Hugging Face's logo

Spaces:
RevolutionCrossroads
/
ocr_correction
Running

App Files Community
MikeTrizna commited on
Commit
5b47a5c
·
1 Parent(s): 9efc1d1

Initial load of web page and supporting files

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. .gitattributes +1 -0
  2. data.json +12 -0
  3. index.html +735 -19
  4. loc_ocr/00001.xml +0 -0
  5. pdfs/00001.pdf +3 -0
  6. pdfs/00547.pdf +3 -0
  7. pdfs/00923.pdf +3 -0
  8. pdfs/01063.pdf +3 -0
  9. pdfs/01405.pdf +3 -0
  10. pdfs/01608.pdf +3 -0
  11. pdfs/01689.pdf +3 -0
  12. pdfs/01718.pdf +3 -0
  13. pdfs/01786.pdf +3 -0
  14. pdfs/02440.pdf +3 -0
  15. pdfs/02999.pdf +3 -0
  16. pdfs/03141.pdf +3 -0
  17. pdfs/03528.pdf +3 -0
  18. pdfs/04184.pdf +3 -0
  19. pdfs/05082.pdf +3 -0
  20. pdfs/06035.pdf +3 -0
  21. pdfs/06145.pdf +3 -0
  22. pdfs/07555.pdf +3 -0
  23. pdfs/07787.pdf +3 -0
  24. pdfs/09426.pdf +3 -0
  25. pdfs/10261.pdf +3 -0
  26. pdfs/10875.pdf +3 -0
  27. pdfs/11901.pdf +3 -0
  28. pdfs/12099.pdf +3 -0
  29. pdfs/12747.pdf +3 -0
  30. pdfs/12827.pdf +3 -0
  31. pdfs/12964.pdf +3 -0
  32. pdfs/13089.pdf +3 -0
  33. pdfs/14825.pdf +3 -0
  34. pdfs/15210.pdf +3 -0
  35. pdfs/16124.pdf +3 -0
  36. pdfs/16999.pdf +3 -0
  37. pdfs/17628.pdf +3 -0
  38. pdfs/17885.pdf +3 -0
  39. pdfs/19652.pdf +3 -0
  40. pdfs/19818.pdf +3 -0
  41. pdfs/19915.pdf +3 -0
  42. pdfs/22785.pdf +3 -0
  43. pdfs/22939.pdf +3 -0
  44. pdfs/23092.pdf +3 -0
  45. pdfs/24047.pdf +3 -0
  46. pdfs/25152.pdf +3 -0
  47. pdfs/25532.pdf +3 -0
  48. pdfs/25615.pdf +3 -0
  49. pdfs/26136.pdf +3 -0
  50. pdfs/28406.pdf +3 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ *.pdf filter=lfs diff=lfs merge=lfs -text
data.json ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ [
2
+ {
3
+ "id": "00001",
4
+ "newspaper_title": "The Delaware gazette",
5
+ "publication_date": "1809-07-08",
6
+ "page": "1",
7
+ "loc_record_url": "https://www.loc.gov/resource/sn82014385/1809-07-08/ed-1/?sp=1",
8
+ "pdf_path": "pdfs/00001.pdf",
9
+ "textract_ocr_file": "textract_ocr/00001_layout_extracted.txt",
10
+ "loc_altoxml_path": "loc_ocr/00001.xml"
11
+ }
12
+ ]
index.html CHANGED
@@ -1,19 +1,735 @@
1
- <!doctype html>
2
- <html>
3
- <head>
4
- <meta charset="utf-8" />
5
- <meta name="viewport" content="width=device-width" />
6
- <title>My static Space</title>
7
- <link rel="stylesheet" href="style.css" />
8
- </head>
9
- <body>
10
- <div class="card">
11
- <h1>Welcome to your static Space!</h1>
12
- <p>You can modify this app directly by editing <i>index.html</i> in the Files and versions tab.</p>
13
- <p>
14
- Also don't forget to check the
15
- <a href="https://huggingface.co/docs/hub/spaces" target="_blank">Spaces documentation</a>.
16
- </p>
17
- </div>
18
- </body>
19
- </html>
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ <!DOCTYPE html>
2
+ <html lang="en">
3
+ <head>
4
+ <meta charset="UTF-8">
5
+ <meta name="viewport" content="width=device-width, initial-scale=1.0">
6
+ <title>Historical Newspaper OCR Viewer</title>
7
+
8
+ <!-- PDF.js Library -->
9
+ <script src="https://cdnjs.cloudflare.com/ajax/libs/pdf.js/3.11.174/pdf.min.js"></script>
10
+
11
+ <style>
12
+ * {
13
+ margin: 0;
14
+ padding: 0;
15
+ box-sizing: border-box;
16
+ }
17
+
18
+ body {
19
+ font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Roboto, Oxygen, Ubuntu, Cantarell, sans-serif;
20
+ height: 100vh;
21
+ overflow: hidden;
22
+ background-color: #f5f5f5;
23
+ }
24
+
25
+ .header {
26
+ background-color: #2c3e50;
27
+ color: white;
28
+ padding: 15px 20px;
29
+ box-shadow: 0 2px 4px rgba(0,0,0,0.1);
30
+ }
31
+
32
+ .header h1 {
33
+ font-size: 24px;
34
+ margin-bottom: 10px;
35
+ }
36
+
37
+ .controls {
38
+ display: flex;
39
+ gap: 15px;
40
+ align-items: center;
41
+ flex-wrap: wrap;
42
+ }
43
+
44
+ .select-container {
45
+ flex: 1;
46
+ max-width: 800px;
47
+ }
48
+
49
+ .select-container select {
50
+ width: 100%;
51
+ padding: 8px 12px;
52
+ border: 1px solid #ddd;
53
+ border-radius: 4px;
54
+ font-size: 14px;
55
+ background-color: white;
56
+ }
57
+
58
+ .metadata {
59
+ background-color: #ecf0f1;
60
+ padding: 15px 20px;
61
+ border-bottom: 1px solid #bdc3c7;
62
+ display: none;
63
+ }
64
+
65
+ .metadata.active {
66
+ display: block;
67
+ }
68
+
69
+ .metadata-content {
70
+ display: flex;
71
+ gap: 20px;
72
+ align-items: center;
73
+ flex-wrap: wrap;
74
+ }
75
+
76
+ .metadata-item {
77
+ font-size: 14px;
78
+ }
79
+
80
+ .metadata-item strong {
81
+ color: #2c3e50;
82
+ }
83
+
84
+ .metadata-item a {
85
+ color: #3498db;
86
+ text-decoration: none;
87
+ }
88
+
89
+ .metadata-item a:hover {
90
+ text-decoration: underline;
91
+ }
92
+
93
+ .main-container {
94
+ display: flex;
95
+ height: calc(100vh - 180px);
96
+ background-color: white;
97
+ }
98
+
99
+ .pdf-panel, .text-panel {
100
+ width: 50%;
101
+ display: flex;
102
+ flex-direction: column;
103
+ border-right: 1px solid #ddd;
104
+ }
105
+
106
+ .text-panel {
107
+ border-right: none;
108
+ }
109
+
110
+ .panel-header {
111
+ background-color: #34495e;
112
+ color: white;
113
+ padding: 12px 20px;
114
+ font-weight: 600;
115
+ font-size: 14px;
116
+ display: flex;
117
+ justify-content: space-between;
118
+ align-items: center;
119
+ }
120
+
121
+ .pdf-controls {
122
+ display: flex;
123
+ gap: 10px;
124
+ align-items: center;
125
+ }
126
+
127
+ .zoom-btn {
128
+ background-color: #2c3e50;
129
+ color: white;
130
+ border: none;
131
+ padding: 5px 12px;
132
+ border-radius: 4px;
133
+ cursor: pointer;
134
+ font-size: 16px;
135
+ font-weight: bold;
136
+ transition: background-color 0.2s;
137
+ min-width: 32px;
138
+ }
139
+
140
+ .zoom-btn:hover:not(:disabled) {
141
+ background-color: #1a252f;
142
+ }
143
+
144
+ .zoom-btn:disabled {
145
+ background-color: #95a5a6;
146
+ cursor: not-allowed;
147
+ opacity: 0.5;
148
+ }
149
+
150
+ .zoom-level {
151
+ font-size: 13px;
152
+ min-width: 50px;
153
+ text-align: center;
154
+ }
155
+
156
+ .ocr-controls {
157
+ display: flex;
158
+ gap: 15px;
159
+ align-items: center;
160
+ }
161
+
162
+ .ocr-toggle {
163
+ display: flex;
164
+ gap: 10px;
165
+ background-color: #2c3e50;
166
+ padding: 5px;
167
+ border-radius: 4px;
168
+ }
169
+
170
+ .ocr-toggle label {
171
+ padding: 5px 12px;
172
+ cursor: pointer;
173
+ border-radius: 3px;
174
+ font-size: 13px;
175
+ transition: background-color 0.2s;
176
+ }
177
+
178
+ .ocr-toggle input[type="radio"] {
179
+ display: none;
180
+ }
181
+
182
+ .ocr-toggle input[type="radio"]:checked + span {
183
+ background-color: #3498db;
184
+ }
185
+
186
+ .ocr-toggle span {
187
+ padding: 5px 12px;
188
+ border-radius: 3px;
189
+ font-size: 13px;
190
+ }
191
+
192
+ .export-btn {
193
+ background-color: #27ae60;
194
+ color: white;
195
+ border: none;
196
+ padding: 6px 16px;
197
+ border-radius: 4px;
198
+ cursor: pointer;
199
+ font-size: 13px;
200
+ font-weight: 600;
201
+ transition: background-color 0.2s;
202
+ }
203
+
204
+ .export-btn:hover {
205
+ background-color: #229954;
206
+ }
207
+
208
+ .export-btn:disabled {
209
+ background-color: #95a5a6;
210
+ cursor: not-allowed;
211
+ }
212
+
213
+ .panel-content {
214
+ flex: 1;
215
+ overflow: auto;
216
+ position: relative;
217
+ }
218
+
219
+ .panel-content.grabbable {
220
+ cursor: grab;
221
+ }
222
+
223
+ .panel-content.grabbing {
224
+ cursor: grabbing;
225
+ }
226
+
227
+ #pdf-canvas {
228
+ display: block;
229
+ margin: 20px auto;
230
+ box-shadow: 0 2px 8px rgba(0,0,0,0.1);
231
+ }
232
+
233
+ #ocr-text {
234
+ width: 100%;
235
+ height: 100%;
236
+ border: none;
237
+ padding: 20px;
238
+ font-family: 'Courier New', monospace;
239
+ font-size: 14px;
240
+ line-height: 1.6;
241
+ resize: none;
242
+ outline: none;
243
+ }
244
+
245
+ .loading {
246
+ position: absolute;
247
+ top: 50%;
248
+ left: 50%;
249
+ transform: translate(-50%, -50%);
250
+ text-align: center;
251
+ color: #7f8c8d;
252
+ }
253
+
254
+ .loading-spinner {
255
+ border: 4px solid #ecf0f1;
256
+ border-top: 4px solid #3498db;
257
+ border-radius: 50%;
258
+ width: 40px;
259
+ height: 40px;
260
+ animation: spin 1s linear infinite;
261
+ margin: 0 auto 10px;
262
+ }
263
+
264
+ @keyframes spin {
265
+ 0% { transform: rotate(0deg); }
266
+ 100% { transform: rotate(360deg); }
267
+ }
268
+
269
+ .error-message {
270
+ color: #e74c3c;
271
+ padding: 20px;
272
+ text-align: center;
273
+ }
274
+
275
+ .empty-state {
276
+ text-align: center;
277
+ color: #95a5a6;
278
+ padding: 40px;
279
+ font-size: 16px;
280
+ }
281
+
282
+ .empty-state-icon {
283
+ font-size: 48px;
284
+ margin-bottom: 10px;
285
+ }
286
+ </style>
287
+ </head>
288
+ <body>
289
+ <div class="header">
290
+ <h1>Historical Newspaper OCR Viewer</h1>
291
+ <div class="controls">
292
+ <div class="select-container">
293
+ <select id="record-select" disabled>
294
+ <option value="">Loading records...</option>
295
+ </select>
296
+ </div>
297
+ </div>
298
+ </div>
299
+
300
+ <div class="metadata" id="metadata">
301
+ <div class="metadata-content">
302
+ <div class="metadata-item">
303
+ <strong>Title:</strong> <span id="meta-title">-</span>
304
+ </div>
305
+ <div class="metadata-item">
306
+ <strong>Date:</strong> <span id="meta-date">-</span>
307
+ </div>
308
+ <div class="metadata-item">
309
+ <strong>Page:</strong> <span id="meta-page">-</span>
310
+ </div>
311
+ <div class="metadata-item">
312
+ <a id="meta-link" href="#" target="_blank">View on Library of Congress →</a>
313
+ </div>
314
+ </div>
315
+ </div>
316
+
317
+ <div class="main-container">
318
+ <div class="pdf-panel">
319
+ <div class="panel-header">
320
+ <span>PDF Viewer</span>
321
+ <div class="pdf-controls">
322
+ <button class="zoom-btn" id="zoom-out-btn" disabled title="Zoom Out">−</button>
323
+ <span class="zoom-level" id="zoom-level">100%</span>
324
+ <button class="zoom-btn" id="zoom-in-btn" disabled title="Zoom In">+</button>
325
+ <button class="zoom-btn" id="zoom-reset-btn" disabled title="Reset Zoom">Reset</button>
326
+ </div>
327
+ </div>
328
+ <div class="panel-content" id="pdf-container">
329
+ <div class="empty-state">
330
+ <div class="empty-state-icon">📄</div>
331
+ <div>Select a record to view the PDF</div>
332
+ </div>
333
+ </div>
334
+ </div>
335
+
336
+ <div class="text-panel">
337
+ <div class="panel-header">
338
+ <span>OCR Text</span>
339
+ <div class="ocr-controls">
340
+ <div class="ocr-toggle">
341
+ <label>
342
+ <input type="radio" name="ocr-source" value="textract" checked>
343
+ <span>Textract OCR</span>
344
+ </label>
345
+ <label>
346
+ <input type="radio" name="ocr-source" value="loc">
347
+ <span>Library of Congress OCR</span>
348
+ </label>
349
+ </div>
350
+ <button class="export-btn" id="export-btn" disabled>Export Text</button>
351
+ </div>
352
+ </div>
353
+ <div class="panel-content">
354
+ <textarea
355
+ id="ocr-text"
356
+ placeholder="Select a record and OCR source to view text..."
357
+ disabled
358
+ ></textarea>
359
+ </div>
360
+ </div>
361
+ </div>
362
+
363
+ <script>
364
+ // Configure PDF.js worker
365
+ pdfjsLib.GlobalWorkerOptions.workerSrc = 'https://cdnjs.cloudflare.com/ajax/libs/pdf.js/3.11.174/pdf.worker.min.js';
366
+
367
+ // Application state
368
+ const state = {
369
+ records: [],
370
+ currentRecord: null,
371
+ currentOcrSource: 'textract',
372
+ originalOcrText: '',
373
+ hasEdits: false,
374
+ isLoading: false,
375
+ currentPdf: null,
376
+ zoomLevel: 1.0
377
+ };
378
+
379
+ // DOM elements
380
+ const recordSelect = document.getElementById('record-select');
381
+ const ocrText = document.getElementById('ocr-text');
382
+ const exportBtn = document.getElementById('export-btn');
383
+ const pdfContainer = document.getElementById('pdf-container');
384
+ const metadata = document.getElementById('metadata');
385
+ const ocrRadios = document.querySelectorAll('input[name="ocr-source"]');
386
+ const zoomInBtn = document.getElementById('zoom-in-btn');
387
+ const zoomOutBtn = document.getElementById('zoom-out-btn');
388
+ const zoomResetBtn = document.getElementById('zoom-reset-btn');
389
+ const zoomLevel = document.getElementById('zoom-level');
390
+
391
+ // Initialize application
392
+ async function init() {
393
+ try {
394
+ const response = await fetch('data.json');
395
+ if (!response.ok) {
396
+ throw new Error('Failed to load data.json');
397
+ }
398
+ state.records = await response.json();
399
+
400
+ populateRecordSelect();
401
+ enableControls();
402
+ setupEventListeners();
403
+ } catch (error) {
404
+ console.error('Initialization error:', error);
405
+ recordSelect.innerHTML = '<option value="">Error loading records</option>';
406
+ alert('Failed to load data.json. Please ensure the file exists in the same directory as index.html.');
407
+ }
408
+ }
409
+
410
+ function enableControls() {
411
+ recordSelect.disabled = false;
412
+ }
413
+
414
+ function populateRecordSelect() {
415
+ recordSelect.innerHTML = '<option value="">Select a record...</option>';
416
+
417
+ state.records.forEach(record => {
418
+ const option = document.createElement('option');
419
+ option.value = record.id;
420
+ option.textContent = `${record.id} - ${record.newspaper_title} ${record.publication_date} Page ${record.page}`;
421
+ recordSelect.appendChild(option);
422
+ });
423
+ }
424
+
425
+ function setupEventListeners() {
426
+ // Record selection
427
+ recordSelect.addEventListener('change', (e) => {
428
+ const recordId = e.target.value;
429
+ if (recordId) {
430
+ const record = state.records.find(r => r.id === recordId);
431
+ loadRecord(record);
432
+ }
433
+ });
434
+
435
+ // OCR source toggle
436
+ ocrRadios.forEach(radio => {
437
+ radio.addEventListener('change', (e) => {
438
+ if (state.hasEdits) {
439
+ const confirmed = confirm('You have unsaved edits. Switching OCR source will discard your changes. Continue?');
440
+ if (!confirmed) {
441
+ // Revert to previous selection
442
+ document.querySelector(`input[name="ocr-source"][value="${state.currentOcrSource}"]`).checked = true;
443
+ return;
444
+ }
445
+ }
446
+ state.currentOcrSource = e.target.value;
447
+ if (state.currentRecord) {
448
+ loadOcrText(state.currentRecord);
449
+ }
450
+ });
451
+ });
452
+
453
+ // Track edits
454
+ ocrText.addEventListener('input', () => {
455
+ state.hasEdits = ocrText.value !== state.originalOcrText;
456
+ });
457
+
458
+ // Export button
459
+ exportBtn.addEventListener('click', exportText);
460
+
461
+ // Zoom controls
462
+ zoomInBtn.addEventListener('click', () => zoomPdf(0.25));
463
+ zoomOutBtn.addEventListener('click', () => zoomPdf(-0.25));
464
+ zoomResetBtn.addEventListener('click', () => resetZoom());
465
+
466
+ // Pan functionality
467
+ setupPanControls();
468
+ }
469
+
470
+ async function loadRecord(record) {
471
+ state.currentRecord = record;
472
+ state.hasEdits = false;
473
+
474
+ // Update metadata
475
+ metadata.classList.add('active');
476
+ document.getElementById('meta-title').textContent = record.newspaper_title;
477
+ document.getElementById('meta-date').textContent = record.publication_date;
478
+ document.getElementById('meta-page').textContent = record.page;
479
+ document.getElementById('meta-link').href = record.loc_record_url;
480
+
481
+ // Load PDF and OCR text in parallel
482
+ await Promise.all([
483
+ loadPdf(record.pdf_path),
484
+ loadOcrText(record)
485
+ ]);
486
+ }
487
+
488
+ async function loadPdf(pdfPath) {
489
+ pdfContainer.innerHTML = '<div class="loading"><div class="loading-spinner"></div><div>Loading PDF...</div></div>';
490
+
491
+ try {
492
+ const loadingTask = pdfjsLib.getDocument(pdfPath);
493
+ const pdf = await loadingTask.promise;
494
+ state.currentPdf = pdf;
495
+ state.zoomLevel = 1.0;
496
+
497
+ await renderPdf();
498
+
499
+ // Enable zoom controls
500
+ zoomInBtn.disabled = false;
501
+ zoomOutBtn.disabled = false;
502
+ zoomResetBtn.disabled = false;
503
+
504
+ // Enable pan controls
505
+ pdfContainer.classList.add('grabbable');
506
+ } catch (error) {
507
+ console.error('PDF loading error:', error);
508
+ pdfContainer.innerHTML = '<div class="error-message">Failed to load PDF: ' + error.message + '</div>';
509
+ }
510
+ }
511
+
512
+ async function renderPdf() {
513
+ if (!state.currentPdf) return;
514
+
515
+ pdfContainer.innerHTML = '<div class="loading"><div class="loading-spinner"></div><div>Rendering PDF...</div></div>';
516
+
517
+ const page = await state.currentPdf.getPage(1);
518
+ const baseScale = 1.5;
519
+ const scale = baseScale * state.zoomLevel;
520
+ const viewport = page.getViewport({ scale });
521
+
522
+ const canvas = document.createElement('canvas');
523
+ canvas.id = 'pdf-canvas';
524
+ const context = canvas.getContext('2d');
525
+ canvas.height = viewport.height;
526
+ canvas.width = viewport.width;
527
+
528
+ const renderContext = {
529
+ canvasContext: context,
530
+ viewport: viewport
531
+ };
532
+
533
+ await page.render(renderContext).promise;
534
+
535
+ pdfContainer.innerHTML = '';
536
+ pdfContainer.appendChild(canvas);
537
+
538
+ // Update zoom level display
539
+ zoomLevel.textContent = Math.round(state.zoomLevel * 100) + '%';
540
+ }
541
+
542
+ async function loadOcrText(record) {
543
+ ocrText.value = '';
544
+ ocrText.disabled = true;
545
+ exportBtn.disabled = true;
546
+ state.hasEdits = false;
547
+
548
+ const loadingDiv = document.createElement('div');
549
+ loadingDiv.className = 'loading';
550
+ loadingDiv.innerHTML = '<div class="loading-spinner"></div><div>Loading OCR text...</div>';
551
+ ocrText.parentElement.appendChild(loadingDiv);
552
+
553
+ try {
554
+ let text = '';
555
+
556
+ if (state.currentOcrSource === 'textract') {
557
+ text = await fetchTextractOcr(record.textract_ocr_file);
558
+ } else {
559
+ text = await fetchLocOcr(record.loc_altoxml_path);
560
+ }
561
+
562
+ state.originalOcrText = text;
563
+ ocrText.value = text;
564
+ ocrText.disabled = false;
565
+ exportBtn.disabled = false;
566
+ loadingDiv.remove();
567
+ } catch (error) {
568
+ console.error('OCR loading error:', error);
569
+ loadingDiv.innerHTML = '<div class="error-message">Failed to load OCR text: ' + error.message + '</div>';
570
+ setTimeout(() => loadingDiv.remove(), 3000);
571
+ }
572
+ }
573
+
574
+ async function fetchTextractOcr(filePath) {
575
+ const response = await fetch(filePath);
576
+ if (!response.ok) {
577
+ throw new Error(`HTTP error! status: ${response.status}`);
578
+ }
579
+ return await response.text();
580
+ }
581
+
582
+ async function fetchLocOcr(xmlPath) {
583
+ const response = await fetch(xmlPath);
584
+ if (!response.ok) {
585
+ throw new Error(`HTTP error! status: ${response.status}`);
586
+ }
587
+ const xmlText = await response.text();
588
+ return parseAltoXml(xmlText);
589
+ }
590
+
591
+ function parseAltoXml(xmlText) {
592
+ const parser = new DOMParser();
593
+ const xmlDoc = parser.parseFromString(xmlText, 'text/xml');
594
+
595
+ // Check for parsing errors
596
+ const parserError = xmlDoc.querySelector('parsererror');
597
+ if (parserError) {
598
+ throw new Error('XML parsing error');
599
+ }
600
+
601
+ const strings = xmlDoc.getElementsByTagName('String');
602
+ const textLines = [];
603
+ let currentLine = [];
604
+ let lastVPos = null;
605
+
606
+ for (let i = 0; i < strings.length; i++) {
607
+ const stringElement = strings[i];
608
+ const content = stringElement.getAttribute('CONTENT');
609
+ const vpos = stringElement.getAttribute('VPOS');
610
+
611
+ if (content) {
612
+ // Check if we're on a new line (VPOS changed significantly)
613
+ if (lastVPos !== null && vpos !== null) {
614
+ const vposDiff = Math.abs(parseInt(vpos) - parseInt(lastVPos));
615
+ if (vposDiff > 50) { // Threshold for new line
616
+ if (currentLine.length > 0) {
617
+ textLines.push(currentLine.join(' '));
618
+ currentLine = [];
619
+ }
620
+ }
621
+ }
622
+
623
+ currentLine.push(content);
624
+ lastVPos = vpos;
625
+ }
626
+ }
627
+
628
+ // Add the last line
629
+ if (currentLine.length > 0) {
630
+ textLines.push(currentLine.join(' '));
631
+ }
632
+
633
+ return textLines.join('\n');
634
+ }
635
+
636
+ function exportText() {
637
+ if (!state.currentRecord) return;
638
+
639
+ const text = ocrText.value;
640
+ const record = state.currentRecord;
641
+
642
+ // Sanitize filename
643
+ const sanitizedTitle = record.newspaper_title
644
+ .replace(/[^a-z0-9]/gi, '_')
645
+ .replace(/_+/g, '_')
646
+ .toLowerCase();
647
+
648
+ const filename = `${record.id}_${sanitizedTitle}_${record.publication_date}.txt`;
649
+
650
+ // Create blob and download
651
+ const blob = new Blob([text], { type: 'text/plain' });
652
+ const url = URL.createObjectURL(blob);
653
+ const a = document.createElement('a');
654
+ a.href = url;
655
+ a.download = filename;
656
+ document.body.appendChild(a);
657
+ a.click();
658
+ document.body.removeChild(a);
659
+ URL.revokeObjectURL(url);
660
+ }
661
+
662
+ function zoomPdf(delta) {
663
+ // Store current scroll position as percentage
664
+ const scrollLeftPercent = pdfContainer.scrollLeft / (pdfContainer.scrollWidth - pdfContainer.clientWidth);
665
+ const scrollTopPercent = pdfContainer.scrollTop / (pdfContainer.scrollHeight - pdfContainer.clientHeight);
666
+
667
+ // Update zoom level (limit between 0.5x and 5x)
668
+ state.zoomLevel = Math.max(0.5, Math.min(5, state.zoomLevel + delta));
669
+
670
+ // Re-render PDF
671
+ renderPdf().then(() => {
672
+ // Restore scroll position
673
+ pdfContainer.scrollLeft = scrollLeftPercent * (pdfContainer.scrollWidth - pdfContainer.clientWidth);
674
+ pdfContainer.scrollTop = scrollTopPercent * (pdfContainer.scrollHeight - pdfContainer.clientHeight);
675
+ });
676
+ }
677
+
678
+ function resetZoom() {
679
+ state.zoomLevel = 1.0;
680
+ renderPdf();
681
+ }
682
+
683
+ function setupPanControls() {
684
+ let isMouseDown = false;
685
+ let startX, startY, scrollLeft, scrollTop;
686
+
687
+ pdfContainer.addEventListener('mousedown', (e) => {
688
+ // Only pan if we clicked on the container or canvas
689
+ if (e.target === pdfContainer || e.target.id === 'pdf-canvas') {
690
+ isMouseDown = true;
691
+ pdfContainer.classList.remove('grabbable');
692
+ pdfContainer.classList.add('grabbing');
693
+
694
+ startX = e.pageX - pdfContainer.offsetLeft;
695
+ startY = e.pageY - pdfContainer.offsetTop;
696
+ scrollLeft = pdfContainer.scrollLeft;
697
+ scrollTop = pdfContainer.scrollTop;
698
+ }
699
+ });
700
+
701
+ pdfContainer.addEventListener('mouseleave', () => {
702
+ if (isMouseDown) {
703
+ isMouseDown = false;
704
+ pdfContainer.classList.remove('grabbing');
705
+ pdfContainer.classList.add('grabbable');
706
+ }
707
+ });
708
+
709
+ pdfContainer.addEventListener('mouseup', () => {
710
+ if (isMouseDown) {
711
+ isMouseDown = false;
712
+ pdfContainer.classList.remove('grabbing');
713
+ pdfContainer.classList.add('grabbable');
714
+ }
715
+ });
716
+
717
+ pdfContainer.addEventListener('mousemove', (e) => {
718
+ if (!isMouseDown) return;
719
+ e.preventDefault();
720
+
721
+ const x = e.pageX - pdfContainer.offsetLeft;
722
+ const y = e.pageY - pdfContainer.offsetTop;
723
+ const walkX = (x - startX) * 1.5; // Multiply for faster scrolling
724
+ const walkY = (y - startY) * 1.5;
725
+
726
+ pdfContainer.scrollLeft = scrollLeft - walkX;
727
+ pdfContainer.scrollTop = scrollTop - walkY;
728
+ });
729
+ }
730
+
731
+ // Initialize on page load
732
+ init();
733
+ </script>
734
+ </body>
735
+ </html>
loc_ocr/00001.xml ADDED
The diff for this file is too large to render. See raw diff
 
pdfs/00001.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:73edc82b6c773bef7b9759de0b96c4d7f7283534b19ea8979d38bf727ccbf2e7
3
+ size 755765
pdfs/00547.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a48221cfc415b70f74d071dfa18988483295099c0d420dd03a6379b969331e39
3
+ size 199392
pdfs/00923.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e7146759a27b60f0b297b23c744da68ce6075031ecba2df6a7995345388ac19f
3
+ size 237203
pdfs/01063.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dfb9199879f777ab5e87c66efe4ce07e31573c91905987181749a5db0d3100d3
3
+ size 261664
pdfs/01405.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bc7b7857a593c7a9817181c9c5344334bd6b99539468c2736f0d8c30bf52eb90
3
+ size 305304
pdfs/01608.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6cd86a0341dcc61e809a8ff6589dab8555ce70d3de59e0727c8aed2656d92ab3
3
+ size 311837
pdfs/01689.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:be30e51aa33b8cec19d02d836a144f45b4574dc05e962328fa6ea02d9f205799
3
+ size 320123
pdfs/01718.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68770ee45704f7e14c110b6122b75be9b126a7d162b03ffeb771d4a5ad909149
3
+ size 308300
pdfs/01786.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5542b99dff7466796e4f1303f4618c5e72dfcc6402a7d918e049b97f1a66a648
3
+ size 305909
pdfs/02440.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:85324033a843db80febe42ee429e2e5bd10f274f14fd918787d8cd17cc9fc5d5
3
+ size 298978
pdfs/02999.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:181e6a25941c5ca517078629900362d7cd687f15129878ad126a0bf6cdb37b80
3
+ size 442692
pdfs/03141.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:de48d073e276fb3029de3b94b5bcf9301d97fe999d5085d1ce27fa7ca50cd0cb
3
+ size 1131187
pdfs/03528.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3000ce0b153b21b00f24e756074dd371cb9b70f4f89851537efb7f11ffbbc927
3
+ size 1079706
pdfs/04184.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:837918fddb438c6b9929b43dd2908bea638546e6456f518bc798983e9447ba0d
3
+ size 1717251
pdfs/05082.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b76abd62e3c6aec2137cca4980e66dab3e1a04f0552e236cb40cc3aff923e4e5
3
+ size 1290013
pdfs/06035.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:21cb9983aa6e1831af1c7610e7a2466392ad0adec6cf79e53ea35838d1d1e4f9
3
+ size 2949080
pdfs/06145.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cc4a2a5cf0434b18cad2894755acd21be598f4a27924f74ad9deb4f437eba960
3
+ size 3149548
pdfs/07555.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a57653f087e3c8092d77f89ee5a3c5d1aaca2f73bda4a341a7182b0f10f2fe03
3
+ size 1874688
pdfs/07787.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c0bfd5400294c8c8c4a4c8b1ea8e11547d91add09866260e5f0e2fa449e97646
3
+ size 1667089
pdfs/09426.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f01af6d3529c7d465bac50bc6f615a816166c26702fd140df3c5109cd2c37800
3
+ size 3234593
pdfs/10261.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1af303780c427a43706db79f795dc3e9f7408a8a339c191237f4dfb57adf4248
3
+ size 4371190
pdfs/10875.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e1df1abc03b86b980d4e600e182ebf05c25f38a39cd4ff8c25c899c5130d8f6f
3
+ size 2062854
pdfs/11901.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9cade9ddeb44a9f158b3d39a95f9ce8e21bd077fd0ba341d39faa41c92290bce
3
+ size 2083303
pdfs/12099.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0186a74e10e3d7229ec6572d4410ad51efa76f7fd1801ac1b31de076b1233e0c
3
+ size 1965570
pdfs/12747.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:356cde18ff536b5de1290752ac2d897e2ab3d44a0e3c9d56303965d6cbf5a2b7
3
+ size 5162920
pdfs/12827.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:eec4dcb7c204ef52bbeb7fe281f1f772b6377e624357adaa27d4adbccfe0e5ca
3
+ size 4786829
pdfs/12964.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f6b75a66ab81568055e410cd1e15834e1e2b5c706865f15d0c9a89baeea180c8
3
+ size 5313714
pdfs/13089.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0e58e9b3638991ae5be837fe944637c6dbbe276f899d6393d3f9250648b61c30
3
+ size 5218234
pdfs/14825.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aba674ffa5124bece8b86d4f145c588d7b6c4b24aff8f65a191cf3e438aa61d6
3
+ size 5778052
pdfs/15210.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d521c7da2bbf7b1c867b6652d77545e879e7cadf68a76ee1acc2df395546136a
3
+ size 5727644
pdfs/16124.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:880569b7b5b414dc2a595782cffcb3751f372a10dbb320cdea52e67a6f4ba2da
3
+ size 2514563
pdfs/16999.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:08a37f3a27981236348de0eb43350d93fa00876c20204e9d846c7a5daeece8e8
3
+ size 3584503
pdfs/17628.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8501e04367d8387c74122647b8cfb1c16649b15303870147d0fd5bf0ecc2e227
3
+ size 3250142
pdfs/17885.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9b8bf4722711d4d0a8cc077284f28c6eb802a4881cab7e5cadd944556d8e9a1
3
+ size 3718836
pdfs/19652.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c6d0978c509df5a101c0fa6ead2be5ec9fcc0db49f51e35469fbb3fa7e82a57
3
+ size 3319886
pdfs/19818.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f86fa4af82dcd3e98bb98f97ec0155fdf0e37240b53a51c70ab1d559d8c7fc8d
3
+ size 2759018
pdfs/19915.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:83a29266dda5027cdda85c2f978fd3f3c31a54307245c59391d5f78ef30f1b90
3
+ size 2251538
pdfs/22785.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7bd027e392232763f19f3753d8231c10e27b200658a5e0f76721370693a12667
3
+ size 2026182
pdfs/22939.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:740238d83b2c6bf71514d218f9d363bf82112e13ec86c4e3066111f8db11b067
3
+ size 1969624
pdfs/23092.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe6a3f6d7680439df7f894b7aaa1a8b07fbbd92ff939bfef343edac8638cbd66
3
+ size 1977563
pdfs/24047.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a0ebe110729e8c98cc937d2bf2243b006f9dfac37e8113a30f4f67626ce7f3b
3
+ size 1952343
pdfs/25152.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5a61e85f7ca2324fcd28a7d1e21c488a955bc898c6742ac068aa981d4d67544e
3
+ size 4972673
pdfs/25532.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dc28f3cafda7483f230bc2801569480fca4fe582ac6309a554194aef2f5f1af2
3
+ size 4092882
pdfs/25615.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fbaf54f85663b010f612e8e3e0e29ef20c15c3d12c89922433742952ce242359
3
+ size 3597146
pdfs/26136.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6d7c095806cbb055e1bbcc4ac5c1cc951a2d870ab96545990a086ccacdec80e5
3
+ size 2082937
pdfs/28406.pdf ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5e0dbd447a35377e365d3ea6f96ac8a189c159c512675015217865712856ecbe
3
+ size 2233416