Download social media videos in Node.js

Node has had fetch built in since 18, so there is no client library to install. AbortSignal.timeout is the part worth copying — without it a slow source keeps a handler alive until the platform kills it.

const API = "https://api.saveapi.org/v1/download";

async function resolve(link) {
  const res = await fetch(API + "?url=" + encodeURIComponent(link), {
    headers: { Authorization: "Bearer sk_live_xxxx" },
    signal: AbortSignal.timeout(30_000),
  });
  return res.json();
}

medias[] holds every file in the post in the order it appears. A TikTok slideshow gives you image entries and a separate audio entry for the background track, so find() by type rather than reaching for medias[0] and hoping.

const data = await resolve("https://www.tiktok.com/@user/video/123");

if (!data.success) throw new Error(data.error.code);

const video = data.medias.find((m) => m.type === "video");
const images = data.medias.filter((m) => m.type === "image");
const audio = data.medias.find((m) => m.type === "audio");

If your users click a download button, do not buffer the file and do not redirect them to the source CDN — the link is signed and expires. Pipe the upstream body straight into your response with a Content-Disposition header and the browser saves it under a name you chose.

import { Readable } from "node:stream";

app.get("/save", async (req, res) => {
  const data = await resolve(req.query.url);
  if (!data.success) return res.status(400).json(data.error);

  const upstream = await fetch(data.medias[0].url);
  res.setHeader("Content-Type", "video/mp4");
  res.setHeader("Content-Disposition", 'attachment; filename="video.mp4"');
  Readable.fromWeb(upstream.body).pipe(res);
});

Timeouts and upstream errors deserve a second attempt. Private, missing and unsupported are final answers — retrying them wastes wall-clock and, where the error is billed, a credit. Keep the retryable codes in a Set so the rule lives in one place.

const RETRY = new Set(["UPSTREAM_TIMEOUT", "UPSTREAM_ERROR"]);

async function withRetry(link) {
  for (let attempt = 0; attempt < 2; attempt++) {
    const data = await resolve(link);
    if (data.success) return data;
    if (!RETRY.has(data.error.code)) throw new Error(data.error.message);
  }
  throw new Error("source unavailable");
}

Every response carries the per-minute limit, what is left of it and the credits remaining. That is enough to back off before you are throttled and to alarm on a pack running out, with no extra request.

const res = await fetch(API + "?url=" + encodeURIComponent(link), { headers });

const perMinute = Number(res.headers.get("X-RateLimit-Limit"));
const left = Number(res.headers.get("X-RateLimit-Remaining"));
const credits = Number(res.headers.get("X-Credits-Remaining"));

When users paste arbitrary text, /v1/detect tells you which platform a URL belongs to without touching your credits. It counts against the rate limit, so it is a validation step rather than a free probe loop.

const check = await fetch(
  "https://api.saveapi.org/v1/detect?url=" + encodeURIComponent(userInput),
  { headers },
);
const { platform } = await check.json();   // kredit sarflanmaydi