Merge pull request #8092 from bpfoley/twitter-thumbnail

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

diff --git a/youtube_dl/extractor/arte.py b/youtube_dl/extractor/arte.py
index a4c786d78a99787cfbf720009ba5534a27925551..3e119e21b39ba2ab6bc504cf1d19a90008bfbd24 100644 (file)
--- a/youtube_dl/extractor/arte.py
+++ b/youtube_dl/extractor/arte.py
@@ -114,23 +114,25 @@ class ArteTVPlus7IE(InfoExtractor):
                     r'program\s*:\s*({.+?["\']embed_html["\'].+?}),?\s*\n',
                     webpage, 'program', default=None)
                 if program:
-                    embed_html = self._parse_json(program,video_id)
+                    embed_html = self._parse_json(program, video_id)
                     if embed_html:
                         iframe_url = find_iframe_url(embed_html['embed_html'])
             if iframe_url:
                 json_url = compat_parse_qs(
                     compat_urllib_parse_urlparse(iframe_url).query)['json_url'][0]
         if json_url:
-            return self._extract_from_json_url(json_url, video_id, lang)
-        # Differend kind of embed URL (e.g.
+            title = self._search_regex(
+                r'<h3[^>]+title=(["\'])(?P<title>.+?)\1',
+                webpage, 'title', default=None, group='title')
+            return self._extract_from_json_url(json_url, video_id, lang, title=title)
+        # Different kind of embed URL (e.g.
         # http://www.arte.tv/magazine/trepalium/fr/episode-0406-replay-trepalium)
         embed_url = self._search_regex(
             r'<iframe[^>]+src=(["\'])(?P<url>.+?)\1',
             webpage, 'embed url', group='url')
         return self.url_result(embed_url)
 
-
-    def _extract_from_json_url(self, json_url, video_id, lang):
+    def _extract_from_json_url(self, json_url, video_id, lang, title=None):
         info = self._download_json(json_url, video_id)
         player_info = info['videoJsonPlayer']
 
@@ -138,7 +140,7 @@ class ArteTVPlus7IE(InfoExtractor):
         if not upload_date_str:
             upload_date_str = (player_info.get('VRA') or player_info.get('VDA') or '').split(' ')[0]
 
-        title = player_info['VTI'].strip()
+        title = (player_info.get('VTI') or title or player_info['VID']).strip()
         subtitle = player_info.get('VSU', '').strip()
         if subtitle:
             title += ' - %s' % subtitle
@@ -240,6 +242,7 @@ class ArteTVFutureIE(ArteTVPlus7IE):
             'id': '050940-028-A',
             'ext': 'mp4',
             'title': 'Les écrevisses aussi peuvent être anxieuses',
+            'upload_date': '20140902',
         },
     }, {
         'url': 'http://future.arte.tv/fr/la-science-est-elle-responsable',
@@ -306,11 +309,12 @@ class ArteTVMagazineIE(ArteTVPlus7IE):
     _TESTS = [{
         # Embedded via <iframe src="http://www.arte.tv/arte_vp/index.php?json_url=..."
         'url': 'http://www.arte.tv/magazine/trepalium/fr/entretien-avec-le-realisateur-vincent-lannoo-trepalium',
-        'md5': '66a093339c1278bb3719157ef07107b2',
+        'md5': '2a9369bcccf847d1c741e51416299f25',
         'info_dict': {
             'id': '065965-000-A',
             'ext': 'mp4',
             'title': 'Trepalium - Extrait Ep.01',
+            'upload_date': '20160121',
         },
     }, {
         # Embedded via <iframe src="http://www.arte.tv/guide/fr/embed/054813-004-A/medium"