knowledge/node_modules/bullmq/dist/esm/commands/includes/requeueDeduplicatedJob.lua

--[[
  Function to create a new job from stored dedup-next data
  when a deduplicated job with keepLastIfActive finishes.
  At most one next job is created per deduplication ID.
  Multiple triggers while active overwrite the dedup-next data,
  so only the latest data is used.
]]

-- Includes
--- @include "getOrSetMaxEvents"
--- @include "setDeduplicationKey"
--- @include "storeAndEnqueueJob"

local function requeueDeduplicatedJob(prefix, deduplicationId, eventStreamKey,
    metaKey, activeKey, waitKey, pausedKey, markerKey, prioritizedKey,
    priorityCounterKey, delayedKey, timestamp)
  local deduplicationNextKey = prefix .. "dn:" .. deduplicationId
  if rcall("EXISTS", deduplicationNextKey) == 1 then
    local nextData = rcall("HMGET", deduplicationNextKey,
        "name", "data", "opts", "pk", "pd", "pdk", "rjk")

    local newJobId = rcall("INCR", prefix .. "id") .. ""
    local newJobIdKey = prefix .. newJobId
    local newOpts = cjson.decode(nextData[3])
    local deduplicationKey = prefix .. "de:" .. deduplicationId

    local parentKey = nextData[4] or nil
    local parentData = nextData[5] or nil
    local parentDependenciesKey = nextData[6] or nil
    local repeatJobKey = nextData[7] or nil

    -- Set dedup key for the new job (without TTL when keepLastIfActive,
    -- so the key outlives the job's active duration)
    local deOpts = newOpts['de']
    if deOpts and deOpts['keepLastIfActive'] then
      rcall('SET', deduplicationKey, newJobId)
    else
      setDeduplicationKey(deduplicationKey, newJobId, deOpts)
    end

    -- Store and enqueue using the shared helper (handles priority/lifo/delayed)
    local maxEvents = getOrSetMaxEvents(metaKey)
    storeAndEnqueueJob(eventStreamKey, newJobIdKey, newJobId, nextData[1], nextData[2],
        newOpts, timestamp, parentKey, parentData, repeatJobKey, maxEvents,
        waitKey, pausedKey, activeKey, metaKey, prioritizedKey,
        priorityCounterKey, delayedKey, markerKey)

    -- Register as child dependency if the job has a parent
    if parentDependenciesKey then
      rcall("SADD", parentDependenciesKey, newJobIdKey)
    end

    -- Only delete the dedup-next hash after the job is fully created,
    -- so that if any step above errors, the data is not permanently lost.
    rcall("DEL", deduplicationNextKey)
  end
end