X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/415fdb62500dca2e22067a05008dfbf87c75b662..14111bdbc4a22585fc51fd4efa3572ea003b5321:/youtube_dl/extractor/infoq.py diff --git a/youtube_dl/extractor/infoq.py b/youtube_dl/extractor/infoq.py index e76dd22..71cfd12 100644 --- a/youtube_dl/extractor/infoq.py +++ b/youtube_dl/extractor/infoq.py @@ -1,18 +1,18 @@ from __future__ import unicode_literals import base64 -import re from .common import InfoExtractor -from ..utils import ( - compat_urllib_parse, +from ..compat import ( + compat_urllib_parse_unquote, + compat_urlparse, ) class InfoQIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?infoq\.com/[^/]+/(?P[^/]+)$' + _VALID_URL = r'https?://(?:www\.)?infoq\.com/(?:[^/]+/)+(?P[^/]+)' - _TEST = { + _TESTS = [{ 'url': 'http://www.infoq.com/presentations/A-Few-of-My-Favorite-Python-Things', 'md5': 'b5ca0e0a8c1fed93b0e65e48e462f9a2', 'info_dict': { @@ -21,12 +21,13 @@ class InfoQIE(InfoExtractor): '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', }, - } + }, { + 'url': 'http://www.infoq.com/fr/presentations/changez-avis-sur-javascript', + 'only_matching': True, + }] def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - video_id = mobj.group('id') - + video_id = self._match_id(url) webpage = self._download_webpage(url, video_id) video_title = self._html_search_regex(r'(.*?)', webpage, 'title') @@ -38,14 +39,14 @@ class InfoQIE(InfoExtractor): # Extract video URL encoded_id = self._search_regex( r"jsclassref\s*=\s*'([^']*)'", webpage, 'encoded id') - real_id = compat_urllib_parse.unquote(base64.b64decode(encoded_id.encode('ascii')).decode('utf-8')) + 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('.') http_base = self._search_regex( - r'EXPRESSINSTALL_SWF\s*=\s*"(https?://[^/"]+/)', webpage, + r'EXPRESSINSTALL_SWF\s*=\s*[^"]*"((?:https?:)?//[^/"]+/)', webpage, 'HTTP base URL') formats = [{ @@ -55,7 +56,7 @@ class InfoQIE(InfoExtractor): 'play_path': playpath, }, { 'format_id': 'http', - 'url': http_base + real_id, + 'url': compat_urlparse.urljoin(url, http_base) + real_id, }] self._sort_formats(formats)