Implement exact match search and rewrite for operator ordering

Author: safwanrahman

readthedocs/search/faceted_search.py

@@ -1,4 +1,5 @@
 from elasticsearch_dsl import FacetedSearch, TermsFacet
+from elasticsearch_dsl.query import SimpleQueryString, Bool
 
 
 class RTDFacetedSearch(FacetedSearch):
@@ -29,3 +30,24 @@ class FileSearch(RTDFacetedSearch):
         'project': TermsFacet(field='project'),
         'version': TermsFacet(field='version')
     }
+
+    def query(self, search, query):
+        """
+        Add query part to ``search``.
+        """
+
+        if query:
+            all_queries = []
+
+            # Need to search for both 'AND' and 'OR' operations
+            # The score of AND should be higher as it comes first
+            for operator in ['AND', 'OR']:
+                query_string = SimpleQueryString(query=query, fields=self.fields,
+                                                 default_operator=operator)
+                all_queries.append(query_string)
+
+            # Run bool query with should, so it returns result where either of the query matches
+            bool_query = Bool(should=all_queries)
+            search = search.query(bool_query)
+
+        return search

readthedocs/search/tests/test_views.py

@@ -101,6 +101,23 @@ def test_file_search_case_insensitive(self, client, project, case):
         # Check the actual text is in the result, not the cased one
         assert query_text in result.text()
 
+    def test_file_search_exact_match(self, client, project):
+        """Check quoted query match exact phrase
+
+        Making a query with quoted text like *"foo bar"* should match
+        exactly *foo bar* phrase.
+        """
+
+        # `Github` word is present both in `kuma` and `pipeline` files
+        # But the phrase Github can is available only in kuma docs.
+        # So search with this phrase to check
+        query = r'"GitHub can"'
+
+        result, _ = self._get_search_result(url=self.url, client=client,
+                                            search_params={'q': query, 'type': 'file'})
+
+        assert len(result) == 1
+
     def test_page_search_not_return_removed_page(self, client, project):
         """Check removed page are not in the search index"""
         query = get_search_query_from_project_file(project_slug=project.slug)

tox.ini

@@ -15,7 +15,6 @@ setenv =
     DJANGO_SETTINGS_MODULE=readthedocs.settings.test
     LANG=C
     LC_CTYPE=C.UTF-8
-    DJANGO_SETTINGS_SKIP_LOCAL=True
 deps = -r{toxinidir}/requirements/testing.txt
 changedir = {toxinidir}/readthedocs
 commands =