]> Raphaƫl G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/screencast.py
debian/copyright: use spaces rather than tabs to start continuation lines.
[youtubedl] / youtube_dl / extractor / screencast.py
index c145f6fc72f1b9eed8a5089dce48dfdd5f106a79..69a0d01f39c24f9fdcef0d31dd4d9d7a2f082c85 100644 (file)
@@ -1,18 +1,18 @@
-# -*- coding: utf-8 -*-
+# coding: utf-8
 from __future__ import unicode_literals
 
 from __future__ import unicode_literals
 
-import re
-
 from .common import InfoExtractor
 from .common import InfoExtractor
-from ..utils import (
-    ExtractorError,
+from ..compat import (
     compat_parse_qs,
     compat_urllib_request,
 )
     compat_parse_qs,
     compat_urllib_request,
 )
+from ..utils import (
+    ExtractorError,
+)
 
 
 class ScreencastIE(InfoExtractor):
 
 
 class ScreencastIE(InfoExtractor):
-    _VALID_URL = r'https?://www\.screencast\.com/t/(?P<id>[a-zA-Z0-9]+)'
+    _VALID_URL = r'https?://(?:www\.)?screencast\.com/t/(?P<id>[a-zA-Z0-9]+)'
     _TESTS = [{
         'url': 'http://www.screencast.com/t/3ZEjQXlT',
         'md5': '917df1c13798a3e96211dd1561fded83',
     _TESTS = [{
         'url': 'http://www.screencast.com/t/3ZEjQXlT',
         'md5': '917df1c13798a3e96211dd1561fded83',
@@ -21,7 +21,7 @@ class ScreencastIE(InfoExtractor):
             'ext': 'm4v',
             'title': 'Color Measurement with Ocean Optics Spectrometers',
             'description': 'md5:240369cde69d8bed61349a199c5fb153',
             'ext': 'm4v',
             'title': 'Color Measurement with Ocean Optics Spectrometers',
             'description': 'md5:240369cde69d8bed61349a199c5fb153',
-            'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
+            'thumbnail': r're:^https?://.*\.(?:gif|jpg)$',
         }
     }, {
         'url': 'http://www.screencast.com/t/V2uXehPJa1ZI',
         }
     }, {
         'url': 'http://www.screencast.com/t/V2uXehPJa1ZI',
@@ -31,7 +31,7 @@ class ScreencastIE(InfoExtractor):
             'ext': 'mov',
             'title': 'The Amadeus Spectrometer',
             'description': 're:^In this video, our friends at.*To learn more about Amadeus, visit',
             'ext': 'mov',
             'title': 'The Amadeus Spectrometer',
             'description': 're:^In this video, our friends at.*To learn more about Amadeus, visit',
-            'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
+            'thumbnail': r're:^https?://.*\.(?:gif|jpg)$',
         }
     }, {
         'url': 'http://www.screencast.com/t/aAB3iowa',
         }
     }, {
         'url': 'http://www.screencast.com/t/aAB3iowa',
@@ -41,7 +41,7 @@ class ScreencastIE(InfoExtractor):
             'ext': 'mp4',
             'title': 'Google Earth Export',
             'description': 'Provides a demo of a CommunityViz export to Google Earth, one of the 3D viewing options.',
             'ext': 'mp4',
             'title': 'Google Earth Export',
             'description': 'Provides a demo of a CommunityViz export to Google Earth, one of the 3D viewing options.',
-            'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
+            'thumbnail': r're:^https?://.*\.(?:gif|jpg)$',
         }
     }, {
         'url': 'http://www.screencast.com/t/X3ddTrYh',
         }
     }, {
         'url': 'http://www.screencast.com/t/X3ddTrYh',
@@ -51,14 +51,15 @@ class ScreencastIE(InfoExtractor):
             'ext': 'wmv',
             'title': 'Toolkit 6 User Group Webinar (2014-03-04) - Default Judgment and First Impression',
             'description': 'md5:7b9f393bc92af02326a5c5889639eab0',
             'ext': 'wmv',
             'title': 'Toolkit 6 User Group Webinar (2014-03-04) - Default Judgment and First Impression',
             'description': 'md5:7b9f393bc92af02326a5c5889639eab0',
-            'thumbnail': 're:^https?://.*\.(?:gif|jpg)$',
+            'thumbnail': r're:^https?://.*\.(?:gif|jpg)$',
         }
         }
-    },
-    ]
+    }, {
+        'url': 'http://screencast.com/t/aAB3iowa',
+        'only_matching': True,
+    }]
 
     def _real_extract(self, url):
 
     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_url = self._html_search_regex(
         webpage = self._download_webpage(url, video_id)
 
         video_url = self._html_search_regex(
@@ -89,14 +90,24 @@ class ScreencastIE(InfoExtractor):
                     r'src=(.*?)(?:$|&)', video_meta,
                     'meta tag video URL', default=None)
 
                     r'src=(.*?)(?:$|&)', video_meta,
                     'meta tag video URL', default=None)
 
+        if video_url is None:
+            video_url = self._html_search_regex(
+                r'MediaContentUrl["\']\s*:(["\'])(?P<url>(?:(?!\1).)+)\1',
+                webpage, 'video url', default=None, group='url')
+
+        if video_url is None:
+            video_url = self._html_search_meta(
+                'og:video', webpage, default=None)
+
         if video_url is None:
             raise ExtractorError('Cannot find video')
 
         title = self._og_search_title(webpage, default=None)
         if title is None:
             title = self._html_search_regex(
         if video_url is None:
             raise ExtractorError('Cannot find video')
 
         title = self._og_search_title(webpage, default=None)
         if title is None:
             title = self._html_search_regex(
-                [r'<b>Title:</b> ([^<]*)</div>',
-                 r'class="tabSeperator">></span><span class="tabText">(.*?)<'],
+                [r'<b>Title:</b> ([^<]+)</div>',
+                 r'class="tabSeperator">></span><span class="tabText">(.+?)<',
+                 r'<title>([^<]+)</title>'],
                 webpage, 'title')
         thumbnail = self._og_search_thumbnail(webpage)
         description = self._og_search_description(webpage, default=None)
                 webpage, 'title')
         thumbnail = self._og_search_thumbnail(webpage)
         description = self._og_search_description(webpage, default=None)