]> Raphaël G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/dvtv.py
debian/control: Canonicalize with `wrap-and-sort -s -a -b`.
[youtubedl] / youtube_dl / extractor / dvtv.py
index e85c58bd5669bd14cb069690d43277f77cea49d3..20996962a7bf58b46d72a262034f63cfd5d049c4 100644 (file)
@@ -32,7 +32,7 @@ class DVTVIE(InfoExtractor):
     }, {
         'url': 'http://video.aktualne.cz/dvtv/dvtv-16-12-2014-utok-talibanu-boj-o-kliniku-uprchlici/r~973eb3bc854e11e498be002590604f2e/',
         'info_dict': {
     }, {
         'url': 'http://video.aktualne.cz/dvtv/dvtv-16-12-2014-utok-talibanu-boj-o-kliniku-uprchlici/r~973eb3bc854e11e498be002590604f2e/',
         'info_dict': {
-            'title': 'DVTV 16. 12. 2014: útok Talibanu, boj o kliniku, uprchlíci',
+            'title': r're:^DVTV 16\. 12\. 2014: útok Talibanu, boj o kliniku, uprchlíci',
             'id': '973eb3bc854e11e498be002590604f2e',
         },
         'playlist': [{
             'id': '973eb3bc854e11e498be002590604f2e',
         },
         'playlist': [{
@@ -93,8 +93,11 @@ class DVTVIE(InfoExtractor):
         'only_matching': True,
     }]
 
         'only_matching': True,
     }]
 
-    def _parse_video_metadata(self, js, video_id):
+    def _parse_video_metadata(self, js, video_id, live_js=None):
         data = self._parse_json(js, video_id, transform_source=js_to_json)
         data = self._parse_json(js, video_id, transform_source=js_to_json)
+        if live_js:
+            data.update(self._parse_json(
+                live_js, video_id, transform_source=js_to_json))
 
         title = unescapeHTML(data['title'])
 
 
         title = unescapeHTML(data['title'])
 
@@ -142,13 +145,18 @@ class DVTVIE(InfoExtractor):
 
         webpage = self._download_webpage(url, video_id)
 
 
         webpage = self._download_webpage(url, video_id)
 
+        # live content
+        live_item = self._search_regex(
+            r'(?s)embedData[0-9a-f]{32}\.asset\.liveStarter\s*=\s*(\{.+?\});',
+            webpage, 'video', default=None)
+
         # single video
         item = self._search_regex(
             r'(?s)embedData[0-9a-f]{32}\[["\']asset["\']\]\s*=\s*(\{.+?\});',
         # single video
         item = self._search_regex(
             r'(?s)embedData[0-9a-f]{32}\[["\']asset["\']\]\s*=\s*(\{.+?\});',
-            webpage, 'video', default=None, fatal=False)
+            webpage, 'video', default=None)
 
         if item:
 
         if item:
-            return self._parse_video_metadata(item, video_id)
+            return self._parse_video_metadata(item, video_id, live_item)
 
         # playlist
         items = re.findall(
 
         # playlist
         items = re.findall(