Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -3,6 +3,7 @@
|
|
| 3 |
Kabyle Translation Hub - Hugging Face Spaces Edition
|
| 4 |
Simple translation interface: MarianMT vs LibreTranslate
|
| 5 |
Users choose their preferred translation - no metrics, no scores.
|
|
|
|
| 6 |
"""
|
| 7 |
|
| 8 |
import warnings
|
|
@@ -34,33 +35,35 @@ model = None
|
|
| 34 |
tokenizer = None
|
| 35 |
device = None
|
| 36 |
|
|
|
|
| 37 |
def load_model():
|
| 38 |
"""Load MarianMT model once and cache it"""
|
| 39 |
global model, tokenizer, device
|
| 40 |
-
|
| 41 |
if model is None:
|
| 42 |
print("Loading MarianMT model...")
|
| 43 |
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
| 44 |
-
|
| 45 |
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, use_fast=False)
|
| 46 |
model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_ID).to(device).eval()
|
| 47 |
-
|
| 48 |
print(f"Model loaded successfully on {device}")
|
| 49 |
-
|
| 50 |
return model, tokenizer, device
|
| 51 |
|
|
|
|
| 52 |
def translate_marian(text):
|
| 53 |
"""Translate using MarianMT with multiple alternatives"""
|
| 54 |
if not text or not text.strip():
|
| 55 |
return ["Please enter text to translate"]
|
| 56 |
-
|
| 57 |
try:
|
| 58 |
model, tokenizer, device = load_model()
|
| 59 |
-
|
| 60 |
# Prepare inputs
|
| 61 |
inputs = tokenizer(text, return_tensors='pt', padding=True, truncation=True, max_length=512)
|
| 62 |
inputs = {k: v.to(device) for k, v in inputs.items()}
|
| 63 |
-
|
| 64 |
with torch.no_grad():
|
| 65 |
# Simple beam search without group beam search
|
| 66 |
outputs = model.generate(
|
|
@@ -71,21 +74,22 @@ def translate_marian(text):
|
|
| 71 |
early_stopping=True,
|
| 72 |
do_sample=False,
|
| 73 |
)
|
| 74 |
-
|
| 75 |
translations = []
|
| 76 |
for output in outputs:
|
| 77 |
trans = tokenizer.decode(output, skip_special_tokens=True)
|
| 78 |
if trans and trans not in translations:
|
| 79 |
translations.append(trans)
|
| 80 |
-
|
| 81 |
return translations if translations else ["[Error: No translation generated]"]
|
| 82 |
-
|
| 83 |
except Exception as e:
|
| 84 |
print(f"MarianMT translation error: {e}")
|
| 85 |
import traceback
|
| 86 |
traceback.print_exc()
|
| 87 |
return [f"[Error: {str(e)}]"]
|
| 88 |
|
|
|
|
| 89 |
def translate_libre_variant(text, variant_code):
|
| 90 |
"""Translate using a specific LibreTranslate variant"""
|
| 91 |
try:
|
|
@@ -101,12 +105,13 @@ def translate_libre_variant(text, variant_code):
|
|
| 101 |
except Exception as e:
|
| 102 |
return {"success": False, "text": f"[Error: {str(e)[:50]}]"}
|
| 103 |
|
|
|
|
| 104 |
def translate_libre_all_variants(text):
|
| 105 |
"""Translate using all LibreTranslate variants in parallel"""
|
| 106 |
results = {}
|
| 107 |
with ThreadPoolExecutor(max_workers=4) as executor:
|
| 108 |
future_to_name = {
|
| 109 |
-
executor.submit(translate_libre_variant, text, code): name
|
| 110 |
for name, code in KABYLE_VARIANTS.items()
|
| 111 |
}
|
| 112 |
for future in as_completed(future_to_name, timeout=15):
|
|
@@ -117,6 +122,7 @@ def translate_libre_all_variants(text):
|
|
| 117 |
results[name] = {"success": False, "text": f"[Error: {e}]"}
|
| 118 |
return results
|
| 119 |
|
|
|
|
| 120 |
HTML_TEMPLATE = """
|
| 121 |
<!DOCTYPE html>
|
| 122 |
<html lang="en">
|
|
@@ -138,9 +144,9 @@ HTML_TEMPLATE = """
|
|
| 138 |
--radius: 12px;
|
| 139 |
--radius-sm: 8px;
|
| 140 |
}
|
| 141 |
-
|
| 142 |
* { margin: 0; padding: 0; box-sizing: border-box; }
|
| 143 |
-
|
| 144 |
body {
|
| 145 |
font-family: system-ui, -apple-system, sans-serif;
|
| 146 |
background: linear-gradient(135deg, var(--bg-primary) 0%, #1a1a2e 100%);
|
|
@@ -148,17 +154,17 @@ HTML_TEMPLATE = """
|
|
| 148 |
min-height: 100vh;
|
| 149 |
padding: 20px;
|
| 150 |
}
|
| 151 |
-
|
| 152 |
.container {
|
| 153 |
max-width: 1000px;
|
| 154 |
margin: 0 auto;
|
| 155 |
}
|
| 156 |
-
|
| 157 |
header {
|
| 158 |
text-align: center;
|
| 159 |
padding: 40px 20px;
|
| 160 |
}
|
| 161 |
-
|
| 162 |
header h1 {
|
| 163 |
font-size: 2.5rem;
|
| 164 |
font-weight: 800;
|
|
@@ -167,12 +173,12 @@ HTML_TEMPLATE = """
|
|
| 167 |
-webkit-text-fill-color: transparent;
|
| 168 |
margin-bottom: 10px;
|
| 169 |
}
|
| 170 |
-
|
| 171 |
header p {
|
| 172 |
color: var(--text-secondary);
|
| 173 |
font-size: 1.1rem;
|
| 174 |
}
|
| 175 |
-
|
| 176 |
.input-card {
|
| 177 |
background: var(--bg-secondary);
|
| 178 |
border-radius: var(--radius);
|
|
@@ -180,13 +186,13 @@ HTML_TEMPLATE = """
|
|
| 180 |
margin-bottom: 24px;
|
| 181 |
border: 1px solid var(--border);
|
| 182 |
}
|
| 183 |
-
|
| 184 |
.input-group {
|
| 185 |
display: flex;
|
| 186 |
flex-direction: column;
|
| 187 |
gap: 16px;
|
| 188 |
}
|
| 189 |
-
|
| 190 |
textarea {
|
| 191 |
width: 100%;
|
| 192 |
padding: 16px;
|
|
@@ -200,16 +206,16 @@ HTML_TEMPLATE = """
|
|
| 200 |
font-family: inherit;
|
| 201 |
transition: border-color 0.2s;
|
| 202 |
}
|
| 203 |
-
|
| 204 |
textarea:focus {
|
| 205 |
outline: none;
|
| 206 |
border-color: var(--accent-primary);
|
| 207 |
}
|
| 208 |
-
|
| 209 |
textarea::placeholder {
|
| 210 |
color: var(--text-secondary);
|
| 211 |
}
|
| 212 |
-
|
| 213 |
.btn-primary {
|
| 214 |
background: linear-gradient(135deg, var(--accent-primary), var(--accent-secondary));
|
| 215 |
color: white;
|
|
@@ -225,31 +231,31 @@ HTML_TEMPLATE = """
|
|
| 225 |
justify-content: center;
|
| 226 |
gap: 8px;
|
| 227 |
}
|
| 228 |
-
|
| 229 |
.btn-primary:hover {
|
| 230 |
transform: translateY(-2px);
|
| 231 |
box-shadow: 0 10px 30px rgba(99, 102, 241, 0.3);
|
| 232 |
}
|
| 233 |
-
|
| 234 |
.btn-primary:disabled {
|
| 235 |
opacity: 0.6;
|
| 236 |
cursor: not-allowed;
|
| 237 |
transform: none;
|
| 238 |
}
|
| 239 |
-
|
| 240 |
.results-container {
|
| 241 |
display: flex;
|
| 242 |
flex-direction: column;
|
| 243 |
gap: 20px;
|
| 244 |
}
|
| 245 |
-
|
| 246 |
.engine-card {
|
| 247 |
background: var(--bg-secondary);
|
| 248 |
border-radius: var(--radius);
|
| 249 |
border: 1px solid var(--border);
|
| 250 |
overflow: hidden;
|
| 251 |
}
|
| 252 |
-
|
| 253 |
.engine-header {
|
| 254 |
padding: 20px 24px;
|
| 255 |
background: var(--bg-tertiary);
|
|
@@ -258,7 +264,7 @@ HTML_TEMPLATE = """
|
|
| 258 |
align-items: center;
|
| 259 |
justify-content: space-between;
|
| 260 |
}
|
| 261 |
-
|
| 262 |
.engine-title {
|
| 263 |
display: flex;
|
| 264 |
align-items: center;
|
|
@@ -266,7 +272,7 @@ HTML_TEMPLATE = """
|
|
| 266 |
font-size: 1.1rem;
|
| 267 |
font-weight: 600;
|
| 268 |
}
|
| 269 |
-
|
| 270 |
.engine-badge {
|
| 271 |
font-size: 0.75rem;
|
| 272 |
padding: 4px 12px;
|
|
@@ -275,11 +281,11 @@ HTML_TEMPLATE = """
|
|
| 275 |
color: var(--text-secondary);
|
| 276 |
border: 1px solid var(--border);
|
| 277 |
}
|
| 278 |
-
|
| 279 |
.translation-list {
|
| 280 |
list-style: none;
|
| 281 |
}
|
| 282 |
-
|
| 283 |
.translation-item {
|
| 284 |
padding: 20px 24px;
|
| 285 |
border-bottom: 1px solid var(--border);
|
|
@@ -290,45 +296,45 @@ HTML_TEMPLATE = """
|
|
| 290 |
justify-content: space-between;
|
| 291 |
gap: 16px;
|
| 292 |
}
|
| 293 |
-
|
| 294 |
.translation-item:last-child {
|
| 295 |
border-bottom: none;
|
| 296 |
}
|
| 297 |
-
|
| 298 |
.translation-item:hover {
|
| 299 |
background: rgba(99, 102, 241, 0.1);
|
| 300 |
}
|
| 301 |
-
|
| 302 |
.translation-item.selected {
|
| 303 |
background: rgba(16, 185, 129, 0.15);
|
| 304 |
border-left: 4px solid var(--accent-success);
|
| 305 |
}
|
| 306 |
-
|
| 307 |
.translation-text {
|
| 308 |
flex: 1;
|
| 309 |
font-size: 1.05rem;
|
| 310 |
line-height: 1.6;
|
| 311 |
color: var(--text-primary);
|
| 312 |
}
|
| 313 |
-
|
| 314 |
.copy-icon {
|
| 315 |
opacity: 0;
|
| 316 |
color: var(--accent-primary);
|
| 317 |
transition: opacity 0.2s;
|
| 318 |
flex-shrink: 0;
|
| 319 |
}
|
| 320 |
-
|
| 321 |
.translation-item:hover .copy-icon {
|
| 322 |
opacity: 1;
|
| 323 |
}
|
| 324 |
-
|
| 325 |
.variant-grid {
|
| 326 |
display: grid;
|
| 327 |
grid-template-columns: repeat(auto-fit, minmax(300px, 1fr));
|
| 328 |
gap: 1px;
|
| 329 |
background: var(--border);
|
| 330 |
}
|
| 331 |
-
|
| 332 |
.variant-item {
|
| 333 |
background: var(--bg-secondary);
|
| 334 |
padding: 16px 20px;
|
|
@@ -339,22 +345,22 @@ HTML_TEMPLATE = """
|
|
| 339 |
gap: 8px;
|
| 340 |
position: relative;
|
| 341 |
}
|
| 342 |
-
|
| 343 |
.variant-item:hover {
|
| 344 |
background: rgba(99, 102, 241, 0.1);
|
| 345 |
}
|
| 346 |
-
|
| 347 |
.variant-item.selected {
|
| 348 |
background: rgba(16, 185, 129, 0.15);
|
| 349 |
box-shadow: inset 4px 0 0 var(--accent-success);
|
| 350 |
}
|
| 351 |
-
|
| 352 |
.variant-header {
|
| 353 |
display: flex;
|
| 354 |
align-items: center;
|
| 355 |
justify-content: space-between;
|
| 356 |
}
|
| 357 |
-
|
| 358 |
.variant-name {
|
| 359 |
font-size: 0.75rem;
|
| 360 |
color: var(--text-secondary);
|
|
@@ -362,7 +368,7 @@ HTML_TEMPLATE = """
|
|
| 362 |
letter-spacing: 0.5px;
|
| 363 |
font-weight: 600;
|
| 364 |
}
|
| 365 |
-
|
| 366 |
.variant-code {
|
| 367 |
font-size: 0.7rem;
|
| 368 |
color: var(--text-secondary);
|
|
@@ -371,23 +377,23 @@ HTML_TEMPLATE = """
|
|
| 371 |
padding: 2px 8px;
|
| 372 |
border-radius: 4px;
|
| 373 |
}
|
| 374 |
-
|
| 375 |
.variant-text {
|
| 376 |
font-size: 1rem;
|
| 377 |
color: var(--text-primary);
|
| 378 |
line-height: 1.5;
|
| 379 |
}
|
| 380 |
-
|
| 381 |
.variant-item.error {
|
| 382 |
opacity: 0.6;
|
| 383 |
}
|
| 384 |
-
|
| 385 |
.variant-item.error .variant-text {
|
| 386 |
color: #ef4444;
|
| 387 |
font-family: monospace;
|
| 388 |
font-size: 0.85rem;
|
| 389 |
}
|
| 390 |
-
|
| 391 |
.source-display {
|
| 392 |
background: var(--bg-tertiary);
|
| 393 |
padding: 20px 24px;
|
|
@@ -395,7 +401,7 @@ HTML_TEMPLATE = """
|
|
| 395 |
margin-bottom: 20px;
|
| 396 |
border: 1px solid var(--border);
|
| 397 |
}
|
| 398 |
-
|
| 399 |
.source-label {
|
| 400 |
font-size: 0.75rem;
|
| 401 |
color: var(--text-secondary);
|
|
@@ -403,24 +409,24 @@ HTML_TEMPLATE = """
|
|
| 403 |
letter-spacing: 0.5px;
|
| 404 |
margin-bottom: 8px;
|
| 405 |
}
|
| 406 |
-
|
| 407 |
.source-text {
|
| 408 |
font-size: 1.1rem;
|
| 409 |
color: var(--text-primary);
|
| 410 |
line-height: 1.6;
|
| 411 |
}
|
| 412 |
-
|
| 413 |
.check-icon {
|
| 414 |
color: var(--accent-success);
|
| 415 |
opacity: 0;
|
| 416 |
transition: opacity 0.2s;
|
| 417 |
}
|
| 418 |
-
|
| 419 |
.translation-item.selected .check-icon,
|
| 420 |
.variant-item.selected .check-icon {
|
| 421 |
opacity: 1;
|
| 422 |
}
|
| 423 |
-
|
| 424 |
.toast {
|
| 425 |
position: fixed;
|
| 426 |
bottom: 24px;
|
|
@@ -436,12 +442,12 @@ HTML_TEMPLATE = """
|
|
| 436 |
z-index: 1000;
|
| 437 |
box-shadow: 0 10px 30px rgba(0,0,0,0.3);
|
| 438 |
}
|
| 439 |
-
|
| 440 |
.toast.show {
|
| 441 |
transform: translateX(-50%) translateY(0);
|
| 442 |
opacity: 1;
|
| 443 |
}
|
| 444 |
-
|
| 445 |
.loading {
|
| 446 |
display: inline-block;
|
| 447 |
width: 18px;
|
|
@@ -451,29 +457,43 @@ HTML_TEMPLATE = """
|
|
| 451 |
border-top-color: white;
|
| 452 |
animation: spin 0.8s linear infinite;
|
| 453 |
}
|
| 454 |
-
|
| 455 |
@keyframes spin {
|
| 456 |
to { transform: rotate(360deg); }
|
| 457 |
}
|
| 458 |
-
|
| 459 |
.empty-state {
|
| 460 |
text-align: center;
|
| 461 |
padding: 60px 20px;
|
| 462 |
color: var(--text-secondary);
|
| 463 |
}
|
| 464 |
-
|
| 465 |
.empty-state svg {
|
| 466 |
width: 64px;
|
| 467 |
height: 64px;
|
| 468 |
margin-bottom: 16px;
|
| 469 |
opacity: 0.5;
|
| 470 |
}
|
| 471 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 472 |
@media (max-width: 640px) {
|
| 473 |
header h1 {
|
| 474 |
font-size: 1.75rem;
|
| 475 |
}
|
| 476 |
-
|
| 477 |
.variant-grid {
|
| 478 |
grid-template-columns: 1fr;
|
| 479 |
}
|
|
@@ -486,14 +506,14 @@ HTML_TEMPLATE = """
|
|
| 486 |
<h1>English → Kabyle</h1>
|
| 487 |
<p>Choose the translation that suits you best</p>
|
| 488 |
</header>
|
| 489 |
-
|
| 490 |
<div class="input-card">
|
| 491 |
<form method="POST" id="translateForm">
|
| 492 |
<div class="input-group">
|
| 493 |
-
<textarea
|
| 494 |
-
name="text"
|
| 495 |
id="inputText"
|
| 496 |
-
placeholder="Enter English text to translate..."
|
| 497 |
required
|
| 498 |
autocomplete="off"
|
| 499 |
>{{ request.form.get('text', '') }}</textarea>
|
|
@@ -507,13 +527,13 @@ HTML_TEMPLATE = """
|
|
| 507 |
</div>
|
| 508 |
</form>
|
| 509 |
</div>
|
| 510 |
-
|
| 511 |
{% if source_text %}
|
| 512 |
<div class="source-display">
|
| 513 |
<div class="source-label">Source Text</div>
|
| 514 |
<div class="source-text">{{ source_text }}</div>
|
| 515 |
</div>
|
| 516 |
-
|
| 517 |
<div class="results-container">
|
| 518 |
<!-- MarianMT Results -->
|
| 519 |
<div class="engine-card">
|
|
@@ -542,7 +562,7 @@ HTML_TEMPLATE = """
|
|
| 542 |
{% endfor %}
|
| 543 |
</ul>
|
| 544 |
</div>
|
| 545 |
-
|
| 546 |
<!-- LibreTranslate Results -->
|
| 547 |
<div class="engine-card">
|
| 548 |
<div class="engine-header">
|
|
@@ -590,13 +610,17 @@ HTML_TEMPLATE = """
|
|
| 590 |
<p>Enter text above and click Translate to see results</p>
|
| 591 |
</div>
|
| 592 |
{% endif %}
|
|
|
|
|
|
|
|
|
|
|
|
|
| 593 |
</div>
|
| 594 |
-
|
| 595 |
<div class="toast" id="toast">Copied to clipboard!</div>
|
| 596 |
-
|
| 597 |
<script>
|
| 598 |
let selectedText = '';
|
| 599 |
-
|
| 600 |
function selectTranslation(element, text) {
|
| 601 |
element.parentElement.querySelectorAll('.translation-item').forEach(item => {
|
| 602 |
item.classList.remove('selected');
|
|
@@ -605,7 +629,7 @@ HTML_TEMPLATE = """
|
|
| 605 |
selectedText = text;
|
| 606 |
copyText(text, false);
|
| 607 |
}
|
| 608 |
-
|
| 609 |
function selectVariant(element, text) {
|
| 610 |
document.querySelectorAll('.variant-item').forEach(item => {
|
| 611 |
item.classList.remove('selected');
|
|
@@ -614,7 +638,7 @@ HTML_TEMPLATE = """
|
|
| 614 |
selectedText = text;
|
| 615 |
copyText(text, false);
|
| 616 |
}
|
| 617 |
-
|
| 618 |
async function copyText(text, showToast = true) {
|
| 619 |
try {
|
| 620 |
await navigator.clipboard.writeText(text);
|
|
@@ -633,14 +657,14 @@ HTML_TEMPLATE = """
|
|
| 633 |
}
|
| 634 |
}
|
| 635 |
}
|
| 636 |
-
|
| 637 |
function showToastMessage(message) {
|
| 638 |
const toast = document.getElementById('toast');
|
| 639 |
toast.textContent = message;
|
| 640 |
toast.classList.add('show');
|
| 641 |
setTimeout(() => toast.classList.remove('show'), 2000);
|
| 642 |
}
|
| 643 |
-
|
| 644 |
document.getElementById('translateForm').addEventListener('submit', function() {
|
| 645 |
const btn = document.getElementById('submitBtn');
|
| 646 |
btn.disabled = true;
|
|
@@ -653,12 +677,13 @@ HTML_TEMPLATE = """
|
|
| 653 |
|
| 654 |
app = Flask(__name__)
|
| 655 |
|
|
|
|
| 656 |
@app.route("/", methods=["GET", "POST"])
|
| 657 |
def index():
|
| 658 |
marian = []
|
| 659 |
libre = {}
|
| 660 |
source_text = ""
|
| 661 |
-
|
| 662 |
if request.method == "POST":
|
| 663 |
source_text = request.form.get("text", "").strip()
|
| 664 |
if source_text:
|
|
@@ -667,14 +692,88 @@ def index():
|
|
| 667 |
for name, data in libre_results.items():
|
| 668 |
data['code'] = KABYLE_VARIANTS[name]
|
| 669 |
libre[name] = data
|
| 670 |
-
|
| 671 |
return render_template_string(
|
| 672 |
-
HTML_TEMPLATE,
|
| 673 |
-
marian=marian,
|
| 674 |
libre=libre,
|
| 675 |
source_text=source_text
|
| 676 |
)
|
| 677 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 678 |
@app.route("/health")
|
| 679 |
def health():
|
| 680 |
return jsonify({
|
|
@@ -682,6 +781,7 @@ def health():
|
|
| 682 |
"model_loaded": model is not None
|
| 683 |
})
|
| 684 |
|
|
|
|
| 685 |
if __name__ == "__main__":
|
| 686 |
port = int(os.environ.get("PORT", 7860))
|
| 687 |
app.run(host="0.0.0.0", port=port, debug=False)
|
|
|
|
| 3 |
Kabyle Translation Hub - Hugging Face Spaces Edition
|
| 4 |
Simple translation interface: MarianMT vs LibreTranslate
|
| 5 |
Users choose their preferred translation - no metrics, no scores.
|
| 6 |
+
Also exposes a JSON API for programmatic access.
|
| 7 |
"""
|
| 8 |
|
| 9 |
import warnings
|
|
|
|
| 35 |
tokenizer = None
|
| 36 |
device = None
|
| 37 |
|
| 38 |
+
|
| 39 |
def load_model():
|
| 40 |
"""Load MarianMT model once and cache it"""
|
| 41 |
global model, tokenizer, device
|
| 42 |
+
|
| 43 |
if model is None:
|
| 44 |
print("Loading MarianMT model...")
|
| 45 |
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
|
| 46 |
+
|
| 47 |
tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, use_fast=False)
|
| 48 |
model = AutoModelForSeq2SeqLM.from_pretrained(MODEL_ID).to(device).eval()
|
| 49 |
+
|
| 50 |
print(f"Model loaded successfully on {device}")
|
| 51 |
+
|
| 52 |
return model, tokenizer, device
|
| 53 |
|
| 54 |
+
|
| 55 |
def translate_marian(text):
|
| 56 |
"""Translate using MarianMT with multiple alternatives"""
|
| 57 |
if not text or not text.strip():
|
| 58 |
return ["Please enter text to translate"]
|
| 59 |
+
|
| 60 |
try:
|
| 61 |
model, tokenizer, device = load_model()
|
| 62 |
+
|
| 63 |
# Prepare inputs
|
| 64 |
inputs = tokenizer(text, return_tensors='pt', padding=True, truncation=True, max_length=512)
|
| 65 |
inputs = {k: v.to(device) for k, v in inputs.items()}
|
| 66 |
+
|
| 67 |
with torch.no_grad():
|
| 68 |
# Simple beam search without group beam search
|
| 69 |
outputs = model.generate(
|
|
|
|
| 74 |
early_stopping=True,
|
| 75 |
do_sample=False,
|
| 76 |
)
|
| 77 |
+
|
| 78 |
translations = []
|
| 79 |
for output in outputs:
|
| 80 |
trans = tokenizer.decode(output, skip_special_tokens=True)
|
| 81 |
if trans and trans not in translations:
|
| 82 |
translations.append(trans)
|
| 83 |
+
|
| 84 |
return translations if translations else ["[Error: No translation generated]"]
|
| 85 |
+
|
| 86 |
except Exception as e:
|
| 87 |
print(f"MarianMT translation error: {e}")
|
| 88 |
import traceback
|
| 89 |
traceback.print_exc()
|
| 90 |
return [f"[Error: {str(e)}]"]
|
| 91 |
|
| 92 |
+
|
| 93 |
def translate_libre_variant(text, variant_code):
|
| 94 |
"""Translate using a specific LibreTranslate variant"""
|
| 95 |
try:
|
|
|
|
| 105 |
except Exception as e:
|
| 106 |
return {"success": False, "text": f"[Error: {str(e)[:50]}]"}
|
| 107 |
|
| 108 |
+
|
| 109 |
def translate_libre_all_variants(text):
|
| 110 |
"""Translate using all LibreTranslate variants in parallel"""
|
| 111 |
results = {}
|
| 112 |
with ThreadPoolExecutor(max_workers=4) as executor:
|
| 113 |
future_to_name = {
|
| 114 |
+
executor.submit(translate_libre_variant, text, code): name
|
| 115 |
for name, code in KABYLE_VARIANTS.items()
|
| 116 |
}
|
| 117 |
for future in as_completed(future_to_name, timeout=15):
|
|
|
|
| 122 |
results[name] = {"success": False, "text": f"[Error: {e}]"}
|
| 123 |
return results
|
| 124 |
|
| 125 |
+
|
| 126 |
HTML_TEMPLATE = """
|
| 127 |
<!DOCTYPE html>
|
| 128 |
<html lang="en">
|
|
|
|
| 144 |
--radius: 12px;
|
| 145 |
--radius-sm: 8px;
|
| 146 |
}
|
| 147 |
+
|
| 148 |
* { margin: 0; padding: 0; box-sizing: border-box; }
|
| 149 |
+
|
| 150 |
body {
|
| 151 |
font-family: system-ui, -apple-system, sans-serif;
|
| 152 |
background: linear-gradient(135deg, var(--bg-primary) 0%, #1a1a2e 100%);
|
|
|
|
| 154 |
min-height: 100vh;
|
| 155 |
padding: 20px;
|
| 156 |
}
|
| 157 |
+
|
| 158 |
.container {
|
| 159 |
max-width: 1000px;
|
| 160 |
margin: 0 auto;
|
| 161 |
}
|
| 162 |
+
|
| 163 |
header {
|
| 164 |
text-align: center;
|
| 165 |
padding: 40px 20px;
|
| 166 |
}
|
| 167 |
+
|
| 168 |
header h1 {
|
| 169 |
font-size: 2.5rem;
|
| 170 |
font-weight: 800;
|
|
|
|
| 173 |
-webkit-text-fill-color: transparent;
|
| 174 |
margin-bottom: 10px;
|
| 175 |
}
|
| 176 |
+
|
| 177 |
header p {
|
| 178 |
color: var(--text-secondary);
|
| 179 |
font-size: 1.1rem;
|
| 180 |
}
|
| 181 |
+
|
| 182 |
.input-card {
|
| 183 |
background: var(--bg-secondary);
|
| 184 |
border-radius: var(--radius);
|
|
|
|
| 186 |
margin-bottom: 24px;
|
| 187 |
border: 1px solid var(--border);
|
| 188 |
}
|
| 189 |
+
|
| 190 |
.input-group {
|
| 191 |
display: flex;
|
| 192 |
flex-direction: column;
|
| 193 |
gap: 16px;
|
| 194 |
}
|
| 195 |
+
|
| 196 |
textarea {
|
| 197 |
width: 100%;
|
| 198 |
padding: 16px;
|
|
|
|
| 206 |
font-family: inherit;
|
| 207 |
transition: border-color 0.2s;
|
| 208 |
}
|
| 209 |
+
|
| 210 |
textarea:focus {
|
| 211 |
outline: none;
|
| 212 |
border-color: var(--accent-primary);
|
| 213 |
}
|
| 214 |
+
|
| 215 |
textarea::placeholder {
|
| 216 |
color: var(--text-secondary);
|
| 217 |
}
|
| 218 |
+
|
| 219 |
.btn-primary {
|
| 220 |
background: linear-gradient(135deg, var(--accent-primary), var(--accent-secondary));
|
| 221 |
color: white;
|
|
|
|
| 231 |
justify-content: center;
|
| 232 |
gap: 8px;
|
| 233 |
}
|
| 234 |
+
|
| 235 |
.btn-primary:hover {
|
| 236 |
transform: translateY(-2px);
|
| 237 |
box-shadow: 0 10px 30px rgba(99, 102, 241, 0.3);
|
| 238 |
}
|
| 239 |
+
|
| 240 |
.btn-primary:disabled {
|
| 241 |
opacity: 0.6;
|
| 242 |
cursor: not-allowed;
|
| 243 |
transform: none;
|
| 244 |
}
|
| 245 |
+
|
| 246 |
.results-container {
|
| 247 |
display: flex;
|
| 248 |
flex-direction: column;
|
| 249 |
gap: 20px;
|
| 250 |
}
|
| 251 |
+
|
| 252 |
.engine-card {
|
| 253 |
background: var(--bg-secondary);
|
| 254 |
border-radius: var(--radius);
|
| 255 |
border: 1px solid var(--border);
|
| 256 |
overflow: hidden;
|
| 257 |
}
|
| 258 |
+
|
| 259 |
.engine-header {
|
| 260 |
padding: 20px 24px;
|
| 261 |
background: var(--bg-tertiary);
|
|
|
|
| 264 |
align-items: center;
|
| 265 |
justify-content: space-between;
|
| 266 |
}
|
| 267 |
+
|
| 268 |
.engine-title {
|
| 269 |
display: flex;
|
| 270 |
align-items: center;
|
|
|
|
| 272 |
font-size: 1.1rem;
|
| 273 |
font-weight: 600;
|
| 274 |
}
|
| 275 |
+
|
| 276 |
.engine-badge {
|
| 277 |
font-size: 0.75rem;
|
| 278 |
padding: 4px 12px;
|
|
|
|
| 281 |
color: var(--text-secondary);
|
| 282 |
border: 1px solid var(--border);
|
| 283 |
}
|
| 284 |
+
|
| 285 |
.translation-list {
|
| 286 |
list-style: none;
|
| 287 |
}
|
| 288 |
+
|
| 289 |
.translation-item {
|
| 290 |
padding: 20px 24px;
|
| 291 |
border-bottom: 1px solid var(--border);
|
|
|
|
| 296 |
justify-content: space-between;
|
| 297 |
gap: 16px;
|
| 298 |
}
|
| 299 |
+
|
| 300 |
.translation-item:last-child {
|
| 301 |
border-bottom: none;
|
| 302 |
}
|
| 303 |
+
|
| 304 |
.translation-item:hover {
|
| 305 |
background: rgba(99, 102, 241, 0.1);
|
| 306 |
}
|
| 307 |
+
|
| 308 |
.translation-item.selected {
|
| 309 |
background: rgba(16, 185, 129, 0.15);
|
| 310 |
border-left: 4px solid var(--accent-success);
|
| 311 |
}
|
| 312 |
+
|
| 313 |
.translation-text {
|
| 314 |
flex: 1;
|
| 315 |
font-size: 1.05rem;
|
| 316 |
line-height: 1.6;
|
| 317 |
color: var(--text-primary);
|
| 318 |
}
|
| 319 |
+
|
| 320 |
.copy-icon {
|
| 321 |
opacity: 0;
|
| 322 |
color: var(--accent-primary);
|
| 323 |
transition: opacity 0.2s;
|
| 324 |
flex-shrink: 0;
|
| 325 |
}
|
| 326 |
+
|
| 327 |
.translation-item:hover .copy-icon {
|
| 328 |
opacity: 1;
|
| 329 |
}
|
| 330 |
+
|
| 331 |
.variant-grid {
|
| 332 |
display: grid;
|
| 333 |
grid-template-columns: repeat(auto-fit, minmax(300px, 1fr));
|
| 334 |
gap: 1px;
|
| 335 |
background: var(--border);
|
| 336 |
}
|
| 337 |
+
|
| 338 |
.variant-item {
|
| 339 |
background: var(--bg-secondary);
|
| 340 |
padding: 16px 20px;
|
|
|
|
| 345 |
gap: 8px;
|
| 346 |
position: relative;
|
| 347 |
}
|
| 348 |
+
|
| 349 |
.variant-item:hover {
|
| 350 |
background: rgba(99, 102, 241, 0.1);
|
| 351 |
}
|
| 352 |
+
|
| 353 |
.variant-item.selected {
|
| 354 |
background: rgba(16, 185, 129, 0.15);
|
| 355 |
box-shadow: inset 4px 0 0 var(--accent-success);
|
| 356 |
}
|
| 357 |
+
|
| 358 |
.variant-header {
|
| 359 |
display: flex;
|
| 360 |
align-items: center;
|
| 361 |
justify-content: space-between;
|
| 362 |
}
|
| 363 |
+
|
| 364 |
.variant-name {
|
| 365 |
font-size: 0.75rem;
|
| 366 |
color: var(--text-secondary);
|
|
|
|
| 368 |
letter-spacing: 0.5px;
|
| 369 |
font-weight: 600;
|
| 370 |
}
|
| 371 |
+
|
| 372 |
.variant-code {
|
| 373 |
font-size: 0.7rem;
|
| 374 |
color: var(--text-secondary);
|
|
|
|
| 377 |
padding: 2px 8px;
|
| 378 |
border-radius: 4px;
|
| 379 |
}
|
| 380 |
+
|
| 381 |
.variant-text {
|
| 382 |
font-size: 1rem;
|
| 383 |
color: var(--text-primary);
|
| 384 |
line-height: 1.5;
|
| 385 |
}
|
| 386 |
+
|
| 387 |
.variant-item.error {
|
| 388 |
opacity: 0.6;
|
| 389 |
}
|
| 390 |
+
|
| 391 |
.variant-item.error .variant-text {
|
| 392 |
color: #ef4444;
|
| 393 |
font-family: monospace;
|
| 394 |
font-size: 0.85rem;
|
| 395 |
}
|
| 396 |
+
|
| 397 |
.source-display {
|
| 398 |
background: var(--bg-tertiary);
|
| 399 |
padding: 20px 24px;
|
|
|
|
| 401 |
margin-bottom: 20px;
|
| 402 |
border: 1px solid var(--border);
|
| 403 |
}
|
| 404 |
+
|
| 405 |
.source-label {
|
| 406 |
font-size: 0.75rem;
|
| 407 |
color: var(--text-secondary);
|
|
|
|
| 409 |
letter-spacing: 0.5px;
|
| 410 |
margin-bottom: 8px;
|
| 411 |
}
|
| 412 |
+
|
| 413 |
.source-text {
|
| 414 |
font-size: 1.1rem;
|
| 415 |
color: var(--text-primary);
|
| 416 |
line-height: 1.6;
|
| 417 |
}
|
| 418 |
+
|
| 419 |
.check-icon {
|
| 420 |
color: var(--accent-success);
|
| 421 |
opacity: 0;
|
| 422 |
transition: opacity 0.2s;
|
| 423 |
}
|
| 424 |
+
|
| 425 |
.translation-item.selected .check-icon,
|
| 426 |
.variant-item.selected .check-icon {
|
| 427 |
opacity: 1;
|
| 428 |
}
|
| 429 |
+
|
| 430 |
.toast {
|
| 431 |
position: fixed;
|
| 432 |
bottom: 24px;
|
|
|
|
| 442 |
z-index: 1000;
|
| 443 |
box-shadow: 0 10px 30px rgba(0,0,0,0.3);
|
| 444 |
}
|
| 445 |
+
|
| 446 |
.toast.show {
|
| 447 |
transform: translateX(-50%) translateY(0);
|
| 448 |
opacity: 1;
|
| 449 |
}
|
| 450 |
+
|
| 451 |
.loading {
|
| 452 |
display: inline-block;
|
| 453 |
width: 18px;
|
|
|
|
| 457 |
border-top-color: white;
|
| 458 |
animation: spin 0.8s linear infinite;
|
| 459 |
}
|
| 460 |
+
|
| 461 |
@keyframes spin {
|
| 462 |
to { transform: rotate(360deg); }
|
| 463 |
}
|
| 464 |
+
|
| 465 |
.empty-state {
|
| 466 |
text-align: center;
|
| 467 |
padding: 60px 20px;
|
| 468 |
color: var(--text-secondary);
|
| 469 |
}
|
| 470 |
+
|
| 471 |
.empty-state svg {
|
| 472 |
width: 64px;
|
| 473 |
height: 64px;
|
| 474 |
margin-bottom: 16px;
|
| 475 |
opacity: 0.5;
|
| 476 |
}
|
| 477 |
+
|
| 478 |
+
.api-note {
|
| 479 |
+
text-align: center;
|
| 480 |
+
margin-top: 32px;
|
| 481 |
+
color: var(--text-secondary);
|
| 482 |
+
font-size: 0.85rem;
|
| 483 |
+
}
|
| 484 |
+
|
| 485 |
+
.api-note code {
|
| 486 |
+
background: var(--bg-tertiary);
|
| 487 |
+
padding: 2px 8px;
|
| 488 |
+
border-radius: 4px;
|
| 489 |
+
font-family: monospace;
|
| 490 |
+
}
|
| 491 |
+
|
| 492 |
@media (max-width: 640px) {
|
| 493 |
header h1 {
|
| 494 |
font-size: 1.75rem;
|
| 495 |
}
|
| 496 |
+
|
| 497 |
.variant-grid {
|
| 498 |
grid-template-columns: 1fr;
|
| 499 |
}
|
|
|
|
| 506 |
<h1>English → Kabyle</h1>
|
| 507 |
<p>Choose the translation that suits you best</p>
|
| 508 |
</header>
|
| 509 |
+
|
| 510 |
<div class="input-card">
|
| 511 |
<form method="POST" id="translateForm">
|
| 512 |
<div class="input-group">
|
| 513 |
+
<textarea
|
| 514 |
+
name="text"
|
| 515 |
id="inputText"
|
| 516 |
+
placeholder="Enter English text to translate..."
|
| 517 |
required
|
| 518 |
autocomplete="off"
|
| 519 |
>{{ request.form.get('text', '') }}</textarea>
|
|
|
|
| 527 |
</div>
|
| 528 |
</form>
|
| 529 |
</div>
|
| 530 |
+
|
| 531 |
{% if source_text %}
|
| 532 |
<div class="source-display">
|
| 533 |
<div class="source-label">Source Text</div>
|
| 534 |
<div class="source-text">{{ source_text }}</div>
|
| 535 |
</div>
|
| 536 |
+
|
| 537 |
<div class="results-container">
|
| 538 |
<!-- MarianMT Results -->
|
| 539 |
<div class="engine-card">
|
|
|
|
| 562 |
{% endfor %}
|
| 563 |
</ul>
|
| 564 |
</div>
|
| 565 |
+
|
| 566 |
<!-- LibreTranslate Results -->
|
| 567 |
<div class="engine-card">
|
| 568 |
<div class="engine-header">
|
|
|
|
| 610 |
<p>Enter text above and click Translate to see results</p>
|
| 611 |
</div>
|
| 612 |
{% endif %}
|
| 613 |
+
|
| 614 |
+
<div class="api-note">
|
| 615 |
+
API available at <code>POST /api/translate</code> — see <code>/api/variants</code> for variant codes.
|
| 616 |
+
</div>
|
| 617 |
</div>
|
| 618 |
+
|
| 619 |
<div class="toast" id="toast">Copied to clipboard!</div>
|
| 620 |
+
|
| 621 |
<script>
|
| 622 |
let selectedText = '';
|
| 623 |
+
|
| 624 |
function selectTranslation(element, text) {
|
| 625 |
element.parentElement.querySelectorAll('.translation-item').forEach(item => {
|
| 626 |
item.classList.remove('selected');
|
|
|
|
| 629 |
selectedText = text;
|
| 630 |
copyText(text, false);
|
| 631 |
}
|
| 632 |
+
|
| 633 |
function selectVariant(element, text) {
|
| 634 |
document.querySelectorAll('.variant-item').forEach(item => {
|
| 635 |
item.classList.remove('selected');
|
|
|
|
| 638 |
selectedText = text;
|
| 639 |
copyText(text, false);
|
| 640 |
}
|
| 641 |
+
|
| 642 |
async function copyText(text, showToast = true) {
|
| 643 |
try {
|
| 644 |
await navigator.clipboard.writeText(text);
|
|
|
|
| 657 |
}
|
| 658 |
}
|
| 659 |
}
|
| 660 |
+
|
| 661 |
function showToastMessage(message) {
|
| 662 |
const toast = document.getElementById('toast');
|
| 663 |
toast.textContent = message;
|
| 664 |
toast.classList.add('show');
|
| 665 |
setTimeout(() => toast.classList.remove('show'), 2000);
|
| 666 |
}
|
| 667 |
+
|
| 668 |
document.getElementById('translateForm').addEventListener('submit', function() {
|
| 669 |
const btn = document.getElementById('submitBtn');
|
| 670 |
btn.disabled = true;
|
|
|
|
| 677 |
|
| 678 |
app = Flask(__name__)
|
| 679 |
|
| 680 |
+
|
| 681 |
@app.route("/", methods=["GET", "POST"])
|
| 682 |
def index():
|
| 683 |
marian = []
|
| 684 |
libre = {}
|
| 685 |
source_text = ""
|
| 686 |
+
|
| 687 |
if request.method == "POST":
|
| 688 |
source_text = request.form.get("text", "").strip()
|
| 689 |
if source_text:
|
|
|
|
| 692 |
for name, data in libre_results.items():
|
| 693 |
data['code'] = KABYLE_VARIANTS[name]
|
| 694 |
libre[name] = data
|
| 695 |
+
|
| 696 |
return render_template_string(
|
| 697 |
+
HTML_TEMPLATE,
|
| 698 |
+
marian=marian,
|
| 699 |
libre=libre,
|
| 700 |
source_text=source_text
|
| 701 |
)
|
| 702 |
|
| 703 |
+
|
| 704 |
+
@app.route("/api/translate", methods=["GET", "POST"])
|
| 705 |
+
def api_translate():
|
| 706 |
+
"""
|
| 707 |
+
JSON API for translation.
|
| 708 |
+
|
| 709 |
+
POST body (application/json):
|
| 710 |
+
{
|
| 711 |
+
"text": "Hello world",
|
| 712 |
+
"engines": ["marian", "libre"], // optional, default both
|
| 713 |
+
"variants": ["kab", "kab_kab"] // optional, default all LibreTranslate variants
|
| 714 |
+
}
|
| 715 |
+
|
| 716 |
+
GET query params (for quick testing):
|
| 717 |
+
/api/translate?text=Hello+world&engines=marian,libre
|
| 718 |
+
"""
|
| 719 |
+
if request.method == "POST":
|
| 720 |
+
payload = request.get_json(silent=True) or {}
|
| 721 |
+
text = payload.get("text") or request.form.get("text", "")
|
| 722 |
+
engines = payload.get("engines") or ["marian", "libre"]
|
| 723 |
+
variant_filter = payload.get("variants")
|
| 724 |
+
else:
|
| 725 |
+
text = request.args.get("text", "")
|
| 726 |
+
engines_param = request.args.get("engines", "marian,libre")
|
| 727 |
+
engines = [e.strip() for e in engines_param.split(",") if e.strip()]
|
| 728 |
+
variants_param = request.args.get("variants")
|
| 729 |
+
variant_filter = [v.strip() for v in variants_param.split(",")] if variants_param else None
|
| 730 |
+
|
| 731 |
+
text = (text or "").strip()
|
| 732 |
+
if not text:
|
| 733 |
+
return jsonify({"error": "Missing 'text' parameter"}), 400
|
| 734 |
+
|
| 735 |
+
engines = [e.lower() for e in engines]
|
| 736 |
+
result = {"source_text": text}
|
| 737 |
+
|
| 738 |
+
if "marian" in engines:
|
| 739 |
+
result["marian"] = translate_marian(text)
|
| 740 |
+
|
| 741 |
+
if "libre" in engines:
|
| 742 |
+
variants_to_use = KABYLE_VARIANTS
|
| 743 |
+
if variant_filter:
|
| 744 |
+
variants_to_use = {
|
| 745 |
+
name: code for name, code in KABYLE_VARIANTS.items()
|
| 746 |
+
if code in variant_filter or name in variant_filter
|
| 747 |
+
}
|
| 748 |
+
if not variants_to_use:
|
| 749 |
+
return jsonify({"error": f"No matching variants for: {variant_filter}"}), 400
|
| 750 |
+
|
| 751 |
+
libre_results = {}
|
| 752 |
+
with ThreadPoolExecutor(max_workers=4) as executor:
|
| 753 |
+
future_to_name = {
|
| 754 |
+
executor.submit(translate_libre_variant, text, code): (name, code)
|
| 755 |
+
for name, code in variants_to_use.items()
|
| 756 |
+
}
|
| 757 |
+
for future in as_completed(future_to_name, timeout=15):
|
| 758 |
+
name, code = future_to_name[future]
|
| 759 |
+
try:
|
| 760 |
+
data = future.result()
|
| 761 |
+
except Exception as e:
|
| 762 |
+
data = {"success": False, "text": f"[Error: {e}]"}
|
| 763 |
+
data["code"] = code
|
| 764 |
+
libre_results[name] = data
|
| 765 |
+
|
| 766 |
+
result["libre"] = libre_results
|
| 767 |
+
|
| 768 |
+
return jsonify(result)
|
| 769 |
+
|
| 770 |
+
|
| 771 |
+
@app.route("/api/variants", methods=["GET"])
|
| 772 |
+
def api_variants():
|
| 773 |
+
"""List available LibreTranslate Kabyle variants."""
|
| 774 |
+
return jsonify(KABYLE_VARIANTS)
|
| 775 |
+
|
| 776 |
+
|
| 777 |
@app.route("/health")
|
| 778 |
def health():
|
| 779 |
return jsonify({
|
|
|
|
| 781 |
"model_loaded": model is not None
|
| 782 |
})
|
| 783 |
|
| 784 |
+
|
| 785 |
if __name__ == "__main__":
|
| 786 |
port = int(os.environ.get("PORT", 7860))
|
| 787 |
app.run(host="0.0.0.0", port=port, debug=False)
|