Merge pull request #4076 from ghedo/direct_type

[youtube-dl] / youtube_dl / extractor / generic.py

diff --git a/youtube_dl/extractor/generic.py b/youtube_dl/extractor/generic.py
index 47580328abc53f1d16f3d45adcdbb16f4f69d8ea..01d6a57f82aedbf8e350ce753872676b559d24ba 100644 (file)
--- a/youtube_dl/extractor/generic.py
+++ b/youtube_dl/extractor/generic.py
@@ -7,11 +7,12 @@ import re
 
 from .common import InfoExtractor
 from .youtube import YoutubeIE
-from ..utils import (
+from ..compat import (
     compat_urllib_parse,
     compat_urlparse,
     compat_xml_parse_error,
-
+)
+from ..utils import (
     determine_ext,
     ExtractorError,
     float_or_none,
@@ -99,6 +100,22 @@ class GenericIE(InfoExtractor):
                 'uploader': 'Championat',
             },
         },
+        {
+            # https://github.com/rg3/youtube-dl/issues/3541
+            'add_ie': ['Brightcove'],
+            'url': 'http://www.kijk.nl/sbs6/leermijvrouwenkennen/videos/jqMiXKAYan2S/aflevering-1',
+            'info_dict': {
+                'id': '3866516442001',
+                'ext': 'mp4',
+                'title': 'Leer mij vrouwen kennen: Aflevering 1',
+                'description': 'Leer mij vrouwen kennen: Aflevering 1',
+                'uploader': 'SBS Broadcasting',
+            },
+            'skip': 'Restricted to Netherlands',
+            'params': {
+                'skip_download': True,  # m3u8 download
+            },
+        },
         # Direct link to a video
         {
             'url': 'http://media.w3.org/2010/05/sintel/trailer.mp4',
@@ -405,6 +422,18 @@ class GenericIE(InfoExtractor):
             'expected_warnings': [
                 r'501.*Not Implemented'
             ],
+        },
+        # Soundcloud embed
+        {
+            'url': 'http://nakedsecurity.sophos.com/2014/10/29/sscc-171-are-you-sure-that-1234-is-a-bad-password-podcast/',
+            'info_dict': {
+                'id': '174391317',
+                'ext': 'mp3',
+                'description': 'md5:ff867d6b555488ad3c52572bb33d432c',
+                'uploader': 'Sophos Security',
+                'title': 'Chet Chat 171 - Oct 29, 2014',
+                'upload_date': '20141029',
+            }
         }
     ]
 
@@ -547,6 +576,7 @@ class GenericIE(InfoExtractor):
             return {
                 'id': video_id,
                 'title': os.path.splitext(url_basename(url))[0],
+                'direct': True,
                 'formats': [{
                     'format_id': m.group('format_id'),
                     'url': url,
@@ -838,7 +868,7 @@ class GenericIE(InfoExtractor):
 
         # Look for embeded soundcloud player
         mobj = re.search(
-            r'<iframe src="(?P<url>https?://(?:w\.)?soundcloud\.com/player[^"]+)"',
+            r'<iframe\s+(?:[a-zA-Z0-9_-]+="[^"]+"\s+)*src="(?P<url>https?://(?:w\.)?soundcloud\.com/player[^"]+)"',
             webpage)
         if mobj is not None:
             url = unescapeHTML(mobj.group('url'))
@@ -875,7 +905,7 @@ class GenericIE(InfoExtractor):
             return self.url_result(mobj.group('url'), 'SBS')
 
         mobj = re.search(
-            r'<iframe[^>]+?src=(["\'])(?P<url>https?://m\.mlb\.com/shared/video/embed/embed\.html\?.+?)\1',
+            r'<iframe[^>]+?src=(["\'])(?P<url>https?://m(?:lb)?\.mlb\.com/shared/video/embed/embed\.html\?.+?)\1',
             webpage)
         if mobj is not None:
             return self.url_result(mobj.group('url'), 'MLB')