Fix "invalid escape sequences" error on Python 3.6

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

diff --git a/youtube_dl/extractor/tudou.py b/youtube_dl/extractor/tudou.py
index 9892e8a62bf9731b8d3e5768333f71baaecd6288..2aae55e7e8f8742b471e4f8ffe94ab2ae79bae25 100644 (file)
--- a/youtube_dl/extractor/tudou.py
+++ b/youtube_dl/extractor/tudou.py
@@ -5,6 +5,7 @@ from __future__ import unicode_literals
 from .common import InfoExtractor
 from ..compat import compat_str
 from ..utils import (
+    ExtractorError,
     int_or_none,
     InAdvancePagedList,
     float_or_none,
@@ -22,7 +23,7 @@ class TudouIE(InfoExtractor):
             'id': '159448201',
             'ext': 'f4v',
             'title': '卡马乔国足开大脚长传冲吊集锦',
-            'thumbnail': 're:^https?://.*\.jpg$',
+            'thumbnail': r're:^https?://.*\.jpg$',
             'timestamp': 1372113489000,
             'description': '卡马乔卡家军，开大脚先进战术不完全集锦！',
             'duration': 289.04,
@@ -35,7 +36,7 @@ class TudouIE(InfoExtractor):
             'id': '117049447',
             'ext': 'f4v',
             'title': 'La Sylphide-Bolshoi-Ekaterina Krysanova & Vyacheslav Lopatin 2012',
-            'thumbnail': 're:^https?://.*\.jpg$',
+            'thumbnail': r're:^https?://.*\.jpg$',
             'timestamp': 1349207518000,
             'description': 'md5:294612423894260f2dcd5c6c04fe248b',
             'duration': 5478.33,
@@ -46,11 +47,27 @@ class TudouIE(InfoExtractor):
 
     _PLAYER_URL = 'http://js.tudouui.com/bin/lingtong/PortalPlayer_177.swf'
 
+    # Translated from tudou/tools/TVCHelper.as in PortalPlayer_193.swf
+    # 0001, 0002 and 4001 are not included as they indicate temporary issues
+    TVC_ERRORS = {
+        '0003': 'The video is deleted or does not exist',
+        '1001': 'This video is unavailable due to licensing issues',
+        '1002': 'This video is unavailable as it\'s under review',
+        '1003': 'This video is unavailable as it\'s under review',
+        '3001': 'Password required',
+        '5001': 'This video is available in Mainland China only due to licensing issues',
+        '7001': 'This video is unavailable',
+        '8001': 'This video is unavailable due to licensing issues',
+    }
+
     def _url_for_id(self, video_id, quality=None):
         info_url = 'http://v2.tudou.com/f?id=' + compat_str(video_id)
         if quality:
             info_url += '&hd' + quality
         xml_data = self._download_xml(info_url, video_id, 'Opening the info XML page')
+        error = xml_data.attrib.get('error')
+        if error is not None:
+            raise ExtractorError('Tudou said: %s' % error, expected=True)
         final_url = xml_data.text
         return final_url
 
@@ -63,6 +80,15 @@ class TudouIE(InfoExtractor):
         if youku_vcode:
             return self.url_result('youku:' + youku_vcode, ie='Youku')
 
+        if not item_data.get('itemSegs'):
+            tvc_code = item_data.get('tvcCode')
+            if tvc_code:
+                err_msg = self.TVC_ERRORS.get(tvc_code)
+                if err_msg:
+                    raise ExtractorError('Tudou said: %s' % err_msg, expected=True)
+                raise ExtractorError('Unexpected error %s returned from Tudou' % tvc_code)
+            raise ExtractorError('Unxpected error returned from Tudou')
+
         title = unescapeHTML(item_data['kw'])
         description = item_data.get('desc')
         thumbnail_url = item_data.get('pic')