Sportbox source fix. HD videos support.

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

diff --git a/youtube_dl/extractor/sportbox.py b/youtube_dl/extractor/sportbox.py
index 19cc976e3d3c9c7cff875f26e63e25a827fb8073..695b3ff825288d620865a8b8cc65a7b378ed8e5b 100644 (file)
--- a/youtube_dl/extractor/sportbox.py
+++ b/youtube_dl/extractor/sportbox.py
@@ -7,15 +7,14 @@ from .common import InfoExtractor
 from ..utils import (
     parse_duration,
     parse_iso8601,
-    int_or_none,
 )
 
 
 class SportBoxIE(InfoExtractor):
-    _VALID_URL = r'https?://news\.sportbox\.ru/Vidy_sporta/(?:[^/]+/)+spbvideo_NI\d+_(?P<display_id>.+)'
+    _VALID_URL = r'https?://news\.sportbox\.ru/(?:[^/]+/)+spbvideo_NI\d+_(?P<display_id>.+)'
     _TESTS = [
         {
-            'url': 'http://news.sportbox.ru/Vidy_sporta/Avtosport/Rossijskij/spbvideo_NI483529_Gonka-2-zaezd-Obyedinenniy-2000-klassi-Turing-i-S',
+            'url': 'http://news.sportbox.ru/Vidy_sporta/Avtosport/Rossijskij/spbvideo_NI483529_Gonka-2-zaezd-Obyedinenniy-2000-klassi-Turing-i-S',            
             'md5': 'ff56a598c2cf411a9a38a69709e97079',
             'info_dict': {
                 'id': '80822',
@@ -26,7 +25,6 @@ class SportBoxIE(InfoExtractor):
                 'timestamp': 1411896237,
                 'upload_date': '20140928',
                 'duration': 4846,
-                'view_count': int,
             },
             'params': {
                 # m3u8 download
@@ -44,15 +42,19 @@ class SportBoxIE(InfoExtractor):
 
         webpage = self._download_webpage(url, display_id)
 
-        video_id = self._search_regex(
-            r'src="/vdl/player/media/(\d+)"', webpage, 'video id')
+        sobj = re.search(r'src="/vdl/player/(?P<media_type>\w+)/(?P<video_id>\d+)"', webpage)
+        if (sobj):
+            video_id = sobj.group('video_id')
+            media_type = sobj.group('media_type')
+        else:
+            raise RegexNotFoundError('Unable to extract video_id')
 
         player = self._download_webpage(
-            'http://news.sportbox.ru/vdl/player/media/%s' % video_id,
+            'http://news.sportbox.ru/vdl/player/%s/%s' % (media_type, video_id),
             display_id, 'Downloading player webpage')
 
         hls = self._search_regex(
-            r"var\s+original_hls_file\s*=\s*'([^']+)'", player, 'hls file')
+            r"sportboxPlayer\.jwplayer_common_params\.file\s*=\s*['\"]+([^\"]+)['\"]+", player, 'hls file')
 
         formats = self._extract_m3u8_formats(hls, display_id, 'mp4')
 
@@ -65,8 +67,6 @@ class SportBoxIE(InfoExtractor):
             r'<span itemprop="uploadDate">([^<]+)</span>', webpage, 'timestamp', fatal=False))
         duration = parse_duration(self._html_search_regex(
             r'<meta itemprop="duration" content="PT([^"]+)">', webpage, 'duration', fatal=False))
-        view_count = int_or_none(self._html_search_regex(
-            r'<span>Просмотров: (\d+)</span>', player, 'view count', fatal=False))
 
         return {
             'id': video_id,
@@ -76,6 +76,5 @@ class SportBoxIE(InfoExtractor):
             'thumbnail': thumbnail,
             'timestamp': timestamp,
             'duration': duration,
-            'view_count': view_count,
             'formats': formats,
         }