aigen/server/api/extend.post.ts

import { copyFileSync, existsSync, readFileSync } from 'node:fs'
import { join } from 'node:path'
import { composeShotPrompt } from '~/utils/globalLocks'

const SAMPLERS = new Set(['res_multistep', 'euler', 'dpmpp_2m'])
const SCHEDULERS = new Set(['simple', 'ddim_uniform', 'sgm_uniform'])

function parseExtendDuration(raw: unknown) {
  const seconds = Number(raw)
  if (!Number.isFinite(seconds)) return 5
  return Math.min(120, Math.max(0.5, Math.round(seconds * 10) / 10))
}

function frameLength(seconds: number, fps: number) {
  return Math.max(5, Math.floor(seconds * fps))
}

function parseFps(raw: unknown) {
  const fps = Number(raw)
  return fps === 12 || fps === 30 || fps === 24 ? fps : 24
}

function parseCfg(raw: unknown, turbo: boolean) {
  const fallback = turbo ? 1.5 : 4
  const value = Number(raw)
  if (!Number.isFinite(value)) return fallback
  const snapped = Math.round(value * 2) / 2
  return Math.min(10, Math.max(1, snapped))
}

function parseSampler(raw: unknown) {
  const value = String(raw || '')
  return SAMPLERS.has(value) ? value : 'res_multistep'
}

function parseScheduler(raw: unknown) {
  const value = String(raw || '')
  return SCHEDULERS.has(value) ? value : 'simple'
}

export default defineEventHandler(async (event) => {
  const body = await readBody<{ clipId?: string; prompt?: string; duration?: number }>(event).catch(() => ({}))
  const clipId = String(body?.clipId || '').trim()
  const prompt = String(body?.prompt || '').trim()
  if (!clipId) {
    throw createError({ statusCode: 400, statusMessage: 'A clip is required to extend' })
  }
  if (!prompt) {
    throw createError({ statusCode: 400, statusMessage: 'An extension prompt is required' })
  }

  const { owner } = assertLibraryOwner(event)
  const source = getClip(owner, clipId)
  assertFolderAccess(event, source.folderId)
  const sourcePath = clipVideoPath(owner, source.id)
  if (!existsSync(sourcePath)) {
    throw createError({ statusCode: 404, statusMessage: 'Source video is missing' })
  }

  const destFolder = publicLibrary(event).folders.find(folder => folder.id === source.folderId)
  const folderLocked = Boolean(destFolder?.protected && !destFolder.unlocked)
  const durationSeconds = parseExtendDuration(body?.duration)
  const composedPrompt = composeShotPrompt({
    globalLocks: source.globalLocks,
    prompt,
    shotIndex: (source.chainIndex || 0) + 1,
    familyRefs: source.permanenceRefs
  })
  const turbo = source.turbo !== false
  const fps = parseFps(source.fps)
  const cfg = parseCfg(source.cfg, turbo)
  const samplerName = parseSampler(source.samplerName)
  const scheduler = parseScheduler(source.scheduler)
  const steps = Number(source.steps || defaultVideoSteps(turbo, parseVideoWorkflow(source.workflow)))
  const seed = Math.floor(Math.random() * 2_147_483_647)
  const length = frameLength(durationSeconds, fps)
  const width = source.width || 1344
  const height = source.height || 768
  const clipName = nextClipPartName(clipTitle(source))

  const job = createJob()
  job.kind = 'video'
  job.maxStep = steps
  job.hideThumbnail = source.hideThumbnail
  const tmpDir = extendTempDir(owner, job.id)
  const part1Path = join(tmpDir, 'part1.mp4')
  const framePath = join(tmpDir, 'last_frame.png')
  job.library = {
    ownerKey: owner,
    folderId: source.folderId,
    hideThumbnail: source.hideThumbnail,
    folderLocked,
    name: clipName,
    prompt: composedPrompt,
    aspect: source.aspect || '16:9',
    width,
    height,
    steps,
    turbo,
    seed,
    cfg,
    fps,
    samplerName,
    scheduler,
    duration: durationSeconds,
    extendTmpDir: tmpDir,
    extendPart1Path: part1Path,
    extendSourceClipId: source.id,
    familyId: source.familyId || crypto.randomUUID(),
    parentClipId: source.id,
    chainIndex: (source.chainIndex || 0) + 1,
    workflow: parseVideoWorkflow(source.workflow),
    globalLocks: source.globalLocks,
    permanenceRefs: source.permanenceRefs
  }

  emitJob(job, { type: 'status', message: 'Extracting last frame & stitching extension...', progress: 2 })

  void runExtend(job, {
    sourcePath,
    part1Path,
    framePath,
    prompt: composedPrompt,
    width,
    height,
    steps,
    seed,
    turbo,
    length,
    cfg,
    fps,
    samplerName,
    scheduler
  }).catch((error) => {
    removeExtendTemp(job.library?.extendTmpDir)
    const message = error instanceof Error ? error.message : String(error)
    job.status = 'error'
    job.error = message
    emitJob(job, { type: 'error', error: message, message })
  })

  return {
    jobId: job.id,
    clientId: job.clientId,
    seed,
    width,
    height,
    steps,
    duration: durationSeconds,
    hideThumbnail: source.hideThumbnail,
    folderLocked
  }
})

async function runExtend(
  job: ReturnType<typeof createJob>,
  params: {
    sourcePath: string
    part1Path: string
    framePath: string
    prompt: string
    width: number
    height: number
    steps: number
    seed: number
    turbo: boolean
    length: number
    cfg: number
    fps: number
    samplerName: string
    scheduler: string
  }
) {
  copyFileSync(params.sourcePath, params.part1Path)
  await extractLastFrame(params.part1Path, params.framePath)
  const frame = readFileSync(params.framePath)
  if (job.library) job.library.thumb = undefined
  const sound = await probeHasAudio(params.part1Path)

  await ensureComfyReady((status) => {
    emitJob(job, {
      type: status.state === 'busy' ? 'busy' : 'status',
      message: status.message,
      progress: status.state === 'online' ? Math.max(job.progress, 3) : Math.max(job.progress, 2),
      busy: status.state === 'busy',
      queueRunning: status.queueRunning,
      queuePending: status.queuePending
    })
  })

  const done = watchComfyJob(job)
  job.status = 'uploading'
  emitJob(job, { type: 'status', message: 'Uploading last frame to ComfyUI...', progress: 4 })
  const uploaded = await uploadImage({ filename: 'last_frame.png', data: frame, type: 'image/png' }, job.id)
  if (job.library) {
    job.library.imageName = uploaded.name
    job.library.imageSubfolder = uploaded.subfolder
  }
  emitJob(job, { type: 'status', message: 'Queueing extension on MiniMax H3...', progress: 6 })
  await waitForComfySocket(job, 4000)

  const graph = buildWorkflow({
    prompt: params.prompt,
    imageName: uploaded.name,
    width: params.width,
    height: params.height,
    steps: params.steps,
    seed: params.seed,
    turbo: params.turbo,
    length: params.length,
    cfg: params.cfg,
    fps: params.fps,
    samplerName: params.samplerName,
    scheduler: params.scheduler,
    filenamePrefix: comfyFilenamePrefix(),
    sound,
    workflow: job.library?.workflow,
    duration: job.library?.duration
  })

  const queued = await queuePrompt(graph, job.clientId)
  job.promptId = queued.prompt_id
  job.status = 'running'
  if (job.library && job.promptId) {
    writePendingJob({
      jobId: job.id,
      promptId: job.promptId,
      clientId: job.clientId,
      ownerKey: job.library.ownerKey,
      folderId: job.library.folderId,
      hideThumbnail: job.library.hideThumbnail,
      folderLocked: job.library.folderLocked,
      name: job.library.name,
      prompt: job.library.prompt,
      aspect: job.library.aspect,
      width: job.library.width,
      height: job.library.height,
      steps: job.library.steps,
      turbo: job.library.turbo,
      seed: job.library.seed,
      startedAt: job.startedAt,
      imageName: job.library.imageName,
      imageSubfolder: job.library.imageSubfolder,
      extendTmpDir: job.library.extendTmpDir,
      extendPart1Path: job.library.extendPart1Path,
      familyId: job.library.familyId,
      parentClipId: job.library.parentClipId,
      chainIndex: job.library.chainIndex,
      stillId: job.library.stillId,
      sound: job.library.sound,
      workflow: job.library.workflow,
      duration: job.library.duration,
      globalLocks: job.library.globalLocks,
      permanenceRefs: job.library.permanenceRefs
    })
  }
  emitJob(job, { type: 'status', message: 'Job queued on ComfyUI', progress: 8 })
  await done
}