Support validated legacy portfolio composition

scripts/combine_belgium_building_portfolio_specs.py

@@ -28,15 +28,18 @@ def combine_specs(paths: list[Path]) -> dict[str, Any]:
     sources = []
     for path in paths:
         payload = json.loads(path.read_text(encoding="utf-8-sig"))
-        if payload.get("status") != "complete":
+        status = payload.get("status")
+        legacy_unstated = status is None and payload.get("schema_version") == 1 and bool(payload.get("samples"))
+        if status != "complete" and not legacy_unstated:
             raise ValueError(f"Portfolio spec is not complete: {path}")
-        current_side = float(payload["side_m"])
-        current_resolution = float(payload["resolution_m"])
-        if side_m is not None and current_side != side_m:
+        current_side = float(payload["side_m"]) if payload.get("side_m") is not None else None
+        current_resolution = float(payload["resolution_m"]) if payload.get("resolution_m") is not None else None
+        if side_m is not None and current_side is not None and current_side != side_m:
             raise ValueError(f"Portfolio side_m differs: {path}")
-        if resolution_m is not None and current_resolution != resolution_m:
+        if resolution_m is not None and current_resolution is not None and current_resolution != resolution_m:
             raise ValueError(f"Portfolio resolution_m differs: {path}")
-        side_m, resolution_m = current_side, current_resolution
+        side_m = current_side if current_side is not None else side_m
+        resolution_m = current_resolution if current_resolution is not None else resolution_m
         for sample in payload.get("samples") or []:
             slug = str(sample.get("sample_slug") or sample.get("slug") or "")
             if not slug:
@@ -48,7 +51,14 @@ def combine_specs(paths: list[Path]) -> dict[str, Any]:
             normalized["sample_slug"] = slug
             normalized.pop("slug", None)
             samples.append(normalized)
-        sources.append({"path": str(path), "sha256": sha256(path), "sample_count": len(payload.get("samples") or [])})
+        sources.append({
+            "path": str(path),
+            "sha256": sha256(path),
+            "sample_count": len(payload.get("samples") or []),
+            "source_status": status or "legacy_unstated",
+        })
+    if side_m is None or resolution_m is None:
+        raise ValueError("Combined portfolio dimensions are unknown")
     return {
         "schema_version": 2,
         "status": "complete",

tests/test_combine_belgium_building_portfolio_specs.py

@@ -36,3 +36,18 @@ def test_combine_specs_rejects_mismatched_resolution(tmp_path: Path) -> None:
 def test_combine_specs_rejects_incomplete_source(tmp_path: Path) -> None:
     with pytest.raises(ValueError, match="not complete"):
         combine_specs([write_spec(tmp_path / "one.json", "one", status="in_progress")])
+
+
+def test_combine_specs_accepts_populated_legacy_v1_when_dimensions_come_from_complete_spec(tmp_path: Path) -> None:
+    legacy = tmp_path / "legacy.json"
+    legacy.write_text(json.dumps({"schema_version": 1, "samples": [{"sample_slug": "legacy"}]}), encoding="utf-8")
+    payload = combine_specs([legacy, write_spec(tmp_path / "current.json", "current")])
+    assert payload["sample_count"] == 2
+    assert payload["source_specs"][0]["source_status"] == "legacy_unstated"
+
+
+def test_combine_specs_rejects_empty_legacy_v1(tmp_path: Path) -> None:
+    legacy = tmp_path / "legacy.json"
+    legacy.write_text(json.dumps({"schema_version": 1, "samples": []}), encoding="utf-8")
+    with pytest.raises(ValueError, match="not complete"):
+        combine_specs([legacy, write_spec(tmp_path / "current.json", "current")])