From 5482248feef2b4c7294d66d33f1758d02c39a9f2 Mon Sep 17 00:00:00 2001
From: j <0x006A@0x2620.org>
Date: Sun, 26 Aug 2012 16:22:18 +0200
Subject: [PATCH] normalize imdb values to make find more reliable for unicode
 titles, fixes #930

---
 oxdata/movie/models.py | 14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)

diff --git a/oxdata/movie/models.py b/oxdata/movie/models.py
index 0336ea2..2e69f87 100644
--- a/oxdata/movie/models.py
+++ b/oxdata/movie/models.py
@@ -2,10 +2,15 @@
 # vi:si:et:sw=4:sts=4:ts=4
 
 import re
+import unicodedata
 
 from django.db import models
 import ox
 
+def normalize_value(value):
+    if isinstance(value, str):
+        value = value.decode('utf-8')
+    return unicodedata.normalize('NFD', value)
 
 def find(info, guess=True):
     q = Imdb.objects.all()
@@ -26,9 +31,10 @@ def find(info, guess=True):
                     fkey = key
                 if isinstance(info[key], list):
                     fkey = '%s__iexact'%key
-                    q = q.filter(**{fkey: '\n'.join(info[key]) + '\n'})
+                    value = normalize_value(u'\n'.join(info[key]) + '\n')
                 else:
-                    q = q.filter(**{fkey:info[key]})
+                    value = normalize_value(info[key])
+                q = q.filter(**{fkey:value})
     if q.count() == 1:
         m = q[0]
         m.update()
@@ -75,7 +81,9 @@ class Imdb(models.Model):
                     if key == 'title' and 'seriesTitle' in info:
                         value = info['seriesTitle']
                     if isinstance(value, list):
-                        value = '\n'.join(value) + '\n'
+                        value = u'\n'.join(value) + '\n'
+                    if isinstance(value, basestring):
+                        value = normalize_value(value)
                     setattr(self, key, value)
             if self.season < 0:
                 self.season = None