printhive/lib/importers/makerworld.js

// MakerWorld file downloader. MakerWorld is behind Cloudflare and its downloads
// are auth-gated, so we go through FlareSolverr (solves Cloudflare) carrying the
// user's `token` cookie (authenticates). The download API returns a short-lived
// signed CDN URL, which we then fetch directly.

const flaresolverr = require('../flaresolverr');

const sleep = (ms) => new Promise((r) => setTimeout(r, ms));

// A model page embeds its printable "instances" in __NEXT_DATA__. Each instance
// id is what the download API takes.
function parseInstances(html) {
  const m = html.match(/<script id="__NEXT_DATA__"[^>]*>([\s\S]*?)<\/script>/);
  if (!m) return [];
  let data;
  try { data = JSON.parse(m[1]); } catch { return []; }

  let result = [];
  (function walk(node) {
    if (result.length || !node || typeof node !== 'object') return;
    if (Array.isArray(node.instances) && node.instances.length && typeof node.instances[0].id === 'number') {
      result = node.instances
        .filter((x) => x && typeof x.id === 'number')
        .map((x) => ({ id: x.id, title: x.title || '' }));
      return;
    }
    for (const key of Object.keys(node)) { walk(node[key]); if (result.length) return; }
  })(data);
  return result;
}

// Cloudflare sometimes answers a download call with its anti-bot check instead of
// the JSON — especially during a run of downloads. It comes back as a normal
// answer, so nothing below retries it on its own.
const CHALLENGE_RE = /not a robot|just a moment|captcha|attention required|cf[- ]challenge/i;
function isChallenge(answer) {
  if (!answer || answer.url) return false;
  return CHALLENGE_RE.test(String(answer.error || answer.message || answer.msg || ''));
}

// Resolve an instance to { name, url } (a signed, short-lived CDN download URL).
// A challenge usually clears if we let FlareSolverr's browser try again after a
// pause, so give it a couple of goes before reporting a problem.
async function resolveDownload(flaresolverrUrl, instanceId, token, { retries = 2, waitMs = 8000 } = {}) {
  const api = `https://makerworld.com/api/v1/design-service/instance/${instanceId}/f3mf?type=download&fileType=&devModelName=N1`;
  const opts = { cookies: [{ name: 'token', value: token, domain: '.makerworld.com' }] };
  let answer;
  for (let attempt = 0; attempt <= retries; attempt++) {
    if (attempt > 0) await sleep(waitMs * attempt);
    answer = await flaresolverr.solveJson(flaresolverrUrl, api, opts);
    if (!isChallenge(answer)) return answer;
  }
  return answer;
}

// Why a download-link answer came back without a link: { message, tokenRejected },
// or null when it has one. Not logged in: {"code":1,"error":"Please log in to download models."}
function downloadProblem(answer) {
  if (answer && answer.url) return null;
  const detail = String((answer && (answer.error || answer.message || answer.msg)) || '').trim();
  if (/log ?in|sign ?in|unauthori[sz]ed|token|expired/i.test(detail)) {
    return { message: 'MakerWorld didn’t accept your auth token (it has probably expired).', tokenRejected: true };
  }
  if (isChallenge(answer)) {
    return {
      message: 'MakerWorld’s anti-bot check blocked the download. Trying again in a few minutes usually works.',
      tokenRejected: false,
      challenged: true,
    };
  }
  return { message: `MakerWorld didn’t give a download link${detail ? `: ${detail}` : ''}`, tokenRejected: false };
}

// The collection id in a MakerWorld collection link, or null.
function collectionIdFromUrl(url) {
  const m = String(url || '').match(/makerworld\.com\/(?:[a-z-]{2,5}\/)?collections\/(\d+)/i);
  return m ? m[1] : null;
}

// Where a collection page loads the rest of its models from (its own "load more"
// call). A stable `seed` keeps the pages from overlapping.
const collectionDesignsUrl = (collectionId, { offset = 0, limit = 64, seed } = {}) =>
  `https://makerworld.com/api/v1/design-service/favorites/${collectionId}/designs?offset=${offset}&limit=${limit}`
  + (seed === undefined || seed === null ? '' : `&seed=${seed}`);

// That call's answer -> { designs, total, seed }.
function parseCollectionDesigns(json) {
  const hits = (json && json.hits) || [];
  return {
    designs: hits
      .filter((d) => d && (typeof d.id === 'number' || typeof d.id === 'string'))
      .map((d) => ({ id: String(d.id), slug: d.slug || '', title: d.title || '' })),
    total: Number(json && json.total) || 0,
    seed: json ? json.seed : undefined,
  };
}

const designUrl = (design) => `https://makerworld.com/en/models/${design.id}${design.slug ? `-${design.slug}` : ''}`;

// A collection page -> { id, title, total, designs: [{ id, slug, title }] }, or null
// when the page holds no collection (a private one read without a signed-in token).
// `total` is how many models the collection holds; the page may list fewer.
function parseCollection(html) {
  const m = String(html || '').match(/<script id="__NEXT_DATA__"[^>]*>([\s\S]*?)<\/script>/);
  if (!m) return null;
  let props;
  try { props = JSON.parse(m[1]).props.pageProps; } catch { return null; }
  // A collection you can't see (private, read without a signed-in token) comes back
  // as a placeholder: id -1, no title, no models.
  const id = Number(props && props.favorite && props.favorite.id);
  if (!Number.isFinite(id) || id <= 0) return null;
  const hits = (props.favoriteDesigns && props.favoriteDesigns.hits) || [];
  const designs = hits
    .filter((d) => d && (typeof d.id === 'number' || typeof d.id === 'string'))
    .map((d) => ({ id: String(d.id), slug: d.slug || '', title: d.title || '' }));
  const total = Number((props.favoriteDesigns && props.favoriteDesigns.total) ?? props.favorite.designCnt);
  return {
    id: String(id),
    title: props.favorite.title || '',
    total: Number.isFinite(total) ? total : designs.length,
    designs,
  };
}

// Pull the license string out of the page data (e.g. "MakerWorld Exclusive License").
function parseLicense(html) {
  const m = html.match(/"license"\s*:\s*"([^"]+)"/);
  return m ? m[1] : null;
}

module.exports = {
  parseInstances, resolveDownload, downloadProblem, isChallenge, parseLicense,
  parseCollection, parseCollectionDesigns, collectionIdFromUrl, collectionDesignsUrl, designUrl,
};