1
0
mirror of https://codeberg.org/polarisfm/youtube-dl synced 2024-12-23 06:57:54 +01:00

[rai] Improve extraction

This commit is contained in:
Sergey M․ 2015-04-10 22:41:26 +06:00
parent 8749477ed0
commit d7c78decb0

View File

@ -82,9 +82,9 @@ class RaiIE(InfoExtractor):
iframe_path = self._search_regex( iframe_path = self._search_regex(
r'<iframe[^>]+src="/?(dl/[^"]+\?iframe\b[^"]*)"', r'<iframe[^>]+src="/?(dl/[^"]+\?iframe\b[^"]*)"',
webpage, 'iframe') webpage, 'iframe')
iframe_page = self._download_webpage( webpage = self._download_webpage(
'%s/%s' % (host, iframe_path), video_id) '%s/%s' % (host, iframe_path), video_id)
relinker_url = self._extract_relinker_url(iframe_page) relinker_url = self._extract_relinker_url(webpage)
relinker = self._download_json( relinker = self._download_json(
'%s&output=47' % relinker_url, video_id) '%s&output=47' % relinker_url, video_id)
@ -112,15 +112,15 @@ class RaiIE(InfoExtractor):
uploader = media.get('author') uploader = media.get('author')
upload_date = unified_strdate(media.get('date')) upload_date = unified_strdate(media.get('date'))
else: else:
title = self._search_regex( title = (self._search_regex(
r'var\s+videoTitolo\s*=\s*"([^"]+)";', r'var\s+videoTitolo\s*=\s*"(.+?)";',
webpage, 'title', default=None) or self._og_search_title(webpage) webpage, 'title', default=None) or self._og_search_title(webpage)).replace('\\"', '"')
description = self._og_search_description(webpage) description = self._og_search_description(webpage)
thumbnail = self._og_search_thumbnail(webpage) thumbnail = self._og_search_thumbnail(webpage)
duration = None duration = None
uploader = self._html_search_meta('Editore', webpage, 'uploader') uploader = self._html_search_meta('Editore', webpage, 'uploader')
upload_date = unified_strdate(self._html_search_meta( upload_date = unified_strdate(self._html_search_meta(
'item-date', webpage, 'upload date')) 'item-date', webpage, 'upload date', default=None))
subtitles = self.extract_subtitles(video_id, webpage) subtitles = self.extract_subtitles(video_id, webpage)