live-translator-nodeapp/index.js

527 lines
17 KiB
JavaScript
Raw Normal View History

2020-06-18 08:12:56 +00:00
const translation_token_generator = require('google-translate-token');
const bodyParser = require('body-parser');
const express = require('express');
const request = require('request');
const cors = require('cors');
const open = require('open');
const fs = require('fs');
2020-06-18 08:12:56 +00:00
var server = express();
server.use(cors());
server.use(express.urlencoded({extended: true}));
server.use(bodyParser.json({extended: true}));
const index_page = `
<!DOCUMENT html>
<html lang="br">
<head>
<meta charset="utf-8">
<title>Tradutor simultâneo</title>
<link rel="stylesheet" type="text/css" href="/main.css" />
<script src="/main.js" defer></script>
</head>
<body>
<div>
<div id="cfgformp">
<h3 class="nospacing">Configurações</h3>
<form id="cfgform" onsubmit="return false;">
<div class="formgroup">
<h5 class="nospacing">Idioma da fala</h5>
<input type="radio" id="ilpt" name="il" value="0" checked>
<label for="ilpt">PT</label><br>
<input type="radio" id="ilen" name="il" value="1">
<label for="ilen">EN</label><br>
<input type="radio" id="iles" name="il" value="2">
<label for="iles">ES</label><br>
</div>
<div class="formgroup">
<h5 class="nospacing">Idioma da 1ª tradução</h5>
<input type="radio" id="1lpt" name="1l" value="0">
<label for="1lpt">PT</label><br>
<input type="radio" id="1len" name="1l" value="1">
<label for="1len">EN</label><br>
<input type="radio" id="1les" name="1l" value="2">
<label for="1les">ES</label><br>
<input type="radio" id="1l_" name="1l" value="-1" checked>
<label for="1l_">Ignorar</label><br>
</div>
<div class="formgroup">
<h5 class="nospacing">Idioma da 2ª tradução</h5>
<input type="radio" id="2lpt" name="2l" value="0">
<label for="2lpt">PT</label><br>
<input type="radio" id="2len" name="2l" value="1">
<label for="2len">EN</label><br>
<input type="radio" id="2les" name="2l" value="2">
<label for="2les">ES</label><br>
<input type="radio" id="2l_" name="2l" value="-1" checked>
<label for="2l_">Ignorar</label><br>
</div>
<div class="formgroup">
<h5 class="nospacing">Controle</h5>
<input type="button" value="Aplicar" onclick="return aplicar();">
</div>
<div class="formgroup">
<h5 class="nospacing">Estatísticas</h5>
<div><span>Traduções solicitadas: </span><span id="trad_count"></span></div>
<div><span>Traduções recebidas: </span><span id="trans_count"></span></div>
</div>
2020-06-18 08:12:56 +00:00
</form>
<hr>
</div>
<div class="loader">
<h2>Configure antes de usar</h2>
<hr>
</div>
<dl id="transcriptions">
<dt id="i1dt">fala</dt>
<dd id="i1dd"><div class="listened"></div></dd>
<dt id="o1dt">trad1</dt>
<dd id="o1dd"><div class="translated1"></div></dd>
<dt id="o2dt">trad2</dt>
<dd id="o2dd"><div class="translated2"></div></dd>
</dl>
<hr>
</div>
</body>
</html>
`.trim();
const main_style = `
body {
font-family: sans-serif;
background-color: black;
color: white;
}
dl#transcriptions {
display: flex;
flex-wrap: wrap;
}
dl#transcriptions>dt {
display: block;
max-width: 4rem;
flex: 0 0 4rem;
border-right: 1px solid white;
border-bottom: 1px solid white;
}
dl#transcriptions>dd {
margin-inline-start: 4em;
max-width: calc( 100% - 9rem );
flex: 0 0 calc( 100% - 9rem );
height: 3.4em;
overflow-y: hidden;
display: flex;
flex-direction: column-reverse;
font-size: 3em;
font-weight: bold;
}
dl#transcriptions>dt, dl#transcriptions>dd {
align-items: baseline;
margin: 1rem;
margin-top: 0.5rem;
}
form {
display: flex;
flex-wrap: wrap;
}
.nospacing {
margin: 0px;
padding: 0px;
}
.formgroup {
padding-left: 1rem;
padding-right: 1rem;
border-right: 1px solid white;
}
form input[type=button] {
background: black;
color: white;
border: white 0.08rem solid;
border-radius: 0.2rem;
padding: 0.2rem;
padding-left: 0.4rem;
padding-right: 0.4rem;
}
2020-06-18 08:12:56 +00:00
`.trim();
const main_script = `
window.SpeechRecognition = window.webkitSpeechRecognition || window.SpeechRecognition;
LANGS_1 = ['pt', 'en', 'es'];
LANGS_2 = ['pt-BR', 'en-US', 'es-AR'];
SPOKEN_LANG_POS = 0;
SELECTED_LANG1_POS = 1;
SELECTED_LANG2_POS = 2;
recognizing = false;
recognition = new Object();
recognition.continuous = false;
recognition.interimResults = true;
recognition.lang = LANGS_2[SPOKEN_LANG_POS];
recog_id = generate_random_int_id();
from_lang = LANGS_1[SPOKEN_LANG_POS];
to_lang1 = LANGS_1[SELECTED_LANG1_POS];
to_lang2 = LANGS_1[SELECTED_LANG2_POS];
loader = document.querySelector('.loader');
listened = document.querySelector('.listened');
translated1 = document.querySelector('.translated1');
translated2 = document.querySelector('.translated2');
cfgform = document.querySelector('#cfgform');
cfgformp = document.querySelector('#cfgformp');
trad_count = document.querySelector('#trad_count');
trans_count = document.querySelector('#trans_count');
2020-06-18 08:12:56 +00:00
bg = document.querySelector('html');
recognition.onstart = function() {
recognizing = true;
console.log('onstart');
}
recognition.onerror = function(event) {
console.log('onerror', event);
aplicar();
}
recognition.onend = function() {
recognizing = false;
console.log('onend');
aplicar();
}
recognition.onresult = function(event) {
var copied = new Object();
copied.lang = from_lang;
copied.resultIndex = event.resultIndex;
copied.identity = recog_id;
copied.results = new Array();
for(var i = 0; i < event.results.length; i++) {
speech_recognition_result = new Object();
speech_recognition_result.isFinal = event.results[i].isFinal;
speech_recognition_result.alternatives = new Array();
for(var j = 0; j < event.results[i].length; j++) {
speech_recognition_alternative = new Object();
speech_recognition_alternative.transcript = event.results[i][j].transcript;
speech_recognition_alternative.confidence = event.results[i][j].confidence;
speech_recognition_result.alternatives.push(speech_recognition_alternative);
}
copied.results.push(speech_recognition_result);
}
doXhr(
'POST',
'/set/speech',
()=>{},
()=>{},
JSON.stringify(copied),
"application/json"
);
}
function aplicar_conf() {
newvalues = Object.fromEntries(
Array.from(
cfgform.querySelectorAll(':checked')
).map(x=>[x.name, parseInt(x.value)])
)
loader.style.display = 'none';
SPOKEN_LANG_POS = newvalues['il']
SELECTED_LANG1_POS = newvalues['1l']
SELECTED_LANG2_POS = newvalues['2l']
from_lang = LANGS_1[SPOKEN_LANG_POS] || null;
to_lang1 = LANGS_1[SELECTED_LANG1_POS] || null;
to_lang2 = LANGS_1[SELECTED_LANG2_POS] || null;
if (from_lang === null) return;
recognition_old = recognition;
recognition = new SpeechRecognition();
recognition.lang = LANGS_2[SPOKEN_LANG_POS] || null;
recognition.onstart = recognition_old.onstart;
recognition.onerror = recognition_old.onerror;
recognition.onend = recognition_old.onend;
recognition.onresult = recognition_old.onresult;
recognition.continuous = recognition_old.continuous;
recognition.interimResults = recognition_old.interimResults;
recognition_old.onend = null;
recognition_old.onerror = null;
recog_id = generate_random_int_id();
}
function aplicar(){
aplicar_conf();
recognition.start();
doXhr(
'POST',
'/set/config',
()=>{},
()=>{},
JSON.stringify({
'from_lang': from_lang,
'to_lang1': to_lang1,
'to_lang2': to_lang2,
}),
"application/json");
}
function doXhr(method, URL, onsuccess, onfailure, body, bodyMime) {
var xhr = new XMLHttpRequest();
xhr.open(method, URL);
xhr.onload = function() {
if (xhr.status != 200)
onfailure(xhr);
else
onsuccess(xhr);
};
if (bodyMime) xhr.setRequestHeader("Content-Type", bodyMime);
xhr.onerror = () => onfailure(xhr);
xhr.send(body);
return xhr;
}
function generate_random_int_id() {
return Math.trunc(Math.random()*Math.pow(2, 22));
}
function check_translations() {
setTimeout(check_translations, 100);
doXhr(
'GET',
'/get/speech',
(xhr)=>{
var translations = JSON.parse(xhr.responseText);
listened.innerText = translations.boxes[0];
translated1.innerText = translations.boxes[1];
translated2.innerText = translations.boxes[2];
trad_count.innerText = translations.translation_count;
trans_count.innerText = translations.translated_count;
2020-06-18 08:12:56 +00:00
},
()=>{}
);
}
setTimeout(check_translations, 100);
`.trim();
var past_input_strings = [];
var latest_input_id = null;
var latest_input_string = '';
var out_strings = ['', '', ''];
var lang_array = ['pt', null, null];
var last_translations = {};
var translation_timeout = 3000;
var translator_locked = false;
const box_names = ['fala', 'trad1', 'trad2'];
var translations_ordered = 0;
var translations_received = 0;
2020-06-18 08:12:56 +00:00
function update_file(location, new_content) {
return new Promise((resolve, reject) => {
var errCallback = (err) => {
if (err) reject(err);
return !!err;
};
var writeFileCallback = (err) => {
if(!errCallback(err)) resolve(true);
}
fs.exists(location, exists => {
if(!exists)
fs.writeFile(
location,
new_content,
undefined,
writeFileCallback);
else {
fs.readFile(location, (err, old_content) => {
if (!errCallback(err)) {
if (old_content==new_content)
resolve(true);
else
fs.writeFile(
location,
new_content,
undefined,
writeFileCallback);
}
});
}
});
});
}
function do_translate(text, lang_from, lang_to) {
return new Promise((resolve, reject) => {
if(!text) return resolve(text);
if(!lang_from) return resolve(text);
if(!lang_to) return resolve(text);
translation_token_generator.get(text).then((token) => {
return request.post(
'https://translate.google.com/translate_a/single?'+
'client=webapp&'+
`sl=${lang_from}&`+
`tl=${lang_to}&`+
`hl=${lang_from}&`+
'dt=at&'+
'dt=bd&'+
'dt=ex&'+
'dt=ld&'+
'dt=md&'+
'dt=qca&'+
'dt=rw&'+
'dt=rm&'+
'dt=ss&'+
'dt=t&'+
'otf=1&'+
'ssel=0&'+
'tsel=0&'+
'kc=1&'+
`${token.name}=${token.value}`,
{form: {q: text}},
(error, response, body) => {
try {
body = JSON.parse(body);
} catch (e) {
2020-06-20 06:19:00 +00:00
return reject([response.statusCode, response.statusMessage, response.request.uri.href]);
2020-06-18 08:12:56 +00:00
}
resolve(get_first_while_array(body));
}
);
}, reject);
});
}
function lock_translator_timedout(){
translator_locked = false;
}
function lock_translator(){
translator_locked = true;
setTimeout(lock_translator_timedout, translation_timeout);
}
function prune_translations() {
for (var k in last_translations)
while (last_translations[k].length > 2048)
last_translations[k].shift();
}
function make_translations(text) {
prune_translations();
var translations_to_get = [];
for(var i = 1; i < lang_array.length; i++){
if (lang_array[i]==null)
out_strings[i] = '';
else if (lang_array[i]==lang_array[0])
out_strings[i] = out_strings[0];
else if (translations_to_get.indexOf(lang_array[i])==-1)
translations_to_get.push(lang_array[i]);
}
if (!translator_locked) {
lock_translator();
for (let translang of translations_to_get){
translations_ordered += 1;
2020-06-18 08:12:56 +00:00
last_translations[translang] = last_translations[translang] || [];
cached = Object.fromEntries(last_translations[translang])[text];
if (cached) {
for(var j = 1; j < lang_array.length; j++){
let v = parseInt(j);
if (lang_array[v]==translang)
out_strings[v] = cached;
}
} else {
do_translate(text, lang_array[0], translang).then((translation) => {
last_translations[translang].push([text, translation]);
for(var k = 1; k < lang_array.length; k++){
let v = parseInt(k);
if (lang_array[v]==translang)
out_strings[v] = translation;
}
translations_received += 1;
2020-06-20 06:19:00 +00:00
}, err => console.log(`Erro ao traduzir para '${translang}': ${err}`));
2020-06-18 08:12:56 +00:00
}
}
}
}
function get_first_while_array(a) {
if (Array.isArray(a))
return get_first_while_array(a[0]);
else
return a;
}
server.get("/", (req, res) => {
res.writeHead(200, "OK", {'Content-Type': 'text/html'});
res.write(index_page);
res.end();
});
server.get("/main.css", (req, res) => {
res.writeHead(200, "OK", {'Content-Type': 'text/css'});
res.write(main_style);
res.end();
});
server.get("/main.js", (req, res) => {
res.writeHead(200, "OK", {'Content-Type': 'application/javascript'});
res.write(main_script);
res.end();
});
server.post("/set/config", (req, res) => {
lang_array[1] = req.body.to_lang1;
lang_array[2] = req.body.to_lang2;
res.end();
});
server.get("/get/speech", (req, res) => {
res.writeHead(200, "OK", {'Content-Type': 'application/json'});
res.write(JSON.stringify({
boxes: out_strings,
translation_count: translations_ordered,
translated_count: translations_received,
}));
2020-06-18 08:12:56 +00:00
res.end();
for (let i in lang_array) {
update_file(
'caixa_'+box_names[i]+'.txt',
out_strings[i]);
}
for (var lang of [...(new Set(lang_array))]) {
let i = lang_array.indexOf(lang);
if (lang_array[i]===null) continue;
update_file(
'idioma_'+lang+'.txt',
out_strings[i]);
}
});
server.post("/set/speech", (req, res) => {
var speech = req.body;
res.end();
if (speech.lang != lang_array[0]) {
last_translations = {};
out_strings[0] = '';
past_input_strings = [];
latest_input_string = '';
latest_input_id = null;
}
if (speech.identity != latest_input_id) {
latest_input_id = speech.identity;
past_input_strings.push(latest_input_string);
if (past_input_strings.length > 5)
past_input_strings.shift();
latest_input_string = '';
}
var s = '';
for (var i = speech.resultIndex; i < speech.results.length; i++)
s += ' ' + speech.results[i].alternatives[0].transcript;
s = s.split(' ').filter(x=>x.length).join(' ');
latest_input_string = s;
out_strings[0] = (
past_input_strings.join(' ')+' '+latest_input_string
).split(' ').filter(x=>x.length).join(' ');
var translation_candidate = out_strings[0].split(' ').splice(-20).join(' ');
make_translations(translation_candidate);
});
server.listen(13653, '127.0.0.1', ()=>{
console.log('Abra http://localhost:13653 no Chrome.');
open('http://localhost:13653');
});
2020-06-20 06:19:00 +00:00