Search query normalisation no longer removes punctuation #5416

CHANGELOG.txt

@@ -65,6 +65,7 @@ Changelog
  * Fix: Move focus to the pages explorer menu when open (Helen Chapman)
  * Fix: Clicking the privacy toggle while the page is still loading no longer loads the wrong data in the page (Helen Chapman)
  * Fix: Added missing `is_stored_locally` method to `AbstractDocument` (jonny5532)
+ * Fix: Query model no longer removes punctuation as part of string normalisation (William Blackie)
 
 
 2.5.1 (07.05.2019)

CONTRIBUTORS.rst

@@ -381,6 +381,7 @@ Contributors
 * Fidel Ramos
 * Quadric
 * jonny5532
+* William Blackie
 
 Translators
 ===========

docs/releases/2.6.rst

@@ -96,6 +96,7 @@ Bug fixes
  * ``Page.copy()`` no longer copies child objects when the accesssor name is included in ``exclude_fields_in_copy`` (Karl Hobley)
  * Clicking the privacy toggle while the page is still loading no longer loads the wrong data in the page (Helen Chapman)
  * Added missing ``is_stored_locally`` method to ``AbstractDocument`` (jonny5532)
+ * Query model no longer removes punctuation as part of string normalisation (William Blackie)
 
 
 Upgrade considerations

wagtail/search/tests/test_queries.py

@@ -33,17 +33,17 @@ class TestHitCounter(TestCase):
 
 class TestQueryStringNormalisation(TestCase):
     def setUp(self):
-        self.query = models.Query.get("Hello World!")
+        self.query = models.Query.get("  Hello  World!  ")
 
     def test_normalisation(self):
-        self.assertEqual(str(self.query), "hello world")
+        self.assertEqual(str(self.query), "hello world!")
 
-    def test_equivilant_queries(self):
+    def test_equivalent_queries(self):
         queries = [
-            "Hello World",
-            "Hello  World!!",
-            "hello world",
-            "Hello' world",
+            "  Hello World!",
+            "Hello World!  ",
+            "hello  world!",
+            "  Hello  world!  ",
         ]
 
         for query in queries:
@@ -52,7 +52,8 @@ class TestQueryStringNormalisation(TestCase):
     def test_different_queries(self):
         queries = [
             "HelloWorld",
-            "Hello orld!!",
+            "HelloWorld!"
+            "  Hello  World!  ",
             "Hello",
         ]
 

wagtail/search/utils.py

@@ -1,6 +1,5 @@
 import operator
 import re
-import string
 from functools import partial, reduce
 
 # Reduce any iterable to a single value using a logical OR e.g. (a | b | ...)
@@ -22,11 +21,8 @@ def normalise_query_string(query_string):
     # Convert query_string to lowercase
     query_string = query_string.lower()
 
-    # Strip punctuation characters
-    query_string = ''.join([c for c in query_string if c not in string.punctuation])
-
-    # Remove double spaces
-    query_string = ' '.join(query_string.split())
+    # Remove leading, trailing and multiple spaces
+    query_string = re.sub(' +', ' ', query_string).strip()
 
     return query_string