78 lines
2.4 KiB
TypeScript
78 lines
2.4 KiB
TypeScript
import "dotenv/config";
|
|
import { writeFile } from "node:fs/promises";
|
|
import { setTimeout as sleep } from "node:timers/promises";
|
|
import { prisma } from "../lib/prisma";
|
|
import { prepareAudio } from "../lib/stt/audio";
|
|
import {
|
|
downloadTranscript,
|
|
getAsrConfig,
|
|
queryTranscription,
|
|
submitTranscription,
|
|
} from "../lib/stt/dashscope";
|
|
|
|
async function main() {
|
|
const videoId = process.argv[2];
|
|
if (!videoId || !/^[A-Za-z0-9_-]+$/.test(videoId))
|
|
throw new Error("用法：npx tsx scripts/stt-smoke.ts <视频 ID>");
|
|
getAsrConfig();
|
|
const video = await prisma.video.findUniqueOrThrow({
|
|
where: { aweme_id: videoId },
|
|
select: { duration_ms: true },
|
|
});
|
|
console.log(
|
|
JSON.stringify({
|
|
videoId,
|
|
durationSeconds: video.duration_ms / 1000,
|
|
phase: "extracting_audio",
|
|
}),
|
|
);
|
|
const audioURL = await prepareAudio(videoId);
|
|
const resultPath = `/tmp/douyin-asr-${videoId}.json`;
|
|
if (!audioURL) {
|
|
await writeFile(
|
|
resultPath,
|
|
JSON.stringify({ videoId, transcript: [] }, null, 2),
|
|
);
|
|
console.log(
|
|
JSON.stringify({ phase: "complete", transcript: [], resultPath }),
|
|
);
|
|
return;
|
|
}
|
|
const taskId = await submitTranscription(audioURL);
|
|
await writeFile(
|
|
resultPath,
|
|
JSON.stringify({ videoId, taskId, status: "submitted" }, null, 2),
|
|
);
|
|
console.log(JSON.stringify({ phase: "submitted", taskId }));
|
|
for (let attempt = 0; attempt < 90; attempt++) {
|
|
await sleep(10_000);
|
|
const task = await queryTranscription(taskId);
|
|
if (task.status === "SUCCEEDED") {
|
|
const transcript = await downloadTranscript(task.resultURL);
|
|
await writeFile(
|
|
resultPath,
|
|
JSON.stringify({ videoId, taskId, transcript }, null, 2),
|
|
);
|
|
console.log(
|
|
JSON.stringify({ phase: "complete", transcript, resultPath }),
|
|
);
|
|
return;
|
|
}
|
|
if (attempt % 6 === 0) console.log(JSON.stringify({ phase: task.status }));
|
|
}
|
|
throw new Error(
|
|
`15 分钟内未完成，任务 ID 已保存在 ${resultPath}，请查询原任务而非重复提交`,
|
|
);
|
|
}
|
|
main()
|
|
.catch((error) => {
|
|
// Do not print arbitrary SDK/network errors: they may contain signed URLs.
|
|
console.error(
|
|
error instanceof Error && !error.name.startsWith("Prisma")
|
|
? error.message
|
|
: "读取视频或转写失败",
|
|
);
|
|
process.exitCode = 1;
|
|
})
|
|
.finally(() => prisma.$disconnect());
|