From 2f40ef1826c4c4701f9b316cf6549798c72b2b79 Mon Sep 17 00:00:00 2001
From: LecygneNoir <git@lecygnenoir.info>
Date: Sat, 9 Mar 2019 11:34:57 +0100
Subject: [PATCH] Simplify cleanString function to prepare python3
 compatibility

---
 lib/utils.py | 17 +++++------------
 1 file changed, 5 insertions(+), 12 deletions(-)

diff --git a/lib/utils.py b/lib/utils.py
index 772ff6b..63083b0 100644
--- a/lib/utils.py
+++ b/lib/utils.py
@@ -3,9 +3,10 @@
 
 from ConfigParser import RawConfigParser, NoOptionError, NoSectionError
 from os.path import dirname, splitext, basename, isfile
+import re
 from os import devnull
 from subprocess import check_call, CalledProcessError, STDOUT
-import unicodedata
+import unidecode
 import logging
 
 ### CATEGORIES ###
@@ -195,16 +196,8 @@ def upcaseFirstLetter(s):
 
 
 def cleanString(toclean):
-    toclean = toclean.split(' ')
-    cleaned = ''
-    for s in toclean:
-        if s == '':
-            continue
-        strtoclean = unicodedata.normalize('NFKD', unicode (s, 'utf-8')).encode('ASCII', 'ignore')
-        strtoclean = ''.join(e for e in strtoclean if e.isalnum())
-        if strtoclean == '':
-            continue
-        strtoclean = upcaseFirstLetter(strtoclean)
-        cleaned = cleaned + strtoclean
+    toclean = toclean.decode('utf-8')
+    toclean = unidecode.unidecode(toclean)
+    cleaned = re.sub('[^A-Za-z0-9]+', '', toclean)
 
     return cleaned