natural-pdf 0.1.4__py3-none-any.whl → 0.1.6__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 (141) hide show
  1. docs/api/index.md +386 -0
  2. docs/assets/favicon.png +3 -0
  3. docs/assets/favicon.svg +3 -0
  4. docs/assets/javascripts/custom.js +17 -0
  5. docs/assets/logo.svg +3 -0
  6. docs/assets/sample-screen.png +0 -0
  7. docs/assets/social-preview.png +17 -0
  8. docs/assets/social-preview.svg +17 -0
  9. docs/assets/stylesheets/custom.css +65 -0
  10. docs/document-qa/index.ipynb +435 -0
  11. docs/document-qa/index.md +79 -0
  12. docs/element-selection/index.ipynb +915 -0
  13. docs/element-selection/index.md +229 -0
  14. docs/index.md +170 -0
  15. docs/installation/index.md +69 -0
  16. docs/interactive-widget/index.ipynb +962 -0
  17. docs/interactive-widget/index.md +12 -0
  18. docs/layout-analysis/index.ipynb +818 -0
  19. docs/layout-analysis/index.md +185 -0
  20. docs/ocr/index.md +209 -0
  21. docs/pdf-navigation/index.ipynb +314 -0
  22. docs/pdf-navigation/index.md +97 -0
  23. docs/regions/index.ipynb +816 -0
  24. docs/regions/index.md +294 -0
  25. docs/tables/index.ipynb +658 -0
  26. docs/tables/index.md +144 -0
  27. docs/text-analysis/index.ipynb +370 -0
  28. docs/text-analysis/index.md +105 -0
  29. docs/text-extraction/index.ipynb +1478 -0
  30. docs/text-extraction/index.md +292 -0
  31. docs/tutorials/01-loading-and-extraction.ipynb +1710 -0
  32. docs/tutorials/01-loading-and-extraction.md +95 -0
  33. docs/tutorials/02-finding-elements.ipynb +340 -0
  34. docs/tutorials/02-finding-elements.md +149 -0
  35. docs/tutorials/03-extracting-blocks.ipynb +147 -0
  36. docs/tutorials/03-extracting-blocks.md +48 -0
  37. docs/tutorials/04-table-extraction.ipynb +114 -0
  38. docs/tutorials/04-table-extraction.md +50 -0
  39. docs/tutorials/05-excluding-content.ipynb +270 -0
  40. docs/tutorials/05-excluding-content.md +109 -0
  41. docs/tutorials/06-document-qa.ipynb +332 -0
  42. docs/tutorials/06-document-qa.md +91 -0
  43. docs/tutorials/07-layout-analysis.ipynb +288 -0
  44. docs/tutorials/07-layout-analysis.md +66 -0
  45. docs/tutorials/07-working-with-regions.ipynb +413 -0
  46. docs/tutorials/07-working-with-regions.md +151 -0
  47. docs/tutorials/08-spatial-navigation.ipynb +508 -0
  48. docs/tutorials/08-spatial-navigation.md +190 -0
  49. docs/tutorials/09-section-extraction.ipynb +2434 -0
  50. docs/tutorials/09-section-extraction.md +256 -0
  51. docs/tutorials/10-form-field-extraction.ipynb +512 -0
  52. docs/tutorials/10-form-field-extraction.md +201 -0
  53. docs/tutorials/11-enhanced-table-processing.ipynb +54 -0
  54. docs/tutorials/11-enhanced-table-processing.md +9 -0
  55. docs/tutorials/12-ocr-integration.ipynb +604 -0
  56. docs/tutorials/12-ocr-integration.md +175 -0
  57. docs/tutorials/13-semantic-search.ipynb +1328 -0
  58. docs/tutorials/13-semantic-search.md +77 -0
  59. docs/visual-debugging/index.ipynb +2970 -0
  60. docs/visual-debugging/index.md +157 -0
  61. docs/visual-debugging/region.png +0 -0
  62. natural_pdf/__init__.py +50 -33
  63. natural_pdf/analyzers/__init__.py +2 -1
  64. natural_pdf/analyzers/layout/base.py +32 -24
  65. natural_pdf/analyzers/layout/docling.py +131 -72
  66. natural_pdf/analyzers/layout/gemini.py +264 -0
  67. natural_pdf/analyzers/layout/layout_analyzer.py +156 -113
  68. natural_pdf/analyzers/layout/layout_manager.py +125 -58
  69. natural_pdf/analyzers/layout/layout_options.py +43 -17
  70. natural_pdf/analyzers/layout/paddle.py +152 -95
  71. natural_pdf/analyzers/layout/surya.py +164 -92
  72. natural_pdf/analyzers/layout/tatr.py +149 -84
  73. natural_pdf/analyzers/layout/yolo.py +89 -45
  74. natural_pdf/analyzers/text_options.py +22 -15
  75. natural_pdf/analyzers/text_structure.py +131 -85
  76. natural_pdf/analyzers/utils.py +30 -23
  77. natural_pdf/collections/pdf_collection.py +146 -97
  78. natural_pdf/core/__init__.py +1 -1
  79. natural_pdf/core/element_manager.py +419 -337
  80. natural_pdf/core/highlighting_service.py +268 -196
  81. natural_pdf/core/page.py +1044 -521
  82. natural_pdf/core/pdf.py +516 -313
  83. natural_pdf/elements/__init__.py +1 -1
  84. natural_pdf/elements/base.py +307 -225
  85. natural_pdf/elements/collections.py +805 -543
  86. natural_pdf/elements/line.py +39 -36
  87. natural_pdf/elements/rect.py +32 -30
  88. natural_pdf/elements/region.py +889 -879
  89. natural_pdf/elements/text.py +127 -99
  90. natural_pdf/exporters/__init__.py +0 -1
  91. natural_pdf/exporters/searchable_pdf.py +261 -102
  92. natural_pdf/ocr/__init__.py +57 -35
  93. natural_pdf/ocr/engine.py +150 -46
  94. natural_pdf/ocr/engine_easyocr.py +146 -150
  95. natural_pdf/ocr/engine_paddle.py +118 -175
  96. natural_pdf/ocr/engine_surya.py +78 -141
  97. natural_pdf/ocr/ocr_factory.py +114 -0
  98. natural_pdf/ocr/ocr_manager.py +122 -124
  99. natural_pdf/ocr/ocr_options.py +16 -20
  100. natural_pdf/ocr/utils.py +98 -0
  101. natural_pdf/qa/__init__.py +1 -1
  102. natural_pdf/qa/document_qa.py +119 -111
  103. natural_pdf/search/__init__.py +37 -31
  104. natural_pdf/search/haystack_search_service.py +312 -189
  105. natural_pdf/search/haystack_utils.py +186 -122
  106. natural_pdf/search/search_options.py +25 -14
  107. natural_pdf/search/search_service_protocol.py +12 -6
  108. natural_pdf/search/searchable_mixin.py +261 -176
  109. natural_pdf/selectors/__init__.py +2 -1
  110. natural_pdf/selectors/parser.py +159 -316
  111. natural_pdf/templates/__init__.py +1 -1
  112. natural_pdf/templates/spa/css/style.css +334 -0
  113. natural_pdf/templates/spa/index.html +31 -0
  114. natural_pdf/templates/spa/js/app.js +472 -0
  115. natural_pdf/templates/spa/words.txt +235976 -0
  116. natural_pdf/utils/debug.py +32 -0
  117. natural_pdf/utils/highlighting.py +8 -2
  118. natural_pdf/utils/identifiers.py +29 -0
  119. natural_pdf/utils/packaging.py +418 -0
  120. natural_pdf/utils/reading_order.py +65 -63
  121. natural_pdf/utils/text_extraction.py +195 -0
  122. natural_pdf/utils/visualization.py +70 -61
  123. natural_pdf/widgets/__init__.py +2 -3
  124. natural_pdf/widgets/viewer.py +749 -718
  125. {natural_pdf-0.1.4.dist-info → natural_pdf-0.1.6.dist-info}/METADATA +53 -17
  126. natural_pdf-0.1.6.dist-info/RECORD +141 -0
  127. {natural_pdf-0.1.4.dist-info → natural_pdf-0.1.6.dist-info}/WHEEL +1 -1
  128. natural_pdf-0.1.6.dist-info/top_level.txt +4 -0
  129. notebooks/Examples.ipynb +1293 -0
  130. pdfs/.gitkeep +0 -0
  131. pdfs/01-practice.pdf +543 -0
  132. pdfs/0500000US42001.pdf +0 -0
  133. pdfs/0500000US42007.pdf +0 -0
  134. pdfs/2014 Statistics.pdf +0 -0
  135. pdfs/2019 Statistics.pdf +0 -0
  136. pdfs/Atlanta_Public_Schools_GA_sample.pdf +0 -0
  137. pdfs/needs-ocr.pdf +0 -0
  138. natural_pdf/templates/ocr_debug.html +0 -517
  139. natural_pdf-0.1.4.dist-info/RECORD +0 -61
  140. natural_pdf-0.1.4.dist-info/top_level.txt +0 -1
  141. {natural_pdf-0.1.4.dist-info → natural_pdf-0.1.6.dist-info}/licenses/LICENSE +0 -0
docs/interactive-widget/index.md ADDED
@@ -0,0 +1,12 @@
1
+ # Interactive widget
2
+
3
+ This is the best possible way, in all of history, to explore a PDF.
4
+
5
+ ```python
6
+ from natural_pdf import PDF
7
+
8
+ pdf = PDF("https://github.com/jsoma/natural-pdf/raw/refs/heads/main/pdfs/01-practice.pdf")
9
+ page = pdf.pages[0]
10
+
11
+ page.viewer()
12
+ ```