Prepare for upload.

[youtubedl] / youtube_dl / extractor / arte.py

diff --git a/youtube_dl/extractor/arte.py b/youtube_dl/extractor/arte.py
index d39b489518f7bc699854d9ddeb7cd4fa357a6c2b..b35a679e3b036d2c573a4f1fc85d53bd793f745b 100644 (file)
--- a/youtube_dl/extractor/arte.py
+++ b/youtube_dl/extractor/arte.py
@@ -10,6 +10,7 @@ from ..utils import (
     unified_strdate,
     determine_ext,
     get_element_by_id,
+    compat_str,
 )
 
 # There are different sources of video in arte.tv, the extraction process 
@@ -158,7 +159,9 @@ class ArteTVPlus7IE(InfoExtractor):
             'thumbnail': player_info.get('programImage') or player_info.get('VTU', {}).get('IUR'),
         }
 
-        formats = player_info['VSR'].values()
+        all_formats = player_info['VSR'].values()
+        # Some formats use the m3u8 protocol
+        all_formats = list(filter(lambda f: f.get('videoFormat') != 'M3U8', all_formats))
         def _match_lang(f):
             if f.get('versionCode') is None:
                 return True
@@ -170,24 +173,39 @@ class ArteTVPlus7IE(InfoExtractor):
             regexes = [r'VO?%s' % l, r'VO?.-ST%s' % l]
             return any(re.match(r, f['versionCode']) for r in regexes)
         # Some formats may not be in the same language as the url
-        formats = filter(_match_lang, formats)
-        # Some formats use the m3u8 protocol
-        formats = filter(lambda f: f.get('videoFormat') != 'M3U8', formats)
-        # We order the formats by quality
+        formats = filter(_match_lang, all_formats)
         formats = list(formats) # in python3 filter returns an iterator
+        if not formats:
+            # Some videos are only available in the 'Originalversion'
+            # they aren't tagged as being in French or German
+            if all(f['versionCode'] == 'VO' for f in all_formats):
+                formats = all_formats
+            else:
+                raise ExtractorError(u'The formats list is empty')
+
         if re.match(r'[A-Z]Q', formats[0]['quality']) is not None:
-            sort_key = lambda f: ['HQ', 'MQ', 'EQ', 'SQ'].index(f['quality'])
+            def sort_key(f):
+                return ['HQ', 'MQ', 'EQ', 'SQ'].index(f['quality'])
         else:
-            sort_key = lambda f: int(f.get('height',-1))
+            def sort_key(f):
+                return (
+                    # Sort first by quality
+                    int(f.get('height',-1)),
+                    int(f.get('bitrate',-1)),
+                    # The original version with subtitles has lower relevance
+                    re.match(r'VO-ST(F|A)', f.get('versionCode', '')) is None,
+                    # The version with sourds/mal subtitles has also lower relevance
+                    re.match(r'VO?(F|A)-STM\1', f.get('versionCode', '')) is None,
+                )
         formats = sorted(formats, key=sort_key)
-        # Prefer videos without subtitles in the same language
-        formats = sorted(formats, key=lambda f: re.match(r'VO(F|A)-STM\1', f.get('versionCode', '')) is None)
-        # Pick the best quality
         def _format(format_info):
-            quality = format_info['quality']
-            m_quality = re.match(r'\w*? - (\d*)p', quality)
-            if m_quality is not None:
-                quality = m_quality.group(1)
+            quality = ''
+            height = format_info.get('height')
+            if height is not None:
+                quality = compat_str(height)
+            bitrate = format_info.get('bitrate')
+            if bitrate is not None:
+                quality += '-%d' % bitrate
             if format_info.get('versionCode') is not None:
                 format_id = u'%s-%s' % (quality, format_info['versionCode'])
             else:
@@ -196,7 +214,7 @@ class ArteTVPlus7IE(InfoExtractor):
                 'format_id': format_id,
                 'format_note': format_info.get('versionLibelle'),
                 'width': format_info.get('width'),
-                'height': format_info.get('height'),
+                'height': height,
             }
             if format_info['mediaType'] == u'rtmp':
                 info['url'] = format_info['streamer']