screenclipper/apps/api-daemon/src/cleanup.ts

import { unlink } from "node:fs/promises";
import { prisma } from "@streamclipper/db";
import { env } from "./env";

const CHECK_INTERVAL_MS = 60 * 60 * 1000; // hourly
const TTL_SETTING_KEY = "raw_segment_ttl_hours";

async function getTtlHours(): Promise<number> {
  const setting = await prisma.appSetting.findUnique({ where: { key: TTL_SETTING_KEY } });
  const parsed = setting ? Number(setting.value) : NaN;
  return Number.isFinite(parsed) && parsed > 0 ? parsed : env.rawSegmentTtlHours;
}

/**
 * PRD's auto-purge lifecycle: processed raw recordings get deleted 24h after
 * capture. Only segments that are done being analyzed (PROCESSED/DISCARDED)
 * and have no candidate still waiting on STT are eligible — deleting a raw
 * file mid-analysis or mid-transcription would break the pipeline for it.
 */
async function cleanupOnce(): Promise<void> {
  const ttlHours = await getTtlHours();
  const cutoff = new Date(Date.now() - ttlHours * 60 * 60 * 1000);

  const staleSegments = await prisma.rawSegment.findMany({
    where: {
      createdAt: { lt: cutoff },
      status: { in: ["PROCESSED", "DISCARDED"] },
      candidates: { none: { status: "PENDING_STT" } },
    },
  });

  for (const segment of staleSegments) {
    await prisma.candidateSegment.deleteMany({ where: { rawSegmentId: segment.id } });
    await prisma.rawSegment.delete({ where: { id: segment.id } });
    await unlink(segment.filePath).catch(() => {});
  }

  if (staleSegments.length > 0) {
    console.log(`[cleanup] removed ${staleSegments.length} raw segment(s) older than ${ttlHours}h`);
  }
}

export function startCleanupLoop(): NodeJS.Timeout {
  console.log(`[cleanup] checking every ${CHECK_INTERVAL_MS}ms, default TTL ${env.rawSegmentTtlHours}h`);
  cleanupOnce().catch((err) => console.error("[cleanup] initial run failed:", err));
  return setInterval(() => {
    cleanupOnce().catch((err) => console.error("[cleanup] run failed:", err));
  }, CHECK_INTERVAL_MS);
}