Make quoted queries behave as described in the API documentation (return exact matches only) (#1012)

* Fix quoted audio search example escaping

* Make quoted queries behave as described in API documentation

* Undo change breaking title match boosting

* Fix and future proof tests against additional test data

Author: sarayourfriend

api/catalog/api/controllers/search_controller.py

@@ -340,18 +340,25 @@ def search(
     search_fields = ["tags.name", "title", "description"]
     if "q" in search_params.data:
         query = _quote_escape(search_params.data["q"])
+        base_query_kwargs = {
+            "query": query,
+            "fields": search_fields,
+            "default_operator": "AND",
+        }
+
+        if '"' in query:
+            base_query_kwargs["quote_field_suffix"] = ".exact"
+
         s = s.query(
             "simple_query_string",
-            query=query,
-            fields=search_fields,
-            default_operator="AND",
+            **base_query_kwargs,
         )
-        # Boost exact matches
+        # Boost exact matches on the title
         quotes_stripped = query.replace('"', "")
         exact_match_boost = Q(
             "simple_query_string",
             fields=["title"],
-            query=f'"{quotes_stripped}"',
+            query=f"{quotes_stripped}",
             boost=10000,
         )
         s = search_client.query(Q("bool", must=s.query, should=exact_match_boost))

api/catalog/api/examples/audio_requests.py

@@ -10,7 +10,7 @@
 syntax_examples = {
     "using single query parameter": "test",
     "using multiple query parameters": "test&license=pdm,by&categories=illustration&page_size=1&page=1",  # noqa: E501
-    "that is an exact match of Giacomo Puccini": '"Giacomo Puccini"',
+    "that is an exact match of Giacomo Puccini": r"%22Giacomo%20Puccini%22",
     "related to both dog and cat": "dog+cat",
     "related to dog or cat, but not necessarily both": "dog|cat",
     "related to dog but won't include results related to 'pug'": "dog -pug",

api/catalog/api/examples/image_requests.py

@@ -10,7 +10,7 @@
 syntax_examples = {
     "using single query parameter": "test",
     "using multiple query parameters": "test&license=pdm,by&categories=illustration&page_size=1&page=1",  # noqa: E501
-    "that are an exact match of Claude Monet": '"Claude Monet"',
+    "that are an exact match of Claude Monet": "%22Claude%20Monet%22",
     "related to both dog and cat": "dog+cat",
     "related to dog or cat, but not necessarily both": "dog|cat",
     "related to dog but won't include results related to 'pug'": "dog -pug",

api/test/audio_integration_test.py

@@ -15,6 +15,7 @@
     search_by_category,
     search_consistency,
     search_quotes,
+    search_quotes_exact,
     search_source_and_excluded,
     search_special_chars,
     stats,
@@ -101,6 +102,11 @@ def test_search_quotes():
     search_quotes("audio", "love")
 
 
+def test_search_quotes_exact():
+    # ``water running`` returns different results when quoted vs unquoted
+    search_quotes_exact("audio", "water running")
+
+
 def test_search_with_special_characters():
     search_special_chars("audio", "love")
 

api/test/image_integration_test.py

@@ -15,6 +15,7 @@
     search_all_excluded,
     search_consistency,
     search_quotes,
+    search_quotes_exact,
     search_source_and_excluded,
     search_special_chars,
     stats,
@@ -53,6 +54,11 @@ def test_search_quotes():
     search_quotes("images", "dog")
 
 
+def test_search_quotes_exact():
+    # ``bird perched`` returns different results when quoted vs unquoted
+    search_quotes_exact("images", "bird perched")
+
+
 def test_search_with_special_characters():
     search_special_chars("images", "dog")
 

api/test/media_integration.py

@@ -45,6 +45,25 @@ def search_quotes(media_path, q="test"):
     assert response.status_code == 200
 
 
+def search_quotes_exact(media_path, q):
+    """Only returns exact matches for the given query"""
+    unquoted_response = requests.get(f"{API_URL}/v1/{media_path}?q={q}", verify=False)
+    assert unquoted_response.status_code == 200
+    unquoted_result_count = unquoted_response.json()["result_count"]
+    assert unquoted_result_count > 0
+
+    quoted_response = requests.get(f'{API_URL}/v1/{media_path}?q="{q}"', verify=False)
+    assert quoted_response.status_code == 200
+    quoted_result_count = quoted_response.json()["result_count"]
+    assert quoted_result_count > 0
+
+    # The rationale here is that the unquoted results will match more records due
+    # to the query being overall less strict. Quoting the query will make it more
+    # strict causing it to return fewer results.
+    # Above we check that the results are not 0 to confirm that we do still get results back.
+    assert quoted_result_count < unquoted_result_count
+
+
 def search_special_chars(media_path, q="test"):
     """Returns a response when query includes special characters."""
     response = requests.get(f"{API_URL}/v1/{media_path}?q={q}!", verify=False)