resource-control/tools/validate.py

#!/usr/bin/env python3
"""Dependency-free validation for resource-control JSON declarations."""

import json
from pathlib import Path

from optimization import validate_case
from portfolio import validate_record
from portfolio_report import build as build_portfolio_report


def load(path: str) -> dict:
    return json.loads(Path(path).read_text())


def main() -> int:
    demand = load("data/demand/platform-audit-storage.json")
    providers = load("data/providers/object-storage.json")
    schema = load("schemas/resource-inventory.schema.json")
    observation_schema = load("schemas/monthly-resource-observation.schema.json")
    planning_schema = load("schemas/planning-evidence.schema.json")
    control_schema = load("schemas/resource-control-cycle.schema.json")
    forecasts = [load(str(path)) for path in Path("data/forecasts").glob("*.json")]
    resource_paths = list(Path("data/resources").glob("*.json"))
    resource_paths += list(Path("examples/portfolio").glob("*.json"))
    resources = [load(str(path)) for path in resource_paths]
    control_records = [load(str(path)) for path in Path("examples/control-cycle").glob("*.json")]
    assert demand["schema_version"] == providers["schema_version"] == "0.1"
    assert demand["retention_days"] >= 30
    assert set(demand["scenarios"]) == {"low", "base", "high"}
    assert len({p["id"] for p in providers["providers"]}) == len(providers["providers"])
    assert {"Host Europe", "Scaleway", "Hetzner", "AWS", "Microsoft Azure", "Google Cloud", "STACKIT"} <= {p["provider"] for p in providers["providers"]}
    assert schema["$schema"].endswith("2020-12/schema")
    assert observation_schema["$schema"].endswith("2020-12/schema")
    assert planning_schema["$schema"].endswith("2020-12/schema")
    assert control_schema["$schema"].endswith("2020-12/schema")
    assert len(planning_schema["oneOf"]) == 5
    for forecast in forecasts:
        assert forecast["record_type"] == "forecast"
        assert len({row["period"] for row in forecast["rows"]}) == len(forecast["rows"])
        assert all(row["total_eur"] == round(row["infrastructure_eur"] + row["internal_labor_eur"], 2) for row in forecast["rows"])
    assert schema["properties"]["schema_version"]["const"] == "0.2"
    required = set(schema["required"])
    for resource in resources:
        assert not required - resource.keys(), f"missing fields: {required - resource.keys()}"
        validate_record(resource)
    operational_records = [load(str(path)) for path in Path("data/control-cycle").glob("*.json")]
    control_records += operational_records
    record_ids = {record["record_id"] for record in control_records}
    assert len(record_ids) == len(control_records)
    assert {record["resource_class"] for record in control_records} >= {"storage", "cluster_compute", "shared_platform_service"}
    for record in control_records:
        assert record["schema_version"] in {"0.1", "0.2"}
        assert record["resource_id"].startswith("resource:")
        costs = record["costs"]
        components = [costs["infrastructure"], costs["internal_labor"], costs["external_labor"]]
        if any(component is None for component in components):
            # An unknown component makes the total unknown; it is never the sum
            # of the parts that happen to be known.
            assert record["schema_version"] == "0.2", "null costs require schema 0.2"
            assert costs["total"] is None
        else:
            assert costs["total"] == round(sum(components), 2)
        if record["record_type"] == "actual":
            assert record["forecast_ref"] in record_ids
    # Operational records assert real facts and must cite the authoritative
    # repository evidence they came from.
    for record in operational_records:
        assert record["evidence"], f"{record['record_id']} cites no evidence"
        assert record["uncertainty"]["notes"], f"{record['record_id']} states no uncertainty"
    case_schema = load("schemas/optimization-case.schema.json")
    assert case_schema["$schema"].endswith("2020-12/schema")
    assert case_schema["properties"]["schema_version"]["const"] == "0.1"
    cases = [load(str(path)) for path in Path("data/optimization").glob("*.json")]
    assert len({case["case_id"] for case in cases}) == len(cases)
    for case in cases:
        assert not set(case_schema["required"]) - case.keys()
        validate_case(case)
    # The optimization process must be validated on the backup case and on at
    # least one non-storage portfolio candidate (RESOURCE-WP-0003-T06).
    case_resources = {rid for case in cases for rid in case["resource_ids"]}
    assert "resource:platform:audit-storage" in case_resources
    assert case_resources - {"resource:platform:audit-storage"}
    report = build_portfolio_report(Path("."))
    assert report["resource_count"] == len(resource_paths) - len(list(Path("examples/portfolio").glob("*.json")))
    assert report["cost"]["known_monthly_spend_eur"] is None
    assert report["next_actions"]
    print("resource-control declarations: valid")
    return 0


if __name__ == "__main__":
    raise SystemExit(main())