Complete live Mol register validation

scripts/provision_buildings_addresses_register.py

@@ -58,7 +58,7 @@ SOURCE_NAME = "digitaal_vlaanderen_buildings_addresses_register"
 DEFAULT_PROJECT_NAME = "Kempen Regional Workbench"
 DEFAULT_AREA_NAME = "Gemeente Mol - officiele grens"
 DEFAULT_API_URL = "http://127.0.0.1:8000"
-DEFAULT_OUTPUT_ROOT = Path("/app/storage/operator-evidence/buildings-addresses-register/mol")
+DEFAULT_OUTPUT_ROOT = Path("/app/storage/operator-evidence/buildings-addresses-register")
 DEFAULT_PAGE_LIMIT = 1000
 DEFAULT_MAX_BUILDINGS = 150_000
 DEFAULT_MAX_UNITS = 200_000
@@ -247,6 +247,22 @@ def normalize_area_name(value: str) -> str:
     return unicodedata.normalize("NFKD", value).encode("ascii", "ignore").decode("ascii").casefold()
 
 
+def area_storage_key(value: str) -> str:
+    normalized = normalize_area_name(value)
+    if normalized.startswith("gemeente "):
+        municipality = normalized.removeprefix("gemeente ").split(" - ", maxsplit=1)[0].strip()
+        if municipality:
+            return "-".join(municipality.split())
+    key = "-".join(part for part in compact_key(value).split() if part)
+    return key or "area"
+
+
+def boundary_checksum(boundary_wgs84) -> str:
+    return sha256_bytes(
+        json.dumps(mapping(boundary_wgs84), sort_keys=True).encode("utf-8")
+    )
+
+
 def locate_workspace(
     session: requests.Session,
     base_url: str,
@@ -763,7 +779,13 @@ def build_output_features(
     }
 
 
-def reusable_snapshot(snapshot_dir: Path) -> tuple[Path, Path, dict[str, Any]] | None:
+def reusable_snapshot(
+    snapshot_dir: Path,
+    *,
+    area_name: str,
+    observed_date: date,
+    expected_boundary_checksum: str,
+) -> tuple[Path, Path, dict[str, Any]] | None:
     artifact_path = snapshot_dir / "buildings_addresses_register.geojson"
     manifest_path = snapshot_dir / "buildings_addresses_register.manifest.json"
     if not artifact_path.is_file() or not manifest_path.is_file():
@@ -774,6 +796,10 @@ def reusable_snapshot(snapshot_dir: Path) -> tuple[Path, Path, dict[str, Any]] |
         return None
     if manifest.get("schema_version") != SCHEMA_VERSION:
         return None
+    if manifest.get("coverage_area") != area_name or manifest.get("observed_at") != observed_date.isoformat():
+        return None
+    if manifest.get("boundary_sha256") != expected_boundary_checksum:
+        return None
     if manifest.get("artifact_sha256") != sha256_file(artifact_path):
         return None
     for collection in manifest.get("collections") or []:
@@ -804,8 +830,14 @@ def prepare_snapshot(
     timeout: int,
     force: bool,
 ) -> tuple[Path, Path, dict[str, Any]]:
+    expected_boundary_checksum = boundary_checksum(boundary_wgs84)
     if not force:
-        reusable = reusable_snapshot(snapshot_dir)
+        reusable = reusable_snapshot(
+            snapshot_dir,
+            area_name=area_name,
+            observed_date=observed_date,
+            expected_boundary_checksum=expected_boundary_checksum,
+        )
         if reusable:
             return reusable
     snapshot_dir.mkdir(parents=True, exist_ok=True)
@@ -892,7 +924,7 @@ def prepare_snapshot(
         "crs_source": "EPSG:4326",
         "crs_clip": "EPSG:31370",
         "crs_persisted": "EPSG:4326",
-        "boundary_sha256": sha256_bytes(json.dumps(mapping(boundary_wgs84), sort_keys=True).encode("utf-8")),
+        "boundary_sha256": expected_boundary_checksum,
         "boundary_bbox_wgs84": list(boundary_wgs84.bounds),
         "collections": [buildings_source, units_source, addresses_source],
         "building_filter": building_filter,
@@ -1113,7 +1145,7 @@ def main() -> int:
         print(json.dumps({"status": "error", "message": "Invalid page or feature safety limits"}), file=sys.stderr)
         return 2
     base_url = args.base_url.rstrip("/")
-    snapshot_dir = args.output_root / args.observed_date.isoformat()
+    snapshot_dir = args.output_root / area_storage_key(args.area_name) / args.observed_date.isoformat()
     try:
         with requests.Session() as api_session:
             project_id, area_id, boundary, datasets = locate_workspace(