add filtering by theme in api

app/search/mappings.py

@@ -51,6 +51,12 @@
             "type": "text",
             "analyzer": TEXT_ANALYZER,
             "search_analyzer": TEXT_ANALYZER,
+            "fields": {
+                "normalized": {
+                    "type": "keyword",
+                    "normalizer": KEYWORD_NORMALIZER,  # simple lowercase normalizer
+                },
+            },
         },
         "identifier": {
             "type": "text",

app/search/queries/filters/__init__.py

@@ -14,6 +14,7 @@
 )
 from app.search.queries.filters.publisher import PUBLISHER_FILTER
 from app.search.queries.filters.spatial_data import SPATIAL_DATA_FILTER
+from app.search.queries.filters.theme import THEME_FILTER
 
 __all__ = [
     "API_CONTEXT",
@@ -32,4 +33,5 @@
     PUBLISHER_FILTER,
     SPATIAL_DATA_FILTER,
     COLLECTION_FILTER,
+    THEME_FILTER,
 )

app/search/queries/filters/theme.py

@@ -0,0 +1,23 @@
+from __future__ import annotations
+
+from app.search.queries.filters.base import (
+    API_CONTEXT,
+    ApiQueryParam,
+    FilterDefinition,
+    get_list,
+)
+
+
+def _clause(criteria, values: list[str]) -> list[dict]:
+    return [{"term": {"theme.normalized": theme.lower()}} for theme in values]
+
+
+THEME_FILTER = FilterDefinition(
+    name="theme",
+    query_params=("theme",),
+    parse_contexts=(API_CONTEXT,),
+    api_query_params=(ApiQueryParam("theme", repeated=True),),
+    parse=lambda args: get_list(args, "theme"),
+    to_query_pairs=lambda values: [("theme", value) for value in values],
+    clause_builder=_clause,
+)

tests/unit/test_routes.py

@@ -270,6 +270,18 @@ def test_search_api_by_org_slug(interface_with_dataset, db_client):
         assert len(response.json["results"]) == 0
 
 
+def test_search_api_by_theme(interface_with_dataset, db_client):
+    with patch("app.routes.interface", interface_with_dataset):
+        response = db_client.get(
+            "/search", query_string={"theme": ["climate", "environment"]}
+        )
+        assert len(response.json["results"]) == 1
+
+        # non-existent theme
+        response = db_client.get("/search", query_string={"theme": ["nonexistent"]})
+        assert len(response.json["results"]) == 0
+
+
 def test_index_page_filters_by_org_slug(db_client):
     mock_interface = Mock()
     mock_org = type(