Function chunkedBatchDelete

apps/sim/lib/cleanup/batch-delete.ts:100–178

({
  tableDef,
  workspaceIds,
  tableName,
  selectChunk,
  onBatch,
  batchSize = DEFAULT_BATCH_SIZE,
  maxBatches = DEFAULT_MAX_BATCHES_PER_TABLE,
  totalRowLimit = DEFAULT_BATCH_SIZE * DEFAULT_MAX_BATCHES_PER_TABLE,
  workspaceChunkSize = DEFAULT_WORKSPACE_CHUNK_SIZE,
}: ChunkedBatchDeleteOptions<TRow>)

 * `DEFAULT_WORKSPACE_CHUNK_SIZE` for why.
 */
export async function chunkedBatchDelete<TRow extends { id: string }>({
  tableDef,
  workspaceIds,
  tableName,
  selectChunk,
  onBatch,
  batchSize = DEFAULT_BATCH_SIZE,
  maxBatches = DEFAULT_MAX_BATCHES_PER_TABLE,
  totalRowLimit = DEFAULT_BATCH_SIZE * DEFAULT_MAX_BATCHES_PER_TABLE,
  workspaceChunkSize = DEFAULT_WORKSPACE_CHUNK_SIZE,
}: ChunkedBatchDeleteOptions<TRow>): Promise<TableCleanupResult> {
  const result: TableCleanupResult = { table: tableName, deleted: 0, failed: 0 }

  if (workspaceIds.length === 0) {
    logger.info(`[${tableName}] Skipped — no workspaces in scope`)
    return result
  }

  const chunks = chunkArray(workspaceIds, workspaceChunkSize)
  let stoppedEarly = false
  let attempted = 0

  for (const [chunkIdx, chunkIds] of chunks.entries()) {
    if (attempted >= totalRowLimit) {
      stoppedEarly = true
      break
    }

    let batchesProcessed = 0
    let hasMore = true

    while (hasMore && batchesProcessed < maxBatches && attempted < totalRowLimit) {
      let rows: TRow[] = []
      try {
        const remainingLimit = totalRowLimit - attempted
        const effectiveBatchSize = Math.min(batchSize, remainingLimit)
        if (effectiveBatchSize <= 0) {
          hasMore = false
          break
        }

        rows = await selectChunk(chunkIds, effectiveBatchSize)

        if (rows.length === 0) {
          hasMore = false
          break
        }

        attempted += rows.length
        if (onBatch) await onBatch(rows)

        const ids = rows.map((r) => r.id)
        const deleted = await db
          .delete(tableDef)
          .where(inArray(sql`id`, ids))
          .returning({ id: sql`id` })

        result.deleted += deleted.length