Never stop coding. Free MIT AI gateway: one endpoint, 375 providers (150+ free), 1,400+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 900+ contributors
TypeScript a2aai-agentsai-gatewayanthropicclaudeclaude-codeclinecodexcopilotcursordeepseekfree-aigeminikimillm-gatewaymcpopenaiopenai-proxyqwentoken-saver
/**
* JobRegistry persistence layer.
*
* CRUD for the `jobs` and `job_runs` tables (migration 136). All timestamps are
* ISO-8601 strings; the registry computes thresholds in JS (never SQL datetime
* arithmetic) so comparisons are plain string compares.
*
* Column naming follows the rest of src/lib/db: snake_case in SQLite, camelCase in
* the returned objects (mapRow). `enabled` is stored INTEGER (0/1), `config` is a
* JSON string parsed/stringified at the boundary.
*/
import { getDbInstance } from "./core";
import type { JobRecord, JobRun } from "../jobRegistry/core";
function mapJob(row: any): JobRecord {
return {
id: row.id,
type: row.type,
cron: row.cron,
intervalMs: row.interval_ms,
enabled: row.enabled === 1,
envFlag: row.env_flag,
config: row.config ? JSON.parse(row.config) : {},
createdAt: row.created_at,
updatedAt: row.updated_at,
};
}
function mapRun(row: any): JobRun {
return {
id: row.id,
jobId: row.job_id,
startedAt: row.started_at,
finishedAt: row.finished_at,
status: row.status,
errorMessage: row.error_message,
recordsAffected: row.records_affected ?? 0,
durationMs: row.duration_ms,
};
}
export function getAllJobs(): JobRecord[] {
const db = getDbInstance();
const rows = db.prepare("SELECT * FROM jobs ORDER BY id").all();
return rows.map(mapJob);
}
export function getJob(id: string): JobRecord | null {
const db = getDbInstance();
const row = db.prepare("SELECT * FROM jobs WHERE id = ?").get(id);
return row ? mapJob(row) : null;
}
/**
* Idempotent register: INSERT OR IGNORE on first sight, then a column-level UPDATE
* that refreshes scheduling fields (type/cron/interval/env_flag/config) and bumps
* updated_at - but NEVER overwrites `enabled` (the user's API-driven toggle) nor
* `created_at`. Pass enabled=true for new jobs; the UPDATE simply skips the column.
*/
export function upsertJob(job: JobRecord): void {
const db = getDbInstance();
db.prepare(
`INSERT INTO jobs (id, type, cron, interval_ms, enabled, env_flag, config, created_at, updated_at)
VALUES (?, ?, ?, ?, ?, ?, ?, ?, datetime('now'))
ON CONFLICT(id) DO UPDATE SET
type = excluded.type,
cron = excluded.cron,
interval_ms = excluded.interval_ms,
env_flag = excluded.env_flag,
config = excluded.config,
updated_at = datetime('now')`
).run(
job.id,
job.type,
job.cron,
job.intervalMs,
job.enabled ? 1 : 0,
job.envFlag,
JSON.stringify(job.config ?? {}),
job.createdAt
);
}
export function updateJobEnabled(id: string, enabled: boolean): void {
const db = getDbInstance();
db.prepare("UPDATE jobs SET enabled = ?, updated_at = datetime('now') WHERE id = ?").run(
enabled ? 1 : 0,
id
);
}
export interface RecordRunOptions {
startedAt?: string;
durationMs?: number;
errorMessage?: string;
recordsAffected?: number;
}
/**
* Insert a completed run. `startedAt` is captured by the registry before the handler
* runs; `finishedAt` is derived from startedAt + durationMs so the two never drift.
* A status='running' insert leaves finishedAt NULL (used to mark in-flight work).
*/
export function recordRun(
jobId: string,
status: JobRun["status"],
opts: RecordRunOptions = {}
): void {
const db = getDbInstance();
const startedAt = opts.startedAt ?? new Date().toISOString();
const finishedAt =
status === "running"
? null
: opts.startedAt && opts.durationMs != null
? new Date(new Date(opts.startedAt).getTime() + opts.durationMs).toISOString()
: new Date().toISOString();
db.prepare(
`INSERT INTO job_runs (job_id, started_at, finished_at, status, error_message, records_affected, duration_ms)
VALUES (?, ?, ?, ?, ?, ?, ?)`
).run(
jobId,
startedAt,
finishedAt,
status,
opts.errorMessage ?? null,
opts.recordsAffected ?? 0,
opts.durationMs ?? null
);
}
/**
* Dual-dimension pruning: keep the most recent `maxRuns` OR anything younger than
* `maxDays`. A run is deleted only when it is BOTH outside the recent-N window AND
* older than maxDays - so a job that runs rarely keeps its full history.
* All thresholds are ISO-8601 string compares (no SQL datetime math).
*/
export function pruneRuns(jobId: string, maxRuns = 100, maxDays = 30): void {
const db = getDbInstance();
const threshold = new Date(Date.now() - maxDays * 86_400_000).toISOString();
db.prepare(
`DELETE FROM job_runs
WHERE job_id = ?
AND id NOT IN (
SELECT id FROM job_runs WHERE job_id = ? ORDER BY started_at DESC LIMIT ?
)
AND started_at < ?`
).run(jobId, jobId, maxRuns, threshold);
}
export function getRuns(jobId: string, limit = 20): JobRun[] {
const db = getDbInstance();
const rows = db
.prepare("SELECT * FROM job_runs WHERE job_id = ? ORDER BY started_at DESC LIMIT ?")
.all(jobId, limit);
return rows.map(mapRun);
}
/**
* Startup repair: any run still marked `running` whose started_at is older than
* `timeoutMinutes` is a leftover from a crashed/hung process. Mark it `failure` so
* the history is accurate and the slot frees up. The 5-minute default avoids
* clobbering a genuinely in-flight run on a slow box.
*/
export function cleanupOrphanedRuns(timeoutMinutes = 5): void {
const db = getDbInstance();
const threshold = new Date(Date.now() - timeoutMinutes * 60_000).toISOString();
db.prepare(
`UPDATE job_runs
SET status = 'failure',
error_message = 'orphaned: exceeded timeout',
finished_at = datetime('now')
WHERE status = 'running' AND started_at < ?`
).run(threshold);
}