search: improve with CompositeSuggestQueryParser

invenio_users_resources/services/users/config.py

@@ -26,10 +26,10 @@
     SortParam,
 )
 from invenio_records_resources.services.records.queryparser import (
+    CompositeSuggestQueryParser,
     FieldValueMapper,
     QueryParser,
     SearchFieldTransformer,
-    SuggestQueryParser,
 )
 from luqum.tree import Word
 
@@ -68,9 +68,16 @@ class UserSearchOptions(SearchOptions, SearchOptionsMixin):
     # The user search needs to be highly restricted to avoid leaking
     # account information, hence do not edit here unless you are
     # absolutely sure what you're doing.
-    suggest_parser_cls = SuggestQueryParser.factory(
+    suggest_parser_cls = CompositeSuggestQueryParser.factory(
         tree_transformer_cls=SearchFieldTransformer,
-        fields=["username^2", "email^2", "profile.full_name^3", "profile.affiliations"],
+        fields=[
+            "username.keyword^2",
+            "username^2",
+            "email.keyword^2",
+            "email^2",
+            "profile.full_name^3",
+            "profile.affiliations",
+        ],
         # Only public emails because hidden emails are stored in email_hidden field.
         allow_list=["username", "email"],
         mapping={
@@ -81,7 +88,6 @@ class UserSearchOptions(SearchOptions, SearchOptionsMixin):
             "name": "profile.full_name",
         },
         type="most_fields",  # https://www.elastic.co/guide/en/elasticsearch/reference/current/query-dsl-multi-match-query.html#multi-match-types
-        fuzziness="AUTO",  # https://www.elastic.co/guide/en/elasticsearch/reference/current/common-options.html#fuzziness
     )
 
     params_interpreters_cls = [

tests/conftest.py

@@ -132,7 +132,7 @@ def users_data():
     """Data for users."""
     return [
         {
-            "username": "pubres",
+            "username": "pub-res",
             "email": "[email protected]",
             "profile": {
                 "full_name": "Tim Smith",
@@ -322,7 +322,7 @@ def user_pub(users):
 @pytest.fixture(scope="module")
 def user_pubres(users):
     """User tjs (public/restricted)."""
-    return users["pubres"]
+    return users["pub-res"]
 
 
 @pytest.fixture(scope="module")

tests/resources/test_resources_users.py

@@ -63,7 +63,7 @@ def test_read_self_serialization(client, headers, users, user_pub):
 
 @pytest.mark.parametrize(
     "username,public_email",
-    [("pub", True), ("pubres", False)],
+    [("pub", True), ("pub-res", False)],
 )
 def test_read_anon_serialization(client, headers, users, username, public_email):
     """Read public user as anon."""

tests/services/users/test_service_users.py

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2022 CERN.
+# Copyright (C) 2022-2024 CERN.
 #
 # Invenio-Users-Resources is free software; you can redistribute it and/or
 # modify it under the terms of the MIT License; see LICENSE file for more
@@ -98,24 +98,36 @@
     assert res["hits"]["total"] == 0
 
 
+USERNAME_JOSE = ["pub"]
+USERNAME_TIM = ["pub-res"]
+USERNAME_BOTH = USERNAME_JOSE + USERNAME_TIM
+
+
+#
+# Read
 @pytest.mark.parametrize(
-    "query",
+    "query,expected_usernames",
     [
-        "CERN",
-        "Jose CERN",
-        "Jose AND CERN",
-        "Tim",
-        "Tim CERN",
-        "Jose",
-        "Jos",
-        "[email protected]",
-        "pub",
+        ("CERN", USERNAME_BOTH),
+        ("Jose", USERNAME_JOSE),
+        ("Jos", USERNAME_JOSE),
+        ("Jose CERN", USERNAME_JOSE),
+        ("Tim", USERNAME_TIM),
+        ("Tim CERN", USERNAME_TIM),
+        ("[email protected]", USERNAME_JOSE),
+        ("[email protected]", USERNAME_JOSE),
+        ("pub@inveniosoft", USERNAME_JOSE),
+        ("pub", USERNAME_BOTH),
+        ("pub-res", USERNAME_TIM),
+        ("re", USERNAME_TIM),
+        ("res", USERNAME_TIM),
     ],
 )
-def test_user_search_field(user_service, user_pub, query):
+def test_user_search_field(user_service, user_pub, query, expected_usernames):
     """Make sure certain fields ARE searchable."""
     res = user_service.search(user_pub.identity, suggest=query).to_dict()
-    assert res["hits"]["total"] > 0
+    usernames = [entry["username"] for entry in res["hits"]["hits"]]
+    assert sorted(usernames) == expected_usernames
 
 
 #
@@ -127,7 +139,7 @@
     assert res["username"] == "pub"
     assert res["email"] == user_pub.email
     res = user_service.read(anon_identity, user_pubres.id).to_dict()
-    assert res["username"] == "pubres"
+    assert res["username"] == "pub-res"
     assert "email" not in res
     pytest.raises(
         # TODO: Should be mapped to a 404
@@ -139,7 +151,7 @@
 
 
 @pytest.mark.parametrize(
-    "username,can_read", [("pub", True), ("pubres", True), ("res", False)]
+    "username,can_read", [("pub", True), ("pub-res", True), ("res", False)]
 )
 def test_read_avatar_with_anon(user_service, anon_identity, users, username, can_read):
     """Anonymous users can read avatar single *public* user."""
@@ -160,7 +172,7 @@
     "username",
     [
         "pub",
-        "pubres",
+        "pub-res",
         "res",
     ],
 )