From 282cb9c7ba918904bfa6f58a5a1bbf44938c76ba Mon Sep 17 00:00:00 2001 From: Kai Weber Date: Sun, 20 Apr 2014 01:01:37 +0200 Subject: [PATCH 1/2] [infoq] Fix extractor --- youtube_dl/extractor/infoq.py | 47 +++++++++++++++++------------------ 1 file changed, 23 insertions(+), 24 deletions(-) diff --git a/youtube_dl/extractor/infoq.py b/youtube_dl/extractor/infoq.py index ed32373a1..55f7608b5 100644 --- a/youtube_dl/extractor/infoq.py +++ b/youtube_dl/extractor/infoq.py @@ -11,16 +11,16 @@ class InfoQIE(InfoExtractor): _VALID_URL = r'https?://(?:www\.)?infoq\.com/[^/]+/(?P[^/]+)$' + _TEST = { - "name": "InfoQ", - "url": "http://www.infoq.com/presentations/A-Few-of-My-Favorite-Python-Things", - "file": "12-jan-pythonthings.mp4", - "info_dict": { - "description": "Mike Pirnat presents some tips and tricks, standard libraries and third party packages that make programming in Python a richer experience.", - "title": "A Few of My Favorite [Python] Things", - }, - "params": { - "skip_download": True, + u'name': u'InfoQ', + u'url': u'http://www.infoq.com/presentations/A-Few-of-My-Favorite-Python-Things', + u'md5': u'fcaa3d995e04080dcb9465d86b5eef62', + u'info_dict': { + u'id': u'12-jan-pythonthings', + u'ext': u'mp4', + u'description': u'Mike Pirnat presents some tips and tricks, standard libraries and third party packages that make programming in Python a richer experience.', + u'title': u'A Few of My Favorite [Python] Things', }, } @@ -30,26 +30,25 @@ def _real_extract(self, url): webpage = self._download_webpage(url, video_id) - # Extract video URL - encoded_id = self._search_regex(r"jsclassref ?= ?'([^']*)'", webpage, 'encoded id') - real_id = compat_urllib_parse.unquote(base64.b64decode(encoded_id.encode('ascii')).decode('utf-8')) - video_url = 'rtmpe://video.infoq.com/cfx/st/' + real_id + self.report_extraction(video_id) - # Extract title - video_title = self._search_regex(r'contentTitle = "(.*?)";', - webpage, 'title') + video_title = self._html_search_regex(r'(.*?)', webpage, 'title') + video_description = self._html_search_meta('description', webpage, 'description') - # Extract description - video_description = self._html_search_regex(r'', - webpage, 'description', fatal=False) + video_url = 'rtmpe://video.infoq.com/cfx/st/' + base64playpath = self._search_regex(r"jsclassref = '([^']*)'", webpage, 'jsclassref') + playpath = 'mp4:' + base64.b64decode(base64playpath).decode('utf-8') - video_filename = video_url.split('/')[-1] + video_filename = playpath.split('/')[-1] video_id, extension = video_filename.split('.') - return { + return [{ 'id': video_id, - 'url': video_url, 'title': video_title, - 'ext': extension, # Extension is always(?) mp4, but seems to be flv 'description': video_description, - } + 'formats': [{ + 'url': video_url, + 'ext': extension, + 'play_path': playpath, + }], + }] From 7560096db5ce0002b3cf3f3bcbbb53ef05e6fb13 Mon Sep 17 00:00:00 2001 From: Kai Weber Date: Sun, 20 Apr 2014 01:10:30 +0200 Subject: [PATCH 2/2] [infoq] Simplify playpath calculation --- youtube_dl/extractor/infoq.py | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/youtube_dl/extractor/infoq.py b/youtube_dl/extractor/infoq.py index 55f7608b5..65f0b76cb 100644 --- a/youtube_dl/extractor/infoq.py +++ b/youtube_dl/extractor/infoq.py @@ -35,9 +35,13 @@ def _real_extract(self, url): video_title = self._html_search_regex(r'(.*?)', webpage, 'title') video_description = self._html_search_meta('description', webpage, 'description') + # The server URL is hardcoded video_url = 'rtmpe://video.infoq.com/cfx/st/' - base64playpath = self._search_regex(r"jsclassref = '([^']*)'", webpage, 'jsclassref') - playpath = 'mp4:' + base64.b64decode(base64playpath).decode('utf-8') + + # Extract video URL + encoded_id = self._search_regex(r"jsclassref ?= ?'([^']*)'", webpage, 'encoded id') + real_id = compat_urllib_parse.unquote(base64.b64decode(encoded_id.encode('ascii')).decode('utf-8')) + playpath = 'mp4:' + real_id video_filename = playpath.split('/')[-1] video_id, extension = video_filename.split('.')