natural-pdf 0.1.33__py3-none-any.whl → 0.1.35__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 (77) hide show
  1. natural_pdf/analyzers/__init__.py +2 -2
  2. natural_pdf/analyzers/guides.py +751 -607
  3. natural_pdf/analyzers/layout/base.py +53 -6
  4. natural_pdf/analyzers/layout/layout_analyzer.py +3 -1
  5. natural_pdf/analyzers/layout/layout_manager.py +18 -14
  6. natural_pdf/analyzers/layout/layout_options.py +1 -0
  7. natural_pdf/analyzers/layout/paddle.py +102 -64
  8. natural_pdf/analyzers/layout/table_structure_utils.py +3 -1
  9. natural_pdf/analyzers/layout/yolo.py +2 -6
  10. natural_pdf/analyzers/shape_detection_mixin.py +15 -6
  11. natural_pdf/classification/manager.py +92 -77
  12. natural_pdf/classification/mixin.py +49 -5
  13. natural_pdf/classification/results.py +1 -1
  14. natural_pdf/cli.py +7 -3
  15. natural_pdf/collections/pdf_collection.py +96 -101
  16. natural_pdf/core/element_manager.py +131 -45
  17. natural_pdf/core/highlighting_service.py +5 -6
  18. natural_pdf/core/page.py +120 -23
  19. natural_pdf/core/pdf.py +477 -75
  20. natural_pdf/describe/__init__.py +18 -12
  21. natural_pdf/describe/base.py +179 -172
  22. natural_pdf/describe/elements.py +155 -155
  23. natural_pdf/describe/mixin.py +27 -19
  24. natural_pdf/describe/summary.py +44 -55
  25. natural_pdf/elements/base.py +134 -18
  26. natural_pdf/elements/collections.py +90 -18
  27. natural_pdf/elements/image.py +2 -1
  28. natural_pdf/elements/line.py +0 -31
  29. natural_pdf/elements/rect.py +0 -14
  30. natural_pdf/elements/region.py +222 -108
  31. natural_pdf/elements/text.py +18 -12
  32. natural_pdf/exporters/__init__.py +4 -1
  33. natural_pdf/exporters/original_pdf.py +12 -4
  34. natural_pdf/extraction/mixin.py +66 -10
  35. natural_pdf/extraction/result.py +1 -1
  36. natural_pdf/flows/flow.py +63 -4
  37. natural_pdf/flows/region.py +4 -4
  38. natural_pdf/ocr/engine.py +83 -2
  39. natural_pdf/ocr/engine_paddle.py +5 -5
  40. natural_pdf/ocr/ocr_factory.py +2 -1
  41. natural_pdf/ocr/ocr_manager.py +24 -13
  42. natural_pdf/ocr/ocr_options.py +3 -10
  43. natural_pdf/qa/document_qa.py +21 -8
  44. natural_pdf/qa/qa_result.py +3 -7
  45. natural_pdf/search/__init__.py +3 -2
  46. natural_pdf/search/lancedb_search_service.py +5 -6
  47. natural_pdf/search/numpy_search_service.py +5 -2
  48. natural_pdf/selectors/parser.py +51 -6
  49. natural_pdf/tables/__init__.py +2 -2
  50. natural_pdf/tables/result.py +7 -6
  51. natural_pdf/utils/bidi_mirror.py +2 -1
  52. natural_pdf/utils/reading_order.py +3 -2
  53. natural_pdf/utils/visualization.py +3 -3
  54. natural_pdf/widgets/viewer.py +0 -1
  55. {natural_pdf-0.1.33.dist-info → natural_pdf-0.1.35.dist-info}/METADATA +1 -1
  56. natural_pdf-0.1.35.dist-info/RECORD +121 -0
  57. optimization/memory_comparison.py +73 -58
  58. optimization/pdf_analyzer.py +141 -96
  59. optimization/performance_analysis.py +111 -110
  60. optimization/test_cleanup_methods.py +47 -36
  61. optimization/test_memory_fix.py +40 -39
  62. tools/bad_pdf_eval/__init__.py +0 -1
  63. tools/bad_pdf_eval/analyser.py +35 -18
  64. tools/bad_pdf_eval/collate_summaries.py +22 -18
  65. tools/bad_pdf_eval/compile_attempts_markdown.py +127 -0
  66. tools/bad_pdf_eval/eval_suite.py +21 -9
  67. tools/bad_pdf_eval/evaluate_quality.py +198 -0
  68. tools/bad_pdf_eval/export_enrichment_csv.py +12 -8
  69. tools/bad_pdf_eval/llm_enrich.py +71 -39
  70. tools/bad_pdf_eval/llm_enrich_with_retry.py +289 -0
  71. tools/bad_pdf_eval/reporter.py +1 -1
  72. tools/bad_pdf_eval/utils.py +7 -4
  73. natural_pdf-0.1.33.dist-info/RECORD +0 -118
  74. {natural_pdf-0.1.33.dist-info → natural_pdf-0.1.35.dist-info}/WHEEL +0 -0
  75. {natural_pdf-0.1.33.dist-info → natural_pdf-0.1.35.dist-info}/entry_points.txt +0 -0
  76. {natural_pdf-0.1.33.dist-info → natural_pdf-0.1.35.dist-info}/licenses/LICENSE +0 -0
  77. {natural_pdf-0.1.33.dist-info → natural_pdf-0.1.35.dist-info}/top_level.txt +0 -0
natural_pdf/describe/__init__.py CHANGED
@@ -4,18 +4,24 @@ Describe functionality for natural-pdf.
4
4
  Provides summary and inspection methods for pages, collections, and regions.
5
5
  """
6
6
 
7
- from .base import describe_page, describe_collection, inspect_collection, describe_region, describe_element
8
- from .summary import ElementSummary, InspectionSummary
7
+ from .base import (
8
+ describe_collection,
9
+ describe_element,
10
+ describe_page,
11
+ describe_region,
12
+ inspect_collection,
13
+ )
9
14
  from .mixin import DescribeMixin, InspectMixin
15
+ from .summary import ElementSummary, InspectionSummary
10
16
 
11
17
  __all__ = [
12
- 'describe_page',
13
- 'describe_collection',
14
- 'inspect_collection',
15
- 'describe_region',
16
- 'describe_element',
17
- 'ElementSummary',
18
- 'InspectionSummary',
19
- 'DescribeMixin',
20
- 'InspectMixin'
21
- ]
18
+ "describe_page",
19
+ "describe_collection",
20
+ "inspect_collection",
21
+ "describe_region",
22
+ "describe_element",
23
+ "ElementSummary",
24
+ "InspectionSummary",
25
+ "DescribeMixin",
26
+ "InspectMixin",
27
+ ]