natural-pdf 0.1.7__py3-none-any.whl → 0.1.9__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (134) hide show
  1. natural_pdf/__init__.py +3 -0
  2. natural_pdf/analyzers/layout/base.py +1 -5
  3. natural_pdf/analyzers/layout/gemini.py +61 -51
  4. natural_pdf/analyzers/layout/layout_analyzer.py +40 -11
  5. natural_pdf/analyzers/layout/layout_manager.py +26 -84
  6. natural_pdf/analyzers/layout/layout_options.py +7 -0
  7. natural_pdf/analyzers/layout/pdfplumber_table_finder.py +142 -0
  8. natural_pdf/analyzers/layout/surya.py +46 -123
  9. natural_pdf/analyzers/layout/tatr.py +51 -4
  10. natural_pdf/analyzers/text_structure.py +3 -5
  11. natural_pdf/analyzers/utils.py +3 -3
  12. natural_pdf/classification/manager.py +422 -0
  13. natural_pdf/classification/mixin.py +163 -0
  14. natural_pdf/classification/results.py +80 -0
  15. natural_pdf/collections/mixins.py +111 -0
  16. natural_pdf/collections/pdf_collection.py +434 -15
  17. natural_pdf/core/element_manager.py +83 -0
  18. natural_pdf/core/highlighting_service.py +13 -22
  19. natural_pdf/core/page.py +578 -93
  20. natural_pdf/core/pdf.py +912 -460
  21. natural_pdf/elements/base.py +134 -40
  22. natural_pdf/elements/collections.py +712 -109
  23. natural_pdf/elements/region.py +722 -69
  24. natural_pdf/elements/text.py +4 -1
  25. natural_pdf/export/mixin.py +137 -0
  26. natural_pdf/exporters/base.py +3 -3
  27. natural_pdf/exporters/paddleocr.py +5 -4
  28. natural_pdf/extraction/manager.py +135 -0
  29. natural_pdf/extraction/mixin.py +279 -0
  30. natural_pdf/extraction/result.py +23 -0
  31. natural_pdf/ocr/__init__.py +5 -5
  32. natural_pdf/ocr/engine_doctr.py +346 -0
  33. natural_pdf/ocr/engine_easyocr.py +6 -3
  34. natural_pdf/ocr/ocr_factory.py +24 -4
  35. natural_pdf/ocr/ocr_manager.py +122 -26
  36. natural_pdf/ocr/ocr_options.py +94 -11
  37. natural_pdf/ocr/utils.py +19 -6
  38. natural_pdf/qa/document_qa.py +0 -4
  39. natural_pdf/search/__init__.py +20 -34
  40. natural_pdf/search/haystack_search_service.py +309 -265
  41. natural_pdf/search/haystack_utils.py +99 -75
  42. natural_pdf/search/search_service_protocol.py +11 -12
  43. natural_pdf/selectors/parser.py +431 -230
  44. natural_pdf/utils/debug.py +3 -3
  45. natural_pdf/utils/identifiers.py +1 -1
  46. natural_pdf/utils/locks.py +8 -0
  47. natural_pdf/utils/packaging.py +8 -6
  48. natural_pdf/utils/text_extraction.py +60 -1
  49. natural_pdf/utils/tqdm_utils.py +51 -0
  50. natural_pdf/utils/visualization.py +18 -0
  51. natural_pdf/widgets/viewer.py +4 -25
  52. {natural_pdf-0.1.7.dist-info → natural_pdf-0.1.9.dist-info}/METADATA +17 -3
  53. natural_pdf-0.1.9.dist-info/RECORD +80 -0
  54. {natural_pdf-0.1.7.dist-info → natural_pdf-0.1.9.dist-info}/WHEEL +1 -1
  55. {natural_pdf-0.1.7.dist-info → natural_pdf-0.1.9.dist-info}/top_level.txt +0 -2
  56. docs/api/index.md +0 -386
  57. docs/assets/favicon.png +0 -3
  58. docs/assets/favicon.svg +0 -3
  59. docs/assets/javascripts/custom.js +0 -17
  60. docs/assets/logo.svg +0 -3
  61. docs/assets/sample-screen.png +0 -0
  62. docs/assets/social-preview.png +0 -17
  63. docs/assets/social-preview.svg +0 -17
  64. docs/assets/stylesheets/custom.css +0 -65
  65. docs/document-qa/index.ipynb +0 -435
  66. docs/document-qa/index.md +0 -79
  67. docs/element-selection/index.ipynb +0 -915
  68. docs/element-selection/index.md +0 -229
  69. docs/finetuning/index.md +0 -176
  70. docs/index.md +0 -170
  71. docs/installation/index.md +0 -69
  72. docs/interactive-widget/index.ipynb +0 -962
  73. docs/interactive-widget/index.md +0 -12
  74. docs/layout-analysis/index.ipynb +0 -818
  75. docs/layout-analysis/index.md +0 -185
  76. docs/ocr/index.md +0 -209
  77. docs/pdf-navigation/index.ipynb +0 -314
  78. docs/pdf-navigation/index.md +0 -97
  79. docs/regions/index.ipynb +0 -816
  80. docs/regions/index.md +0 -294
  81. docs/tables/index.ipynb +0 -658
  82. docs/tables/index.md +0 -144
  83. docs/text-analysis/index.ipynb +0 -370
  84. docs/text-analysis/index.md +0 -105
  85. docs/text-extraction/index.ipynb +0 -1478
  86. docs/text-extraction/index.md +0 -292
  87. docs/tutorials/01-loading-and-extraction.ipynb +0 -194
  88. docs/tutorials/01-loading-and-extraction.md +0 -95
  89. docs/tutorials/02-finding-elements.ipynb +0 -340
  90. docs/tutorials/02-finding-elements.md +0 -149
  91. docs/tutorials/03-extracting-blocks.ipynb +0 -147
  92. docs/tutorials/03-extracting-blocks.md +0 -48
  93. docs/tutorials/04-table-extraction.ipynb +0 -114
  94. docs/tutorials/04-table-extraction.md +0 -50
  95. docs/tutorials/05-excluding-content.ipynb +0 -270
  96. docs/tutorials/05-excluding-content.md +0 -109
  97. docs/tutorials/06-document-qa.ipynb +0 -332
  98. docs/tutorials/06-document-qa.md +0 -91
  99. docs/tutorials/07-layout-analysis.ipynb +0 -288
  100. docs/tutorials/07-layout-analysis.md +0 -66
  101. docs/tutorials/07-working-with-regions.ipynb +0 -413
  102. docs/tutorials/07-working-with-regions.md +0 -151
  103. docs/tutorials/08-spatial-navigation.ipynb +0 -508
  104. docs/tutorials/08-spatial-navigation.md +0 -190
  105. docs/tutorials/09-section-extraction.ipynb +0 -2434
  106. docs/tutorials/09-section-extraction.md +0 -256
  107. docs/tutorials/10-form-field-extraction.ipynb +0 -512
  108. docs/tutorials/10-form-field-extraction.md +0 -201
  109. docs/tutorials/11-enhanced-table-processing.ipynb +0 -54
  110. docs/tutorials/11-enhanced-table-processing.md +0 -9
  111. docs/tutorials/12-ocr-integration.ipynb +0 -604
  112. docs/tutorials/12-ocr-integration.md +0 -175
  113. docs/tutorials/13-semantic-search.ipynb +0 -1328
  114. docs/tutorials/13-semantic-search.md +0 -77
  115. docs/visual-debugging/index.ipynb +0 -2970
  116. docs/visual-debugging/index.md +0 -157
  117. docs/visual-debugging/region.png +0 -0
  118. natural_pdf/templates/finetune/fine_tune_paddleocr.md +0 -415
  119. natural_pdf/templates/spa/css/style.css +0 -334
  120. natural_pdf/templates/spa/index.html +0 -31
  121. natural_pdf/templates/spa/js/app.js +0 -472
  122. natural_pdf/templates/spa/words.txt +0 -235976
  123. natural_pdf/widgets/frontend/viewer.js +0 -88
  124. natural_pdf-0.1.7.dist-info/RECORD +0 -145
  125. notebooks/Examples.ipynb +0 -1293
  126. pdfs/.gitkeep +0 -0
  127. pdfs/01-practice.pdf +0 -543
  128. pdfs/0500000US42001.pdf +0 -0
  129. pdfs/0500000US42007.pdf +0 -0
  130. pdfs/2014 Statistics.pdf +0 -0
  131. pdfs/2019 Statistics.pdf +0 -0
  132. pdfs/Atlanta_Public_Schools_GA_sample.pdf +0 -0
  133. pdfs/needs-ocr.pdf +0 -0
  134. {natural_pdf-0.1.7.dist-info → natural_pdf-0.1.9.dist-info}/licenses/LICENSE +0 -0
@@ -1,334 +0,0 @@
1
- /* Styles adapted from ocr_debug.css */
2
- body {
3
- font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, Helvetica, Arial, sans-serif;
4
- line-height: 1.6;
5
- color: #333;
6
- margin: 0;
7
- display: flex;
8
- flex-direction: column;
9
- min-height: 100vh;
10
- }
11
-
12
- header {
13
- background-color: #f8f9fa;
14
- padding: 10px 20px;
15
- border-bottom: 1px solid #ddd;
16
- }
17
-
18
- main {
19
- flex-grow: 1;
20
- max-width: 1400px; /* Wider for SPA layout */
21
- width: 100%;
22
- margin: 20px auto;
23
- padding: 0 20px; /* Add horizontal padding */
24
- box-sizing: border-box;
25
- }
26
-
27
- footer {
28
- background-color: #f8f9fa;
29
- padding: 10px 20px;
30
- border-top: 1px solid #ddd;
31
- text-align: center;
32
- font-size: 0.9em;
33
- color: #666;
34
- margin-top: auto;
35
- }
36
-
37
- h1, h2 {
38
- color: #2c3e50;
39
- margin-top: 0;
40
- }
41
-
42
- .loading-message,
43
- .initial-message {
44
- text-align: center;
45
- padding: 40px;
46
- font-size: 1.2em;
47
- color: #666;
48
- }
49
-
50
- .task-loader {
51
- margin-bottom: 20px;
52
- padding: 15px;
53
- border: 1px dashed #ccc;
54
- background-color: #f0f0f0;
55
- text-align: center;
56
- }
57
-
58
- .page-section {
59
- margin-bottom: 30px;
60
- border: 1px solid #ddd;
61
- border-radius: 4px;
62
- padding: 15px;
63
- background: #f8f9fa;
64
- }
65
-
66
- .page-title {
67
- display: flex;
68
- justify-content: space-between;
69
- align-items: center;
70
- margin-bottom: 15px;
71
- }
72
-
73
- .page-controls {
74
- display: flex;
75
- gap: 10px;
76
- align-items: center;
77
- }
78
-
79
- .controls-container {
80
- margin-bottom: 20px;
81
- display: flex;
82
- justify-content: space-between;
83
- flex-wrap: wrap;
84
- gap: 15px; /* Increased gap */
85
- padding: 15px;
86
- background: #eee;
87
- border-radius: 4px;
88
- position: sticky; /* Make controls sticky */
89
- top: 0;
90
- z-index: 100;
91
- border-bottom: 1px solid #ddd;
92
- }
93
-
94
- .filter-control {
95
- display: flex;
96
- align-items: center;
97
- gap: 8px;
98
- }
99
-
100
- input, select, button {
101
- padding: 8px 12px; /* Slightly larger padding */
102
- border: 1px solid #ccc;
103
- border-radius: 4px;
104
- font-size: 14px;
105
- vertical-align: middle; /* Align form elements */
106
- }
107
-
108
- input[type="range"] {
109
- padding: 0; /* Remove padding for range */
110
- }
111
-
112
- button {
113
- background: #4b6bfb;
114
- color: white;
115
- cursor: pointer;
116
- border: none; /* Remove border */
117
- }
118
-
119
- button:hover {
120
- background: #3b5de7;
121
- }
122
-
123
- button.secondary {
124
- background: #6c757d;
125
- }
126
- button.secondary:hover {
127
- background: #5a6268;
128
- }
129
-
130
- /* --- NEW FLEXBOX BASED LAYOUT --- */
131
- .region-list {
132
- /* Container for all region items */
133
- margin-top: 15px;
134
- border: 1px solid #ddd; /* Add border around the list */
135
- border-radius: 4px;
136
- overflow: hidden; /* Ensures border radius clips children */
137
- }
138
-
139
- .region-item {
140
- display: flex;
141
- align-items: stretch; /* Make cells same height */
142
- border-bottom: 1px solid #ddd; /* Separator line */
143
- background-color: #fff;
144
- }
145
- .region-item:last-child {
146
- border-bottom: none; /* Remove border for the last item */
147
- }
148
-
149
- .confidence-cell {
150
- width: 80px; /* Fixed width for confidence */
151
- padding: 10px;
152
- text-align: center;
153
- white-space: nowrap;
154
- border-right: 1px solid #ddd; /* Separator line */
155
- display: flex; /* Center content vertically */
156
- align-items: center;
157
- justify-content: center;
158
- flex-shrink: 0; /* Prevent shrinking */
159
- }
160
-
161
- /* Confidence background colors (moved from .confidence) */
162
- .confidence-cell[data-level="high"] {
163
- background-color: rgba(40, 167, 69, 0.1);
164
- }
165
- .confidence-cell[data-level="medium"] {
166
- background-color: rgba(255, 193, 7, 0.1);
167
- }
168
- .confidence-cell[data-level="low"] {
169
- background-color: rgba(220, 53, 69, 0.1);
170
- }
171
-
172
- .region-content-cell {
173
- flex-grow: 1; /* Take remaining width */
174
- padding: 10px;
175
- display: flex; /* Use flex to control children */
176
- flex-direction: column; /* Stack image and text */
177
- /* background-color: #fafafa; /* Optional: Slight background difference */
178
- }
179
-
180
- .image-clip {
181
- /* Styles for the image snippet div */
182
- position: relative;
183
- overflow: hidden;
184
- background-repeat: no-repeat;
185
- border-radius: 3px;
186
- box-shadow: 0 1px 3px rgba(0,0,0,0.2);
187
- margin-bottom: 10px; /* Space below image */
188
- /* Use inline style for width/height/background */
189
- max-width: 350px; /* Max width within the cell */
190
- max-height: 250px; /* Max height */
191
- transform-origin: top left;
192
- border: 1px solid #ccc;
193
- background-color: #f8f8f8; /* Placeholder background */
194
- align-self: flex-start; /* Align to start if cell is wider */
195
- /* display: inline-block; /* Remove this if using flex parent */
196
- }
197
-
198
- .editing-content {
199
- font-size: 18px;
200
- text-align: center;
201
- }
202
-
203
- .text-content-input {
204
- /* Styles for the textarea */
205
- text-align: center;
206
- width: 100%; /* Full width of the cell */
207
- font-family: monospace;
208
- padding: 8px;
209
- font-size: 16px;
210
- line-height: 1.4;
211
- white-space: pre-wrap;
212
- word-break: break-all;
213
- border: 1px solid #ddd;
214
- border-radius: 4px;
215
- resize: vertical;
216
- min-height: 50px; /* Slightly taller */
217
- box-sizing: border-box;
218
- }
219
-
220
- .text-content-input.modified {
221
- border-color: #4b6bfb;
222
- background-color: rgba(75, 107, 251, 0.05);
223
- }
224
-
225
- .text-content-input:focus {
226
- border-color: #4b6bfb;
227
- outline: none;
228
- box-shadow: 0 0 0 2px rgba(75, 107, 251, 0.25);
229
- }
230
- /* --- END NEW FLEXBOX BASED LAYOUT --- */
231
-
232
-
233
- /* --- REMOVED OLD TABLE STYLES --- */
234
- /*
235
- .region-table { ... }
236
- .region-table th, .region-table td { ... }
237
- .region-table th { ... }
238
- .region-image { ... }
239
- .confidence { ... }
240
- .text-content { ... }
241
- */
242
- /* --- END REMOVED OLD TABLE STYLES --- */
243
-
244
-
245
- .hidden {
246
- display: none !important;
247
- }
248
-
249
- .toggle-btn {
250
- background: #eee;
251
- color: #333;
252
- border: 1px solid #ccc;
253
- padding: 3px 8px;
254
- border-radius: 3px;
255
- cursor: pointer;
256
- font-size: 12px;
257
- }
258
-
259
- .toggle-btn:hover {
260
- background: #ddd;
261
- }
262
-
263
- .export-btn {
264
- margin-left: auto;
265
- }
266
-
267
- .page-image {
268
- max-width: 100%;
269
- height: auto;
270
- margin-bottom: 15px;
271
- border: 1px solid #ddd;
272
- display: none; /* Hidden by default */
273
- }
274
-
275
- .page-image.show {
276
- display: block;
277
- }
278
-
279
- /* Responsive adjustments */
280
- @media (max-width: 800px) {
281
- /* REMOVED table-specific responsive styles */
282
- /*
283
- .region-table thead { ... }
284
- .region-table, .region-table tbody, .region-table tr, .region-table td { ... }
285
- .region-table tr { ... }
286
- .region-table td { ... }
287
- .region-table td::before { ... }
288
- .region-table td:last-child { ... }
289
- .region-image, .text-content { ... }
290
- .confidence { ... }
291
- */
292
-
293
- /* NEW Flexbox Responsive Styles */
294
- .region-item {
295
- flex-direction: column; /* Stack columns vertically on mobile */
296
- align-items: stretch; /* Make items full width */
297
- }
298
-
299
- .confidence-cell {
300
- width: auto; /* Allow confidence to take full width */
301
- border-right: none; /* Remove right border */
302
- border-bottom: 1px solid #ddd; /* Add bottom border */
303
- justify-content: flex-start; /* Align text left */
304
- padding: 5px 10px; /* Adjust padding */
305
- }
306
- /* Optional: Add label back using ::before */
307
- .confidence-cell::before {
308
- content: "Confidence: ";
309
- font-weight: bold;
310
- margin-right: 5px;
311
- }
312
-
313
- .region-content-cell {
314
- padding: 10px; /* Reset padding */
315
- }
316
-
317
- .image-clip {
318
- max-width: 100%; /* Allow image full width */
319
- align-self: center; /* Center image */
320
- /* Remove fixed max-height? */
321
- /* max-height: none; */
322
- }
323
-
324
- .controls-container {
325
- flex-direction: column;
326
- align-items: stretch;
327
- position: static; /* Unstick controls */
328
- }
329
-
330
- .export-btn {
331
- margin-left: 0;
332
- margin-top: 10px;
333
- }
334
- }
@@ -1,31 +0,0 @@
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>OCR Correction Tool</title>
7
- <link rel="stylesheet" href="css/style.css">
8
- </head>
9
- <body>
10
- <header>
11
- <h1>OCR Correction Tool</h1>
12
- </header>
13
-
14
- <main id="app">
15
- <p>Loading application...</p>
16
- </main>
17
-
18
- <footer>
19
- <p>Generated by natural-pdf</p>
20
- </footer>
21
-
22
- <script src="js/lib/react.development.js"></script>
23
- <script src="js/lib/react-dom.development.js"></script>
24
- <script src="js/lib/babel.min.js"></script>
25
- <script src="js/lib/jszip.min.js"></script>
26
- <script src="js/lib/FileSaver.min.js"></script>
27
-
28
- <script type="text/babel" src="js/app.js"></script>
29
-
30
- </body>
31
- </html>