Optimize assistant context by requested themes

2026-07-16 03:48:36 +02:00
parent a19bff9d14
commit 1e23d30eaf
5 changed files with 206 additions and 4 deletions
@@ -23,6 +23,7 @@ from app.services.geo_assistant_service import GeoAssistantService
 from app.services.thematic_raster_acquisition_service import ThematicRasterAcquisitionService
 from app.services.thematic_raster_analysis_service import ThematicRasterAnalysisService
 from app.services.dataset_service import DatasetService
+from app.services.vector_feature_service import VectorFeatureService


 ROOT = Path(__file__).resolve().parents[2]
@@ -277,6 +278,104 @@ def test_assistant_context_receives_persisted_thematic_metrics(tmp_path) -> None
    assert context["rules"]["thematic_policy_rasters_available"] is True


+def test_assistant_context_skips_unrequested_expensive_themes(monkeypatch) -> None:
+    project_id = uuid4()
+    soil_id, agriculture_id = uuid4(), uuid4()
+    population_id, space_id = uuid4(), uuid4()
+    project = Project(id=project_id, name="Kempen", region="Kempen")
+    datasets = [
+        Dataset(
+            id=soil_id,
+            project_id=project_id,
+            name="soil.geojson",
+            dataset_type="vector",
+            source="official",
+            source_name="dov",
+            source_metadata={"theme": "soil"},
+            status="ready",
+        ),
+        Dataset(
+            id=agriculture_id,
+            project_id=project_id,
+            name="agriculture.geojson",
+            dataset_type="vector",
+            source="official",
+            source_name="lv",
+            source_metadata={"theme": "agriculture"},
+            status="ready",
+        ),
+        Dataset(
+            id=population_id,
+            project_id=project_id,
+            name="population.tif",
+            dataset_type="raster",
+            source="official",
+            source_name=ThematicRasterAcquisitionService.PROVIDER,
+            source_metadata={"product_key": "population_density_2019"},
+            status="ready",
+        ),
+        Dataset(
+            id=space_id,
+            project_id=project_id,
+            name="space.tif",
+            dataset_type="raster",
+            source="official",
+            source_name=ThematicRasterAcquisitionService.PROVIDER,
+            source_metadata={"product_key": "space_occupation_2025"},
+            status="ready",
+        ),
+    ]
+    db = FakeSession({(Project, project_id): project}, query_result=datasets)
+    summarized: list = []
+    analyzed: list = []
+
+    def summarize(_db, *, dataset, **_kwargs):
+        summarized.append(dataset.id)
+        return {
+            "metric_label": "Gekarteerde bodemoppervlakte",
+            "metric_value": 12.5,
+            "metric_unit": "ha",
+            "is_estimate": False,
+            "warning": "Historische bodemkaart",
+        }
+
+    def analyze(_db, _project_id, dataset_id, _payload, **_kwargs):
+        analyzed.append(dataset_id)
+        return {
+            "theme": "population",
+            "summary": {
+                "metrics": [
+                    {
+                        "metric_label": "Geraamd aantal inwoners (2019)",
+                        "metric_value": 100.0,
+                        "metric_unit": "inwoners",
+                        "is_estimate": True,
+                    }
+                ]
+            },
+            "unsupported_metrics": ["current_population"],
+            "limitation_message": "Rasterraming",
+        }
+
+    monkeypatch.setattr(VectorFeatureService, "summarize_features_by_bbox", summarize)
+    monkeypatch.setattr(ThematicRasterAnalysisService, "analyze", analyze)
+
+    context, metrics, _series, dataset_ids, _warnings, _scope = GeoAssistantService(Settings(_env_file=None))._build_context(
+        db,
+        project_id=project_id,
+        payload=AssistantQueryRequest(
+            question="Hoeveel inwoners zijn er en welke bodemtypes komen voor?",
+            bbox=payload("population_density_2019", side_m=200.0).bbox,
+        ),
+    )
+
+    assert summarized == [soil_id]
+    assert analyzed == [population_id]
+    assert {metric.theme for metric in metrics} == {"soil", "population"}
+    assert set(dataset_ids) == {soil_id, population_id}
+    assert context["scope"]["requested_themes"] == ["population", "soil"]
+
+
 def test_index_renderer_returns_browser_png(tmp_path) -> None:
    project_id, dataset_id = uuid4(), uuid4()
    values = np.linspace(0.1, 4.0, 100, dtype="float32").reshape((10, 10))