videoserver>[^']+)'", playerdata)
if mobj:
videoserver = mobj.group('videoserver')
mobj = re.search(r'\'vidid\'\s*:\s*"(?P<vidid>[^\']+)"', playerdata)
vidid = mobj.group('vidid') if mobj else full_video_id
videolist_url = 'http://%s/vod/smil:%s.smil/jwplayer.smil' % (videoserver, vidid)
else:
mobj = re.search(r"file\s*:\s*'(?P<smil>http.+?/jwplayer\.smil)'", playerdata)
if mobj:
videolist_url = mobj.group('smil')
if videolist_url:
videolist = self._download_xml(videolist_url, video_id, 'Downloading videolist XML')
formats = []
baseurl = vidurl[:vidurl.rfind('/')+1]
for video in videolist.findall('.//video'):
src = video.get('src')
if not src:
continue
file_ = src.partition(':')[-1]
width = int_or_none(video.get('width'))
height = int_or_none(video.get('height'))
bitrate = int_or_none(video.get('system-bitrate'))
format = {
'url': baseurl + file_,
'format_id': src.rpartition('.')[0].rpartition('_')[-1],
}
if width or height:
format.update({
'tbr': bitrate // 1000 if bitrate else None,
'width': width,
'height': height,
})
else:
format.update({
'abr': bitrate // 1000 if bitrate else None,
'vcodec': 'none',
})
formats.append(format)
self._sort_formats(formats)
else:
formats = [{
'url': vidurl,
}]
2013-10-05 07:37:11 +00:00
return {
'id': video_id,
'title': video_title,
'formats': formats,
'description': video_description,
'upload_date': video_date,
'thumbnail': video_thumbnail,
}