From ac2547f5ffc30a352207336194e7bbb0435d01a7 Mon Sep 17 00:00:00 2001 From: Alex Van't Hof Date: Thu, 31 Oct 2013 01:57:22 -0400 Subject: [teamcoco] Fix video url extraction for some videos Video url extraction failed for some videos, e.g. http://teamcoco.com/video/old-time-baseball The url extracted was also occasionally suboptimal quality, e.g. http://teamcoco.com/video/louis-ck-interview-george-w-bush --- youtube_dl/extractor/teamcoco.py | 20 +++++++++++++++++--- 1 file changed, 17 insertions(+), 3 deletions(-) (limited to 'youtube_dl/extractor/teamcoco.py') diff --git a/youtube_dl/extractor/teamcoco.py b/youtube_dl/extractor/teamcoco.py index c910110ca..76246c7cc 100644 --- a/youtube_dl/extractor/teamcoco.py +++ b/youtube_dl/extractor/teamcoco.py @@ -3,6 +3,7 @@ import re from .common import InfoExtractor from ..utils import ( ExtractorError, + RegexNotFoundError, ) @@ -11,7 +12,7 @@ class TeamcocoIE(InfoExtractor): _TEST = { u'url': u'http://teamcoco.com/video/louis-ck-interview-george-w-bush', u'file': u'19705.mp4', - u'md5': u'27b6f7527da5acf534b15f21b032656e', + u'md5': u'cde9ba0fa3506f5f017ce11ead928f9a', u'info_dict': { u"description": u"Louis C.K. got starstruck by George W. Bush, so what? Part one.", u"title": u"Louis C.K. Interview Pt. 1 11/3/11" @@ -33,8 +34,21 @@ class TeamcocoIE(InfoExtractor): data_url = 'http://teamcoco.com/cvp/2.0/%s.xml' % video_id data = self._download_webpage(data_url, video_id, 'Downloading data webpage') - video_url = self._html_search_regex(r']*type="high".*?>(.*?)', - data, u'video URL') + + qualities = [ '1080p', '720p', '1000k', '480p', '500k' ] + best_quality_idx = len(qualities)+1 # First regex match may not be optimal + for idx, quality in enumerate(qualities): + regex = r']*type="(?:high|standard)".*?>(.*%s.*)' % quality + try: + url = self._html_search_regex(regex, data, u'video URL') + if idx < best_quality_idx: + video_url = url + best_quality_idx = idx + except RegexNotFoundError: + # Just catch fatal exc. Don't want the fatal=False warning + continue + if not video_url: + raise RegexNotFoundError(u'Unable to extract video URL') return [{ 'id': video_id, -- cgit v1.2.3 From e7e6b54d8a02f7f76bd56e35bcd87a8723f71642 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaime=20Marqui=CC=81nez=20Ferra=CC=81ndiz?= Date: Sun, 3 Nov 2013 17:48:12 +0100 Subject: [teamcoco] Parse the xml file and extract all the formats --- youtube_dl/extractor/teamcoco.py | 47 +++++++++++++++++++++++++--------------- 1 file changed, 29 insertions(+), 18 deletions(-) (limited to 'youtube_dl/extractor/teamcoco.py') diff --git a/youtube_dl/extractor/teamcoco.py b/youtube_dl/extractor/teamcoco.py index 76246c7cc..bc48620f0 100644 --- a/youtube_dl/extractor/teamcoco.py +++ b/youtube_dl/extractor/teamcoco.py @@ -1,9 +1,9 @@ import re +import xml.etree.ElementTree from .common import InfoExtractor from ..utils import ( ExtractorError, - RegexNotFoundError, ) @@ -32,29 +32,40 @@ class TeamcocoIE(InfoExtractor): self.report_extraction(video_id) data_url = 'http://teamcoco.com/cvp/2.0/%s.xml' % video_id - data = self._download_webpage(data_url, video_id, 'Downloading data webpage') + data_xml = self._download_webpage(data_url, video_id, 'Downloading data webpage') + data = xml.etree.ElementTree.fromstring(data_xml.encode('utf-8')) - qualities = [ '1080p', '720p', '1000k', '480p', '500k' ] - best_quality_idx = len(qualities)+1 # First regex match may not be optimal - for idx, quality in enumerate(qualities): - regex = r']*type="(?:high|standard)".*?>(.*%s.*)' % quality + qualities = ['500k', '480p', '1000k', '720p', '1080p'] + formats = [] + for file in data.findall('files/file'): + if file.attrib.get('playmode') == 'all': + # it just duplicates one of the entries + break + file_url = file.text + m_format = re.search(r'(\d+(k|p))\.mp4', file_url) + if m_format is not None: + format_id = m_format.group(1) + else: + format_id = file.attrib['bitrate'] + formats.append({ + 'url': file_url, + 'ext': 'mp4', + 'format_id': format_id, + }) + def sort_key(f): try: - url = self._html_search_regex(regex, data, u'video URL') - if idx < best_quality_idx: - video_url = url - best_quality_idx = idx - except RegexNotFoundError: - # Just catch fatal exc. Don't want the fatal=False warning - continue - if not video_url: + return qualities.index(f['format_id']) + except ValueError: + return -1 + formats.sort(key=sort_key) + if not formats: raise RegexNotFoundError(u'Unable to extract video URL') - return [{ + return { 'id': video_id, - 'url': video_url, - 'ext': 'mp4', + 'formats': formats, 'title': self._og_search_title(webpage), 'thumbnail': self._og_search_thumbnail(webpage), 'description': self._og_search_description(webpage), - }] + } -- cgit v1.2.3