X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/6d835d4d6903cf5e18ad01844736929e06d16004..5dafebb045625934d6c679b05132c61c5f9793c5:/youtube_dl/extractor/nba.py

diff --git a/youtube_dl/extractor/nba.py b/youtube_dl/extractor/nba.py
index 78e650b..a071378 100644
--- a/youtube_dl/extractor/nba.py
+++ b/youtube_dl/extractor/nba.py
@@ -4,46 +4,99 @@ import re
 
 from .common import InfoExtractor
 from ..utils import (
-    remove_end,
     parse_duration,
+    int_or_none,
+    xpath_text,
+    xpath_attr,
 )
 
 
 class NBAIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:watch\.|www\.)?nba\.com/(?:nba/)?video(?P<id>/[^?]*?)(?:/index\.html)?(?:\?.*)?$'
-    _TEST = {
+    _VALID_URL = r'https?://(?:watch\.|www\.)?nba\.com/(?P<path>(?:[^/]+/)?video/(?P<id>[^?]*?))/?(?:/index\.html)?(?:\?.*)?$'
+    _TESTS = [{
         'url': 'http://www.nba.com/video/games/nets/2012/12/04/0021200253-okc-bkn-recap.nba/index.html',
-        'md5': 'c0edcfc37607344e2ff8f13c378c88a4',
+        'md5': '9e7729d3010a9c71506fd1248f74e4f4',
         'info_dict': {
-            'id': '0021200253-okc-bkn-recap.nba',
+            'id': '0021200253-okc-bkn-recap',
             '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,
+            'timestamp': 1354638466,
+            'upload_date': '20121204',
         },
-    }
+        'params': {
+            # m3u8 download
+            'skip_download': True,
+        },
+    }, {
+        'url': 'http://www.nba.com/video/games/hornets/2014/12/05/0021400276-nyk-cha-play5.nba/',
+        'only_matching': True,
+    }, {
+        'url': 'http://watch.nba.com/video/channels/playoffs/2015/05/20/0041400301-cle-atl-recap.nba',
+        'md5': 'b2b39b81cf28615ae0c3360a3f9668c4',
+        'info_dict': {
+            'id': '0041400301-cle-atl-recap',
+            'ext': 'mp4',
+            'title': 'Hawks vs. Cavaliers Game 1',
+            'description': 'md5:8094c3498d35a9bd6b1a8c396a071b4d',
+            'duration': 228,
+            'timestamp': 1432134543,
+            'upload_date': '20150520',
+        }
+    }]
 
     def _real_extract(self, url):
-        mobj = re.match(self._VALID_URL, url)
-        video_id = mobj.group('id')
-
-        webpage = self._download_webpage(url, video_id)
-
-        video_url = 'http://ht-mobile.cdn.turner.com/nba/big' + video_id + '_nba_1280x720.mp4'
-
-        shortened_video_id = video_id.rpartition('/')[2]
-        title = remove_end(
-            self._og_search_title(webpage, default=shortened_video_id), ' : NBA.com')
+        path, video_id = re.match(self._VALID_URL, url).groups()
+        if path.startswith('nba/'):
+            path = path[3:]
+        video_info = self._download_xml('http://www.nba.com/%s.xml' % path, video_id)
+        video_id = xpath_text(video_info, 'slug')
+        title = xpath_text(video_info, 'headline')
+        description = xpath_text(video_info, 'description')
+        duration = parse_duration(xpath_text(video_info, 'length'))
+        timestamp = int_or_none(xpath_attr(video_info, 'dateCreated', 'uts'))
 
-        description = self._og_search_description(webpage)
-        duration = parse_duration(
-            self._html_search_meta('duration', webpage, 'duration', fatal=False))
+        thumbnails = []
+        for image in video_info.find('images'):
+            thumbnails.append({
+                'id': image.attrib.get('cut'),
+                'url': image.text,
+                'width': int_or_none(image.attrib.get('width')),
+                'height': int_or_none(image.attrib.get('height')),
+            })
 
+        formats = []
+        for video_file in video_info.findall('.//file'):
+            video_url = video_file.text
+            if video_url.startswith('/'):
+                continue
+            if video_url.endswith('.m3u8'):
+                formats.extend(self._extract_m3u8_formats(video_url, video_id, ext='mp4', m3u8_id='hls', fatal=False))
+            elif video_url.endswith('.f4m'):
+                formats.extend(self._extract_f4m_formats(video_url + '?hdcore=3.4.1.1', video_id, f4m_id='hds', fatal=False))
+            else:
+                key = video_file.attrib.get('bitrate')
+                format_info = {
+                    'format_id': key,
+                    'url': video_url,
+                }
+                mobj = re.search(r'(\d+)x(\d+)(?:_(\d+))?', key)
+                if mobj:
+                    format_info.update({
+                        'width': int(mobj.group(1)),
+                        'height': int(mobj.group(2)),
+                        'tbr': int_or_none(mobj.group(3)),
+                    })
+                formats.append(format_info)
+        self._sort_formats(formats)
 
         return {
-            'id': shortened_video_id,
-            'url': video_url,
+            'id': video_id,
             'title': title,
             'description': description,
             'duration': duration,
+            'timestamp': timestamp,
+            'thumbnails': thumbnails,
+            'formats': formats,
         }