Merge tag 'upstream/2017.02.07'

[youtubedl] / youtube_dl / extractor / safari.py

diff --git a/youtube_dl/extractor/safari.py b/youtube_dl/extractor/safari.py
index 08ddbe3c4222879cade93fd9a5433e1d9c4e6e49..c3aec1edde5e9d02efb377fa39941ae01d2f04b4 100644 (file)
--- a/youtube_dl/extractor/safari.py
+++ b/youtube_dl/extractor/safari.py
@@ -1,4 +1,4 @@
-# encoding: utf-8
+# coding: utf-8

 from __future__ import unicode_literals
 
 import re
 from __future__ import unicode_literals
 
 import re


@@ -103,13 +103,13 @@ class SafariIE(SafariBaseIE):
 
         webpage = self._download_webpage(url, video_id)
         reference_id = self._search_regex(
 
         webpage = self._download_webpage(url, video_id)
         reference_id = self._search_regex(

-            r'data-reference-id=(["\'])(?P<id>.+?)\1',
+            r'data-reference-id=(["\'])(?P<id>(?:(?!\1).)+)\1',

             webpage, 'kaltura reference id', group='id')
         partner_id = self._search_regex(
             webpage, 'kaltura reference id', group='id')
         partner_id = self._search_regex(

-            r'data-partner-id=(["\'])(?P<id>.+?)\1',
+            r'data-partner-id=(["\'])(?P<id>(?:(?!\1).)+)\1',

             webpage, 'kaltura widget id', group='id')
         ui_id = self._search_regex(
             webpage, 'kaltura widget id', group='id')
         ui_id = self._search_regex(

-            r'data-ui-id=(["\'])(?P<id>.+?)\1',
+            r'data-ui-id=(["\'])(?P<id>(?:(?!\1).)+)\1',

             webpage, 'kaltura uiconf id', group='id')
 
         query = {
             webpage, 'kaltura uiconf id', group='id')
 
         query = {


@@ -157,7 +157,14 @@ class SafariCourseIE(SafariBaseIE):
     IE_NAME = 'safari:course'
     IE_DESC = 'safaribooksonline.com online courses'
 
     IE_NAME = 'safari:course'
     IE_DESC = 'safaribooksonline.com online courses'
 

-    _VALID_URL = r'https?://(?:www\.)?safaribooksonline\.com/(?:library/view/[^/]+|api/v1/book)/(?P<id>[^/]+)/?(?:[#?]|$)'
+    _VALID_URL = r'''(?x)
+                    https?://
+                        (?:
+                            (?:www\.)?safaribooksonline\.com/(?:library/view/[^/]+|api/v1/book)|
+                            techbus\.safaribooksonline\.com
+                        )
+                        /(?P<id>[^/]+)/?(?:[#?]|$)
+                    '''

 
     _TESTS = [{
         'url': 'https://www.safaribooksonline.com/library/view/hadoop-fundamentals-livelessons/9780133392838/',
 
     _TESTS = [{
         'url': 'https://www.safaribooksonline.com/library/view/hadoop-fundamentals-livelessons/9780133392838/',


@@ -170,6 +177,9 @@ class SafariCourseIE(SafariBaseIE):
     }, {
         'url': 'https://www.safaribooksonline.com/api/v1/book/9781449396459/?override_format=json',
         'only_matching': True,
     }, {
         'url': 'https://www.safaribooksonline.com/api/v1/book/9781449396459/?override_format=json',
         'only_matching': True,

+    }, {
+        'url': 'http://techbus.safaribooksonline.com/9780134426365',
+        'only_matching': True,

     }]
 
     def _real_extract(self, url):
     }]
 
     def _real_extract(self, url):