sort cleanup

pandora/item/models.py

@@ -282,6 +282,9 @@ class Item(models.Model):
                 data['releasedate'] = min(data.pop('release date'))
             if 'plot' in data:
                 data['summary'] = data.pop('plot')
+            if isinstance(data['cast'][0], basestring):
+                data['actor'] = [data['cast'][0]]
+            else:
                 data['actor'] = [c[0] for c in data['cast']]
             self.external_data = data
             self.save()
@@ -562,24 +565,23 @@ class Item(models.Model):
             if name not in base_keys:
                 if field_type == 'title':
                     value = utils.sort_title(canonicalTitle(self.get(name)))
-                    value = unicodedata.normalize('NFKD', value)
+                    value = utils.sort_string(value)
                     setattr(s, '%s_desc'%name, value)
                     if not value:
                         value = 'zzzzzzzzzzzzzzzzzzzzzzzzz'
                     setattr(s, name, value)
                 elif field_type == 'person':
                     value = sortNames(self.get(name, []))
-                    value = unicodedata.normalize('NFKD', value)[:255]
+                    value = utils.sort_string(value)[:955]
                     setattr(s, '%s_desc'%name, value)
                     if not value:
                         value = 'zzzzzzzzzzzzzzzzzzzzzzzzz'
                     setattr(s, name, value)
-                elif field_type == 'text':
+                elif field_type == 'string':
-                    #FIXME: what use pural_key?
                     value = self.get(name, u'')
                     if isinstance(value, list):
                         value = u','.join(value)
-                    value = unicodedata.normalize('NFKD', value)
+                    value = utils.sort_string(value)[:955]
                     setattr(s, '%s_desc'%name, value)
                     if not value:
                         value = 'zzzzzzzzzzzzzzzzzzzzzzzzz'
@@ -896,15 +898,12 @@ attrs = {
 for key in site_config['sortKeys']:
     name = key['id']
     field_type = key['type']
-    if field_type in ('string', 'title'):
+    if field_type in ('string', 'title', 'person'):
         attrs[name] = models.CharField(max_length=1000, db_index=True)
         attrs['%s_desc'%name] = models.CharField(max_length=1000, db_index=True)
     elif field_type == 'year':
         attrs[name] = models.CharField(max_length=4, db_index=True)
         attrs['%s_desc'%name] = models.CharField(max_length=4, db_index=True)
-    elif field_type in ('text', 'person'):
-        attrs[name] = models.TextField(blank=True, db_index=True)
-        attrs['%s_desc'%name] = models.TextField(blank=True, db_index=True)
     elif field_type in ('integer', 'words', 'length'):
         attrs[name] = models.BigIntegerField(blank=True, db_index=True)
         attrs['%s_desc'%name] = models.BigIntegerField(blank=True, db_index=True)

pandora/item/utils.py

@@ -189,6 +189,11 @@ def parse_path(path):
     return r
 
 
+def sort_string(string):
+    string = string.replace(u'Þ', 'Th')
+    return unicodedata.normalize('NFKD', string)
+
+
 def sort_title(title):
     #title
     title = re.sub(u'[\'!¿¡,\.;\-"\:\*\[\]]', '', title)
@@ -196,9 +201,8 @@ def sort_title(title):
     #title = title.replace(u'Æ', 'Ae')
     if isinstance(title, str):
         title = unicode(title)
-    title = unicodedata.normalize('NFKD', title)
+    title = sort_string(title)
 
     #pad numbered titles
     title = re.sub('(\d+)', lambda x: '%010d' % int(x.group(0)), title)
     return title.strip()
-

pandora/templates/site.json

@@ -75,16 +75,16 @@
     "sortKeys": [
         {"id": "title", "title": "Title", "width": 180, "removable": false, "type": "title", "autocompleteSortKey": "votes"},
         {"id": "director", "title": "Director", "width": 180, "removable": false, "type": "person"},
-        {"id": "country", "title": "Country", "width": 120, "type": "text"},
+        {"id": "country", "title": "Country", "width": 120, "type": "string"},
         {"id": "year", "title": "Year", "width": 60, "type": "year"},
-        {"id": "language", "title": "Language", "width": 120, "type": "text"},
+        {"id": "language", "title": "Language", "width": 120, "type": "string"},
         {"id": "runtime", "title": "Runtime", "width": 60, "type": "integer"},
         {"id": "writer", "title": "Writer", "width": 180, "type": "person"},
         {"id": "producer", "title": "Producer", "width": 180, "type": "person"},
         {"id": "cinematographer", "title": "Cinematographer", "width": 180, "type": "person"},
         {"id": "editor", "title": "Editor", "width": 180, "type": "person"},
         {"id": "actor", "title": "Number of Actors", "width": 60, "type": "person"},
-        {"id": "genre", "title": "Genre", "width": 120, "type": "text"},
+        {"id": "genre", "title": "Genre", "width": 120, "type": "string"},
         {"id": "keyword", "title": "Number of Keywords", "width": 60, "type": "integer"},
         {"id": "summary", "title": "Words in Summary", "width": 60, "type": "words"},
         {"id": "trivia", "title": "Words in Trivia", "width": 60, "type": "words"},
@@ -111,7 +111,7 @@
         {"id": "size", "title": "Size", "width": 90, "type": "integer"},
         {"id": "bitrate", "title": "Bitrate", "width": 90, "type": "integer"},
         {"id": "files", "title": "Files", "width": 60, "type": "integer"},
-        {"id": "filename", "title": "Filename", "width": 180, "type": "text"},
+        {"id": "filename", "title": "Filename", "width": 180, "type": "string"},
         {"id": "published", "title": "Date Published", "width": 90, "type": "date"},
         {"id": "modified", "title": "Date Modified", "width": 90, "type": "date"},
         {"id": "popularity", "title": "Popularity", "type": "float"}