Open-source manufacturing ERP, MES and QMS. Quoting, MRP, inventory, shop floor, quality and lot/serial traceability on one Postgres schema, with a REST API and MCP server. Self-host or use Carbon Cloud.
TypeScript ai-agentsbill-of-materialserpinventory-managementleanmanufacturingmanufacturing-execution-systemmcp-servermesmrpopen-source-erppostgresqlproduction-planningqmsquality-management-systemreact-routerself-hostedsupabasetraceabilitytypescript
// SPDX-License-Identifier: AGPL-3.0-only
// Carbon (github.com/crbnos/carbon). Modified or adapted versions of this file,
// including ports, remain AGPLv3; serving them over a network requires releasing their source.
/**
* Encodes a value for a `json` / `jsonb` column written through Kysely.
*
* The Postgres driver serialises query parameters by JS type, not
* by column type: an object becomes `JSON.stringify(value)`, but a string is
* sent as raw text and an array as a Postgres array literal (`{a,b}`). For a
* JSON column that means a stored JSON string scalar (`"some text"`) is read
* back by supabase-js as the JS string `some text`, re-sent unquoted, and
* rejected with `invalid input syntax for type json`; a JSON array is mangled
* the same way. Only object-shaped values survive the round trip.
*
* Pre-serialising here makes the wire value valid JSON text for every shape,
* so a row copied from one document to another (quote → sales order, RFQ →
* quote, …) keeps working whatever a previous writer stored. `null` and
* `undefined` pass through so the column's NULL / DEFAULT semantics are kept.
*/
export function toJson(value: unknown): string | null | undefined {
if (value === null || value === undefined) return value;
return JSON.stringify(value);
}
/**
* Runs `toJson` over a fixed list of jsonb columns on a row read back from
* Supabase, for re-inserting that row (or a spread of it) through Kysely —
* e.g. `{ ...line, ...toJsonColumns(line, QUOTE_LINE_JSON_COLUMNS) }`.
*
* A document-copy that spreads a source row raw (`{...line, quoteId, companyId}`)
* silently regresses to the unserialised bug the day a new jsonb column is
* added to that table and the spread picks it up untouched. Naming the
* column list here, next to the call site, makes that list something a
* future column addition has to be checked against, and something a test can
* assert against directly.
*/
export function toJsonColumns<
T extends Record<string, unknown>,
K extends keyof T
>(row: T, keys: readonly K[]): { [P in K]: string | null | undefined } {
const out = {} as { [P in K]: string | null | undefined };
for (const key of keys) {
out[key] = toJson(row[key]);
}
return out;
}