Properly support simple titles in the newest InfoExtractors
authorRicardo Garcia <sarbalap+freshmeat@gmail.com>
Sat, 20 Feb 2010 23:13:34 +0000 (00:13 +0100)
committerRicardo Garcia <sarbalap+freshmeat@gmail.com>
Sun, 31 Oct 2010 10:26:34 +0000 (11:26 +0100)
youtube-dl

index fe8a591606a38c243cb1214372d90b505f41e207..9e15a6c695c1ead283abf0dd0d190102661d376a 100755 (executable)
@@ -1034,6 +1034,7 @@ class GoogleIE(InfoExtractor):
                        return
                video_title = mobj.group(1).decode('utf-8')
                video_title = sanitize_title(video_title)
+               simple_title = re.sub(ur'(?u)([^%s]+)' % simple_title_chars, ur'_', video_title)
 
                # Google Video doesn't show uploader nicknames?
                video_uploader = 'NA'
@@ -1045,7 +1046,7 @@ class GoogleIE(InfoExtractor):
                                'url':          video_url.decode('utf-8'),
                                'uploader':     video_uploader.decode('utf-8'),
                                'title':        video_title,
-                               'stitle':       video_title,
+                               'stitle':       simple_title,
                                'ext':          video_extension.decode('utf-8'),
                        })
                except UnavailableFormatError:
@@ -1111,6 +1112,7 @@ class PhotobucketIE(InfoExtractor):
                        return
                video_title = mobj.group(1).decode('utf-8')
                video_title = sanitize_title(video_title)
+               simple_title = re.sub(ur'(?u)([^%s]+)' % simple_title_chars, ur'_', video_title)
 
                video_uploader = mobj.group(2).decode('utf-8')
 
@@ -1121,7 +1123,7 @@ class PhotobucketIE(InfoExtractor):
                                'url':          video_url.decode('utf-8'),
                                'uploader':     video_uploader,
                                'title':        video_title,
-                               'stitle':       video_title,
+                               'stitle':       simple_title,
                                'ext':          video_extension.decode('utf-8'),
                        })
                except UnavailableFormatError:
@@ -1199,6 +1201,7 @@ class GenericIE(InfoExtractor):
                        return
                video_title = mobj.group(1).decode('utf-8')
                video_title = sanitize_title(video_title)
+               simple_title = re.sub(ur'(?u)([^%s]+)' % simple_title_chars, ur'_', video_title)
 
                # video uploader is domain name
                mobj = re.match(r'(?:https?://)?([^/]*)/.*', url)
@@ -1214,7 +1217,7 @@ class GenericIE(InfoExtractor):
                                'url':          video_url.decode('utf-8'),
                                'uploader':     video_uploader,
                                'title':        video_title,
-                               'stitle':       video_title,
+                               'stitle':       simple_title,
                                'ext':          video_extension.decode('utf-8'),
                        })
                except UnavailableFormatError: