]> Raphaƫl G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/nba.py
debian/copyright: Fix missing-license-paragraph-in-dep5-copyright.
[youtubedl] / youtube_dl / extractor / nba.py
index 296d4cd36e9e295ccbb6f88af0c3b73e7a78ce3b..f69fe0925ee4d5d68699f09d84a568ce34f787ec 100644 (file)
@@ -1,40 +1,48 @@
+from __future__ import unicode_literals
+
 import re
 
 from .common import InfoExtractor
 from ..utils import (
-    ExtractorError,
+    remove_end,
+    parse_duration,
 )
 
 
 class NBAIE(InfoExtractor):
-    _VALID_URL = r'^(?:https?://)?(?:watch\.|www\.)?nba\.com/(?:nba/)?video(/[^?]*?)(?:/index\.html)?(?:\?.*)?$'
+    _VALID_URL = r'https?://(?:watch\.|www\.)?nba\.com/(?:nba/)?video(?P<id>/[^?]*?)(?:/index\.html)?(?:\?.*)?$'
+    _TEST = {
+        'url': 'http://www.nba.com/video/games/nets/2012/12/04/0021200253-okc-bkn-recap.nba/index.html',
+        'md5': 'c0edcfc37607344e2ff8f13c378c88a4',
+        'info_dict': {
+            'id': '0021200253-okc-bkn-recap.nba',
+            'ext': 'mp4',
+            'title': 'Thunder vs. Nets',
+            'description': 'Kevin Durant scores 32 points and dishes out six assists as the Thunder beat the Nets in Brooklyn.',
+            'duration': 181,
+        },
+    }
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
-        if mobj is None:
-            raise ExtractorError(u'Invalid URL: %s' % url)
-
-        video_id = mobj.group(1)
+        video_id = mobj.group('id')
 
         webpage = self._download_webpage(url, video_id)
 
-        video_url = u'http://ht-mobile.cdn.turner.com/nba/big' + video_id + '_nba_1280x720.mp4'
+        video_url = 'http://ht-mobile.cdn.turner.com/nba/big' + video_id + '_nba_1280x720.mp4'
 
         shortened_video_id = video_id.rpartition('/')[2]
-        title = self._html_search_regex(r'<meta property="og:title" content="(.*?)"',
-            webpage, 'title', default=shortened_video_id).replace('NBA.com: ', '')
+        title = remove_end(
+            self._og_search_title(webpage, default=shortened_video_id), ' : NBA.com')
 
-        # It isn't there in the HTML it returns to us
-        # uploader_date = self._html_search_regex(r'<b>Date:</b> (.*?)</div>', webpage, 'upload_date', fatal=False)
+        description = self._og_search_description(webpage)
+        duration = parse_duration(
+            self._html_search_meta('duration', webpage, 'duration', fatal=False))
 
-        description = self._html_search_regex(r'<meta name="description" (?:content|value)="(.*?)" />', webpage, 'description', fatal=False)
-
-        info = {
+        return {
             'id': shortened_video_id,
             'url': video_url,
-            'ext': 'mp4',
             'title': title,
-            # 'uploader_date': uploader_date,
             'description': description,
+            'duration': duration,
         }
-        return [info]