X-Git-Url: http://git.cielonegro.org/gitweb.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fgoshgay.py;h=18474cbb72684ac6327957fb3385fb9de3bc4480;hb=28e614de5ccfd846fe4dafe37d4834ef3488b403;hp=3f31ec896f48e5de0cee24c265e38467dca3d684;hpb=6f66eedc5dc84ac72b36722771ec625f4ccecef7;p=youtube-dl.git
diff --git a/youtube_dl/extractor/goshgay.py b/youtube_dl/extractor/goshgay.py
index 3f31ec896..18474cbb7 100644
--- a/youtube_dl/extractor/goshgay.py
+++ b/youtube_dl/extractor/goshgay.py
@@ -1,15 +1,11 @@
# -*- coding: utf-8 -*-
from __future__ import unicode_literals
-import re
-
from .common import InfoExtractor
from ..utils import (
compat_urlparse,
- str_to_int,
ExtractorError,
)
-import json
class GoshgayIE(InfoExtractor):
@@ -22,40 +18,32 @@ class GoshgayIE(InfoExtractor):
'ext': 'flv',
'title': 'md5:089833a4790b5e103285a07337f245bf',
'thumbnail': 're:http://.*\.jpg',
+ 'age_limit': 18,
}
}
def _real_extract(self, url):
- mobj = re.match(self._VALID_URL, url)
- video_id = mobj.group('id')
+ video_id = self._match_id(url)
webpage = self._download_webpage(url, video_id)
- title = self._search_regex(r'class="video-title">
(.+?)<', webpage, 'title')
+ title = self._og_search_title(webpage)
+ thumbnail = self._og_search_thumbnail(webpage)
+ family_friendly = self._html_search_meta(
+ 'isFamilyFriendly', webpage, default='false')
+ config_url = self._search_regex(
+ r"'config'\s*:\s*'([^']+)'", webpage, 'config URL')
- player_config = self._search_regex(r'jwplayer\("player"\)\.setup\(({.+?})\)', webpage, 'config settings',
- fatal=True, flags=re.S)
- player_vars = json.loads(player_config.replace("'", '"'))
- width = str_to_int(player_vars.get('width'))
- height = str_to_int(player_vars.get('height'))
- config_uri = player_vars.get('config')
+ config = self._download_xml(
+ config_url, video_id, 'Downloading player config XML')
- if config_uri is None:
- raise ExtractorError('Missing config URI')
- node = self._download_xml(config_uri, video_id, 'Downloading player config XML',
- errnote='Unable to download XML')
- if node is None:
+ if config is None:
raise ExtractorError('Missing config XML')
- if node.tag != 'config':
+ if config.tag != 'config':
raise ExtractorError('Missing config attribute')
- fns = node.findall('file')
- imgs = node.findall('image')
- if len(fns) != 1:
+ fns = config.findall('file')
+ if len(fns) < 1:
raise ExtractorError('Missing media URI')
video_url = fns[0].text
- if len(imgs) < 1:
- thumbnail = None
- else:
- thumbnail = imgs[0].text
url_comp = compat_urlparse.urlparse(url)
ref = "%s://%s%s" % (url_comp[0], url_comp[1], url_comp[2])
@@ -64,9 +52,7 @@ class GoshgayIE(InfoExtractor):
'id': video_id,
'url': video_url,
'title': title,
- 'width': width,
- 'height': height,
'thumbnail': thumbnail,
'http_referer': ref,
- 'age_limit': 18,
+ 'age_limit': 0 if family_friendly == 'true' else 18,
}