Add bug number that we are closing.

[youtubedl] / youtube_dl / extractor / youku.py

diff --git a/youtube_dl/extractor/youku.py b/youtube_dl/extractor/youku.py
index eb98298019c04334276688a7d9c6a5db8bd90664..00fa2ccb5469329685a00ada37194219c31cb79f 100644 (file)
--- a/youtube_dl/extractor/youku.py
+++ b/youtube_dl/extractor/youku.py
@@ -13,7 +13,7 @@ from ..utils import (
 
 
 class YoukuIE(InfoExtractor):
 
 
 class YoukuIE(InfoExtractor):

-    _VALID_URL =  r'(?:http://)?v\.youku\.com/v_show/id_(?P<ID>[A-Za-z0-9]+)\.html'
+    _VALID_URL =  r'(?:http://)?(v|player)\.youku\.com/(v_show/id_|player\.php/sid/)(?P<ID>[A-Za-z0-9]+)(\.html|/v.swf)'

     _TEST =   {
         u"url": u"http://v.youku.com/v_show/id_XNDgyMDQ2NTQw.html",
         u"file": u"XNDgyMDQ2NTQw_part00.flv",
     _TEST =   {
         u"url": u"http://v.youku.com/v_show/id_XNDgyMDQ2NTQw.html",
         u"file": u"XNDgyMDQ2NTQw_part00.flv",


@@ -66,6 +66,12 @@ class YoukuIE(InfoExtractor):
         self.report_extraction(video_id)
         try:
             config = json.loads(jsondata)
         self.report_extraction(video_id)
         try:
             config = json.loads(jsondata)

+            error_code = config['data'][0].get('error_code')
+            if error_code:
+                # -8 means blocked outside China.
+                error = config['data'][0].get('error')  # Chinese and English, separated by newline.
+                raise ExtractorError(error or u'Server reported error %i' % error_code,
+                    expected=True)

 
             video_title =  config['data'][0]['title']
             seed = config['data'][0]['seed']
 
             video_title =  config['data'][0]['title']
             seed = config['data'][0]['seed']


@@ -89,6 +95,7 @@ class YoukuIE(InfoExtractor):
 
             fileid = config['data'][0]['streamfileids'][format]
             keys = [s['k'] for s in config['data'][0]['segs'][format]]
 
             fileid = config['data'][0]['streamfileids'][format]
             keys = [s['k'] for s in config['data'][0]['segs'][format]]

+            # segs is usually a dictionary, but an empty *list* if an error occured.

         except (UnicodeDecodeError, ValueError, KeyError):
             raise ExtractorError(u'Unable to extract info section')
 
         except (UnicodeDecodeError, ValueError, KeyError):
             raise ExtractorError(u'Unable to extract info section')