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.
- docs/api/index.md +386 -0
- docs/assets/favicon.png +3 -0
- docs/assets/favicon.svg +3 -0
- docs/assets/javascripts/custom.js +17 -0
- docs/assets/logo.svg +3 -0
- docs/assets/sample-screen.png +0 -0
- docs/assets/social-preview.png +17 -0
- docs/assets/social-preview.svg +17 -0
- docs/assets/stylesheets/custom.css +65 -0
- docs/document-qa/index.ipynb +435 -0
- docs/document-qa/index.md +79 -0
- docs/element-selection/index.ipynb +915 -0
- docs/element-selection/index.md +229 -0
- docs/index.md +170 -0
- docs/installation/index.md +69 -0
- docs/interactive-widget/index.ipynb +962 -0
- docs/interactive-widget/index.md +12 -0
- docs/layout-analysis/index.ipynb +818 -0
- docs/layout-analysis/index.md +185 -0
- docs/ocr/index.md +209 -0
- docs/pdf-navigation/index.ipynb +314 -0
- docs/pdf-navigation/index.md +97 -0
- docs/regions/index.ipynb +816 -0
- docs/regions/index.md +294 -0
- docs/tables/index.ipynb +658 -0
- docs/tables/index.md +144 -0
- docs/text-analysis/index.ipynb +370 -0
- docs/text-analysis/index.md +105 -0
- docs/text-extraction/index.ipynb +1478 -0
- docs/text-extraction/index.md +292 -0
- docs/tutorials/01-loading-and-extraction.ipynb +1710 -0
- docs/tutorials/01-loading-and-extraction.md +95 -0
- docs/tutorials/02-finding-elements.ipynb +340 -0
- docs/tutorials/02-finding-elements.md +149 -0
- docs/tutorials/03-extracting-blocks.ipynb +147 -0
- docs/tutorials/03-extracting-blocks.md +48 -0
- docs/tutorials/04-table-extraction.ipynb +114 -0
- docs/tutorials/04-table-extraction.md +50 -0
- docs/tutorials/05-excluding-content.ipynb +270 -0
- docs/tutorials/05-excluding-content.md +109 -0
- docs/tutorials/06-document-qa.ipynb +332 -0
- docs/tutorials/06-document-qa.md +91 -0
- docs/tutorials/07-layout-analysis.ipynb +288 -0
- docs/tutorials/07-layout-analysis.md +66 -0
- docs/tutorials/07-working-with-regions.ipynb +413 -0
- docs/tutorials/07-working-with-regions.md +151 -0
- docs/tutorials/08-spatial-navigation.ipynb +508 -0
- docs/tutorials/08-spatial-navigation.md +190 -0
- docs/tutorials/09-section-extraction.ipynb +2434 -0
- docs/tutorials/09-section-extraction.md +256 -0
- docs/tutorials/10-form-field-extraction.ipynb +512 -0
- docs/tutorials/10-form-field-extraction.md +201 -0
- docs/tutorials/11-enhanced-table-processing.ipynb +54 -0
- docs/tutorials/11-enhanced-table-processing.md +9 -0
- docs/tutorials/12-ocr-integration.ipynb +604 -0
- docs/tutorials/12-ocr-integration.md +175 -0
- docs/tutorials/13-semantic-search.ipynb +1328 -0
- docs/tutorials/13-semantic-search.md +77 -0
- docs/visual-debugging/index.ipynb +2970 -0
- docs/visual-debugging/index.md +157 -0
- docs/visual-debugging/region.png +0 -0
- natural_pdf/__init__.py +50 -33
- natural_pdf/analyzers/__init__.py +2 -1
- natural_pdf/analyzers/layout/base.py +32 -24
- natural_pdf/analyzers/layout/docling.py +131 -72
- natural_pdf/analyzers/layout/gemini.py +264 -0
- natural_pdf/analyzers/layout/layout_analyzer.py +156 -113
- natural_pdf/analyzers/layout/layout_manager.py +125 -58
- natural_pdf/analyzers/layout/layout_options.py +43 -17
- natural_pdf/analyzers/layout/paddle.py +152 -95
- natural_pdf/analyzers/layout/surya.py +164 -92
- natural_pdf/analyzers/layout/tatr.py +149 -84
- natural_pdf/analyzers/layout/yolo.py +89 -45
- natural_pdf/analyzers/text_options.py +22 -15
- natural_pdf/analyzers/text_structure.py +131 -85
- natural_pdf/analyzers/utils.py +30 -23
- natural_pdf/collections/pdf_collection.py +146 -97
- natural_pdf/core/__init__.py +1 -1
- natural_pdf/core/element_manager.py +419 -337
- natural_pdf/core/highlighting_service.py +268 -196
- natural_pdf/core/page.py +1044 -521
- natural_pdf/core/pdf.py +516 -313
- natural_pdf/elements/__init__.py +1 -1
- natural_pdf/elements/base.py +307 -225
- natural_pdf/elements/collections.py +805 -543
- natural_pdf/elements/line.py +39 -36
- natural_pdf/elements/rect.py +32 -30
- natural_pdf/elements/region.py +889 -879
- natural_pdf/elements/text.py +127 -99
- natural_pdf/exporters/__init__.py +0 -1
- natural_pdf/exporters/searchable_pdf.py +261 -102
- natural_pdf/ocr/__init__.py +57 -35
- natural_pdf/ocr/engine.py +150 -46
- natural_pdf/ocr/engine_easyocr.py +146 -150
- natural_pdf/ocr/engine_paddle.py +118 -175
- natural_pdf/ocr/engine_surya.py +78 -141
- natural_pdf/ocr/ocr_factory.py +114 -0
- natural_pdf/ocr/ocr_manager.py +122 -124
- natural_pdf/ocr/ocr_options.py +16 -20
- natural_pdf/ocr/utils.py +98 -0
- natural_pdf/qa/__init__.py +1 -1
- natural_pdf/qa/document_qa.py +119 -111
- natural_pdf/search/__init__.py +37 -31
- natural_pdf/search/haystack_search_service.py +312 -189
- natural_pdf/search/haystack_utils.py +186 -122
- natural_pdf/search/search_options.py +25 -14
- natural_pdf/search/search_service_protocol.py +12 -6
- natural_pdf/search/searchable_mixin.py +261 -176
- natural_pdf/selectors/__init__.py +2 -1
- natural_pdf/selectors/parser.py +159 -316
- natural_pdf/templates/__init__.py +1 -1
- natural_pdf/templates/spa/css/style.css +334 -0
- natural_pdf/templates/spa/index.html +31 -0
- natural_pdf/templates/spa/js/app.js +472 -0
- natural_pdf/templates/spa/words.txt +235976 -0
- natural_pdf/utils/debug.py +32 -0
- natural_pdf/utils/highlighting.py +8 -2
- natural_pdf/utils/identifiers.py +29 -0
- natural_pdf/utils/packaging.py +418 -0
- natural_pdf/utils/reading_order.py +65 -63
- natural_pdf/utils/text_extraction.py +195 -0
- natural_pdf/utils/visualization.py +70 -61
- natural_pdf/widgets/__init__.py +2 -3
- natural_pdf/widgets/viewer.py +749 -718
- {natural_pdf-0.1.4.dist-info → natural_pdf-0.1.6.dist-info}/METADATA +53 -17
- natural_pdf-0.1.6.dist-info/RECORD +141 -0
- {natural_pdf-0.1.4.dist-info → natural_pdf-0.1.6.dist-info}/WHEEL +1 -1
- natural_pdf-0.1.6.dist-info/top_level.txt +4 -0
- notebooks/Examples.ipynb +1293 -0
- pdfs/.gitkeep +0 -0
- pdfs/01-practice.pdf +543 -0
- pdfs/0500000US42001.pdf +0 -0
- pdfs/0500000US42007.pdf +0 -0
- pdfs/2014 Statistics.pdf +0 -0
- pdfs/2019 Statistics.pdf +0 -0
- pdfs/Atlanta_Public_Schools_GA_sample.pdf +0 -0
- pdfs/needs-ocr.pdf +0 -0
- natural_pdf/templates/ocr_debug.html +0 -517
- natural_pdf-0.1.4.dist-info/RECORD +0 -61
- natural_pdf-0.1.4.dist-info/top_level.txt +0 -1
- {natural_pdf-0.1.4.dist-info → natural_pdf-0.1.6.dist-info}/licenses/LICENSE +0 -0
@@ -1,6 +1,6 @@
|
|
1
1
|
Metadata-Version: 2.4
|
2
2
|
Name: natural-pdf
|
3
|
-
Version: 0.1.
|
3
|
+
Version: 0.1.6
|
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
|
@@ -16,38 +16,70 @@ Requires-Dist: Pillow
|
|
16
16
|
Requires-Dist: colour
|
17
17
|
Requires-Dist: numpy
|
18
18
|
Requires-Dist: urllib3
|
19
|
-
Requires-Dist:
|
20
|
-
Requires-Dist: torchvision
|
21
|
-
Requires-Dist: transformers
|
22
|
-
Requires-Dist: huggingface_hub
|
23
|
-
Requires-Dist: ocrmypdf
|
24
|
-
Requires-Dist: pikepdf
|
19
|
+
Requires-Dist: tqdm
|
25
20
|
Provides-Extra: interactive
|
26
21
|
Requires-Dist: ipywidgets<9.0.0,>=7.0.0; extra == "interactive"
|
27
22
|
Provides-Extra: haystack
|
28
23
|
Requires-Dist: haystack-ai; extra == "haystack"
|
29
24
|
Requires-Dist: chroma-haystack; extra == "haystack"
|
30
25
|
Requires-Dist: sentence-transformers; extra == "haystack"
|
26
|
+
Requires-Dist: protobuf<4; extra == "haystack"
|
27
|
+
Requires-Dist: natural-pdf[core-ml]; extra == "haystack"
|
31
28
|
Provides-Extra: easyocr
|
32
29
|
Requires-Dist: easyocr; extra == "easyocr"
|
30
|
+
Requires-Dist: natural-pdf[core-ml]; extra == "easyocr"
|
33
31
|
Provides-Extra: paddle
|
34
32
|
Requires-Dist: paddlepaddle; extra == "paddle"
|
35
33
|
Requires-Dist: paddleocr; extra == "paddle"
|
36
34
|
Provides-Extra: layout-yolo
|
37
35
|
Requires-Dist: doclayout_yolo; extra == "layout-yolo"
|
36
|
+
Requires-Dist: natural-pdf[core-ml]; extra == "layout-yolo"
|
38
37
|
Provides-Extra: surya
|
39
38
|
Requires-Dist: surya-ocr; extra == "surya"
|
39
|
+
Requires-Dist: natural-pdf[core-ml]; extra == "surya"
|
40
40
|
Provides-Extra: qa
|
41
|
+
Requires-Dist: natural-pdf[core-ml]; extra == "qa"
|
42
|
+
Provides-Extra: docling
|
43
|
+
Requires-Dist: docling; extra == "docling"
|
44
|
+
Requires-Dist: natural-pdf[core-ml]; extra == "docling"
|
45
|
+
Provides-Extra: llm
|
46
|
+
Requires-Dist: openai>=1.0; extra == "llm"
|
47
|
+
Requires-Dist: pydantic; extra == "llm"
|
48
|
+
Provides-Extra: test
|
49
|
+
Requires-Dist: pytest; extra == "test"
|
50
|
+
Provides-Extra: dev
|
51
|
+
Requires-Dist: black; extra == "dev"
|
52
|
+
Requires-Dist: isort; extra == "dev"
|
53
|
+
Requires-Dist: mypy; extra == "dev"
|
54
|
+
Requires-Dist: pytest; extra == "dev"
|
55
|
+
Requires-Dist: nox; extra == "dev"
|
56
|
+
Requires-Dist: nox-uv; extra == "dev"
|
57
|
+
Requires-Dist: build; extra == "dev"
|
58
|
+
Requires-Dist: uv; extra == "dev"
|
59
|
+
Requires-Dist: pipdeptree; extra == "dev"
|
60
|
+
Requires-Dist: nbformat; extra == "dev"
|
61
|
+
Requires-Dist: jupytext; extra == "dev"
|
62
|
+
Requires-Dist: nbclient; extra == "dev"
|
41
63
|
Provides-Extra: all
|
42
|
-
Requires-Dist:
|
43
|
-
Requires-Dist:
|
44
|
-
Requires-Dist:
|
45
|
-
Requires-Dist:
|
46
|
-
Requires-Dist:
|
47
|
-
Requires-Dist: surya
|
48
|
-
Requires-Dist:
|
49
|
-
Requires-Dist:
|
50
|
-
Requires-Dist:
|
64
|
+
Requires-Dist: natural-pdf[interactive]; extra == "all"
|
65
|
+
Requires-Dist: natural-pdf[haystack]; extra == "all"
|
66
|
+
Requires-Dist: natural-pdf[easyocr]; extra == "all"
|
67
|
+
Requires-Dist: natural-pdf[paddle]; extra == "all"
|
68
|
+
Requires-Dist: natural-pdf[layout_yolo]; extra == "all"
|
69
|
+
Requires-Dist: natural-pdf[surya]; extra == "all"
|
70
|
+
Requires-Dist: natural-pdf[qa]; extra == "all"
|
71
|
+
Requires-Dist: natural-pdf[ocr-export]; extra == "all"
|
72
|
+
Requires-Dist: natural-pdf[docling]; extra == "all"
|
73
|
+
Requires-Dist: natural-pdf[llm]; extra == "all"
|
74
|
+
Requires-Dist: natural-pdf[test]; extra == "all"
|
75
|
+
Provides-Extra: core-ml
|
76
|
+
Requires-Dist: torch; extra == "core-ml"
|
77
|
+
Requires-Dist: torchvision; extra == "core-ml"
|
78
|
+
Requires-Dist: transformers; extra == "core-ml"
|
79
|
+
Requires-Dist: huggingface_hub; extra == "core-ml"
|
80
|
+
Provides-Extra: ocr-export
|
81
|
+
Requires-Dist: ocrmypdf; extra == "ocr-export"
|
82
|
+
Requires-Dist: pikepdf; extra == "ocr-export"
|
51
83
|
Dynamic: license-file
|
52
84
|
|
53
85
|
# Natural PDF
|
@@ -75,6 +107,10 @@ pip install natural-pdf[easyocr]
|
|
75
107
|
pip install natural-pdf[surya]
|
76
108
|
pip install natural-pdf[paddle]
|
77
109
|
|
110
|
+
# Example: Install support for features using Large Language Models (e.g., via OpenAI-compatible APIs)
|
111
|
+
pip install natural-pdf[llm]
|
112
|
+
# (May require setting API key environment variables, e.g., GOOGLE_API_KEY for Gemini)
|
113
|
+
|
78
114
|
# Example: Install with interactive viewer support
|
79
115
|
pip install natural-pdf[interactive]
|
80
116
|
|
@@ -127,7 +163,7 @@ Natural PDF offers a range of features for working with PDFs:
|
|
127
163
|
* **Spatial Navigation:** Select content relative to other elements (`heading.below()`, `element.select_until(...)`).
|
128
164
|
* **Text & Table Extraction:** Get clean text or structured table data, automatically handling exclusions.
|
129
165
|
* **OCR Integration:** Extract text from scanned documents using engines like EasyOCR, PaddleOCR, or Surya.
|
130
|
-
* **Layout Analysis:** Detect document structures (titles, paragraphs, tables) using
|
166
|
+
* **Layout Analysis:** Detect document structures (titles, paragraphs, tables) using various engines (e.g., YOLO, Paddle, LLM via API).
|
131
167
|
* **Document QA:** Ask natural language questions about your document's content.
|
132
168
|
* **Semantic Search:** Index PDFs and find relevant pages or documents based on semantic meaning using Haystack.
|
133
169
|
* **Visual Debugging:** Highlight elements and use an interactive viewer or save images to understand your selections.
|
@@ -0,0 +1,141 @@
|
|
1
|
+
docs/index.md,sha256=P1kXZc8aefnxH0bBjvBgj1o3puRiezjUiBLqS4bcUhM,4889
|
2
|
+
docs/api/index.md,sha256=4bn8nYklWJuNDrnY-Kt7sf7IejeAEDhcnqYmjH9GJTA,22405
|
3
|
+
docs/assets/favicon.png,sha256=nxca8jM2Y4GxZKzkmagUHO1GpUREK-GRA5LEFue9OOU,284
|
4
|
+
docs/assets/favicon.svg,sha256=nxca8jM2Y4GxZKzkmagUHO1GpUREK-GRA5LEFue9OOU,284
|
5
|
+
docs/assets/logo.svg,sha256=DdSmjHISSLt20mhf7YDXztigE6w_Reja_ATUYNAPL0M,286
|
6
|
+
docs/assets/sample-screen.png,sha256=qfsrThVoPOksn1ACVbLGe_gdPQmY1EsespXPs1Ocm2Y,182330
|
7
|
+
docs/assets/social-preview.png,sha256=AvyzzM8dC0j5SPFF63bvQrxU4GE1f9j-GUNUv0oA9ts,1085
|
8
|
+
docs/assets/social-preview.svg,sha256=AvyzzM8dC0j5SPFF63bvQrxU4GE1f9j-GUNUv0oA9ts,1085
|
9
|
+
docs/assets/javascripts/custom.js,sha256=0NVHGprwiLPFYdYunJcHjOphzk_EhBSNuOUz5Uzdv_k,594
|
10
|
+
docs/assets/stylesheets/custom.css,sha256=PbTp3k77gzUBUQQ01pDXzpNwo4wUv3aJD-SMBQvQItY,1156
|
11
|
+
docs/document-qa/index.ipynb,sha256=MXJoFhi8TUKK6ZnRFiUBglLGpMbzwdb7LJYfzw8Gp48,528713
|
12
|
+
docs/document-qa/index.md,sha256=mwuO4tothg0OzBXewnj73QEJu46Udq7f1pQBYrKOHwM,2131
|
13
|
+
docs/element-selection/index.ipynb,sha256=-7PwKw1RbPlZ4stzN1Rd1GJ8mwjOD4ySsLcpqVX7chc,1193628
|
14
|
+
docs/element-selection/index.md,sha256=_1P8vI64Y0aSVwUzdRJD4ayb80BJWBLED9TvVpveFx8,6979
|
15
|
+
docs/installation/index.md,sha256=nd4RZrQFR8_vv7Xm3xAzp7z-CQQr9ffAcGa7yuEYn2U,1594
|
16
|
+
docs/interactive-widget/index.ipynb,sha256=zY1rz5N34OUW-OtgcbI6iiOjlIJqXjVcx9OoNWMjuyU,293111
|
17
|
+
docs/interactive-widget/index.md,sha256=tZbq0uYI7Zwo9mLbhXpqeBriuAjazkIyEJeP-jasJ-Q,259
|
18
|
+
docs/layout-analysis/index.ipynb,sha256=dkS_-cu-KGir5G2LGRcxBThKnW0dfA5nPPnwpoYGFtU,1869093
|
19
|
+
docs/layout-analysis/index.md,sha256=ZnH5yd7B_eOLgGxW_4rNlzQs4Tn3Xx1cK3jX43CSpSM,5390
|
20
|
+
docs/ocr/index.md,sha256=uuzTqcAgUmMN7jZVq8VkVcbRDHn8Yg2nJVvHJ-bDK-Y,8177
|
21
|
+
docs/pdf-navigation/index.ipynb,sha256=h6yew0HePXK1_c5FmETqzjBQceUBT0MU-vnXx_y91mo,8018
|
22
|
+
docs/pdf-navigation/index.md,sha256=P3b3tsmOcmRtnfRxpsMeTgwm7vApnH_4le_QIwJd51M,2391
|
23
|
+
docs/regions/index.ipynb,sha256=5A-N5A4v4lcXNptOAeI4i7i9Gx66To-Yus8B816dHBk,1303347
|
24
|
+
docs/regions/index.md,sha256=e4aS_vV2FUFHPc5-Up60Ip8PYBIwT9qkjZcNwaS3JbY,8197
|
25
|
+
docs/tables/index.ipynb,sha256=61I9GwJlOM02Mx5aUtzJpRMh0OkpVlqMuEsO1J2s4go,763784
|
26
|
+
docs/tables/index.md,sha256=MVQpkhcWiFJwhMjfPouRVV0nZIUG-PNwFdspc-E8Xow,4428
|
27
|
+
docs/text-analysis/index.ipynb,sha256=iaup8pcQXGp0ZK3IWi-HHssQLdIzWYGYfvZK5i8yjjg,538024
|
28
|
+
docs/text-analysis/index.md,sha256=02pfZemOgV37izV7H-XzKmHu7AedDKLidQ-sKhYaMVw,3527
|
29
|
+
docs/text-extraction/index.ipynb,sha256=809y9ZamXT3bc3GhwwFyoDnlyEpO-kUZ3tIsZZWyrj8,2537087
|
30
|
+
docs/text-extraction/index.md,sha256=b1KfQpvIEelc8cPbFETUnK92az7iB4b7-LqK2DRH8vw,6985
|
31
|
+
docs/tutorials/01-loading-and-extraction.ipynb,sha256=-9hFAVQtHmuXsR9Ge3A80wKr-t9wxouAAlW4_Iotdwo,544610
|
32
|
+
docs/tutorials/01-loading-and-extraction.md,sha256=g40J8GhKz-ikM2URj5MqIatKKj4l5kTFozHeVjxDJQA,2191
|
33
|
+
docs/tutorials/02-finding-elements.ipynb,sha256=k1CSz47_atA9D6DXfQzVS64t5-L-KjssU2VuFvdy7oU,524374
|
34
|
+
docs/tutorials/02-finding-elements.md,sha256=qOkjcWUzem05of54aKzKvy-MMzRX_S4CyZisVV-73QM,4162
|
35
|
+
docs/tutorials/03-extracting-blocks.ipynb,sha256=1UjdP0j3kPCE3aU8p1jBCBqflG-xRLli2Ltx80DhOVk,260729
|
36
|
+
docs/tutorials/03-extracting-blocks.md,sha256=_kqvhk6rSL7cGp2MSwTJk8LYlJGbK_r_umnCSBdR8XU,1665
|
37
|
+
docs/tutorials/04-table-extraction.ipynb,sha256=u92Wppw1qHG__Mx3ZKtETm4AWuGF8X-Ln3kvmF8zCSo,3973
|
38
|
+
docs/tutorials/04-table-extraction.md,sha256=4q4v17VX8K-ZBtWYy0nbWPccyqB_ybd5Vl_IROmxz6Q,2130
|
39
|
+
docs/tutorials/05-excluding-content.ipynb,sha256=oSg8ll_nuWOfQHGLp0fNKVeyYyn_L8a-F7HJADjjdq8,336857
|
40
|
+
docs/tutorials/05-excluding-content.md,sha256=U52SPlc5knqxiyhRokmxrj06T54r2ENyTfP7BMGykhY,3907
|
41
|
+
docs/tutorials/06-document-qa.ipynb,sha256=Facyqns8jw2bTvsOSbNnsLskFH8kg1JTz4kmJ16dpcE,10303
|
42
|
+
docs/tutorials/06-document-qa.md,sha256=PzPPgw0Rkkfe6sfz3XyKD9S9JbQ40qf4bDzCBvwH1P0,3026
|
43
|
+
docs/tutorials/07-layout-analysis.ipynb,sha256=tdNnMro1V66YPx0h96HZnujSm-zDpy7o78euQix4lyU,559517
|
44
|
+
docs/tutorials/07-layout-analysis.md,sha256=NAYVzJTecDnXjo_isbPCSUBSn3c-xM1tELct1Zn5GmI,2533
|
45
|
+
docs/tutorials/07-working-with-regions.ipynb,sha256=s4BFKKbKUemmURCpg6j91rNI8eFFOJUgxY4QN4alK4I,69584
|
46
|
+
docs/tutorials/07-working-with-regions.md,sha256=oanbTFSQ-topAVd9kjfkaPiMjHcx6Y8cqyxVbmxLhgs,4365
|
47
|
+
docs/tutorials/08-spatial-navigation.ipynb,sha256=jfwF6OHLvrMvaaknp-9AfUvr-pPXjPljUyGnFKF9wsw,194523
|
48
|
+
docs/tutorials/08-spatial-navigation.md,sha256=IMbOYBjayXKE7pHfBjApTxOoKRD8WYj7opf8fsJCtzA,4855
|
49
|
+
docs/tutorials/09-section-extraction.ipynb,sha256=Aqcy08oXTJ1pkJCmVVumndje-4WXnbkl_QfJPhps7f8,1100736
|
50
|
+
docs/tutorials/09-section-extraction.md,sha256=Jy_be8ftAl_VPBWl5nEv7_5sKSZPx22DLUcBVHMD3Nc,7832
|
51
|
+
docs/tutorials/10-form-field-extraction.ipynb,sha256=yyopvBoS5vkKKtUQ6rZ4Kyo5E0Olp2WYnmunhfzSQkQ,281491
|
52
|
+
docs/tutorials/10-form-field-extraction.md,sha256=t9tPlW36vJEhDrKIsHGg_f3P_MK62DT4-ZK1thKFs4Y,5494
|
53
|
+
docs/tutorials/11-enhanced-table-processing.ipynb,sha256=BWpVUhtjaAX7r4OOdiy5gQgrSqREaoB0L5TuHqoHEn8,1278
|
54
|
+
docs/tutorials/11-enhanced-table-processing.md,sha256=2HK-r1UwU7FLn7zWr_pMG7iLk-i0L4U4-t6ubOEeduc,282
|
55
|
+
docs/tutorials/12-ocr-integration.ipynb,sha256=xurkoPwgk2p6mhmPdCehy9ccuYHrAhBCb1zGnjRbZ7Y,26724
|
56
|
+
docs/tutorials/12-ocr-integration.md,sha256=wU90sfnm1R6BoMFq-orbGpl8OUVcm-wEBTlK0bLgJC4,4572
|
57
|
+
docs/tutorials/13-semantic-search.ipynb,sha256=5h806AIal3EwXPVuXJESbXwdUImCx7fo0mo5-f3Dj44,42817
|
58
|
+
docs/tutorials/13-semantic-search.md,sha256=nsNjv0ipYUC3YPSqT5d6dga9ZjObEc04Mc8c0-gsRnU,2914
|
59
|
+
docs/visual-debugging/index.ipynb,sha256=MJ92u3Q9sfRCyDAQM4KWmCrs4QhKwIagbn6ytPF83L4,2175800
|
60
|
+
docs/visual-debugging/index.md,sha256=ueGD2kNFhEAgIHt7qxCfrLRLjHcR7NTD3AU9okBhX9k,4176
|
61
|
+
docs/visual-debugging/region.png,sha256=ULAJs3ZTxMjpD9F4w1DKaZXmhxga3KRq3NrUsXgw28s,67835
|
62
|
+
natural_pdf/__init__.py,sha256=IjeAfnDU9fpkVc9YvU2DltNtBtIsZm8_DUyXGuyHGB8,2669
|
63
|
+
natural_pdf/analyzers/__init__.py,sha256=dIXjsMqoxKmd9OOnSBzn12wvdIz7D7YNQRAnXslpJSM,142
|
64
|
+
natural_pdf/analyzers/text_options.py,sha256=nE2E1pp4psDPpxmtarvNtEQsgozPkyFRjv0TVP2HTyU,2865
|
65
|
+
natural_pdf/analyzers/text_structure.py,sha256=9h8hKRz0JWnr13xQr3b4FFr_-hDIjue07WvG7LmT8nc,12827
|
66
|
+
natural_pdf/analyzers/utils.py,sha256=Lgub1kYSTOnNxeLO1klStHLwH-GIuT4vpdqyVRF-Mcg,2084
|
67
|
+
natural_pdf/analyzers/layout/__init__.py,sha256=oq1uJ5UkGGMbBKGirV1aRKK3hxAUyjTLywYkPCQH1f0,33
|
68
|
+
natural_pdf/analyzers/layout/base.py,sha256=9dCR758mAuz7ExlHJ-gwnPnETaM4GZV3W1IRei_t13s,6815
|
69
|
+
natural_pdf/analyzers/layout/docling.py,sha256=4BJYyNVR6VegZGxyisvNIBBRvVk6YKPyDVs7ZdVfzEU,12676
|
70
|
+
natural_pdf/analyzers/layout/gemini.py,sha256=Dslt6rOar9S-QSahyjjXN7Z0KEp7CoVStAA4tvB06X8,10894
|
71
|
+
natural_pdf/analyzers/layout/layout_analyzer.py,sha256=6aed1qz5jpndOiakXCBRZAcnyG_waeXi3WPuP5fRvh4,14046
|
72
|
+
natural_pdf/analyzers/layout/layout_manager.py,sha256=Vh8EKiszKqjELofxQ1eiVLKVjibyjBsZpLFzTf0_21E,11179
|
73
|
+
natural_pdf/analyzers/layout/layout_options.py,sha256=s7xr4brE3OutE6aYNAi2PniRy1p2w8a342C2xGpvX2s,3777
|
74
|
+
natural_pdf/analyzers/layout/paddle.py,sha256=gTI9ZqNd5-t4H5IByGfL32WgcE6JrdchW6jRiGI6ulM,13375
|
75
|
+
natural_pdf/analyzers/layout/surya.py,sha256=vhji6ynHPMyQLHuYRPQcplNi7m_lG4P4NYtWv6MzcME,13556
|
76
|
+
natural_pdf/analyzers/layout/tatr.py,sha256=-GJhMy4d0yx6egkO9-ULAIdQkkQRyAKExoIta-b256U,12971
|
77
|
+
natural_pdf/analyzers/layout/yolo.py,sha256=ANo2U4EZgeN2eYKM1bZIuysiuJLgwl4JeQchrRxOKwA,8388
|
78
|
+
natural_pdf/collections/pdf_collection.py,sha256=Qfauo7fskmw3MSzG0C_OjIGLskyFSQxWQkbHrPDYpno,13240
|
79
|
+
natural_pdf/core/__init__.py,sha256=QC8H4M3KbXwMFiQORZ0pdPlzx1Ix6oKKQSS7Ib2KEaA,38
|
80
|
+
natural_pdf/core/element_manager.py,sha256=rPTkppvU5nQ6Q1_ZtTbcY-KpcRBZV_uBKuNwllHBks0,22100
|
81
|
+
natural_pdf/core/highlighting_service.py,sha256=CTVd7y-fpIreFSe70cTpMu1Pwl6HKMtTHp0bh2U7VXk,32609
|
82
|
+
natural_pdf/core/page.py,sha256=Fw01hvK7ekj3iGpj9YbbsVygCmOtTC3qLNvzGBHS4iQ,84784
|
83
|
+
natural_pdf/core/pdf.py,sha256=dq2bcxbrbY0zS8t3NKkvqi63Oj2nNzaCHUyHz81OoBI,46383
|
84
|
+
natural_pdf/elements/__init__.py,sha256=S8XeiNWJ1WcgnyYKdYV1yxQlAxCCO3FfITT8MQwNbyk,41
|
85
|
+
natural_pdf/elements/base.py,sha256=hgXpcftlvyUOMGZ8EHEu0WhJz2weqcLrCOFgMEhfm-w,35171
|
86
|
+
natural_pdf/elements/collections.py,sha256=Fy5l1Rso-LMunZ6D6Y0n90Uk89h8dzJ2Js7r-8r8NSY,68885
|
87
|
+
natural_pdf/elements/line.py,sha256=7cow3xMUKhAj7zoQz7OaB1eIH2_a8B__LB7iGJ4Mb0o,4612
|
88
|
+
natural_pdf/elements/rect.py,sha256=kiVa3e377ZnqIOXc89d9ZSY4EcmDxtccdtUw-HOQzpw,3796
|
89
|
+
natural_pdf/elements/region.py,sha256=-mBZnWoYMB7cLm5RZhHWCnJkJjsfcgH-f2ESktl-lNk,69656
|
90
|
+
natural_pdf/elements/text.py,sha256=8PNKSLUgXUhEu9IFfbNbSSpuu0Slm11T6UH8jn4O6hQ,11078
|
91
|
+
natural_pdf/exporters/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
92
|
+
natural_pdf/exporters/searchable_pdf.py,sha256=qsaPsnbOOaZHA_aplfZbwQnBoK9KghWm-wzbyRRomeY,16859
|
93
|
+
natural_pdf/ocr/__init__.py,sha256=8EvoA2poYbtkS8Jmbkgn4hFEwOHbN5fIXdJc9tT4NzI,2477
|
94
|
+
natural_pdf/ocr/engine.py,sha256=oTABzt20pEvKadTv1U7zOIv_jSTwv5dty2Q_6vf6HQs,8767
|
95
|
+
natural_pdf/ocr/engine_easyocr.py,sha256=zki3-fd-u01E9VPdaBD0Rkel18Nk9xYBg477MXY0C8w,8487
|
96
|
+
natural_pdf/ocr/engine_paddle.py,sha256=8OYbcYrjIaYop7h7W4R7bXWHKOCYhJQpqT2cVccNb5E,6209
|
97
|
+
natural_pdf/ocr/engine_surya.py,sha256=PtzvDOqxNzhAEeh0rIhTwPXDAbTHMR2IRs0tXb-4pag,4693
|
98
|
+
natural_pdf/ocr/ocr_factory.py,sha256=3fxIpgGi6NzxhfYpl5Kblufyx_-RRoT7dlRGktYjB_Y,4425
|
99
|
+
natural_pdf/ocr/ocr_manager.py,sha256=sMQrpvAaA4PbjhZOf7G_KGW3KBhXhx1l26ig_Xqf_0g,9190
|
100
|
+
natural_pdf/ocr/ocr_options.py,sha256=3njghlnnkcxOL4Td7l6mt-mTEFmL8Z-ryf_BX-EM3i4,3289
|
101
|
+
natural_pdf/ocr/utils.py,sha256=AIBkGQExetjl2wyuQLpWJSy0HSxaOBmuCo2QsKmY7Rc,3404
|
102
|
+
natural_pdf/qa/__init__.py,sha256=Pjo62JTnUNEjGNsC437mvsS5KQ5m7X_BibGvavR9AW0,108
|
103
|
+
natural_pdf/qa/document_qa.py,sha256=W4E4vS_Eox_IBsYpVb0ifQbJb0FP-PYEIG93CU3rUkE,15246
|
104
|
+
natural_pdf/search/__init__.py,sha256=EB_HRwlktJn5WGPVtSaRbOQNjLAZTxujeYf_eN-zd2U,4191
|
105
|
+
natural_pdf/search/haystack_search_service.py,sha256=6RjTFWbTo3gaO-90IF6PEuo_9WRwOdj232eWn3OT0BQ,29270
|
106
|
+
natural_pdf/search/haystack_utils.py,sha256=UI4eu3SVieGR_QnBtLhP8Fjtt2AJgeLgxrpa_dBmD6k,19289
|
107
|
+
natural_pdf/search/search_options.py,sha256=sq_e8_jSROicD94b_xtDtLnjEr_Zsy4icjzPkK0a8QA,3566
|
108
|
+
natural_pdf/search/search_service_protocol.py,sha256=ybNcF_NoLZuIx0rb4XB1dsDl3o_LAaWR1fVVKld2TxI,6818
|
109
|
+
natural_pdf/search/searchable_mixin.py,sha256=M2a6FaFVM0vcfh7FgjDH6BLhS-7ggeVpcfft4OOBDxY,26390
|
110
|
+
natural_pdf/selectors/__init__.py,sha256=oZGeqSv53EqmIZOhcnawuaGGlRg1h79vArXuZCWKm4A,123
|
111
|
+
natural_pdf/selectors/parser.py,sha256=59_GSsTApM6MFvtqhrrmbKaBfODPbGXMluvvQJcrqhE,15754
|
112
|
+
natural_pdf/templates/__init__.py,sha256=jYBxzfi73vew0f6yhIh1MlRxw4F_TVN2hKQR0YXOFe0,20
|
113
|
+
natural_pdf/templates/spa/index.html,sha256=6hLTp07OeV5Q4jUMp5Sgl-dwfBs3oPzBxqphG4kEs24,787
|
114
|
+
natural_pdf/templates/spa/words.txt,sha256=vkGtl5Y7-Nq-3Vhx1daRWWF1Jp1UCVaw-ZZaiFwrurk,2493885
|
115
|
+
natural_pdf/templates/spa/css/style.css,sha256=Qdl0U3L5HMyhBDNzyRPklfb3OxW6rMxCfQbzO8i8IW4,7643
|
116
|
+
natural_pdf/templates/spa/js/app.js,sha256=Efb7NmcTN9RLdLwKpDcU6CG5Ix0laHtzRHmfUlDMJXw,19679
|
117
|
+
natural_pdf/utils/__init__.py,sha256=s3M8FggaK1P3EBYn6R_-HgSDjNc9C73gyKe1hihtNWg,43
|
118
|
+
natural_pdf/utils/debug.py,sha256=56dk0OcDUwjRbYU2g1k0_G_3hfHkAF9Z6vnMVt3myMU,992
|
119
|
+
natural_pdf/utils/highlighting.py,sha256=EIY6ihVGtUTS_DjWyxpnr_UXpcR4btC1KhSGQ9VUfKg,698
|
120
|
+
natural_pdf/utils/identifiers.py,sha256=IvXj2q-NW1cmivWAedDOZBFPTRbVl0_J8BP4phREt9U,1092
|
121
|
+
natural_pdf/utils/packaging.py,sha256=mNPEqCtc0CVWjKg1RcSldYTgspUWQOLAR_ZEllyKxs4,20175
|
122
|
+
natural_pdf/utils/reading_order.py,sha256=s3DsYq_3g_1YA07qhd4BGEjeIRTeyGtnwc_hNtSzwBY,7290
|
123
|
+
natural_pdf/utils/text_extraction.py,sha256=VlbkXg14GlvwYTjRJWa8FVUigETY3Hq0v8NlIRnzYkM,8619
|
124
|
+
natural_pdf/utils/visualization.py,sha256=ir5PgpptRuVuVeRT9IcdTsNeEpdOYD_69rByjHQ7JhI,8592
|
125
|
+
natural_pdf/widgets/__init__.py,sha256=O2fSDo604wDAP6UwUkmBq3eT91RSqHwBpAOQXq92S8s,214
|
126
|
+
natural_pdf/widgets/viewer.py,sha256=Aiw6kuBc0WkhcZrPNKyLNzzWbmtmU6rvOmHV0IuXCBk,40862
|
127
|
+
natural_pdf/widgets/frontend/viewer.js,sha256=w8ywfz_IOAAv2nP_qaf2VBUkF1KhjT3zorhJxM1-CfU,4371
|
128
|
+
natural_pdf-0.1.6.dist-info/licenses/LICENSE,sha256=9zfwINwJlarbDmdh6iJV4QUG54QSJlSAUcnC1YiC_Ns,1074
|
129
|
+
notebooks/Examples.ipynb,sha256=l4YMtMEx_DWBzWIjl9CmBkWTo0g_nK8l_XWOyzYooQM,4275170
|
130
|
+
pdfs/.gitkeep,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
131
|
+
pdfs/01-practice.pdf,sha256=dxWyJIa2cm7bALE3BWDJ2dg3inyFlo1n8ntVyy0hkTo,7906
|
132
|
+
pdfs/0500000US42001.pdf,sha256=VHn5gxlysgD6oJUCndeWXe_RwOrOraO1uIRt_fu8YNY,315883
|
133
|
+
pdfs/0500000US42007.pdf,sha256=pTfu_IVKvHpv9WUyN3QSiGevAOpPZmnr4QL1z-rYQ4E,1168633
|
134
|
+
pdfs/2014 Statistics.pdf,sha256=B-30OQVjqj_3718-G9cGUefNddnz-MosPdHAzfGfkcc,9559
|
135
|
+
pdfs/2019 Statistics.pdf,sha256=reuSJxvAlx9_P-pW7IPqzox0jFCxSPbK1i1-WFu-uGA,511439
|
136
|
+
pdfs/Atlanta_Public_Schools_GA_sample.pdf,sha256=PLBh_uWJQH0MnBaSm5ng5Ima63_m6Mi11CjdravB_S8,137689
|
137
|
+
pdfs/needs-ocr.pdf,sha256=vusKiLxSOlELUTetfZfaotNU54RtMj9PCzGfLc2cuNs,139305
|
138
|
+
natural_pdf-0.1.6.dist-info/METADATA,sha256=Yd92-6FRKc6KdEuTpCOWyxBcewnhAoYTAhtbq6qIB0Q,6668
|
139
|
+
natural_pdf-0.1.6.dist-info/WHEEL,sha256=pxyMxgL8-pra_rKaQ4drOZAegBVuX-G_4nRHjjgWbmo,91
|
140
|
+
natural_pdf-0.1.6.dist-info/top_level.txt,sha256=7nDKUnpkN7B8cBI7DEpW5JM8S7OcOgHw3jXH-1iCX2o,32
|
141
|
+
natural_pdf-0.1.6.dist-info/RECORD,,
|