From e1fc584f1c00bc093cd41a82a0e42bd6516198e0 Mon Sep 17 00:00:00 2001
From: Mikalai Radchuk <m1kola@users.noreply.github.com>
Date: Mon, 30 Oct 2017 10:34:31 +0000
Subject: [PATCH] Implementation of separate_filters_from_query (#2766)

* Implementation of separate_filters_from_query

* Adds test with unclosed quotation mark
---
 wagtail/wagtailsearch/tests/test_queries.py | 48 ++++++++++++++++++++-
 wagtail/wagtailsearch/utils.py              | 14 ++++++
 2 files changed, 60 insertions(+), 2 deletions(-)

diff --git a/wagtail/wagtailsearch/tests/test_queries.py b/wagtail/wagtailsearch/tests/test_queries.py
index afc0f02175..400d2494a0 100644
--- a/wagtail/wagtailsearch/tests/test_queries.py
+++ b/wagtail/wagtailsearch/tests/test_queries.py
@@ -3,13 +3,13 @@ from __future__ import absolute_import, unicode_literals
 import datetime
 
 from django.core import management
-from django.test import TestCase
+from django.test import SimpleTestCase, TestCase
 from django.utils.six import StringIO
 
 from wagtail.contrib.wagtailsearchpromotions.models import SearchPromotion
 from wagtail.tests.utils import WagtailTestUtils
 from wagtail.wagtailsearch import models
-from wagtail.wagtailsearch.utils import normalise_query_string
+from wagtail.wagtailsearch.utils import normalise_query_string, separate_filters_from_query
 
 
 class TestHitCounter(TestCase):
@@ -176,3 +176,47 @@ class TestQueryChooserView(TestCase, WagtailTestUtils):
         for page in pages:
             response = self.get({'p': page})
             self.assertEqual(response.status_code, 200)
+
+
+class TestSeparateFiltersFromQuery(SimpleTestCase):
+    def test_only_query(self):
+        filters, query = separate_filters_from_query('hello world')
+
+        self.assertDictEqual(filters, {})
+        self.assertEquals(query, 'hello world')
+
+    def test_filter(self):
+        filters, query = separate_filters_from_query('author:foo')
+
+        self.assertDictEqual(filters, {'author': 'foo'})
+        self.assertEquals(query, '')
+
+    def test_filter_with_quotation_mark(self):
+        filters, query = separate_filters_from_query('author:"foo bar"')
+
+        self.assertDictEqual(filters, {'author': 'foo bar'})
+        self.assertEquals(query, '')
+
+    def test_filter_and_query(self):
+        filters, query = separate_filters_from_query('author:foo hello world')
+
+        self.assertDictEqual(filters, {'author': 'foo'})
+        self.assertEquals(query, 'hello world')
+
+    def test_filter_with_quotation_mark_and_query(self):
+        filters, query = separate_filters_from_query('author:"foo bar" hello world')
+
+        self.assertDictEqual(filters, {'author': 'foo bar'})
+        self.assertEquals(query, 'hello world')
+
+    def test_filter_with_unclosed_quotation_mark_and_query(self):
+        filters, query = separate_filters_from_query('author:"foo bar hello world')
+
+        self.assertDictEqual(filters, {})
+        self.assertEquals(query, 'author:"foo bar hello world')
+
+    def test_two_filters_and_query(self):
+        filters, query = separate_filters_from_query('author:"foo bar" hello world bar:beer')
+
+        self.assertDictEqual(filters, {'author': 'foo bar', 'bar': 'beer'})
+        self.assertEquals(query, 'hello world')
diff --git a/wagtail/wagtailsearch/utils.py b/wagtail/wagtailsearch/utils.py
index 1036829aa1..495a58833f 100644
--- a/wagtail/wagtailsearch/utils.py
+++ b/wagtail/wagtailsearch/utils.py
@@ -1,5 +1,6 @@
 from __future__ import absolute_import, unicode_literals
 
+import re
 import string
 
 MAX_QUERY_STRING_LENGTH = 255
@@ -19,3 +20,16 @@ def normalise_query_string(query_string):
     query_string = ' '.join(query_string.split())
 
     return query_string
+
+
+def separate_filters_from_query(query_string):
+    filters_regexp = r'(\w+):(\w+|".+")'
+
+    filters = {}
+    for match_object in re.finditer(filters_regexp, query_string):
+        key, value = match_object.groups()
+        filters[key] = value.strip("\"")
+
+    query_string = re.sub(filters_regexp, '', query_string).strip()
+
+    return filters, query_string