ai-chatbot/lib/video-canvas/audio-generation.ts

import "server-only";
import { mkdtemp, readFile, rm } from "node:fs/promises";
import os from "node:os";
import path from "node:path";
import { getEnabledAiModelsWithProvider } from "@/lib/db/queries";
import { uploadFile } from "@/lib/storage/service";
import { downloadCanvasMedia } from "./download-media";
import {
  audioContract,
  audioCheckpointSchema,
  audioRequestSchema,
  buildAudioRequest,
  officialAudioBase,
} from "./audio-contract";
import { ProviderTaskUncertainError } from "./video-task-lifecycle";
export async function listCanvasAudioModels() {
  return (await getEnabledAiModelsWithProvider()).filter(({ model, provider }) => {
    try {
      return (
        provider.type === "alibaba" &&
        !!audioContract(model.modelId) &&
        !!officialAudioBase(provider.baseUrl, model.modelId)
      );
    } catch {
      return false;
    }
  });
}
export async function executeCanvasAudio(
  params: unknown,
  lifecycle: {
    resume?: unknown;
    assertActive: () => Promise<void>;
    checkpoint: (value: Record<string, unknown>) => Promise<void>;
  },
) {
  const input = audioRequestSchema.parse(params);
  let checkpoint = lifecycle.resume ? audioCheckpointSchema.parse(lifecycle.resume) : undefined;
  if (checkpoint && checkpoint.modelRef !== input.modelRef) throw Error("音频记录与原模型不匹配");
  if (!checkpoint) {
    const item = (await listCanvasAudioModels()).find((x) => x.model.id === input.modelRef);
    if (!item) throw Error("音频模型未启用，请先在阿里云和模型设置中开通");
    const request = buildAudioRequest(input, item.model.modelId);
    await lifecycle.assertActive();
    let response: Response;
    try {
      response = await fetch(
        officialAudioBase(item.provider.baseUrl, item.model.modelId) + request.path,
        {
          method: "POST",
          headers: {
            Authorization: `Bearer ${item.provider.apiKey}`,
            "Content-Type": "application/json",
          },
          body: JSON.stringify(request.body),
          signal: AbortSignal.timeout(input.mode === "music" ? 600000 : 180000),
        },
      );
    } catch {
      throw new ProviderTaskUncertainError(
        "音频请求连接中断，无法确认是否已受理；请核对服务商记录，系统不会自动重发",
      );
    }
    let data: any;
    try {
      data = await response.json();
    } catch {
      throw new ProviderTaskUncertainError("音频响应无法解析，请核对服务商记录，系统不会自动重发");
    }
    if (!response.ok)
      throw Error(
        `阿里云音频请求失败（${response.status}，${String(data.code || data.error?.code || "UnknownError").slice(0, 120)}）${input.mode === "music" ? "，请确认 Fun-Music 邀测权限已开通" : ""}`,
      );
    const url = data.output?.audio?.url;
    if (typeof url !== "string" || !/^https?:\/\//i.test(url))
      throw new ProviderTaskUncertainError("音频服务未返回有效文件地址，请核对原请求，勿重复生成");
    checkpoint = {
      operation: "audio",
      url,
      format: input.format,
      modelRef: item.model.id,
      modelId: item.model.modelId,
      requestId: data.request_id,
      expiresAt:
        typeof data.output.audio.expires_at === "number" ? data.output.audio.expires_at : undefined,
    };
    try {
      await lifecycle.assertActive();
      await lifecycle.checkpoint({ upstreamAudio: checkpoint });
    } catch {
      throw new ProviderTaskUncertainError(
        "音频已生成，但保存任务记录失败；请核对原请求，系统不会自动重发",
      );
    }
  }
  if (checkpoint.expiresAt && checkpoint.expiresAt * 1000 < Date.now())
    throw Error("阿里云音频临时地址已过期，无法继续转存");
  await lifecycle.assertActive();
  const dir = await mkdtemp(path.join(os.tmpdir(), "canvas-audio-"));
  try {
    const file = path.join(dir, "audio." + checkpoint.format);
    await downloadCanvasMedia(checkpoint.url, file);
    const bytes = await readFile(file);
    if (bytes.length < 12) throw Error("音频内容为空");
    const wav =
      bytes.toString("ascii", 0, 4) === "RIFF" && bytes.toString("ascii", 8, 12) === "WAVE";
    const mp3 =
      bytes.toString("ascii", 0, 3) === "ID3" || (bytes[0] === 0xff && (bytes[1] & 0xe0) === 0xe0);
    if (!(checkpoint.format === "wav" ? wav : mp3)) throw Error("返回文件不是有效音频格式");
    await lifecycle.assertActive();
    const saved = await uploadFile(
      `audio-${crypto.randomUUID()}.${checkpoint.format}`,
      Uint8Array.from(bytes).buffer,
    );
    return {
      url: saved.url,
      modelRef: checkpoint.modelRef,
      modelId: checkpoint.modelId,
      metadata: {
        storageStatus: "archived",
        format: checkpoint.format,
        requestId: checkpoint.requestId,
      },
    };
  } catch (error) {
    throw new ProviderTaskUncertainError(
      `音频已生成，转存暂未完成，将继续转存原文件：${error instanceof Error ? error.message : "存储失败"}`,
    );
  } finally {
    await rm(dir, { recursive: true, force: true });
  }
}