Simplifying normalization of strings used to sort songs

songbook_core/content/sorted.py

@@ -7,18 +7,28 @@ from songbook_core.content.song import OnlySongsError, process_songs
 
 DEFAULT_SORT = ['by', 'album', '@title']
 
+def normalize_string(string):
+    return locale.strxfrm(string.lower().strip())
+
+def normalize_field(field):
+    if isinstance(field, basestring):
+        return normalize_string(field)
+    elif isinstance(field, list):
+        return [normalize_string(string) for string in field]
+
 def key_generator(sort):
     def ordered_song_keys(song):
         songkey = []
         for key in sort:
             if key == "@title":
-                songkey.append(song.normalized_titles)
+                field = song.unprefixed_titles
             elif key == "@path":
-                songkey.append(locale.strxfrm(song.path))
+                field = song.path
             elif key == "by":
-                songkey.append(song.normalized_authors)
+                field = song.authors
             else:
-                songkey.append(locale.strxfrm(song.args.get(key, "")))
+                field = song.args.get(key, "")
+            songkey.append(normalize_field(field))
         return songkey
     return ordered_song_keys
 

songbook_core/index.py

@@ -30,7 +30,7 @@ def sortkey(value):
     don't forget to call locale.setlocale(locale.LC_ALL, '')). It also handles
     the sort with  latex escape sequences.
     """
-    return locale.strxfrm(unidecode(simpleparse(value).replace(' ', 'A')))
+    return locale.strxfrm(unidecode(simpleparse(value).replace(' ', 'A')).lower())
 
 
 def process_sxd(filename):

songbook_core/songs.py

@@ -18,12 +18,10 @@ class Song(object):
         # Data extraction from the song with plastex
         data = parsetex(filename)
         self.titles = data['titles']
-        self.normalized_titles = [
+        self.unprefixed_titles = [
-                locale.strxfrm(
+                unprefixed_title(
-                    unprefixed_title(
+                    unidecode(unicode(title, "utf-8")),
-                        unidecode(unicode(title, "utf-8")),
+                    config['titleprefixwords']
-                        config['titleprefixwords']
-                        )
                     )
                 for title
                 in self.titles
@@ -32,13 +30,9 @@ class Song(object):
         self.path = filename
         self.languages = data['languages']
         if "by" in self.args.keys():
-            self.normalized_authors = [
+            self.authors = processauthors(self.args["by"], **config["authwords"])
-                locale.strxfrm(author)
-                for author
-                in processauthors(self.args["by"], **config["authwords"])
-                ]
         else:
-            self.normalized_authors = []
+            self.authors = []
 
     def __repr__(self):
         return repr((self.titles, self.args, self.path))