From df015e10e09ff2aa110e4b83ea792064cbf4a820 Mon Sep 17 00:00:00 2001
From: Codex <codex@geointel.local>
Date: Wed, 15 Jul 2026 23:49:30 +0200
Subject: [PATCH] Align historical metrics with polygon semantics

---
 .../app/services/vector_feature_service.py    | 13 +++++++++++-
 ...st_sprint201_semantic_selection_metrics.py | 20 +++++++++++++++++++
 ...t_sprint209_regional_historical_landuse.py |  3 +++
 .../provision_regional_historical_landuse.py  |  4 +++-
 4 files changed, 38 insertions(+), 2 deletions(-)

diff --git a/backend/app/services/vector_feature_service.py b/backend/app/services/vector_feature_service.py
index 1fe1f498..626f5e6d 100644
--- a/backend/app/services/vector_feature_service.py
+++ b/backend/app/services/vector_feature_service.py
@@ -30,6 +30,12 @@ FULL_AREA_CLIPPED_OPERATOR_TOOLS = {
     "provision_buildings_addresses_register.py",
 }
 
+SEMANTIC_METRICS_DISABLED_OPERATOR_TOOLS = {
+    # Historical land-use themes are polygon map classes. Generic live-theme
+    # line metrics (road/watercourse length) would therefore be meaningless.
+    "provision_regional_historical_landuse.py",
+}
+
 
 SEMANTIC_SELECTION_METRICS: dict[str, tuple[dict[str, Any], ...]] = {
     "buildings": (
@@ -379,9 +385,14 @@ class VectorFeatureService:
             "is_estimate": bool(config.get("is_estimate", False)),
             **({"property": config.get("property")} if config.get("property") else {}),
         }
+        provenance = dataset.provenance_metadata if isinstance(dataset.provenance_metadata, dict) else {}
+        semantic_metrics_disabled = (
+            source_metadata.get("semantic_metrics") is False
+            or provenance.get("operator_tool") in SEMANTIC_METRICS_DISABLED_OPERATOR_TOOLS
+        )
         semantic_metrics = (
             []
-            if source_metadata.get("semantic_metrics") is False
+            if semantic_metrics_disabled
             else [dict(metric) for metric in SEMANTIC_SELECTION_METRICS.get(theme or "", ())]
         )
         primary_config = configured_metric
diff --git a/backend/tests/test_sprint201_semantic_selection_metrics.py b/backend/tests/test_sprint201_semantic_selection_metrics.py
index dce17eb5..2aa22054 100644
--- a/backend/tests/test_sprint201_semantic_selection_metrics.py
+++ b/backend/tests/test_sprint201_semantic_selection_metrics.py
@@ -144,6 +144,26 @@ def test_station_measurement_uses_numeric_mean_without_area_extrapolation() -> N
     assert result["warning"] == "Puntmeting; geen gebiedsdekkend watervolume."
 
 
+def test_regional_historical_polygons_do_not_emit_irrelevant_line_metrics() -> None:
+    dataset = themed_dataset("water", method="intersection_area")
+    dataset.source_metadata["selection_aggregation"].update(
+        {"metric_key": "water_area", "label": "Historische wateroppervlakte", "unit": "ha"}
+    )
+    dataset.provenance_metadata = {"operator_tool": "provision_regional_historical_landuse.py"}
+
+    result = VectorFeatureService.summarize_features_by_bbox(
+        SequenceScalarSession([52_500.0]),
+        dataset=dataset,
+        bbox=BBOX,
+        total_feature_count=23,
+    )
+
+    assert [(item["metric_key"], item["metric_unit"]) for item in result["metrics"]] == [
+        ("water_area", "ha"),
+        ("feature_count", "objecten"),
+    ]
+
+
 def test_future_regional_imports_persist_semantic_aggregation_configuration() -> None:
     buildings = (ROOT / "scripts/provision_regional_grb_buildings.py").read_text(encoding="utf-8")
     context = (ROOT / "scripts/provision_regional_grb_context.py").read_text(encoding="utf-8")
diff --git a/backend/tests/test_sprint209_regional_historical_landuse.py b/backend/tests/test_sprint209_regional_historical_landuse.py
index 94aa5227..d72caafc 100644
--- a/backend/tests/test_sprint209_regional_historical_landuse.py
+++ b/backend/tests/test_sprint209_regional_historical_landuse.py
@@ -228,6 +228,9 @@ def test_upload_contract_is_regional_temporal_and_partition_audited(tmp_path: Pa
     assert source_metadata["partitioned_source_audit"] is True
     assert source_metadata["geometry_clipped_to_area"] is True
     assert source_metadata["identity_stable"] is False
+    assert source_metadata["semantic_metrics"] is False
+    assert source_metadata["selection_aggregation"]["metric_key"] == "roads_area"
+    assert source_metadata["selection_aggregation"]["label"] == "Oppervlakte historische wegen"
     assert provenance["partition_count"] == 28
     assert provenance["raw_source_responses_retained"] is True
     assert provenance["geometry_clipped_to_area"] is True
diff --git a/scripts/provision_regional_historical_landuse.py b/scripts/provision_regional_historical_landuse.py
index e58f8e23..1b727f3e 100644
--- a/scripts/provision_regional_historical_landuse.py
+++ b/scripts/provision_regional_historical_landuse.py
@@ -566,10 +566,12 @@ def upload_snapshot(
         "attribution": ATTRIBUTION,
         "source_catalog_url": SOURCE_CATALOG_URL,
         "identity_stable": False,
+        "semantic_metrics": False,
         "geometry_simplification_tolerance_degrees": simplify_tolerance_degrees,
         "selection_aggregation": {
+            "metric_key": f"{definition.key}_area",
             "method": "intersection_area",
-            "label": "Oppervlakte",
+            "label": f"Oppervlakte {definition.label.lower()}",
             "unit": "ha",
             "is_estimate": False,
             "warning": (