NODEDC_PLATFORM/services/external-data-plane/src/data-product-policy.mjs

import { isBoundedGeoJsonGeometry, isBoundedTelemetryReadings } from "@nodedc/external-provider-contract/data-plane";

const IDENTIFIER = /^[a-z][a-z0-9._:-]{2,127}$/;
const SEMVER = /^\d+\.\d+\.\d+(?:[-+][a-z0-9.-]+)?$/i;
const DELIVERY_MODES = new Set(["snapshot", "snapshot+patch", "query"]);
const HISTORY_MODES = new Set(["none", "all", "sampled"]);
const FIELD_CONTRACT_TYPES = new Set(["string", "number", "boolean", "string_array", "telemetry_readings", "point", "geometry"]);
const DEFINITION_KEYS = new Set([
  "id", "version", "ontologyRevision", "deliveryMode", "semanticTypes", "fields", "fieldContracts", "history",
]);
const HISTORY_KEYS = new Set(["mode", "intervalMs", "retentionDays", "strategy"]);
const FIELD_CONTRACT_KEYS = new Set(["type", "required", "enum", "minimum", "maximum"]);

export function normalizeDataProductDefinition(value) {
  if (!isPlainObject(value) || !hasOnlyKeys(value, DEFINITION_KEYS)) {
    throw policyError("data_product_definition_invalid");
  }
  const id = identifier(value.id);
  const version = string(value.version);
  const ontologyRevision = identifier(value.ontologyRevision);
  const deliveryMode = string(value.deliveryMode);
  const semanticTypes = identifierSet(
    value.semanticTypes,
    "data_product_definition_semantic_types_invalid",
    "data_product_definition_semantic_types_duplicate",
  );
  const fields = identifierSet(
    value.fields,
    "data_product_definition_fields_invalid",
    "data_product_definition_fields_duplicate",
  );
  if (!id || !SEMVER.test(version) || !ontologyRevision || !DELIVERY_MODES.has(deliveryMode)) {
    throw policyError("data_product_definition_identity_invalid");
  }
  if (!semanticTypes.length || !fields.length) throw policyError("data_product_definition_shape_invalid");

  const fieldContracts = normalizeFieldContracts(value.fieldContracts, fields);
  const history = normalizeHistoryPolicy(value.history);
  return Object.freeze({ id, version, ontologyRevision, deliveryMode, semanticTypes, fields, fieldContracts, history });
}

export function normalizeFieldContracts(value, fields) {
  if (value === undefined) return Object.freeze({});
  if (!isPlainObject(value)) throw policyError("data_product_field_contracts_invalid");
  const names = Object.keys(value);
  if (names.length === 0) return Object.freeze({});
  if (names.length !== fields.length || fields.some((field) => !Object.hasOwn(value, field))) {
    throw policyError("data_product_field_contracts_must_match_fields");
  }

  const normalized = {};
  for (const field of fields) normalized[field] = normalizeFieldContract(value[field]);
  return Object.freeze(normalized);
}

function normalizeFieldContract(value) {
  if (!isPlainObject(value) || !hasOnlyKeys(value, FIELD_CONTRACT_KEYS)) {
    throw policyError("data_product_field_contract_invalid");
  }
  const type = string(value.type);
  if (!FIELD_CONTRACT_TYPES.has(type) || typeof value.required !== "boolean") {
    throw policyError("data_product_field_contract_shape_invalid");
  }
  const contract = { type, required: value.required };
  if (value.enum !== undefined) {
    if (!Array.isArray(value.enum) || value.enum.length === 0
      || new Set(value.enum.map(stableLiteral)).size !== value.enum.length
      || new Set(["point", "geometry", "string_array", "telemetry_readings"]).has(type)
      || value.enum.some((item) => !fieldContractValueMatchesType(item, type))) {
      throw policyError("data_product_field_contract_enum_invalid");
    }
    contract.enum = Object.freeze([...value.enum].sort((left, right) => stableLiteral(left).localeCompare(stableLiteral(right))));
  }
  if (value.minimum !== undefined || value.maximum !== undefined) {
    if (type !== "number"
      || (value.minimum !== undefined && !Number.isFinite(value.minimum))
      || (value.maximum !== undefined && !Number.isFinite(value.maximum))
      || (Number.isFinite(value.minimum) && Number.isFinite(value.maximum) && value.minimum > value.maximum)) {
      throw policyError("data_product_field_contract_range_invalid");
    }
    if (value.minimum !== undefined) contract.minimum = value.minimum;
    if (value.maximum !== undefined) contract.maximum = value.maximum;
    if (contract.enum?.some((item) => (
      (contract.minimum !== undefined && item < contract.minimum)
      || (contract.maximum !== undefined && item > contract.maximum)
    ))) throw policyError("data_product_field_contract_enum_out_of_range");
  }
  return Object.freeze(contract);
}

export function normalizeHistoryPolicy(value = { mode: "none" }) {
  if (!isPlainObject(value) || !hasOnlyKeys(value, HISTORY_KEYS)) throw policyError("history_policy_invalid");
  const mode = string(value.mode);
  if (!HISTORY_MODES.has(mode)) throw policyError("history_policy_mode_invalid");
  const retentionDays = integer(value.retentionDays, mode === "none" ? 1 : 90, 1, 3650);
  if (mode === "none") {
    if (value.intervalMs !== undefined || value.strategy !== undefined) throw policyError("history_policy_none_has_sampling_fields");
    return Object.freeze({ mode, retentionDays });
  }
  if (mode === "all") {
    if (value.intervalMs !== undefined || value.strategy !== undefined) throw policyError("history_policy_all_has_sampling_fields");
    return Object.freeze({ mode, retentionDays });
  }

  const intervalMs = integer(value.intervalMs, 60_000, 1_000, 24 * 60 * 60 * 1000);
  const strategy = string(value.strategy || "latest-per-entity-per-bucket");
  if (strategy !== "latest-per-entity-per-bucket") throw policyError("history_policy_strategy_invalid");
  return Object.freeze({ mode, intervalMs, strategy, retentionDays });
}

export function safeDataProductDefinition(row) {
  return {
    id: row.id,
    version: row.version,
    ontologyRevision: row.ontologyRevision,
    deliveryMode: row.deliveryMode,
    semanticTypes: array(row.semanticTypes),
    fields: array(row.fields),
    fieldContracts: isPlainObject(row.fieldContracts) ? row.fieldContracts : {},
    history: isPlainObject(row.historyPolicy) ? row.historyPolicy : {},
    active: row.active === true,
    createdAt: iso(row.createdAt),
    updatedAt: iso(row.updatedAt),
  };
}

function policyError(code) {
  return Object.assign(new Error(code), { status: 400, code });
}

function identifier(value) {
  const normalized = string(value);
  return IDENTIFIER.test(normalized) ? normalized : "";
}

function identifierSet(value, invalidCode, duplicateCode) {
  if (!Array.isArray(value)) throw policyError(invalidCode);

  const normalized = value.map((entry) => {
    const result = identifier(entry);
    if (!result) throw policyError(invalidCode);
    return result;
  });
  if (new Set(normalized).size !== normalized.length) throw policyError(duplicateCode);

  return Object.freeze(normalized.sort());
}

function integer(value, fallback, min, max) {
  const number = value === undefined ? fallback : Number(value);
  if (!Number.isInteger(number) || number < min || number > max) throw policyError("history_policy_number_invalid");
  return number;
}

function string(value) {
  return typeof value === "string" ? value.trim() : "";
}

function array(value) {
  return Array.isArray(value) ? value : [];
}

function iso(value) {
  return value ? new Date(value).toISOString() : undefined;
}

function isPlainObject(value) {
  return Boolean(value) && typeof value === "object" && !Array.isArray(value);
}

function hasOnlyKeys(value, allowed) {
  return Object.keys(value).every((key) => allowed.has(key));
}

function fieldContractValueMatchesType(value, type) {
  if (type === "string_array") return Array.isArray(value) && value.every((item) => typeof item === "string");
  if (type === "telemetry_readings") return isBoundedTelemetryReadings(value);
  if (type === "point") return isBoundedGeoJsonGeometry(value, new Set(["Point"]));
  if (type === "geometry") return isBoundedGeoJsonGeometry(value);
  return typeof value === type && (type !== "number" || Number.isFinite(value));
}

function stableLiteral(value) {
  return `${typeof value}:${JSON.stringify(value)}`;
}