natural-pdf 0.1.14__py3-none-any.whl → 0.1.16__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 (39) hide show
  1. natural_pdf/__init__.py +31 -0
  2. natural_pdf/analyzers/layout/gemini.py +137 -162
  3. natural_pdf/analyzers/layout/layout_manager.py +9 -5
  4. natural_pdf/analyzers/layout/layout_options.py +77 -7
  5. natural_pdf/analyzers/layout/paddle.py +318 -165
  6. natural_pdf/analyzers/layout/table_structure_utils.py +78 -0
  7. natural_pdf/analyzers/shape_detection_mixin.py +770 -405
  8. natural_pdf/classification/mixin.py +2 -8
  9. natural_pdf/collections/pdf_collection.py +25 -30
  10. natural_pdf/core/highlighting_service.py +47 -32
  11. natural_pdf/core/page.py +226 -70
  12. natural_pdf/core/pdf.py +19 -22
  13. natural_pdf/elements/base.py +9 -9
  14. natural_pdf/elements/collections.py +105 -50
  15. natural_pdf/elements/region.py +320 -113
  16. natural_pdf/exporters/paddleocr.py +38 -13
  17. natural_pdf/flows/__init__.py +3 -3
  18. natural_pdf/flows/collections.py +303 -132
  19. natural_pdf/flows/element.py +277 -132
  20. natural_pdf/flows/flow.py +33 -16
  21. natural_pdf/flows/region.py +142 -79
  22. natural_pdf/ocr/engine_doctr.py +37 -4
  23. natural_pdf/ocr/engine_easyocr.py +23 -3
  24. natural_pdf/ocr/engine_paddle.py +281 -30
  25. natural_pdf/ocr/engine_surya.py +8 -3
  26. natural_pdf/ocr/ocr_manager.py +75 -76
  27. natural_pdf/ocr/ocr_options.py +52 -87
  28. natural_pdf/search/__init__.py +25 -12
  29. natural_pdf/search/lancedb_search_service.py +91 -54
  30. natural_pdf/search/numpy_search_service.py +86 -65
  31. natural_pdf/search/searchable_mixin.py +2 -2
  32. natural_pdf/selectors/parser.py +125 -81
  33. natural_pdf/widgets/__init__.py +1 -1
  34. natural_pdf/widgets/viewer.py +205 -449
  35. {natural_pdf-0.1.14.dist-info → natural_pdf-0.1.16.dist-info}/METADATA +27 -45
  36. {natural_pdf-0.1.14.dist-info → natural_pdf-0.1.16.dist-info}/RECORD +39 -38
  37. {natural_pdf-0.1.14.dist-info → natural_pdf-0.1.16.dist-info}/WHEEL +0 -0
  38. {natural_pdf-0.1.14.dist-info → natural_pdf-0.1.16.dist-info}/licenses/LICENSE +0 -0
  39. {natural_pdf-0.1.14.dist-info → natural_pdf-0.1.16.dist-info}/top_level.txt +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: natural-pdf
3
- Version: 0.1.14
3
+ Version: 0.1.16
4
4
  Summary: A more intuitive interface for working with PDFs
5
5
  Author-email: Jonathan Soma <jonathan.soma@gmail.com>
6
6
  License-Expression: MIT
@@ -12,6 +12,7 @@ Requires-Python: >=3.9
12
12
  Description-Content-Type: text/markdown
13
13
  License-File: LICENSE
14
14
  Requires-Dist: pdfplumber
15
+ Requires-Dist: colormath2
15
16
  Requires-Dist: pillow
16
17
  Requires-Dist: colour
17
18
  Requires-Dist: numpy
@@ -21,47 +22,31 @@ Requires-Dist: pydantic
21
22
  Requires-Dist: jenkspy
22
23
  Requires-Dist: pikepdf>=9.7.0
23
24
  Requires-Dist: scipy
24
- Provides-Extra: viewer
25
- Requires-Dist: ipywidgets<9.0.0,>=7.0.0; extra == "viewer"
26
- Provides-Extra: easyocr
27
- Requires-Dist: easyocr; extra == "easyocr"
28
- Requires-Dist: natural-pdf[core-ml]; extra == "easyocr"
29
- Provides-Extra: paddle
30
- Requires-Dist: paddlepaddle; extra == "paddle"
31
- Requires-Dist: paddleocr; extra == "paddle"
32
- Provides-Extra: layout-yolo
33
- Requires-Dist: doclayout_yolo; extra == "layout-yolo"
34
- Requires-Dist: natural-pdf[core-ml]; extra == "layout-yolo"
35
- Provides-Extra: surya
36
- Requires-Dist: surya-ocr; extra == "surya"
37
- Requires-Dist: natural-pdf[core-ml]; extra == "surya"
38
- Provides-Extra: doctr
39
- Requires-Dist: python-doctr[torch]; extra == "doctr"
40
- Requires-Dist: natural-pdf[core-ml]; extra == "doctr"
41
- Provides-Extra: docling
42
- Requires-Dist: docling; extra == "docling"
43
- Requires-Dist: natural-pdf[core-ml]; extra == "docling"
44
- Provides-Extra: llm
45
- Requires-Dist: openai>=1.0; extra == "llm"
25
+ Requires-Dist: torch
26
+ Requires-Dist: torchvision
27
+ Requires-Dist: transformers[sentencepiece]<=4.34.1
28
+ Requires-Dist: huggingface_hub>=0.29.3
29
+ Requires-Dist: sentence-transformers
30
+ Requires-Dist: timm
46
31
  Provides-Extra: test
47
32
  Requires-Dist: pytest; extra == "test"
33
+ Requires-Dist: pytest-xdist; extra == "test"
34
+ Requires-Dist: setuptools; extra == "test"
48
35
  Provides-Extra: search
49
36
  Requires-Dist: lancedb; extra == "search"
50
37
  Requires-Dist: pyarrow; extra == "search"
51
38
  Provides-Extra: favorites
52
39
  Requires-Dist: natural-pdf[deskew]; extra == "favorites"
53
- Requires-Dist: natural-pdf[llm]; extra == "favorites"
54
- Requires-Dist: natural-pdf[surya]; extra == "favorites"
55
- Requires-Dist: natural-pdf[easyocr]; extra == "favorites"
56
- Requires-Dist: natural-pdf[layout_yolo]; extra == "favorites"
57
40
  Requires-Dist: natural-pdf[ocr-export]; extra == "favorites"
58
- Requires-Dist: natural-pdf[viewer]; extra == "favorites"
59
41
  Requires-Dist: natural-pdf[search]; extra == "favorites"
42
+ Requires-Dist: ipywidgets; extra == "favorites"
43
+ Requires-Dist: surya-ocr; extra == "favorites"
60
44
  Provides-Extra: dev
61
45
  Requires-Dist: black; extra == "dev"
62
46
  Requires-Dist: isort; extra == "dev"
63
47
  Requires-Dist: mypy; extra == "dev"
64
48
  Requires-Dist: pytest; extra == "dev"
49
+ Requires-Dist: pytest-xdist; extra == "dev"
65
50
  Requires-Dist: nox; extra == "dev"
66
51
  Requires-Dist: nox-uv; extra == "dev"
67
52
  Requires-Dist: build; extra == "dev"
@@ -71,31 +56,28 @@ Requires-Dist: nbformat; extra == "dev"
71
56
  Requires-Dist: jupytext; extra == "dev"
72
57
  Requires-Dist: nbclient; extra == "dev"
73
58
  Requires-Dist: ipykernel; extra == "dev"
59
+ Requires-Dist: pre-commit; extra == "dev"
60
+ Requires-Dist: setuptools; extra == "dev"
74
61
  Provides-Extra: deskew
75
62
  Requires-Dist: deskew>=1.5; extra == "deskew"
76
63
  Requires-Dist: img2pdf; extra == "deskew"
64
+ Provides-Extra: addons
65
+ Requires-Dist: surya-ocr; extra == "addons"
66
+ Requires-Dist: doclayout_yolo; extra == "addons"
67
+ Requires-Dist: paddlepaddle>=3.0.0; extra == "addons"
68
+ Requires-Dist: paddleocr>=3.0.0; extra == "addons"
69
+ Requires-Dist: ipywidgets>=7.0.0; extra == "addons"
70
+ Requires-Dist: easyocr; extra == "addons"
71
+ Requires-Dist: surya-ocr; extra == "addons"
72
+ Requires-Dist: doclayout_yolo; extra == "addons"
73
+ Requires-Dist: python-doctr[torch]; extra == "addons"
74
+ Requires-Dist: docling; extra == "addons"
77
75
  Provides-Extra: all
78
- Requires-Dist: natural-pdf[viewer]; extra == "all"
79
- Requires-Dist: natural-pdf[easyocr]; extra == "all"
80
- Requires-Dist: natural-pdf[paddle]; extra == "all"
81
- Requires-Dist: natural-pdf[layout_yolo]; extra == "all"
82
- Requires-Dist: natural-pdf[surya]; extra == "all"
83
- Requires-Dist: natural-pdf[doctr]; extra == "all"
84
76
  Requires-Dist: natural-pdf[ocr-export]; extra == "all"
85
- Requires-Dist: natural-pdf[docling]; extra == "all"
86
- Requires-Dist: natural-pdf[llm]; extra == "all"
87
- Requires-Dist: natural-pdf[core-ml]; extra == "all"
88
77
  Requires-Dist: natural-pdf[deskew]; extra == "all"
89
78
  Requires-Dist: natural-pdf[test]; extra == "all"
90
79
  Requires-Dist: natural-pdf[search]; extra == "all"
91
- Provides-Extra: core-ml
92
- Requires-Dist: torch; extra == "core-ml"
93
- Requires-Dist: torchvision; extra == "core-ml"
94
- Requires-Dist: transformers[sentencepiece]; extra == "core-ml"
95
- Requires-Dist: huggingface_hub; extra == "core-ml"
96
- Requires-Dist: sentence-transformers; extra == "core-ml"
97
- Requires-Dist: numpy; extra == "core-ml"
98
- Requires-Dist: timm; extra == "core-ml"
80
+ Requires-Dist: natural-pdf[addons]; extra == "all"
99
81
  Provides-Extra: ocr-export
100
82
  Requires-Dist: pikepdf; extra == "ocr-export"
101
83
  Provides-Extra: export-extras
@@ -1,37 +1,38 @@
1
- natural_pdf/__init__.py,sha256=0sCYgb9BAV5OnpD_1AswMuOLuXNmpe3OLJpv_6p3tgw,2449
1
+ natural_pdf/__init__.py,sha256=qDFJNF8sbEDO-2WSFAxoWEM8updOUP6dB-ckya0kxfs,3275
2
2
  natural_pdf/analyzers/__init__.py,sha256=dIXjsMqoxKmd9OOnSBzn12wvdIz7D7YNQRAnXslpJSM,142
3
- natural_pdf/analyzers/shape_detection_mixin.py,sha256=6IXKuifC2QDurW8_gMakZCetTHSdOUK5Ye_B-P4EQMA,75381
3
+ natural_pdf/analyzers/shape_detection_mixin.py,sha256=HHefg-v7CJMxYiJHxdGOdqdtbWe9yk4OBoW3a_aRrjM,81798
4
4
  natural_pdf/analyzers/text_options.py,sha256=qEkDaYWla0rIM_gszEOsu52q7C_dAfV81P2HLJZM2sw,3333
5
5
  natural_pdf/analyzers/text_structure.py,sha256=VfKTsTFrK877sC0grsis9jK3rrgp0Mbp13VWEbukTcs,28437
6
6
  natural_pdf/analyzers/utils.py,sha256=PYbzJzSAHZ7JsMes84WIrSbA0zkjJGs0CLvIeINsf_k,2100
7
7
  natural_pdf/analyzers/layout/__init__.py,sha256=oq1uJ5UkGGMbBKGirV1aRKK3hxAUyjTLywYkPCQH1f0,33
8
8
  natural_pdf/analyzers/layout/base.py,sha256=bYawhmc_0xqKG-xbxUSiazIU1om-aBox5Jh8qDqv-eM,6451
9
9
  natural_pdf/analyzers/layout/docling.py,sha256=4BJYyNVR6VegZGxyisvNIBBRvVk6YKPyDVs7ZdVfzEU,12676
10
- natural_pdf/analyzers/layout/gemini.py,sha256=iuq-zZYkTS7fdAjD3ULRhqYTP9Ky2NgVHaXSLppDidw,11751
10
+ natural_pdf/analyzers/layout/gemini.py,sha256=ldECVCQ5HNQA3Omjg2NOsTrJXslyYb0vErDncmLIiuE,10510
11
11
  natural_pdf/analyzers/layout/layout_analyzer.py,sha256=n327Zjuf7aSzKQKChPHeiCVHinzeDGaWNyKiwQ-DkJk,15571
12
- natural_pdf/analyzers/layout/layout_manager.py,sha256=RiVq6gUA8t9OLj-HojdzQkJtabM32iBWEBoLtS7_TjY,8115
13
- natural_pdf/analyzers/layout/layout_options.py,sha256=Jsm4MfD_vedXvS7NCpVmuIRsIuyNyKOjvdgoRYOKZpI,4133
14
- natural_pdf/analyzers/layout/paddle.py,sha256=gTI9ZqNd5-t4H5IByGfL32WgcE6JrdchW6jRiGI6ulM,13375
12
+ natural_pdf/analyzers/layout/layout_manager.py,sha256=NkOZJCFiaZuvDN4Jgu1GsYRtlSJEy2tU_I-v-4EQQqU,8322
13
+ natural_pdf/analyzers/layout/layout_options.py,sha256=-Nv6bcu4_pqSCN6uNhCZ9mvoCBtRDZIUkO6kjkuLXsg,7703
14
+ natural_pdf/analyzers/layout/paddle.py,sha256=tX2bI1yayAdmRhvsfZ_Ygs7zAG5e9eW-pLJkw4NUpBQ,21325
15
15
  natural_pdf/analyzers/layout/pdfplumber_table_finder.py,sha256=Tk0Q7wv7nGYPo69lh6RoezjdepTnMl90SaNIrP29Pwc,5902
16
16
  natural_pdf/analyzers/layout/surya.py,sha256=4RdnhRxSS3i3Ns5mFhOA9-P0xd7Ms19uZuKvUGQfEBI,9789
17
+ natural_pdf/analyzers/layout/table_structure_utils.py,sha256=nISZDBd46RPYkFHxbQyIHwg9WweG4DslpoYJ31OMJYA,2768
17
18
  natural_pdf/analyzers/layout/tatr.py,sha256=cVr0ZyhY2mNLAKZ4DGMm-b7XNJpILKh8x8ZpyDeUhLk,15032
18
19
  natural_pdf/analyzers/layout/yolo.py,sha256=ANo2U4EZgeN2eYKM1bZIuysiuJLgwl4JeQchrRxOKwA,8388
19
20
  natural_pdf/classification/manager.py,sha256=-rdZzGP_JK4RDDxIEgdY8_gHRNS0cNHhpOSodjxbd84,17853
20
- natural_pdf/classification/mixin.py,sha256=hhX9qWPShpOq_-mgoEq0GUWnutBnNMo3YdUlxwyNWMA,6781
21
+ natural_pdf/classification/mixin.py,sha256=_XtoqCMqj1nxZYskIV2RbVYiVVcEWzFwae4s5vpzC74,6566
21
22
  natural_pdf/classification/results.py,sha256=El1dY7cBQVOB5lP-uj52dWgH6Y7TeQgJOVcZD-OLjes,2778
22
23
  natural_pdf/collections/mixins.py,sha256=sj76Cn6EdBtb5f-bdAV-1qpdixX8tI4BzPccPiYLI1w,5117
23
- natural_pdf/collections/pdf_collection.py,sha256=nsbrzcsXAD2qVLLXhDYpljAb_WnjMNanHJ6J7UtYzGA,31165
24
+ natural_pdf/collections/pdf_collection.py,sha256=HLlyakM--23ZOeHDPucoM6Tw3yUyMXm0SSoqJwxRc2E,30744
24
25
  natural_pdf/core/__init__.py,sha256=QC8H4M3KbXwMFiQORZ0pdPlzx1Ix6oKKQSS7Ib2KEaA,38
25
26
  natural_pdf/core/element_manager.py,sha256=_UdXu51sLi6STzc8Pj4k8R721G3yJixXDLuRHn3hmr8,25731
26
- natural_pdf/core/highlighting_service.py,sha256=tjMJpdJj2oaMGpdqiNHPcTJqID4nd-uBZ5v7KtPmoc0,36762
27
- natural_pdf/core/page.py,sha256=hg7EoYMbvgo9dXivBl6xb6dENobhSHt0Wuu36O5J900,111119
28
- natural_pdf/core/pdf.py,sha256=395aBTg4Le4vABvQWgBhPm669nGJ8JdMToTs1UtQ2Vg,69575
27
+ natural_pdf/core/highlighting_service.py,sha256=_kQUS6_BBvsLBuSZloFrVag6jN90KzHa0ULyGBjufSs,36955
28
+ natural_pdf/core/page.py,sha256=ciwBf-SoI431SJjp2VRfLxdtqgO2L6p044kXXjlNtjo,118231
29
+ natural_pdf/core/pdf.py,sha256=bAoGPiKIrFaebLwULMT-9VkHQ_wkE_zNl4hlbMLk-2w,69325
29
30
  natural_pdf/elements/__init__.py,sha256=S8XeiNWJ1WcgnyYKdYV1yxQlAxCCO3FfITT8MQwNbyk,41
30
- natural_pdf/elements/base.py,sha256=NNF-iUzkip0UgfKTuqLE1jVJsq2yD7LUTvOQWMi_Jpc,39631
31
- natural_pdf/elements/collections.py,sha256=qd58tD3f-eojz90ICytlqu4Ej0OQoWgsxV4umQDhUvA,120809
31
+ natural_pdf/elements/base.py,sha256=tEyCInUc6wxbUtnXVaBa21Qpr591Sgu4yi7tKxWb-3U,39607
32
+ natural_pdf/elements/collections.py,sha256=_lWL-W-RKlYikkGJU66dskGCZ8-7WfMyUx2G0IgjhlQ,121965
32
33
  natural_pdf/elements/line.py,sha256=300kSFBDUBIudfeQtH_tzW9gTYRgRKUDPiTABw6J-BE,4782
33
34
  natural_pdf/elements/rect.py,sha256=kiVa3e377ZnqIOXc89d9ZSY4EcmDxtccdtUw-HOQzpw,3796
34
- natural_pdf/elements/region.py,sha256=wBBAcuudRqL1b9ojLdrXiwUIcQbTWEWTky_RbBuCgnU,115798
35
+ natural_pdf/elements/region.py,sha256=nCXyI0vq9-MIQ4Zk90q5Nn-U6gDGv22NY6ime6qG1MY,123330
35
36
  natural_pdf/elements/text.py,sha256=13HvVZGinj2Vm_fFCAnqi7hohtoKvnpCp3VCfkpeAbc,11146
36
37
  natural_pdf/export/mixin.py,sha256=L1q3MIEFWuvie4j4_EmW7GT3NerbZ1as0XMUoqTS7gM,5083
37
38
  natural_pdf/exporters/__init__.py,sha256=7MnvRLLQdwtg-ULu-8uK8C84GsKiJamyhRw_GgWhw7k,151
@@ -39,7 +40,7 @@ natural_pdf/exporters/base.py,sha256=XhR1xlkHOh7suOuX7mWbsj1h2o1pZNet-OAS5YCJyeI
39
40
  natural_pdf/exporters/hocr.py,sha256=MOb5sTxe-GlMSOtmqp3p4SY_ZigwOtmd4sj_zMRCIQY,19907
40
41
  natural_pdf/exporters/hocr_font.py,sha256=1wsGOMj6zoaRN2rxCwrv4MMLGawpNz984WgXpmWekgw,4574
41
42
  natural_pdf/exporters/original_pdf.py,sha256=zsZPg_lUoEerKIzzoEw-qGdM5XBg_LZhFJeVKnCUp4o,5054
42
- natural_pdf/exporters/paddleocr.py,sha256=srwk_N10wVqtEU5bI8B3XGfXr54gaaJ0Q5zpq4-cSVY,18361
43
+ natural_pdf/exporters/paddleocr.py,sha256=IAG2p9YeImYcsIvb6a_L5mMrKarvaMaDvRrvdlY6bX4,19489
43
44
  natural_pdf/exporters/searchable_pdf.py,sha256=G2Tc4tpDXSYIufXJlkA8ppW_3DuzHAaweYKae33pI_c,16290
44
45
  natural_pdf/exporters/data/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
45
46
  natural_pdf/exporters/data/pdf.ttf,sha256=x4RUIJJaI9iO2DCmOVe4r4Wmao2vjZ_JDoQ2c7LvGlk,572
@@ -47,31 +48,31 @@ natural_pdf/exporters/data/sRGB.icc,sha256=KpLUuuRQt22LCqQhk9-XTXX2Jzjs6_dPAcXnW
47
48
  natural_pdf/extraction/manager.py,sha256=mUBbfgLG5Pl31wmajXwyipdEJb_dZ5I-y8GnWw7IzGo,4969
48
49
  natural_pdf/extraction/mixin.py,sha256=eKbr70VibpbtfjvCE80lTFuYHzq_BoVtOHjznL_GMRA,11719
49
50
  natural_pdf/extraction/result.py,sha256=c1vLguCR6l95cvg-BJJmZvL_MPg2McJaczge55bKZMg,934
50
- natural_pdf/flows/__init__.py,sha256=82ibI0eNJfVergEsTyom9Nxe_T6pnWQsr4-CISGQlz0,277
51
- natural_pdf/flows/collections.py,sha256=iOmRqM5K74kqioh7-UAbNgkpXMr9nkZZ5oW4_sQ1Alo,26433
52
- natural_pdf/flows/element.py,sha256=HMlSBjnQH3CF89wDwc0qBpAfmMmHHK_5LeY1TYeqtTs,20564
53
- natural_pdf/flows/flow.py,sha256=ft07Ou0uRodF_gTgumVlU9YUquE3LTZz5LEAoQGErEs,10375
54
- natural_pdf/flows/region.py,sha256=5xAnePZjs292oKrGG5El3pwhpxaHQYLzse35ilswhqI,21298
51
+ natural_pdf/flows/__init__.py,sha256=cUN4A8hTDLZSRr4PO2W_lR4z6hWpbNG8Seox-IIcrLU,277
52
+ natural_pdf/flows/collections.py,sha256=qGuSPFSPQF-wiYquG6STiSzg_o951MSsFEq_B44Jef8,28441
53
+ natural_pdf/flows/element.py,sha256=mKzk3B7A7sWNvu4CDvAjLr3_ZFLt--ktrSNoLfLpFxU,23940
54
+ natural_pdf/flows/flow.py,sha256=I61BpFVDQyo6ORsmoqoYiOEP1DBRp0vgDJjm_V8frhc,10562
55
+ natural_pdf/flows/region.py,sha256=OJNt1nVt_w89kCw4sQmZRZE5SHgiFWYb8ebcX49jq2A,22192
55
56
  natural_pdf/ocr/__init__.py,sha256=VY8hhvDPf7Gh2lB-d2QRmghLLyTy6ydxlgo1cS4dOSk,2482
56
57
  natural_pdf/ocr/engine.py,sha256=ZBC1tZNM5EDbGDJJmZI9mNHr4nCMLEZvUFhiJq8GdF4,8741
57
- natural_pdf/ocr/engine_doctr.py,sha256=519WpvSHgwP6Hv24tci_YHFX7XPlaxOnlREN_YG-Yys,16331
58
- natural_pdf/ocr/engine_easyocr.py,sha256=9TbxJjmhWFrzM8mcNnZjoRtIDr6gwpuwKm4-Zfub2-8,9281
59
- natural_pdf/ocr/engine_paddle.py,sha256=2nIrvLBBAiZG1BxVo3eFVJulA6YGoOTXw_RN98p_BUk,6184
60
- natural_pdf/ocr/engine_surya.py,sha256=CQHpPecCYsJsr7pEvEyubAf5FJFs7vFHAm_0cGGr-A4,4839
58
+ natural_pdf/ocr/engine_doctr.py,sha256=ptKrupMWoulZb-R93zr9btoe94JPWU7vlJuN7OBJEIM,17740
59
+ natural_pdf/ocr/engine_easyocr.py,sha256=bWz6kHUgAJfe3rqdnZBAF-IPvw3B35DlvX5KDdFUtzo,9888
60
+ natural_pdf/ocr/engine_paddle.py,sha256=ZUtyjso_UjjAPnJt5ac-AtOpR6PfOhO76iOyjngGzr0,16198
61
+ natural_pdf/ocr/engine_surya.py,sha256=Qc3geQQzJ1-9WS1aho38jfvd7yxbYOUVeIpzpapHLRg,5159
61
62
  natural_pdf/ocr/ocr_factory.py,sha256=gBFXdFs7E4aCynHz06sQsAhaO3s8yhgoFgN5nyxtg9c,5221
62
- natural_pdf/ocr/ocr_manager.py,sha256=f0q68ynGYVPkF4D3WnufxmHWD5R1jW5Z_1czTEi9JVU,13931
63
- natural_pdf/ocr/ocr_options.py,sha256=ZvtnFn1kPkFEoWveQ13uy6B-ofquP0gHEi4tBHrjqCE,6438
63
+ natural_pdf/ocr/ocr_manager.py,sha256=O-wSx50k9pcf0M8N_5nKVefS55r6tMJWRF8KjktA8ts,13664
64
+ natural_pdf/ocr/ocr_options.py,sha256=l33QKu_93r-uwi3t_v8UH8pEgHo6HTVzP4tfmQFRF1w,5488
64
65
  natural_pdf/ocr/utils.py,sha256=OxuHwDbHWj6setvnC0QYwMHrAjxGkhmLzWHpMqqGupA,4397
65
66
  natural_pdf/qa/__init__.py,sha256=Pjo62JTnUNEjGNsC437mvsS5KQ5m7X_BibGvavR9AW0,108
66
67
  natural_pdf/qa/document_qa.py,sha256=Jw4yyq3Vifn57D0ANmOfUlZeG8CJjBkItZBV-8ZAmos,15111
67
- natural_pdf/search/__init__.py,sha256=72n_Mj_AhF_RCIoBBhZ6EZKjbILM8omelXZ99fXw7n4,3688
68
- natural_pdf/search/lancedb_search_service.py,sha256=tW7ONPcWGY1HKle_7OqCXRnMCI-aKL-AqneKz2YbLlM,13706
69
- natural_pdf/search/numpy_search_service.py,sha256=3_8fx7NV-15jBokOU73mcxrznxPxzVQnOlDHf3dpo28,10117
68
+ natural_pdf/search/__init__.py,sha256=RHP1E-5m3hhLXz__g7EvZihBJjPTDtUYh_bZr_NwDo0,3724
69
+ natural_pdf/search/lancedb_search_service.py,sha256=kgm-nYXjPQBkEkWE0gkdpL4V53xm_CEX4rZ5KBpxgfM,14190
70
+ natural_pdf/search/numpy_search_service.py,sha256=5zkkZds-Dcp8PsrvTJdyW15fS1ffHDLVjeiXTGWoRsY,10006
70
71
  natural_pdf/search/search_options.py,sha256=sq_e8_jSROicD94b_xtDtLnjEr_Zsy4icjzPkK0a8QA,3566
71
72
  natural_pdf/search/search_service_protocol.py,sha256=Dl-Q-CrutkhZwI69scbW9EWPeYM63qxB60_EA7YqIYo,6699
72
- natural_pdf/search/searchable_mixin.py,sha256=dZbaHv8Go3TJNqxoPtnp9Dr0Ftxuf_44RpBeIRXkPxc,23534
73
+ natural_pdf/search/searchable_mixin.py,sha256=hqQ_AuID5eTGRCtKYdFLZ1zF35y73uk3x1M1VW9Il8U,23514
73
74
  natural_pdf/selectors/__init__.py,sha256=oZGeqSv53EqmIZOhcnawuaGGlRg1h79vArXuZCWKm4A,123
74
- natural_pdf/selectors/parser.py,sha256=EZsNRjn40qjiQ-KlQOmGJVrkvlVd8wez9v3SGdoFYSo,30226
75
+ natural_pdf/selectors/parser.py,sha256=0EJskGjRfwXiqjedY6siIso4TA4dACHb-_DOyogW9TY,31970
75
76
  natural_pdf/templates/__init__.py,sha256=jYBxzfi73vew0f6yhIh1MlRxw4F_TVN2hKQR0YXOFe0,20
76
77
  natural_pdf/utils/__init__.py,sha256=s3M8FggaK1P3EBYn6R_-HgSDjNc9C73gyKe1hihtNWg,43
77
78
  natural_pdf/utils/debug.py,sha256=RN7H3E6ph-GtxubCW6psW7TO8o2BxcNLiEzByTVR9fk,995
@@ -82,10 +83,10 @@ natural_pdf/utils/packaging.py,sha256=Jshxp6S1zfcqoZmFhdd7WOpL--b6rBSz-Y9mYqELXI
82
83
  natural_pdf/utils/reading_order.py,sha256=s3DsYq_3g_1YA07qhd4BGEjeIRTeyGtnwc_hNtSzwBY,7290
83
84
  natural_pdf/utils/text_extraction.py,sha256=z6Jhy11pakYCsEpkvh8ldw6DkUFsYF1hCL9YDmfXWL4,9605
84
85
  natural_pdf/utils/visualization.py,sha256=30pRWQdsRJh2pSObh-brKVsFgC1n8tHmSrta_UDnVPw,8989
85
- natural_pdf/widgets/__init__.py,sha256=O2fSDo604wDAP6UwUkmBq3eT91RSqHwBpAOQXq92S8s,214
86
- natural_pdf/widgets/viewer.py,sha256=ekgXTEfA48GrR-JjpCpgyBCXdf4IubV0pAXDJozcU7A,39196
87
- natural_pdf-0.1.14.dist-info/licenses/LICENSE,sha256=9zfwINwJlarbDmdh6iJV4QUG54QSJlSAUcnC1YiC_Ns,1074
88
- natural_pdf-0.1.14.dist-info/METADATA,sha256=NzaR_hcSyFH22knKZ-NMCct_XOo2nPUk83XHspTncyE,7674
89
- natural_pdf-0.1.14.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
90
- natural_pdf-0.1.14.dist-info/top_level.txt,sha256=Cyw1zmNDlUZfb5moU-WUWGprrwH7ln_8LDGdmMHF1xI,17
91
- natural_pdf-0.1.14.dist-info/RECORD,,
86
+ natural_pdf/widgets/__init__.py,sha256=QTVaUmsw__FCweFYZebwPssQxxUFUMd0wpm_cUbGZJY,181
87
+ natural_pdf/widgets/viewer.py,sha256=2VUY1TzWMDe9I-IVNOosKZ2LaqpjLB62ftMAdk-s6_8,24952
88
+ natural_pdf-0.1.16.dist-info/licenses/LICENSE,sha256=9zfwINwJlarbDmdh6iJV4QUG54QSJlSAUcnC1YiC_Ns,1074
89
+ natural_pdf-0.1.16.dist-info/METADATA,sha256=ncvnNI_PubS4q4v29OKp5UXyanEZNVWqsCanu-xGCOA,6753
90
+ natural_pdf-0.1.16.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
91
+ natural_pdf-0.1.16.dist-info/top_level.txt,sha256=Cyw1zmNDlUZfb5moU-WUWGprrwH7ln_8LDGdmMHF1xI,17
92
+ natural_pdf-0.1.16.dist-info/RECORD,,