Update ocr.html
Browse files
ocr.html
CHANGED
|
@@ -1,514 +1,602 @@
|
|
| 1 |
<!DOCTYPE html>
|
| 2 |
-
<html lang="
|
| 3 |
<head>
|
| 4 |
-
<title>UFastPro OCR Technology</title>
|
| 5 |
<meta charset="UTF-8">
|
| 6 |
-
<meta name="viewport" content="width=device-width, initial-scale=1">
|
| 7 |
-
<
|
| 8 |
-
|
| 9 |
<script src="https://cdnjs.cloudflare.com/ajax/libs/pdf.js/3.4.120/pdf.min.js"></script>
|
| 10 |
<style>
|
| 11 |
-
:root {
|
| 12 |
-
--primary: #4361ee;
|
| 13 |
-
--primary-hover: #3a56d4;
|
| 14 |
-
--secondary: #4cc9f0;
|
| 15 |
-
--secondary-hover: #3db8dd;
|
| 16 |
-
}
|
| 17 |
-
|
| 18 |
body {
|
|
|
|
|
|
|
|
|
|
| 19 |
background-color: #f8f9fa;
|
| 20 |
}
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
max-width: 900px;
|
| 24 |
-
}
|
| 25 |
-
|
| 26 |
-
header {
|
| 27 |
-
margin-bottom: 2rem;
|
| 28 |
-
padding: 1rem 0;
|
| 29 |
-
background: linear-gradient(135deg, var(--primary), var(--secondary));
|
| 30 |
color: white;
|
| 31 |
-
|
| 32 |
-
|
| 33 |
-
|
| 34 |
-
|
| 35 |
-
header h1 {
|
| 36 |
-
margin: 0;
|
| 37 |
-
font-weight: bold;
|
| 38 |
}
|
| 39 |
-
|
| 40 |
-
|
| 41 |
-
margin: 0
|
| 42 |
-
opacity: 0.9;
|
| 43 |
}
|
| 44 |
-
|
| 45 |
-
article {
|
| 46 |
-
margin: 2rem 0;
|
| 47 |
-
padding: 2rem;
|
| 48 |
background-color: white;
|
| 49 |
border-radius: 10px;
|
| 50 |
-
box-shadow: 0 2px
|
|
|
|
|
|
|
| 51 |
}
|
| 52 |
-
|
| 53 |
-
|
| 54 |
-
|
| 55 |
-
|
|
|
|
| 56 |
}
|
| 57 |
-
|
| 58 |
-
|
| 59 |
-
height: 20px;
|
| 60 |
-
background: linear-gradient(to right, var(--primary), var(--secondary));
|
| 61 |
-
width: 0%;
|
| 62 |
border-radius: 5px;
|
| 63 |
-
|
| 64 |
-
|
| 65 |
-
|
| 66 |
-
|
| 67 |
-
.page-result {
|
| 68 |
-
margin-bottom: 20px;
|
| 69 |
-
border: 1px solid #e9ecef;
|
| 70 |
-
padding: 20px;
|
| 71 |
-
border-radius: 8px;
|
| 72 |
-
background-color: white;
|
| 73 |
-
box-shadow: 0 2px 4px rgba(0, 0, 0, 0.05);
|
| 74 |
-
transition: transform 0.2s;
|
| 75 |
}
|
| 76 |
-
|
| 77 |
-
|
| 78 |
-
|
| 79 |
-
box-shadow: 0 4px 8px rgba(0, 0, 0, 0.1);
|
| 80 |
}
|
| 81 |
-
|
| 82 |
-
|
| 83 |
-
margin-top: 0;
|
| 84 |
-
color: var(--primary);
|
| 85 |
-
border-bottom: 2px solid #f1f3f5;
|
| 86 |
-
padding-bottom: 0.5rem;
|
| 87 |
}
|
| 88 |
-
|
| 89 |
-
|
| 90 |
-
|
| 91 |
-
|
| 92 |
-
margin-top: 20px;
|
| 93 |
-
padding-right: 10px;
|
| 94 |
}
|
| 95 |
-
|
| 96 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 97 |
text-align: center;
|
| 98 |
-
|
| 99 |
-
|
| 100 |
-
color: #495057;
|
| 101 |
-
}
|
| 102 |
-
|
| 103 |
-
button {
|
| 104 |
-
background-color: var(--primary);
|
| 105 |
-
transition: all 0.2s;
|
| 106 |
}
|
| 107 |
-
|
| 108 |
-
|
| 109 |
-
|
| 110 |
-
transform: translateY(-2px);
|
| 111 |
-
}
|
| 112 |
-
|
| 113 |
-
#cancelProcessingBtn {
|
| 114 |
-
background-color: #e63946;
|
| 115 |
}
|
| 116 |
-
|
| 117 |
-
|
| 118 |
-
|
| 119 |
}
|
| 120 |
-
|
| 121 |
-
|
| 122 |
-
|
| 123 |
-
|
|
|
|
|
|
|
| 124 |
padding: 10px;
|
| 125 |
-
|
|
|
|
| 126 |
}
|
| 127 |
-
|
| 128 |
-
|
| 129 |
-
|
| 130 |
-
|
| 131 |
-
font-weight: 500;
|
| 132 |
}
|
| 133 |
-
|
| 134 |
-
footer {
|
| 135 |
-
text-align: center;
|
| 136 |
-
margin-top: 2rem;
|
| 137 |
-
padding: 1rem 0;
|
| 138 |
-
color: #6c757d;
|
| 139 |
-
font-size: 0.9rem;
|
| 140 |
-
}
|
| 141 |
-
|
| 142 |
.logo {
|
| 143 |
font-weight: bold;
|
| 144 |
-
font-size:
|
| 145 |
-
|
| 146 |
-
|
| 147 |
}
|
| 148 |
-
|
| 149 |
.logo span {
|
| 150 |
-
color:
|
| 151 |
-
}
|
| 152 |
-
|
| 153 |
-
.upload-container {
|
| 154 |
-
border: 2px dashed #ced4da;
|
| 155 |
-
border-radius: 8px;
|
| 156 |
-
padding: 2rem;
|
| 157 |
-
text-align: center;
|
| 158 |
-
transition: all 0.2s;
|
| 159 |
-
cursor: pointer;
|
| 160 |
-
margin-bottom: 1rem;
|
| 161 |
-
}
|
| 162 |
-
|
| 163 |
-
.upload-container:hover {
|
| 164 |
-
border-color: var(--primary);
|
| 165 |
-
background-color: rgba(67, 97, 238, 0.05);
|
| 166 |
-
}
|
| 167 |
-
|
| 168 |
-
.upload-icon {
|
| 169 |
-
font-size: 2rem;
|
| 170 |
-
color: var(--primary);
|
| 171 |
-
margin-bottom: 1rem;
|
| 172 |
-
}
|
| 173 |
-
|
| 174 |
-
.export-button {
|
| 175 |
-
display: none;
|
| 176 |
-
margin-top: 20px;
|
| 177 |
-
}
|
| 178 |
-
|
| 179 |
-
@media (max-width: 768px) {
|
| 180 |
-
article {
|
| 181 |
-
padding: 1rem;
|
| 182 |
-
}
|
| 183 |
}
|
| 184 |
</style>
|
| 185 |
</head>
|
| 186 |
<body>
|
| 187 |
-
<
|
| 188 |
-
<div
|
| 189 |
-
<
|
| 190 |
-
<
|
|
|
|
| 191 |
</div>
|
| 192 |
-
|
| 193 |
-
|
| 194 |
-
|
| 195 |
-
|
| 196 |
-
<
|
| 197 |
-
|
| 198 |
-
|
| 199 |
-
<div class="upload-container" id="uploadContainer">
|
| 200 |
-
<div class="upload-icon">📄</div>
|
| 201 |
-
<p>Drag & drop your PDF here or click to browse</p>
|
| 202 |
-
<input type="file" id="pdfFile" name="pdfFile" accept=".pdf" style="display: none;">
|
| 203 |
</div>
|
| 204 |
-
|
| 205 |
-
|
| 206 |
-
|
| 207 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 208 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 209 |
|
| 210 |
-
<div
|
| 211 |
-
<button id="
|
| 212 |
-
<button id="
|
| 213 |
</div>
|
| 214 |
|
| 215 |
-
<
|
| 216 |
-
|
| 217 |
-
|
| 218 |
-
|
| 219 |
-
|
| 220 |
-
|
| 221 |
-
|
| 222 |
-
|
| 223 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 224 |
|
| 225 |
<script>
|
| 226 |
-
//
|
| 227 |
pdfjsLib.GlobalWorkerOptions.workerSrc = 'https://cdnjs.cloudflare.com/ajax/libs/pdf.js/3.4.120/pdf.worker.min.js';
|
| 228 |
|
| 229 |
-
//
|
| 230 |
-
const
|
| 231 |
-
|
| 232 |
-
'091dbfbf0emsh6606eb2165cfd97p191257jsnd017aa229e09',
|
| 233 |
-
'2175cafd75msh4465ea00b022145p1e062ajsn05e78cebc495',
|
| 234 |
-
'eb11693cddmshb8bd157e05b74acp1f6aa4jsn4369fa546e55',
|
| 235 |
-
'e9c1cd0db3mshb54648cf3400243p1d0a2bjsnf0d54e43fca2',
|
| 236 |
-
'fdb95540bamsh786e3ec8083bb22p1b4c1ajsn44fecbb812bb'
|
| 237 |
-
];
|
| 238 |
|
| 239 |
-
//
|
| 240 |
-
|
| 241 |
-
|
| 242 |
-
|
| 243 |
-
}
|
| 244 |
|
| 245 |
-
|
| 246 |
-
const
|
| 247 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 248 |
document.addEventListener('DOMContentLoaded', function() {
|
| 249 |
-
|
| 250 |
-
|
| 251 |
-
|
| 252 |
-
|
| 253 |
-
|
| 254 |
-
|
| 255 |
-
|
| 256 |
-
|
| 257 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 258 |
|
| 259 |
-
|
| 260 |
-
let isProcessing = false;
|
| 261 |
-
let shouldCancel = false;
|
| 262 |
-
let allExtractedText = [];
|
| 263 |
|
| 264 |
-
//
|
| 265 |
-
|
| 266 |
-
|
| 267 |
-
|
| 268 |
-
|
| 269 |
-
|
| 270 |
-
|
| 271 |
-
|
| 272 |
-
|
| 273 |
-
|
| 274 |
-
|
| 275 |
-
|
| 276 |
-
|
| 277 |
-
|
| 278 |
-
|
| 279 |
-
|
| 280 |
-
uploadContainer.style.backgroundColor = '';
|
| 281 |
|
| 282 |
-
if (
|
| 283 |
-
|
| 284 |
-
|
|
|
|
|
|
|
|
|
|
| 285 |
}
|
| 286 |
-
|
| 287 |
-
|
| 288 |
-
|
| 289 |
-
|
| 290 |
-
|
| 291 |
-
|
| 292 |
-
|
| 293 |
-
|
| 294 |
-
|
| 295 |
-
|
| 296 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 297 |
}
|
| 298 |
|
| 299 |
-
|
| 300 |
-
|
| 301 |
-
|
| 302 |
-
|
| 303 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 304 |
|
| 305 |
-
//
|
| 306 |
-
|
| 307 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 308 |
|
| 309 |
-
//
|
| 310 |
-
|
| 311 |
-
statusText.textContent = `PDF loaded with ${pdfDocument.numPages} pages. Click "Start Processing" to begin.`;
|
| 312 |
-
progressBar.style.width = '20%';
|
| 313 |
-
} catch (error) {
|
| 314 |
-
console.error('Error loading PDF:', error);
|
| 315 |
-
alert('Error loading PDF file. Please try again.');
|
| 316 |
-
progressContainer.hidden = true;
|
| 317 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
| 318 |
}
|
| 319 |
-
|
| 320 |
-
|
| 321 |
-
|
| 322 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 323 |
|
| 324 |
-
|
| 325 |
-
|
| 326 |
-
startProcessingBtn.hidden = true;
|
| 327 |
-
cancelProcessingBtn.hidden = false;
|
| 328 |
-
resultsContainer.innerHTML = '';
|
| 329 |
-
allExtractedText = [];
|
| 330 |
-
exportAllTextBtn.style.display = 'none';
|
| 331 |
|
| 332 |
-
|
|
|
|
|
|
|
| 333 |
|
| 334 |
-
|
| 335 |
-
|
| 336 |
-
|
| 337 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 338 |
}
|
| 339 |
-
|
| 340 |
-
|
| 341 |
-
|
| 342 |
-
|
| 343 |
-
try {
|
| 344 |
-
// Process the current page
|
| 345 |
-
await processPage(pageNum, totalPages);
|
| 346 |
-
} catch (error) {
|
| 347 |
-
console.error(`Error processing page ${pageNum}:`, error);
|
| 348 |
-
addPageResult(pageNum, `Error: ${error.message}`, null);
|
| 349 |
-
allExtractedText.push(`[PAGE ${pageNum}]\nError: ${error.message}\n`);
|
| 350 |
}
|
| 351 |
}
|
| 352 |
-
|
| 353 |
-
|
| 354 |
-
|
| 355 |
-
|
| 356 |
-
|
| 357 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 358 |
}
|
| 359 |
-
|
| 360 |
-
|
| 361 |
-
|
| 362 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 363 |
}
|
| 364 |
|
| 365 |
-
//
|
| 366 |
-
|
| 367 |
-
|
| 368 |
-
|
| 369 |
-
}
|
| 370 |
|
| 371 |
-
|
| 372 |
-
|
| 373 |
-
|
| 374 |
-
progressBar.style.width = `${percentage}%`;
|
| 375 |
-
statusText.textContent = `Processing page ${current} of ${total}`;
|
| 376 |
}
|
| 377 |
|
| 378 |
-
//
|
| 379 |
-
|
| 380 |
-
if (allExtractedText.length === 0) return;
|
| 381 |
-
|
| 382 |
-
const combinedText = allExtractedText.join('\n\n');
|
| 383 |
-
const blob = new Blob([combinedText], { type: 'text/plain' });
|
| 384 |
-
const url = URL.createObjectURL(blob);
|
| 385 |
-
|
| 386 |
const a = document.createElement('a');
|
| 387 |
-
a.href =
|
| 388 |
-
a.download =
|
| 389 |
document.body.appendChild(a);
|
| 390 |
a.click();
|
| 391 |
document.body.removeChild(a);
|
| 392 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 393 |
}
|
| 394 |
|
| 395 |
-
//
|
| 396 |
-
|
| 397 |
-
|
| 398 |
-
|
|
|
|
|
|
|
|
|
|
| 399 |
|
| 400 |
-
//
|
| 401 |
-
|
| 402 |
-
|
| 403 |
-
|
| 404 |
-
|
| 405 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 406 |
|
| 407 |
-
//
|
| 408 |
-
|
| 409 |
-
canvasContext: context,
|
| 410 |
-
viewport: viewport
|
| 411 |
-
}).promise;
|
| 412 |
|
| 413 |
-
//
|
| 414 |
-
|
| 415 |
-
|
| 416 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 417 |
|
| 418 |
-
|
| 419 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 420 |
|
| 421 |
-
//
|
| 422 |
const formData = new FormData();
|
| 423 |
-
formData.append('image', blob, `
|
| 424 |
|
| 425 |
-
|
|
|
|
| 426 |
method: 'POST',
|
| 427 |
headers: {
|
| 428 |
-
'X-RapidAPI-Key':
|
| 429 |
-
'X-RapidAPI-Host':
|
| 430 |
},
|
| 431 |
body: formData
|
| 432 |
});
|
| 433 |
|
| 434 |
-
|
| 435 |
-
|
| 436 |
-
// Extract text from the response
|
| 437 |
-
let extractedText = '';
|
| 438 |
-
try {
|
| 439 |
-
extractedText = data.results[0].entities[0].objects[0].entities[0].text;
|
| 440 |
-
} catch (error) {
|
| 441 |
-
extractedText = 'No text could be extracted from this page.';
|
| 442 |
}
|
| 443 |
|
| 444 |
-
|
| 445 |
-
allExtractedText.push(`[PAGE ${pageNum}]\n${extractedText}`);
|
| 446 |
|
| 447 |
-
|
| 448 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 449 |
}
|
| 450 |
|
| 451 |
-
|
| 452 |
-
|
| 453 |
-
|
| 454 |
-
|
| 455 |
-
|
| 456 |
-
|
| 457 |
-
|
| 458 |
-
|
| 459 |
-
|
| 460 |
-
|
| 461 |
-
|
| 462 |
-
const textHeader = document.createElement('h5');
|
| 463 |
-
textHeader.textContent = 'Extracted Text:';
|
| 464 |
-
pageResult.appendChild(textHeader);
|
| 465 |
-
|
| 466 |
-
const textContent = document.createElement('pre');
|
| 467 |
-
textContent.style.whiteSpace = 'pre-wrap';
|
| 468 |
-
textContent.textContent = text || 'No text extracted';
|
| 469 |
-
pageResult.appendChild(textContent);
|
| 470 |
-
|
| 471 |
-
// Add copy button
|
| 472 |
-
const copyButton = document.createElement('button');
|
| 473 |
-
copyButton.textContent = 'Copy Text';
|
| 474 |
-
copyButton.setAttribute('aria-label', 'Copy text to clipboard');
|
| 475 |
-
copyButton.style.marginBottom = '10px';
|
| 476 |
-
copyButton.addEventListener('click', () => {
|
| 477 |
-
navigator.clipboard.writeText(text || '')
|
| 478 |
-
.then(() => {
|
| 479 |
-
const originalText = copyButton.textContent;
|
| 480 |
-
copyButton.textContent = 'Copied!';
|
| 481 |
-
setTimeout(() => {
|
| 482 |
-
copyButton.textContent = originalText;
|
| 483 |
-
}, 2000);
|
| 484 |
-
})
|
| 485 |
-
.catch(err => console.error('Failed to copy text:', err));
|
| 486 |
-
});
|
| 487 |
-
pageResult.appendChild(copyButton);
|
| 488 |
-
|
| 489 |
-
// Add raw data toggle (optional, can be removed if not needed)
|
| 490 |
-
if (rawData) {
|
| 491 |
-
const rawToggle = document.createElement('details');
|
| 492 |
-
const rawSummary = document.createElement('summary');
|
| 493 |
-
rawSummary.textContent = 'Show Raw API Response';
|
| 494 |
-
rawToggle.appendChild(rawSummary);
|
| 495 |
-
|
| 496 |
-
const rawContent = document.createElement('pre');
|
| 497 |
-
rawContent.style.maxHeight = '200px';
|
| 498 |
-
rawContent.style.overflow = 'auto';
|
| 499 |
-
rawContent.textContent = JSON.stringify(rawData, null, 2);
|
| 500 |
-
rawToggle.appendChild(rawContent);
|
| 501 |
-
|
| 502 |
-
pageResult.appendChild(rawToggle);
|
| 503 |
-
}
|
| 504 |
-
|
| 505 |
-
// Add to results container
|
| 506 |
-
resultsContainer.appendChild(pageResult);
|
| 507 |
-
|
| 508 |
-
// Scroll to the new result
|
| 509 |
-
pageResult.scrollIntoView({ behavior: 'smooth', block: 'end' });
|
| 510 |
}
|
| 511 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 512 |
</script>
|
| 513 |
</body>
|
| 514 |
</html>
|
|
|
|
| 1 |
<!DOCTYPE html>
|
| 2 |
+
<html lang="ar" dir="rtl">
|
| 3 |
<head>
|
|
|
|
| 4 |
<meta charset="UTF-8">
|
| 5 |
+
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
| 6 |
+
<title>موندو لينجوا - نظام OCR</title>
|
| 7 |
+
<link href="https://cdn.jsdelivr.net/npm/bootstrap@5.2.3/dist/css/bootstrap.min.css" rel="stylesheet">
|
| 8 |
<script src="https://cdnjs.cloudflare.com/ajax/libs/pdf.js/3.4.120/pdf.min.js"></script>
|
| 9 |
<style>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
body {
|
| 11 |
+
font-family: Arial, sans-serif;
|
| 12 |
+
margin: 0;
|
| 13 |
+
padding: 20px;
|
| 14 |
background-color: #f8f9fa;
|
| 15 |
}
|
| 16 |
+
.header {
|
| 17 |
+
background-color: #3b82f6;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
color: white;
|
| 19 |
+
padding: 15px;
|
| 20 |
+
border-radius: 10px;
|
| 21 |
+
margin-bottom: 20px;
|
| 22 |
+
text-align: center;
|
|
|
|
|
|
|
|
|
|
| 23 |
}
|
| 24 |
+
.container {
|
| 25 |
+
max-width: 800px;
|
| 26 |
+
margin: 0 auto;
|
|
|
|
| 27 |
}
|
| 28 |
+
.card {
|
|
|
|
|
|
|
|
|
|
| 29 |
background-color: white;
|
| 30 |
border-radius: 10px;
|
| 31 |
+
box-shadow: 0 2px 5px rgba(0,0,0,0.1);
|
| 32 |
+
padding: 20px;
|
| 33 |
+
margin-bottom: 20px;
|
| 34 |
}
|
| 35 |
+
.pdf-grid {
|
| 36 |
+
display: grid;
|
| 37 |
+
grid-template-columns: repeat(auto-fill, minmax(120px, 1fr));
|
| 38 |
+
gap: 10px;
|
| 39 |
+
margin-top: 15px;
|
| 40 |
}
|
| 41 |
+
.pdf-page {
|
| 42 |
+
border: 1px solid #ddd;
|
|
|
|
|
|
|
|
|
|
| 43 |
border-radius: 5px;
|
| 44 |
+
padding: 5px;
|
| 45 |
+
position: relative;
|
| 46 |
+
cursor: pointer;
|
| 47 |
+
transition: all 0.2s;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
}
|
| 49 |
+
.pdf-page:hover {
|
| 50 |
+
transform: translateY(-3px);
|
| 51 |
+
box-shadow: 0 3px 10px rgba(0,0,0,0.1);
|
|
|
|
| 52 |
}
|
| 53 |
+
.pdf-page.selected {
|
| 54 |
+
border: 2px solid #3b82f6;
|
|
|
|
|
|
|
|
|
|
|
|
|
| 55 |
}
|
| 56 |
+
.pdf-page img {
|
| 57 |
+
width: 100%;
|
| 58 |
+
height: auto;
|
| 59 |
+
border-radius: 3px;
|
|
|
|
|
|
|
| 60 |
}
|
| 61 |
+
.page-number {
|
| 62 |
+
position: absolute;
|
| 63 |
+
bottom: 0;
|
| 64 |
+
left: 0;
|
| 65 |
+
right: 0;
|
| 66 |
+
background-color: rgba(0,0,0,0.6);
|
| 67 |
+
color: white;
|
| 68 |
text-align: center;
|
| 69 |
+
font-size: 12px;
|
| 70 |
+
padding: 2px;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 71 |
}
|
| 72 |
+
.btn-primary {
|
| 73 |
+
background-color: #3b82f6;
|
| 74 |
+
border-color: #3b82f6;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 75 |
}
|
| 76 |
+
.btn-success {
|
| 77 |
+
background-color: #10b981;
|
| 78 |
+
border-color: #10b981;
|
| 79 |
}
|
| 80 |
+
.result-text {
|
| 81 |
+
max-height: 300px;
|
| 82 |
+
overflow-y: auto;
|
| 83 |
+
white-space: pre-wrap;
|
| 84 |
+
direction: rtl;
|
| 85 |
+
border: 1px solid #ddd;
|
| 86 |
padding: 10px;
|
| 87 |
+
border-radius: 5px;
|
| 88 |
+
background-color: #f8f9fa;
|
| 89 |
}
|
| 90 |
+
.spinner-border {
|
| 91 |
+
width: 1.5rem;
|
| 92 |
+
height: 1.5rem;
|
| 93 |
+
margin-left: 0.5rem;
|
|
|
|
| 94 |
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 95 |
.logo {
|
| 96 |
font-weight: bold;
|
| 97 |
+
font-size: 24px;
|
| 98 |
+
display: inline-block;
|
| 99 |
+
margin-bottom: 5px;
|
| 100 |
}
|
|
|
|
| 101 |
.logo span {
|
| 102 |
+
color: #bfdbfe;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 103 |
}
|
| 104 |
</style>
|
| 105 |
</head>
|
| 106 |
<body>
|
| 107 |
+
<div class="container">
|
| 108 |
+
<div class="header">
|
| 109 |
+
<div class="logo">مـوندو <span>لينجـوا</span></div>
|
| 110 |
+
<h1>نظام التعرف الضوئي على النصوص</h1>
|
| 111 |
+
<p class="mb-0">استخراج النصوص من الصور والملفات متعددة الصفحات</p>
|
| 112 |
</div>
|
| 113 |
+
|
| 114 |
+
<!-- بطاقة تحميل الملف -->
|
| 115 |
+
<div class="card">
|
| 116 |
+
<h3>تحميل الملف</h3>
|
| 117 |
+
<div class="mb-3">
|
| 118 |
+
<input type="file" class="form-control" id="fileInput" accept=".pdf,.jpg,.jpeg,.png">
|
| 119 |
+
<div class="form-text">يمكنك تحميل ملف PDF (حتى 30 صفحة) أو صورة</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
| 120 |
</div>
|
| 121 |
+
<button id="processBtn" class="btn btn-primary">معالجة الملف</button>
|
| 122 |
+
</div>
|
| 123 |
+
|
| 124 |
+
<!-- عرض حالة المعالجة -->
|
| 125 |
+
<div id="processingStatus" class="card d-none">
|
| 126 |
+
<h3>حالة المعالجة</h3>
|
| 127 |
+
<div class="alert alert-info">
|
| 128 |
+
<div class="d-flex align-items-center">
|
| 129 |
+
<div class="spinner-border text-primary" role="status"></div>
|
| 130 |
+
<span id="statusText" class="ms-2">جاري معالجة الملف...</span>
|
| 131 |
+
</div>
|
| 132 |
+
</div>
|
| 133 |
+
<div class="progress mt-2">
|
| 134 |
+
<div id="progressBar" class="progress-bar" role="progressbar" style="width: 0%"></div>
|
| 135 |
</div>
|
| 136 |
+
</div>
|
| 137 |
+
|
| 138 |
+
<!-- عرض صفحات PDF -->
|
| 139 |
+
<div id="pdfPagesCard" class="card d-none">
|
| 140 |
+
<h3>صفحات الملف</h3>
|
| 141 |
+
<p>اختر الصفحات التي تريد معالجتها (انقر للتحديد)</p>
|
| 142 |
|
| 143 |
+
<div class="mb-3">
|
| 144 |
+
<button id="selectAllBtn" class="btn btn-sm btn-outline-primary me-2">تحديد الكل</button>
|
| 145 |
+
<button id="deselectAllBtn" class="btn btn-sm btn-outline-secondary">إلغاء تحديد الكل</button>
|
| 146 |
</div>
|
| 147 |
|
| 148 |
+
<div id="pdfPagesContainer" class="pdf-grid"></div>
|
| 149 |
+
|
| 150 |
+
<div class="mt-3">
|
| 151 |
+
<button id="extractImagesBtn" class="btn btn-success me-2">تحويل إلى صور</button>
|
| 152 |
+
<button id="extractTextBtn" class="btn btn-primary">استخراج النص</button>
|
| 153 |
+
</div>
|
| 154 |
+
</div>
|
| 155 |
+
|
| 156 |
+
<!-- عرض النتائج -->
|
| 157 |
+
<div id="resultsCard" class="card d-none">
|
| 158 |
+
<h3>النص المستخرج</h3>
|
| 159 |
+
<div id="resultText" class="result-text mt-3">
|
| 160 |
+
لم يتم استخراج نص بعد.
|
| 161 |
+
</div>
|
| 162 |
+
<div class="mt-3">
|
| 163 |
+
<button id="copyTextBtn" class="btn btn-outline-primary me-2">نسخ النص</button>
|
| 164 |
+
<button id="downloadTextBtn" class="btn btn-success">تنزيل النص</button>
|
| 165 |
+
</div>
|
| 166 |
+
</div>
|
| 167 |
+
</div>
|
| 168 |
|
| 169 |
<script>
|
| 170 |
+
// تهيئة PDF.js
|
| 171 |
pdfjsLib.GlobalWorkerOptions.workerSrc = 'https://cdnjs.cloudflare.com/ajax/libs/pdf.js/3.4.120/pdf.worker.min.js';
|
| 172 |
|
| 173 |
+
// إعدادات API4AI OCR
|
| 174 |
+
const RAPIDAPI_KEY = 'eb11693cddmshb8bd157e05b74acp1f6aa4jsn4369fa546e55';
|
| 175 |
+
const OCR_API_URL = 'https://ocr43.p.rapidapi.com/v1/results';
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 176 |
|
| 177 |
+
// متغيرات عامة
|
| 178 |
+
let documentPages = [];
|
| 179 |
+
let selectedPages = [];
|
| 180 |
+
let extractedTexts = [];
|
|
|
|
| 181 |
|
| 182 |
+
// عناصر DOM
|
| 183 |
+
const fileInput = document.getElementById('fileInput');
|
| 184 |
+
const processBtn = document.getElementById('processBtn');
|
| 185 |
+
const processingStatus = document.getElementById('processingStatus');
|
| 186 |
+
const statusText = document.getElementById('statusText');
|
| 187 |
+
const progressBar = document.getElementById('progressBar');
|
| 188 |
+
const pdfPagesCard = document.getElementById('pdfPagesCard');
|
| 189 |
+
const pdfPagesContainer = document.getElementById('pdfPagesContainer');
|
| 190 |
+
const selectAllBtn = document.getElementById('selectAllBtn');
|
| 191 |
+
const deselectAllBtn = document.getElementById('deselectAllBtn');
|
| 192 |
+
const extractImagesBtn = document.getElementById('extractImagesBtn');
|
| 193 |
+
const extractTextBtn = document.getElementById('extractTextBtn');
|
| 194 |
+
const resultsCard = document.getElementById('resultsCard');
|
| 195 |
+
const resultText = document.getElementById('resultText');
|
| 196 |
+
const copyTextBtn = document.getElementById('copyTextBtn');
|
| 197 |
+
const downloadTextBtn = document.getElementById('downloadTextBtn');
|
| 198 |
+
|
| 199 |
+
// إضافة مستمعات الأحداث
|
| 200 |
document.addEventListener('DOMContentLoaded', function() {
|
| 201 |
+
processBtn.addEventListener('click', processFile);
|
| 202 |
+
selectAllBtn.addEventListener('click', selectAllPages);
|
| 203 |
+
deselectAllBtn.addEventListener('click', deselectAllPages);
|
| 204 |
+
extractImagesBtn.addEventListener('click', extractImages);
|
| 205 |
+
extractTextBtn.addEventListener('click', extractText);
|
| 206 |
+
copyTextBtn.addEventListener('click', copyText);
|
| 207 |
+
downloadTextBtn.addEventListener('click', downloadText);
|
| 208 |
+
});
|
| 209 |
+
|
| 210 |
+
// معالجة الملف
|
| 211 |
+
async function processFile() {
|
| 212 |
+
if (!fileInput.files || fileInput.files.length === 0) {
|
| 213 |
+
alert('الرجاء اختيار ملف أولاً');
|
| 214 |
+
return;
|
| 215 |
+
}
|
| 216 |
|
| 217 |
+
const file = fileInput.files[0];
|
|
|
|
|
|
|
|
|
|
| 218 |
|
| 219 |
+
// إظهار حالة المعالجة
|
| 220 |
+
processingStatus.classList.remove('d-none');
|
| 221 |
+
statusText.textContent = 'جاري معالجة الملف...';
|
| 222 |
+
progressBar.style.width = '0%';
|
| 223 |
+
|
| 224 |
+
// إخفاء بطاقات أخرى
|
| 225 |
+
pdfPagesCard.classList.add('d-none');
|
| 226 |
+
resultsCard.classList.add('d-none');
|
| 227 |
+
|
| 228 |
+
// إفراغ مصفوفات الصفحا��
|
| 229 |
+
documentPages = [];
|
| 230 |
+
selectedPages = [];
|
| 231 |
+
extractedTexts = [];
|
| 232 |
+
|
| 233 |
+
try {
|
| 234 |
+
const fileType = file.name.split('.').pop().toLowerCase();
|
|
|
|
| 235 |
|
| 236 |
+
if (fileType === 'pdf') {
|
| 237 |
+
await processPdf(file);
|
| 238 |
+
} else if (['jpg', 'jpeg', 'png'].includes(fileType)) {
|
| 239 |
+
await processImage(file);
|
| 240 |
+
} else {
|
| 241 |
+
throw new Error('نوع الملف غير مدعوم. يرجى اختيار ملف PDF أو صورة.');
|
| 242 |
}
|
| 243 |
+
|
| 244 |
+
// إظهار بطاقة صفحات PDF
|
| 245 |
+
pdfPagesCard.classList.remove('d-none');
|
| 246 |
+
|
| 247 |
+
// إخفاء حالة المعالجة
|
| 248 |
+
processingStatus.classList.add('d-none');
|
| 249 |
+
|
| 250 |
+
} catch (error) {
|
| 251 |
+
console.error('Error processing file:', error);
|
| 252 |
+
statusText.textContent = `خطأ: ${error.message}`;
|
| 253 |
+
// لا نخفي حالة المعالجة لإظهار الخطأ
|
| 254 |
+
}
|
| 255 |
+
}
|
| 256 |
+
|
| 257 |
+
// معالجة ملف PDF
|
| 258 |
+
async function processPdf(file) {
|
| 259 |
+
try {
|
| 260 |
+
// تحميل ملف PDF
|
| 261 |
+
const arrayBuffer = await file.arrayBuffer();
|
| 262 |
+
const pdfDoc = await pdfjsLib.getDocument({ data: arrayBuffer }).promise;
|
| 263 |
+
|
| 264 |
+
// التحقق من عدد الصفحات
|
| 265 |
+
const numPages = pdfDoc.numPages;
|
| 266 |
+
if (numPages > 30) {
|
| 267 |
+
alert('هذا الملف يحتوي على أكثر من 30 صفحة. سيتم معالجة أول 30 صفحة فقط.');
|
| 268 |
}
|
| 269 |
|
| 270 |
+
// مسح حاوية الصفحات
|
| 271 |
+
pdfPagesContainer.innerHTML = '';
|
| 272 |
+
|
| 273 |
+
// عرض أول 30 صفحة كصور مصغرة
|
| 274 |
+
const maxPages = Math.min(numPages, 30);
|
| 275 |
+
for (let i = 1; i <= maxPages; i++) {
|
| 276 |
+
// تحديث شريط التقدم
|
| 277 |
+
progressBar.style.width = `${(i / maxPages) * 100}%`;
|
| 278 |
+
statusText.textContent = `جاري معالجة الصفحة ${i} من ${maxPages}...`;
|
| 279 |
+
|
| 280 |
+
// تحويل الصفحة إلى صورة
|
| 281 |
+
const pageImage = await convertPdfPageToImage(pdfDoc, i);
|
| 282 |
|
| 283 |
+
// إضافة الصفحة إلى المصفوفة
|
| 284 |
+
documentPages.push({
|
| 285 |
+
pageNumber: i,
|
| 286 |
+
imageData: pageImage.imageData,
|
| 287 |
+
width: pageImage.width,
|
| 288 |
+
height: pageImage.height
|
| 289 |
+
});
|
| 290 |
|
| 291 |
+
// إنشاء عنصر للصفحة في واجهة المستخدم
|
| 292 |
+
createPageElement(pageImage.imageData, i);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 293 |
}
|
| 294 |
+
|
| 295 |
+
} catch (error) {
|
| 296 |
+
console.error('Error processing PDF:', error);
|
| 297 |
+
throw error;
|
| 298 |
}
|
| 299 |
+
}
|
| 300 |
+
|
| 301 |
+
// معالجة ملف صورة
|
| 302 |
+
async function processImage(file) {
|
| 303 |
+
try {
|
| 304 |
+
// قراءة الصورة كـ Data URL
|
| 305 |
+
const imageData = await readFileAsDataURL(file);
|
| 306 |
+
|
| 307 |
+
// إنشاء صورة لقياس أبعادها
|
| 308 |
+
const img = new Image();
|
| 309 |
+
await new Promise((resolve, reject) => {
|
| 310 |
+
img.onload = resolve;
|
| 311 |
+
img.onerror = reject;
|
| 312 |
+
img.src = imageData;
|
| 313 |
+
});
|
| 314 |
+
|
| 315 |
+
// مسح حاوية الصفحات
|
| 316 |
+
pdfPagesContainer.innerHTML = '';
|
| 317 |
+
|
| 318 |
+
// إضافة الصورة إلى المصفوفة
|
| 319 |
+
documentPages = [{
|
| 320 |
+
pageNumber: 1,
|
| 321 |
+
imageData: imageData,
|
| 322 |
+
width: img.width,
|
| 323 |
+
height: img.height
|
| 324 |
+
}];
|
| 325 |
|
| 326 |
+
// إنشاء عنصر للصورة في واجهة المستخدم
|
| 327 |
+
createPageElement(imageData, 1);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 328 |
|
| 329 |
+
// تحديث شريط التقدم
|
| 330 |
+
progressBar.style.width = '100%';
|
| 331 |
+
statusText.textContent = 'تمت معالجة الصورة بنجاح';
|
| 332 |
|
| 333 |
+
} catch (error) {
|
| 334 |
+
console.error('Error processing image:', error);
|
| 335 |
+
throw error;
|
| 336 |
+
}
|
| 337 |
+
}
|
| 338 |
+
|
| 339 |
+
// تحويل صفحة PDF إلى صورة
|
| 340 |
+
async function convertPdfPageToImage(pdfDoc, pageNumber, scale = 1.5) {
|
| 341 |
+
try {
|
| 342 |
+
// الحصول على الصفحة من مستند PDF
|
| 343 |
+
const page = await pdfDoc.getPage(pageNumber);
|
| 344 |
+
|
| 345 |
+
// إنشاء عنصر canvas لرسم الصفحة
|
| 346 |
+
const canvas = document.createElement('canvas');
|
| 347 |
+
const context = canvas.getContext('2d');
|
| 348 |
+
|
| 349 |
+
// ضبط حجم الـ canvas بناءً على حجم الصفحة ومقياس التكبير
|
| 350 |
+
const viewport = page.getViewport({ scale });
|
| 351 |
+
canvas.width = viewport.width;
|
| 352 |
+
canvas.height = viewport.height;
|
| 353 |
+
|
| 354 |
+
// رسم الصفحة على الـ canvas
|
| 355 |
+
await page.render({
|
| 356 |
+
canvasContext: context,
|
| 357 |
+
viewport: viewport
|
| 358 |
+
}).promise;
|
| 359 |
+
|
| 360 |
+
// تحويل الـ canvas إلى صورة بصيغة PNG
|
| 361 |
+
return {
|
| 362 |
+
imageData: canvas.toDataURL('image/png'),
|
| 363 |
+
width: viewport.width,
|
| 364 |
+
height: viewport.height,
|
| 365 |
+
pageNumber: pageNumber
|
| 366 |
+
};
|
| 367 |
+
} catch (error) {
|
| 368 |
+
console.error(`Error converting PDF page ${pageNumber} to image:`, error);
|
| 369 |
+
throw error;
|
| 370 |
+
}
|
| 371 |
+
}
|
| 372 |
+
|
| 373 |
+
// إنشاء عنصر للصفحة في واجهة المستخدم
|
| 374 |
+
function createPageElement(imageData, pageNumber) {
|
| 375 |
+
const pageDiv = document.createElement('div');
|
| 376 |
+
pageDiv.className = 'pdf-page';
|
| 377 |
+
pageDiv.dataset.page = pageNumber;
|
| 378 |
+
|
| 379 |
+
const img = document.createElement('img');
|
| 380 |
+
img.src = imageData;
|
| 381 |
+
img.alt = `صفحة ${pageNumber}`;
|
| 382 |
+
|
| 383 |
+
const pageNumberDiv = document.createElement('div');
|
| 384 |
+
pageNumberDiv.className = 'page-number';
|
| 385 |
+
pageNumberDiv.textContent = `صفحة ${pageNumber}`;
|
| 386 |
+
|
| 387 |
+
pageDiv.appendChild(img);
|
| 388 |
+
pageDiv.appendChild(pageNumberDiv);
|
| 389 |
+
|
| 390 |
+
// إضافة حدث النقر للتحديد
|
| 391 |
+
pageDiv.addEventListener('click', function() {
|
| 392 |
+
this.classList.toggle('selected');
|
| 393 |
+
|
| 394 |
+
// تحديث مصفوفة الصفحات المحددة
|
| 395 |
+
const page = parseInt(this.dataset.page);
|
| 396 |
+
if (this.classList.contains('selected')) {
|
| 397 |
+
if (!selectedPages.includes(page)) {
|
| 398 |
+
selectedPages.push(page);
|
| 399 |
}
|
| 400 |
+
} else {
|
| 401 |
+
const index = selectedPages.indexOf(page);
|
| 402 |
+
if (index > -1) {
|
| 403 |
+
selectedPages.splice(index, 1);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 404 |
}
|
| 405 |
}
|
| 406 |
+
});
|
| 407 |
+
|
| 408 |
+
pdfPagesContainer.appendChild(pageDiv);
|
| 409 |
+
}
|
| 410 |
+
|
| 411 |
+
// قراءة ملف كـ Data URL
|
| 412 |
+
function readFileAsDataURL(file) {
|
| 413 |
+
return new Promise((resolve, reject) => {
|
| 414 |
+
const reader = new FileReader();
|
| 415 |
+
reader.onload = e => resolve(e.target.result);
|
| 416 |
+
reader.onerror = reject;
|
| 417 |
+
reader.readAsDataURL(file);
|
| 418 |
+
});
|
| 419 |
+
}
|
| 420 |
+
|
| 421 |
+
// تحديد كل الصفحات
|
| 422 |
+
function selectAllPages() {
|
| 423 |
+
document.querySelectorAll('.pdf-page').forEach(page => {
|
| 424 |
+
page.classList.add('selected');
|
| 425 |
+
const pageNumber = parseInt(page.dataset.page);
|
| 426 |
+
if (!selectedPages.includes(pageNumber)) {
|
| 427 |
+
selectedPages.push(pageNumber);
|
| 428 |
}
|
| 429 |
+
});
|
| 430 |
+
}
|
| 431 |
+
|
| 432 |
+
// إلغاء تحديد كل الصفحات
|
| 433 |
+
function deselectAllPages() {
|
| 434 |
+
document.querySelectorAll('.pdf-page').forEach(page => {
|
| 435 |
+
page.classList.remove('selected');
|
| 436 |
+
});
|
| 437 |
+
selectedPages = [];
|
| 438 |
+
}
|
| 439 |
+
|
| 440 |
+
// استخراج الصور
|
| 441 |
+
function extractImages() {
|
| 442 |
+
if (documentPages.length === 0) {
|
| 443 |
+
alert('لا توجد صفحات للمعالجة');
|
| 444 |
+
return;
|
| 445 |
}
|
| 446 |
|
| 447 |
+
// اختيار الصفحات المحددة فقط أو كل الصفحات إذا لم يتم تحديد أي صفحة
|
| 448 |
+
const pagesToExtract = selectedPages.length > 0
|
| 449 |
+
? documentPages.filter(page => selectedPages.includes(page.pageNumber))
|
| 450 |
+
: documentPages;
|
|
|
|
| 451 |
|
| 452 |
+
if (pagesToExtract.length === 0) {
|
| 453 |
+
alert('الرجاء تحديد صفحة واحدة على الأقل');
|
| 454 |
+
return;
|
|
|
|
|
|
|
| 455 |
}
|
| 456 |
|
| 457 |
+
// إنشاء وتنزيل كل صورة
|
| 458 |
+
pagesToExtract.forEach(page => {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 459 |
const a = document.createElement('a');
|
| 460 |
+
a.href = page.imageData;
|
| 461 |
+
a.download = `mondo_lingua_page_${page.pageNumber}.png`;
|
| 462 |
document.body.appendChild(a);
|
| 463 |
a.click();
|
| 464 |
document.body.removeChild(a);
|
| 465 |
+
});
|
| 466 |
+
|
| 467 |
+
// إظهار رسالة نجاح
|
| 468 |
+
alert(`تم استخراج ${pagesToExtract.length} صورة بنجاح`);
|
| 469 |
+
}
|
| 470 |
+
|
| 471 |
+
// استخراج النص
|
| 472 |
+
async function extractText() {
|
| 473 |
+
if (documentPages.length === 0) {
|
| 474 |
+
alert('لا توجد صفحات للمعالجة');
|
| 475 |
+
return;
|
| 476 |
+
}
|
| 477 |
+
|
| 478 |
+
// اختيار الصفحات المحددة فقط أو كل الصفحات إذا لم يتم تحديد أي صفحة
|
| 479 |
+
const pagesToProcess = selectedPages.length > 0
|
| 480 |
+
? documentPages.filter(page => selectedPages.includes(page.pageNumber))
|
| 481 |
+
: documentPages;
|
| 482 |
+
|
| 483 |
+
if (pagesToProcess.length === 0) {
|
| 484 |
+
alert('الرجاء تحديد صفحة واحدة على الأقل');
|
| 485 |
+
return;
|
| 486 |
}
|
| 487 |
|
| 488 |
+
// إظهار حالة المعالجة
|
| 489 |
+
processingStatus.classList.remove('d-none');
|
| 490 |
+
statusText.textContent = 'جاري استخراج النص...';
|
| 491 |
+
progressBar.style.width = '0%';
|
| 492 |
+
|
| 493 |
+
try {
|
| 494 |
+
extractedTexts = [];
|
| 495 |
|
| 496 |
+
// معالجة كل صفحة
|
| 497 |
+
for (let i = 0; i < pagesToProcess.length; i++) {
|
| 498 |
+
const page = pagesToProcess[i];
|
| 499 |
+
|
| 500 |
+
// تحديث التقدم
|
| 501 |
+
progressBar.style.width = `${((i + 1) / pagesToProcess.length) * 100}%`;
|
| 502 |
+
statusText.textContent = `جاري معالجة الصفحة ${i + 1} من ${pagesToProcess.length}...`;
|
| 503 |
+
|
| 504 |
+
// استخراج النص من الصورة باستخدام OCR API
|
| 505 |
+
const pageText = await extractTextFromImage(page.imageData, page.pageNumber);
|
| 506 |
+
extractedTexts.push(pageText);
|
| 507 |
+
}
|
| 508 |
|
| 509 |
+
// جمع النصوص المستخرجة
|
| 510 |
+
const combinedText = extractedTexts.join('\n\n');
|
|
|
|
|
|
|
|
|
|
| 511 |
|
| 512 |
+
// عرض النص المستخرج
|
| 513 |
+
resultText.textContent = combinedText;
|
| 514 |
+
resultsCard.classList.remove('d-none');
|
| 515 |
+
|
| 516 |
+
// إخفاء حالة المعالجة
|
| 517 |
+
processingStatus.classList.add('d-none');
|
| 518 |
+
|
| 519 |
+
// التمرير إلى النتائج
|
| 520 |
+
resultsCard.scrollIntoView({ behavior: 'smooth' });
|
| 521 |
|
| 522 |
+
} catch (error) {
|
| 523 |
+
console.error('Error extracting text:', error);
|
| 524 |
+
statusText.textContent = `خطأ: ${error.message}`;
|
| 525 |
+
}
|
| 526 |
+
}
|
| 527 |
+
|
| 528 |
+
// استخراج النص من صورة باستخدام API4AI OCR
|
| 529 |
+
async function extractTextFromImage(imageData, pageNumber) {
|
| 530 |
+
try {
|
| 531 |
+
// تحويل Data URL إلى Blob
|
| 532 |
+
const response = await fetch(imageData);
|
| 533 |
+
const blob = await response.blob();
|
| 534 |
|
| 535 |
+
// إنشاء FormData وإضافة الصورة
|
| 536 |
const formData = new FormData();
|
| 537 |
+
formData.append('image', blob, `page_${pageNumber}.png`);
|
| 538 |
|
| 539 |
+
// طلب OCR
|
| 540 |
+
const ocrResponse = await fetch(OCR_API_URL, {
|
| 541 |
method: 'POST',
|
| 542 |
headers: {
|
| 543 |
+
'X-RapidAPI-Key': RAPIDAPI_KEY,
|
| 544 |
+
'X-RapidAPI-Host': 'ocr43.p.rapidapi.com'
|
| 545 |
},
|
| 546 |
body: formData
|
| 547 |
});
|
| 548 |
|
| 549 |
+
if (!ocrResponse.ok) {
|
| 550 |
+
throw new Error(`فشل في طلب OCR: ${ocrResponse.status}`);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 551 |
}
|
| 552 |
|
| 553 |
+
const data = await ocrResponse.json();
|
|
|
|
| 554 |
|
| 555 |
+
try {
|
| 556 |
+
// استخراج النص من الاستجابة
|
| 557 |
+
const text = data.results[0].entities[0].objects[0].entities[0].text;
|
| 558 |
+
return `=== صفحة ${pageNumber} ===\n${text}`;
|
| 559 |
+
} catch (e) {
|
| 560 |
+
console.error('Error parsing OCR response:', e);
|
| 561 |
+
return `=== صفحة ${pageNumber} ===\n[خطأ في معالجة النص]`;
|
| 562 |
+
}
|
| 563 |
+
} catch (error) {
|
| 564 |
+
console.error(`Error in OCR for page ${pageNumber}:`, error);
|
| 565 |
+
throw error;
|
| 566 |
+
}
|
| 567 |
+
}
|
| 568 |
+
|
| 569 |
+
// نسخ النص
|
| 570 |
+
function copyText() {
|
| 571 |
+
const text = resultText.textContent;
|
| 572 |
+
if (!text || text === 'لم يتم استخراج نص بعد.') {
|
| 573 |
+
alert('لا يوجد نص للنسخ');
|
| 574 |
+
return;
|
| 575 |
}
|
| 576 |
|
| 577 |
+
navigator.clipboard.writeText(text)
|
| 578 |
+
.then(() => alert('تم نسخ النص بنجاح'))
|
| 579 |
+
.catch(err => alert('حدث خطأ أثناء نسخ النص: ' + err));
|
| 580 |
+
}
|
| 581 |
+
|
| 582 |
+
// تنزيل النص
|
| 583 |
+
function downloadText() {
|
| 584 |
+
const text = resultText.textContent;
|
| 585 |
+
if (!text || text === 'لم يتم استخراج نص بعد.') {
|
| 586 |
+
alert('لا يوجد نص للتنزيل');
|
| 587 |
+
return;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 588 |
}
|
| 589 |
+
|
| 590 |
+
const blob = new Blob([text], { type: 'text/plain;charset=utf-8' });
|
| 591 |
+
const url = URL.createObjectURL(blob);
|
| 592 |
+
const a = document.createElement('a');
|
| 593 |
+
a.href = url;
|
| 594 |
+
a.download = 'mondo_lingua_ocr_text.txt';
|
| 595 |
+
document.body.appendChild(a);
|
| 596 |
+
a.click();
|
| 597 |
+
document.body.removeChild(a);
|
| 598 |
+
URL.revokeObjectURL(url);
|
| 599 |
+
}
|
| 600 |
</script>
|
| 601 |
</body>
|
| 602 |
</html>
|