X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Finfoq.py;h=71cfd12c56549d0be540c9daee6a2732959039de;hb=55801fc76e2813de9a84eaa830d70ed73cb44463;hp=f25f43664e262b25473557c5f11dae91e697e3f6;hpb=ff21a8e0ee43d4ce0b75cd938f9bdfab664dd579;p=youtube-dl diff --git a/youtube_dl/extractor/infoq.py b/youtube_dl/extractor/infoq.py index f25f43664..71cfd12c5 100644 --- a/youtube_dl/extractor/infoq.py +++ b/youtube_dl/extractor/infoq.py @@ -4,14 +4,15 @@ import base64 from .common import InfoExtractor from ..compat import ( - compat_urllib_parse, + 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': { @@ -20,7 +21,10 @@ 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): video_id = self._match_id(url) @@ -35,14 +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 = [{ @@ -52,7 +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)