From: Jan Szejko <jan.szejko@gmail.com>
Date: Mon, 11 Jul 2016 16:21:07 +0000 (+0200)
Subject: don't escape non-ascii for regular expressions
X-Git-Url: https://git.mdrn.pl/wolnelektury.git/commitdiff_plain/2e7bf0fd9796ec992d1d9dfde2e6824961429723?hp=--cc

don't escape non-ascii for regular expressions
---

2e7bf0fd9796ec992d1d9dfde2e6824961429723
diff --git a/src/catalogue/views.py b/src/catalogue/views.py
index 6782e5369..44a17b412 100644
--- a/src/catalogue/views.py
+++ b/src/catalogue/views.py
@@ -365,7 +365,14 @@ def _no_diacritics_regexp(query):
 
 def unicode_re_escape(query):
     """ Unicode-friendly version of re.escape """
-    return re.sub(r'(?u)(\W)', r'\\\1', query)
+    s = list(query)
+    for i, c in enumerate(query):
+        if re.match(r'(?u)(\W)', c) and re.match(r'[\x00-\x7e]', c):
+            if c == "\000":
+                s[i] = "\\000"
+            else:
+                s[i] = "\\" + c
+    return query[:0].join(s)
 
 
 def _word_starts_with(name, prefix):