import { Communicate } from "edge-tts-ts";
import { NextResponse } from "next/server";

export const runtime = "nodejs";
export const dynamic = "force-dynamic";

const DEFAULT_VOICE = process.env.EDGE_TTS_VOICE || "zh-CN-XiaoxiaoNeural";
const MAX_TEXT_LENGTH = 12_000;

const jsonError = (message: string, status: number) =>
  NextResponse.json({ error: message }, { status });

export async function POST(request: Request) {
  let payload: { text?: unknown; voice?: unknown };
  try {
    payload = await request.json() as { text?: unknown; voice?: unknown };
  } catch {
    return jsonError("请求内容不是有效 JSON", 400);
  }

  const text = typeof payload.text === "string" ? payload.text.trim() : "";
  if (!text) return jsonError("缺少待朗读文本", 400);
  if (text.length > MAX_TEXT_LENGTH) return jsonError(`单次文本不能超过 ${MAX_TEXT_LENGTH} 字`, 413);

  const voice = typeof payload.voice === "string" && payload.voice.trim()
    ? payload.voice.trim()
    : DEFAULT_VOICE;

  try {
    const communicate = new Communicate(text, { voice });
    const chunks: Uint8Array[] = [];
    let byteLength = 0;

    for await (const chunk of communicate.stream()) {
      if (chunk.type !== "audio") continue;
      chunks.push(chunk.data);
      byteLength += chunk.data.byteLength;
    }

    if (!byteLength) return jsonError("语音服务返回了空音频", 502);

    const audio = new Uint8Array(byteLength);
    let offset = 0;
    for (const chunk of chunks) {
      audio.set(chunk, offset);
      offset += chunk.byteLength;
    }

    return new Response(audio.buffer.slice(audio.byteOffset, audio.byteOffset + audio.byteLength), {
      headers: {
        "Content-Type": "audio/mpeg",
        "Cache-Control": "no-store"
      }
    });
  } catch (error) {
    console.error("[Agent TTS] 语音生成失败", error);
    return jsonError("语音生成失败", 502);
  }
}