Spaces:
Running
Running
Fix load hang in private Space: forward __sign to worker/module/SW requests, main-thread fallback, timeouts, size checks
6bfdc94 verified | // AksharaLP Space — engine host, shared by the Web Worker (js/worker.js) and the main-thread fallback in js/app.js. | |
| // Two-tier load: the small core (model + reranker) first, so typing works immediately; then the dictionary pack | |
| // (lexicon + sentence LM), read from the Cache API after the first visit. If the pack cannot be fetched, or does not | |
| // arrive within DICT_TIMEOUT_MS, the engine keeps running core-only. | |
| // | |
| // `search` is the page's query string. On a private Hugging Face Space the page is opened as index.html?__sign=<JWT> | |
| // inside a cross-site iframe, where the SameSite=Lax auth cookie is not sent; same-origin requests are authorised only | |
| // through their Referer. A worker's (and a module's) requests carry the worker's / module's own URL as Referer, so | |
| // the query is forwarded onto those URLs. Cross-origin requests (the CDN) only ever see the origin. | |
| const DICT_TIMEOUT_MS = 180000; | |
| export function startHost(post, search = "") { | |
| let engine = null; | |
| const q = new URLSearchParams(search); | |
| const fwd = new URLSearchParams(search); fwd.delete("dict"); const qs = fwd.toString() ? "?" + fwd.toString() : ""; | |
| const ready = (async () => { | |
| try { | |
| const { default: BhashaCore } = await import("./aksharalp.mjs" + qs); | |
| const base = new URL("../model/", import.meta.url).href; | |
| engine = await BhashaCore.load(base, { dict: false, onProgress: (file) => post({ type: "progress", file }) }); | |
| const t0 = performance.now(); | |
| engine.transliterate("main ghar ja raha hoon", { k: 5 }); | |
| post({ type: "ready", info: engine.info(), loadMs: engine.loadMs, warmMs: performance.now() - t0, note: engine.manifest && engine.manifest.note, name: engine.manifest && engine.manifest.name }); | |
| if (q.get("dict") === "0") { post({ type: "dict", ok: false, skipped: true, info: engine.info() }); return; } | |
| const t1 = performance.now(); let src = null, timer; | |
| const timeout = new Promise((res) => { timer = setTimeout(() => res("timeout"), DICT_TIMEOUT_MS); }); | |
| const ok = await Promise.race([engine.loadDict({ onProgress: (file, from) => { src = from; post({ type: "progress", file: "dictionary/" + file, from }); } }), timeout]); | |
| clearTimeout(timer); | |
| if (ok === "timeout") { post({ type: "dict", ok: false, error: `dictionary pack did not arrive within ${DICT_TIMEOUT_MS / 1000} s; running core-only`, info: engine.info() }); return; } | |
| if (ok) engine.transliterate("main ghar ja raha hoon", { k: 5 }); | |
| post({ type: "dict", ok, from: src, ms: performance.now() - t1, error: engine.dictError, info: engine.info() }); | |
| } catch (e) { | |
| post({ type: "error", message: String(e && e.message || e) }); | |
| } | |
| })(); | |
| return (data) => { | |
| if (!engine) return post({ type: "error", id: data.id, message: "model not loaded" }); | |
| try { | |
| if (data.type === "run") { | |
| const res = engine.transliterate(data.text, data.opts || {}); | |
| post({ type: "result", id: data.id, tag: data.tag, res }); | |
| } else if (data.type === "batch") { | |
| const out = data.items.map((it) => engine.transliterate(it.text, it.opts || {})); | |
| post({ type: "batch", id: data.id, out }); | |
| } | |
| } catch (e) { | |
| post({ type: "error", id: data.id, message: String(e && e.message || e) }); | |
| } | |
| }; | |
| } | |