From 9d0b581feaa3adc72ed3a53b7e8c86fadfddb6e1 Mon Sep 17 00:00:00 2001 From: =?utf8?q?Sergey=20M=E2=80=A4?= Date: Tue, 30 Jun 2015 20:52:26 +0600 Subject: [PATCH] [youtube] Prefer meta for upload date and modernize --- youtube_dl/extractor/youtube.py | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-) diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index ed382e10b..6769a009d 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -999,15 +999,16 @@ class YoutubeIE(YoutubeBaseInfoExtractor): video_thumbnail = compat_urllib_parse.unquote_plus(video_info['thumbnail_url'][0]) # upload date - upload_date = None - mobj = re.search(r'(?s)id="eow-date.*?>(.*?)', video_webpage) - if mobj is None: - mobj = re.search( - r'id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live|Started) on (.*?)', - video_webpage) - if mobj is not None: - upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split()) - upload_date = unified_strdate(upload_date) + upload_date = self._html_search_meta( + 'datePublished', video_webpage, 'upload date', default=None) + if not upload_date: + upload_date = self._search_regex( + [r'(?s)id="eow-date.*?>(.*?)', + r'id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live|Started) on (.+?)'], + video_webpage, 'upload date', default=None) + if upload_date: + upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split()) + upload_date = unified_strdate(upload_date) m_cat_container = self._search_regex( r'(?s)]*>\s*Category\s*\s*]*>(.*?)', -- 2.22.2