diff options
| -rw-r--r-- | names/compose-libretranslate.yml | 26 | ||||
| -rw-r--r-- | names/server.js | 17 | ||||
| -rw-r--r-- | names/translate.js | 82 | ||||
| -rw-r--r-- | tools/translate-test.js | 116 |
4 files changed, 241 insertions, 0 deletions
diff --git a/names/compose-libretranslate.yml b/names/compose-libretranslate.yml new file mode 100644 index 0000000..80ed9c6 --- /dev/null +++ b/names/compose-libretranslate.yml @@ -0,0 +1,26 @@ +# LibreTranslate for post translation (coinos v3). Reached only by the +# names registrar over the docker network (POST /translate there); no port +# on the host, no web UI. Models live in the lt-data volume. +services: + libretranslate: + image: libretranslate/libretranslate:v1.9.6 + container_name: libretranslate + environment: + LT_LOAD_ONLY: en,es,fr,de,pt,id,tr,vi,ru,ar,ur,hi,bn,ja,ko,zh,it,nl,pl,uk,fa,he,th,sv,cs,el,ro,hu,fi,da,nb,ca + LT_DISABLE_WEB_UI: "true" + LT_CHAR_LIMIT: "5000" + LT_THREADS: "4" + LT_UPDATE_MODELS: "true" + volumes: + - lt-data:/home/libretranslate/.local + networks: [net] + deploy: + resources: + limits: + memory: 8g + restart: unless-stopped +volumes: + lt-data: +networks: + net: + external: true diff --git a/names/server.js b/names/server.js index 6cd82a1..463ab8e 100644 --- a/names/server.js +++ b/names/server.js @@ -28,6 +28,7 @@ import { mnemonicToSeedSync } from '@scure/bip39'; import { lnBackend } from '../bridge/ln.js'; import { ArkManager } from '../src/ark/manager.js'; import { fetchPreview } from './preview.js'; +import { translate, languages } from './translate.js'; import { decodeAddress } from '../src/ark/proto.js'; import { decodeNoffer } from '../src/noffer.js'; import { decodeBolt11 } from '../src/ark/lightning.js'; @@ -1036,6 +1037,22 @@ Bun.serve({ }); } + // A post in another language, in the reader's (LibreTranslate behind a + // cache — see translate.js). POST { text, target, source? } → + // { text, source }; GET /translate/languages lists what it can do. + if (url.pathname === '/translate/languages' && req.method === 'GET') { + try { return new Response(JSON.stringify(await languages()), { headers: { 'content-type': 'application/json', 'access-control-allow-origin': '*', 'cache-control': 'public, max-age=3600' } }); } + catch { return json({ error: 'translation unavailable' }, 503); } + } + if (url.pathname === '/translate' && req.method === 'POST') { + if (!rateOk(ip, 60)) return json({ error: 'rate limited' }, 429); + let body = null; + try { body = await req.json(); } catch {} + if (!body || typeof body.text !== 'string' || typeof body.target !== 'string') return json({ error: 'bad request' }, 400); + try { return json(await translate(body.text, body.target, body.source)); } + catch (e) { return json({ error: e.message || 'translation failed' }, e.status || 503); } + } + if (url.pathname === '/health') { return json({ ok: true, domain: DOMAIN, names: Object.keys(state.names).length, diff --git a/names/translate.js b/names/translate.js new file mode 100644 index 0000000..7d6181d --- /dev/null +++ b/names/translate.js @@ -0,0 +1,82 @@ +// Post translation for coinos v3: a thin, cached front for a LibreTranslate +// container on the same docker network. Notes are public, so nothing private +// passes through here; the cache keys on the text itself, so a popular post +// is translated once whoever asks. +import { createHash } from 'node:crypto'; +import { splitProse, hasWords, translatePieces } from '../src/translate-split.js'; + +const LT = process.env.LT_URL || 'http://libretranslate:5000'; +const MAX_CHARS = 5000; +const CACHE_MAX = 20000; +const BUSY_MAX = 4; // LibreTranslate is CPU-bound: a few at once, the rest wait their turn +const cache = new Map(); // sha256(target|text) -> { text, source } +let busy = 0; +const waiting = []; +let langs = null, langsAt = 0; + +async function slot() { + if (busy < BUSY_MAX) { busy++; return; } + await new Promise((r) => waiting.push(r)); + busy++; +} +function release() { busy--; const next = waiting.shift(); if (next) next(); } + +// The languages the container has models for: { code: [targets] } +export async function languages() { + if (langs && Date.now() - langsAt < 3600_000) return langs; + const r = await fetch(LT + '/languages', { signal: AbortSignal.timeout(5000) }); + if (!r.ok) throw new Error('languages ' + r.status); + const list = await r.json(); + langs = Object.fromEntries(list.map((l) => [l.code, l.targets || []])); + if (langs['zh-Hans']) langs.zh = langs['zh-Hans']; // the app says 'zh' + langsAt = Date.now(); + return langs; +} + +// The app's locale codes ↔ LibreTranslate's ('zh' is Simplified there). +const norm = (c) => String(c || '').toLowerCase().split(/[-_]/)[0]; +const toLt = (c) => (c === 'zh' ? 'zh-Hans' : c); + +export async function translate(text, target, source = 'auto') { + text = String(text || ''); + target = norm(target); + source = source && source !== 'auto' ? norm(source) : 'auto'; + if (!text.trim()) return { text: '', source }; + if (text.length > MAX_CHARS) throw Object.assign(new Error('too long'), { status: 413 }); + const have = await languages(); + if (!have[toLt(target)]) throw Object.assign(new Error('unsupported language'), { status: 422 }); + const key = createHash('sha256').update(target + '|' + source + '|' + text).digest('hex'); + const hit = cache.get(key); + if (hit) { cache.delete(key); cache.set(key, hit); return hit; } + const prose = splitProse(text).filter((p) => p.prose && hasWords(p.prose)); + if (!prose.length) return { text, source: source === 'auto' ? null : source }; + await slot(); + try { + const call = async (path, body) => { + const r = await fetch(LT + path, { + method: 'POST', headers: { 'content-type': 'application/json' }, + body: JSON.stringify(body), signal: AbortSignal.timeout(30_000), + }); + const out = await r.json().catch(() => ({})); + if (!r.ok) throw Object.assign(new Error(out.error || path + ' ' + r.status), { status: r.status === 400 ? 422 : 502 }); + return out; + }; + let from = source, confidence = null; + if (from === 'auto') { + const d = await call('/detect', { q: prose.map((p) => p.prose).join('\n') }); + from = norm(d[0]?.language) || null; + confidence = d[0]?.confidence ?? null; + if (!from) throw Object.assign(new Error('unknown language'), { status: 422 }); + } + if (from === target) return { text, source: from, confidence }; + if (!have[toLt(from)]) throw Object.assign(new Error('unsupported language'), { status: 422 }); + const translated = await translatePieces(text, async (qs) => { + const r = await call('/translate', { q: qs, source: toLt(from), target: toLt(target), format: 'text' }); + return Array.isArray(r.translatedText) ? r.translatedText : [r.translatedText]; + }); + const out = { text: translated, source: from, confidence }; + cache.set(key, out); + if (cache.size > CACHE_MAX) cache.delete(cache.keys().next().value); + return out; + } finally { release(); } +} diff --git a/tools/translate-test.js b/tools/translate-test.js new file mode 100644 index 0000000..2760693 --- /dev/null +++ b/tools/translate-test.js @@ -0,0 +1,116 @@ +// Translation: posts that read foreign are shown in the app's language +// ("Translated from Spanish · Show original"), automatically by default; +// with the switch off they offer Translate instead; ⋯ → Translate asks for +// any post. The registrar's /translate is stubbed here (window.fetch shim). +// Run: bun tools/translate-test.js +import puppeteer from './puppeteer.js'; +import { cacheKeyFor } from '../src/wallet.js'; +import { generateSecretKey, getPublicKey, finalizeEvent } from 'nostr-tools/pure'; +import { buildHtml } from '../build.js'; +const sleep = (ms) => new Promise((r) => setTimeout(r, ms)); +let ok = true; +const check = (n, c, d = '') => { console.log(` ${c ? '✓' : '✗'} ${n}${d ? ' — ' + d : ''}`); if (!c) ok = false; }; +const AUTHOR = (() => { const sk = generateSecretKey(); return { sk, pk: getPublicKey(sk) }; })(); +const now = Math.floor(Date.now() / 1000); +const post = (content, ago) => finalizeEvent({ kind: 1, created_at: now - ago, tags: [], content }, AUTHOR.sk); +const ES = post('Hola a todos, hoy compré café con bitcoin en el mercado y fue muy fácil para todos los que estaban allí', 100); +const JA = post('ビットコインは素晴らしいです。明日また会いましょう。', 200); +const EN = post('An ordinary post in English about the weather and the market today', 300); +const ANSWERS = { + [ES.content]: { text: 'Hello everyone, today I bought coffee with bitcoin at the market and it was very easy for everyone there', source: 'es' }, + [JA.content]: { text: 'Bitcoin is great. Let’s meet again tomorrow.', source: 'ja' }, + [EN.content]: { text: EN.content, source: 'en' }, +}; + +const html = await buildHtml({ minify: true, pwa: false }); +const server = Bun.serve({ port: 5303, fetch: () => new Response(html, { headers: { 'content-type': 'text/html' } }) }); +const browser = await puppeteer.launch({ executablePath: '/usr/bin/google-chrome', headless: 'new', args: ['--no-sandbox'] }); +const page = await browser.newPage(); +await page.evaluateOnNewDocument((answers) => { + window.__tr = []; + try { delete self.Translator; } catch {} // the registrar path, every time + const real = window.fetch; + window.fetch = async (url, opts) => { + if (String(url) === 'https://names.coinos.io/translate') { + const body = JSON.parse(opts.body); + window.__tr.push(body); + const a = answers[body.text]; + await new Promise((r) => setTimeout(r, 150)); + return new Response(JSON.stringify(a || { error: 'nope' }), { status: a ? 200 : 503, headers: { 'content-type': 'application/json' } }); + } + return real(url, opts); + }; + const Real = window.WebSocket; + window.WebSocket = function (...a) { + const ws = new Real(...a); + const send = ws.send.bind(ws); + ws.send = (data) => { + try { const m = JSON.parse(data); if (m[0] === 'REQ') { setTimeout(() => ws.dispatchEvent(new MessageEvent('message', { data: JSON.stringify(['EOSE', m[1]]) })), 0); return; } if (m[0] === 'EVENT') return; } catch {} + return send(data); + }; + return ws; + }; + window.WebSocket.prototype = Real.prototype; Object.assign(window.WebSocket, Real); +}, ANSWERS); +const errs = []; +page.on('pageerror', (e) => errs.push(String(e))); +const click = (t) => page.evaluate((x) => { const e = [...document.querySelectorAll('button')].find((n) => n.textContent.trim().toLowerCase().includes(x)); if (e) { e.click(); return true; } return false; }, t); +const text = () => page.evaluate(() => document.body.innerText); +const waitText = async (x, ms = 25000) => { for (let i = 0; i < ms / 250; i++) { if ((await text()).toLowerCase().includes(x.toLowerCase())) return true; await sleep(250); } return false; }; +const has = async (x) => (await text()).includes(x); +const row = (m) => page.evaluate((m) => { const r = [...document.querySelectorAll('[data-zap-post]')].find((n) => n.textContent.includes(m)); return r ? r.innerText : null; }, m); +const openFeed = async () => { await page.evaluate(() => document.querySelector('.app-nav-button[aria-label="Feed"]')?.click()); await sleep(1200); }; + +try { + await page.setViewport({ width: 390, height: 844 }); + await page.goto('http://localhost:5303/', { waitUntil: 'domcontentloaded' }); + await page.evaluate(() => { localStorage.setItem('btc-wallet-network', 'regtest'); localStorage.setItem('btc-wallet-lang', 'en'); }); + await page.reload({ waitUntil: 'domcontentloaded' }); + await sleep(400); + await click('create a new wallet'); await sleep(600); + await page.waitForSelector('.words .w .t'); + const mn = (await page.$$eval('.words .w .t', (l) => l.map((e) => e.textContent.trim()))).join(' '); + await click('skip verification'); + await waitText('receive', 20000); + const base = cacheKeyFor(mn + '\n'); + await page.evaluate(([k, pk, ns]) => { + localStorage.setItem(k + ':follows', JSON.stringify({ tags: [['p', pk]], c: '', at: Math.floor(Date.now() / 1000) })); + localStorage.setItem(k + ':feedNotes', JSON.stringify(ns)); + localStorage.setItem(k + ':profiles', JSON.stringify({ [pk]: { name: 'Globetrotter', t: Date.now() } })); + localStorage.setItem('btc-wallet-feed:' + k, 'following'); + }, [base, AUTHOR.pk, [ES, JA, EN]]); + await page.reload({ waitUntil: 'domcontentloaded' }); + await waitText('receive', 20000); + await openFeed(); + + console.log('\n[automatic, the default]'); + check('the Spanish post is shown in English', await waitText('today I bought coffee with bitcoin', 10000) && !(await has('compré café'))); + check('...saying where it came from', /Translated from Spanish/.test(await row('today I bought coffee') || ''), (await row('today I bought coffee') || '').split('\n').slice(-6).join(' | ')); + check('the Japanese post too', await waitText('Bitcoin is great', 5000) && /Translated from Japanese/.test(await row('Bitcoin is great') || '')); + const asked = await page.evaluate(() => window.__tr.map((b) => b.text.slice(0, 12) + '>' + b.target)); + check('the English post is left alone (never sent)', asked.length === 2 && !asked.some((x) => x.startsWith('An ordinary')), asked.join(', ')); + await page.evaluate(() => { const r = [...document.querySelectorAll('[data-zap-post]')].find((n) => n.textContent.includes('today I bought coffee')); [...r.querySelectorAll('.note-translate-btn')].find((b) => /original/i.test(b.textContent)).click(); }); + await sleep(400); + check('Show original brings the Spanish back', await has('compré café') && !(await has('today I bought coffee'))); + check('...with the offer to translate again', /Translate/.test(await row('compré café') || '')); + + console.log('\n[⋯ → Translate on a post the guess calls English]'); + await page.evaluate(() => { const r = [...document.querySelectorAll('[data-zap-post]')].find((n) => n.textContent.includes('An ordinary post')); [...r.querySelectorAll('button')].find((b) => b.getAttribute('aria-label') === 'More').click(); }); + await sleep(300); + check('the menu offers Translate', await page.evaluate(() => [...document.querySelectorAll('.confirm-pop button')].some((b) => /Translate/.test(b.textContent)))); + await page.evaluate(() => [...document.querySelectorAll('.confirm-pop button')].find((b) => /Translate/.test(b.textContent)).click()); + check('it says the post is already in English', await waitText('already in your language', 4000)); + + console.log('\n[automatic switched off]'); + await page.evaluate(() => localStorage.setItem('coinos-auto-translate', '0')); + await page.reload({ waitUntil: 'domcontentloaded' }); + await waitText('receive', 20000); + await openFeed(); + check('foreign posts stay as written', await waitText('明日また会いましょう', 10000) && !(await has('Bitcoin is great'))); + check('...offering Translate', /Translate/.test(await row('明日また') || '')); + await page.evaluate(() => { const r = [...document.querySelectorAll('[data-zap-post]')].find((n) => n.textContent.includes('明日また')); r.querySelector('.note-translate-btn').click(); }); + check('tapping it translates (remembered from before: no new request)', await waitText('Bitcoin is great', 5000)); + check('no page errors', errs.length === 0, errs.slice(0, 2).join(' | ')); +} finally { await browser.close(); server.stop(true); } +console.log(ok ? '\n✅ posts translate into the reader\'s language' : '\n❌ failed'); +process.exit(ok ? 0 : 1); |
