From 13a1fca02b5cc1f73b154d4f2b694e7aa53c3a7c Mon Sep 17 00:00:00 2001
From: Pablo Tamarit <pablo.tamarit@cern.ch>
Date: Mon, 9 Dec 2024 09:10:07 +0100
Subject: [PATCH] subjects: improve search with CompositeSuggestQueryParser

---
 .../contrib/subjects/config.py                |  9 ++---
 invenio_vocabularies/services/querystr.py     |  5 +++
 .../subjects/test_subjects_resource.py        | 36 +++++++++++--------
 3 files changed, 31 insertions(+), 19 deletions(-)

diff --git a/invenio_vocabularies/contrib/subjects/config.py b/invenio_vocabularies/contrib/subjects/config.py
index 888362ff..eb7c3959 100644
--- a/invenio_vocabularies/contrib/subjects/config.py
+++ b/invenio_vocabularies/contrib/subjects/config.py
@@ -15,10 +15,12 @@
 from invenio_i18n import lazy_gettext as _
 from invenio_records_resources.services import SearchOptions
 from invenio_records_resources.services.records.components import DataComponent
+from invenio_records_resources.services.records.queryparser import (
+    CompositeSuggestQueryParser,
+)
 from werkzeug.local import LocalProxy
 
 from ...services.components import PIDComponent
-from ...services.querystr import FilteredSuggestQueryParser
 
 subject_schemes = LocalProxy(
     lambda: current_app.config["VOCABULARIES_SUBJECTS_SCHEMES"]
@@ -38,9 +40,8 @@
 class SubjectsSearchOptions(SearchOptions):
     """Search options."""
 
-    suggest_parser_cls = FilteredSuggestQueryParser.factory(
-        filter_field="scheme",
-        fields=[  # suggest fields
+    suggest_parser_cls = CompositeSuggestQueryParser.factory(
+        fields=[
             "subject^100",
             localized_title,
             "synonyms^20",
diff --git a/invenio_vocabularies/services/querystr.py b/invenio_vocabularies/services/querystr.py
index b84cf6be..1886f5f4 100644
--- a/invenio_vocabularies/services/querystr.py
+++ b/invenio_vocabularies/services/querystr.py
@@ -8,6 +8,7 @@
 
 """Querystring parsing."""
 
+import warnings
 from functools import partial
 
 from invenio_records_resources.services.records.params import SuggestQueryParser
@@ -20,6 +21,10 @@ class FilteredSuggestQueryParser(SuggestQueryParser):
     @classmethod
     def factory(cls, filter_field=None, **extra_params):
         """Create a prepared instance of the query parser."""
+        warnings.warn(
+            "FilteredSuggestQueryParser is deprecated, use SuggestQueryParser or CompositeSuggestQueryParser instead",
+            DeprecationWarning,
+        )
         return partial(cls, filter_field=filter_field, extra_params=extra_params)
 
     def __init__(self, identity=None, filter_field=None, extra_params=None):
diff --git a/tests/contrib/subjects/test_subjects_resource.py b/tests/contrib/subjects/test_subjects_resource.py
index 77abd01b..f1c88592 100644
--- a/tests/contrib/subjects/test_subjects_resource.py
+++ b/tests/contrib/subjects/test_subjects_resource.py
@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 #
-# Copyright (C) 2021 CERN.
+# Copyright (C) 2021-2024 CERN.
 # Copyright (C) 2021 Northwestern University.
 #
 # Invenio-Vocabularies is free software; you can redistribute it and/or
@@ -129,22 +129,28 @@ def example_subjects(app, db, search_clear, identity, service):
 
 
 def test_suggest(client, h, prefix, example_subjects):
-    """Test FilteredSuggestParam."""
-    # No filter
+    """Test the subjects query parser."""
+    # Full word
+    res = client.get(f"{prefix}?suggest=abdomen", headers=h)
+    assert res.json["hits"]["total"] == 4
+    # The less specific text appears at the end of the list
+    assert res.json["hits"]["hits"][3]["subject"] == "Abdomen, Acute"
+
+    # Incomplete word -> finds the same results
     res = client.get(f"{prefix}?suggest=abdo", headers=h)
     assert res.json["hits"]["total"] == 4
 
-    # Single filter
-    res = client.get(f"{prefix}?suggest=MeSH:abdo", headers=h)
-    assert res.status_code == 200
-    assert res.json["hits"]["total"] == 2
+    # Multiple words -> narrows down the results
+    res = client.get(f"{prefix}?suggest=abdomen%20acute", headers=h)
+    assert res.json["hits"]["total"] == 1
+    assert res.json["hits"]["hits"][0]["subject"] == "Abdomen, Acute"
 
-    # Multiple filters
-    res = client.get(f"{prefix}?suggest=MeSH,Other:abdo", headers=h)
-    assert res.status_code == 200
-    assert res.json["hits"]["total"] == 3
+    # Multiple words with incomplete last word -> still finds the same result
+    res = client.get(f"{prefix}?suggest=abdomen%20acu", headers=h)
+    assert res.json["hits"]["total"] == 1
+    assert res.json["hits"]["hits"][0]["subject"] == "Abdomen, Acute"
 
-    # Ignore non existing filter
-    res = client.get(f"{prefix}?suggest=MeSH,Foo:abdo", headers=h)
-    assert res.status_code == 200
-    assert res.json["hits"]["total"] == 2
+    # Multiple words with incomplete last word in other order -> still finds the same result
+    res = client.get(f"{prefix}?suggest=acute%20abdo", headers=h)
+    assert res.json["hits"]["total"] == 1
+    assert res.json["hits"]["hits"][0]["subject"] == "Abdomen, Acute"