[downloader/http] Simplify

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

diff --git a/youtube_dl/extractor/nbc.py b/youtube_dl/extractor/nbc.py
index 395f53df37daac4da38ca37cd60d548e316bb517..ccdbfb6c9c42c3eb22101eb2f7080b4bda1df00d 100644 (file)
--- a/youtube_dl/extractor/nbc.py
+++ b/youtube_dl/extractor/nbc.py
@@ -10,6 +10,8 @@ from ..compat import (
 from ..utils import (
     ExtractorError,
     find_xpath_attr,
+    lowercase_escape,
+    unescapeHTML,
 )
 
 
@@ -37,14 +39,32 @@ class NBCIE(InfoExtractor):
             },
             'skip': 'Only works from US',
         },
+        {
+            'url': 'http://www.nbc.com/saturday-night-live/video/star-wars-teaser/2832821',
+            'info_dict': {
+                'id': '8iUuyzWDdYUZ',
+                'ext': 'flv',
+                'title': 'Star Wars Teaser',
+                'description': 'md5:0b40f9cbde5b671a7ff62fceccc4f442',
+            },
+            'skip': 'Only works from US',
+        },
+        {
+            # This video has expired but with an escaped embedURL
+            'url': 'http://www.nbc.com/parenthood/episode-guide/season-5/just-like-at-home/515',
+            'skip': 'Expired'
+        }
     ]
 
     def _real_extract(self, url):
         video_id = self._match_id(url)
         webpage = self._download_webpage(url, video_id)
-        theplatform_url = self._search_regex(
-            '(?:class="video-player video-player-full" data-mpx-url|class="player" src)="(.*?)"',
-            webpage, 'theplatform url').replace('_no_endcard', '')
+        theplatform_url = unescapeHTML(lowercase_escape(self._html_search_regex(
+            [
+                r'(?:class="video-player video-player-full" data-mpx-url|class="player" src)="(.*?)"',
+                r'"embedURL"\s*:\s*"([^"]+)"'
+            ],
+            webpage, 'theplatform url').replace('_no_endcard', '').replace('\\/', '/')))
         if theplatform_url.startswith('//'):
             theplatform_url = 'http:' + theplatform_url
         return self.url_result(theplatform_url)
@@ -55,7 +75,6 @@ class NBCSportsVPlayerIE(InfoExtractor):
 
     _TESTS = [{
         'url': 'https://vplayer.nbcsports.com/p/BxmELC/nbcsports_share/select/9CsDKds0kvHI',
-        'md5': 'ceae8dced5c14a1c1ffcb7a32194cca5',
         'info_dict': {
             'id': '9CsDKds0kvHI',
             'ext': 'flv',
@@ -63,7 +82,6 @@ class NBCSportsVPlayerIE(InfoExtractor):
             'title': 'Tyler Kalinoski hits buzzer-beater to lift Davidson',
         }
     }, {
-        'note': 'This video is already expired. It\'s for testing _VALID_URL',
         'url': 'http://vplayer.nbcsports.com/p/BxmELC/nbc_embedshare/select/_hqLjQ95yx8Z',
         'only_matching': True,
     }]
@@ -88,7 +106,6 @@ class NBCSportsIE(InfoExtractor):
 
     _TEST = {
         'url': 'http://www.nbcsports.com//college-basketball/ncaab/tom-izzo-michigan-st-has-so-much-respect-duke',
-        'md5': 'ba6c93f96b67bf05344f78bd523dac0f',
         'info_dict': {
             'id': 'PHJSaFWbrTY9',
             'ext': 'flv',
@@ -107,7 +124,7 @@ class NBCSportsIE(InfoExtractor):
 class NBCNewsIE(InfoExtractor):
     _VALID_URL = r'''(?x)https?://(?:www\.)?nbcnews\.com/
         (?:video/.+?/(?P<id>\d+)|
-        (?:feature|nightly-news)/[^/]+/(?P<title>.+))
+        (?:watch|feature|nightly-news)/[^/]+/(?P<title>.+))
         '''
 
     _TESTS = [
@@ -152,6 +169,10 @@ class NBCNewsIE(InfoExtractor):
                 'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5',
             },
         },
+        {
+            'url': 'http://www.nbcnews.com/watch/dateline/full-episode--deadly-betrayal-386250819952',
+            'only_matching': True,
+        },
     ]
 
     def _real_extract(self, url):