[beeg] PEP8 and additional tests

pull/3655/head
Philipp Hagemeister 2014-09-01 00:00:01 +02:00
parent 2d7af09487
commit d169e36f5c
1 changed files with 10 additions and 5 deletions

View File

@ -15,6 +15,8 @@ class BeegIE(InfoExtractor):
'ext': 'mp4',
'title': 'Sultry Striptease',
'description': 'md5:6db3c6177972822aaba18652ff59c773',
'categories': list, # NSFW
'thumbnail': 're:https?://.*\.jpg$',
}
}
@ -24,15 +26,18 @@ class BeegIE(InfoExtractor):
webpage = self._download_webpage(url, video_id)
video_url = self._html_search_regex(r"'480p'\s*:\s*'([^']+)'", webpage, 'video URL')
video_url = self._html_search_regex(
r"'480p'\s*:\s*'([^']+)'", webpage, 'video URL')
title = self._html_search_regex(r'<title>([^<]+)\s*-\s*beeg\.?</title>', webpage, 'title')
title = self._html_search_regex(
r'<title>([^<]+)\s*-\s*beeg\.?</title>', webpage, 'title')
description = self._html_search_regex(
r'<meta name="description" content="([^"]*)"', webpage, 'description', fatal=False)
r'<meta name="description" content="([^"]*)"',
webpage, 'description', fatal=False)
thumbnail = self._html_search_regex(
r'\'previewer.url\'\s*:\s*"([^"]*)"', webpage, 'thumbnail', fatal=False)
r'\'previewer.url\'\s*:\s*"([^"]*)"',
webpage, 'thumbnail', fatal=False)
categories_str = self._html_search_regex(
r'<meta name="keywords" content="([^"]+)"', webpage, 'categories', fatal=False)