Merge branch 'qqmusic-format-fix' of https://github.com/ping/youtube-dl into ping...

[youtube-dl.git] / youtube_dl / extractor / infoq.py

diff --git a/youtube_dl/extractor/infoq.py b/youtube_dl/extractor/infoq.py
index e76dd222d1ee81dc0e0b2d5b1b3c28ef22e1bd83..91a1b3ccb70d41027d7ecfa41f1b606ead8f056a 100644 (file)
--- 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 (
+from ..compat import (
     compat_urllib_parse,
+    compat_urlparse,
 )
 
 
 class InfoQIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?infoq\.com/[^/]+/(?P<id>[^/]+)$'
+    _VALID_URL = r'https?://(?:www\.)?infoq\.com/(?:[^/]+/)+(?P<id>[^/]+)'
 
-    _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'<title>(.*?)</title>', webpage, 'title')
@@ -45,7 +46,7 @@ class InfoQIE(InfoExtractor):
         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)