agent-pdf-workspace 0.1.1__tar.gz

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 (61) hide show
  1. agent_pdf_workspace-0.1.1/.github/workflows/ci.yml +76 -0
  2. agent_pdf_workspace-0.1.1/.gitignore +11 -0
  3. agent_pdf_workspace-0.1.1/CONTRIBUTING.md +11 -0
  4. agent_pdf_workspace-0.1.1/LICENSE +201 -0
  5. agent_pdf_workspace-0.1.1/PKG-INFO +173 -0
  6. agent_pdf_workspace-0.1.1/README.md +148 -0
  7. agent_pdf_workspace-0.1.1/SECURITY.md +29 -0
  8. agent_pdf_workspace-0.1.1/benchmarks/parsebench/README.md +13 -0
  9. agent_pdf_workspace-0.1.1/benchmarks/parsebench/agent_pdf_workspace.py +312 -0
  10. agent_pdf_workspace-0.1.1/benchmarks/parsebench/results/2026-07-15-chart-test.md +53 -0
  11. agent_pdf_workspace-0.1.1/benchmarks/parsebench/results/2026-07-15-test-suite.md +47 -0
  12. agent_pdf_workspace-0.1.1/docs/format-v1.md +54 -0
  13. agent_pdf_workspace-0.1.1/docs/release.md +18 -0
  14. agent_pdf_workspace-0.1.1/pyproject.toml +79 -0
  15. agent_pdf_workspace-0.1.1/scripts/generate_schemas.py +94 -0
  16. agent_pdf_workspace-0.1.1/scripts/no_network/sitecustomize.py +24 -0
  17. agent_pdf_workspace-0.1.1/scripts/smoke_wheel.py +105 -0
  18. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/__init__.py +48 -0
  19. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/_io.py +97 -0
  20. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/assets/opencode/agent_pdf_workspace.ts +237 -0
  21. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/cli.py +322 -0
  22. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/config.py +39 -0
  23. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/errors.py +33 -0
  24. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/extraction/__init__.py +1 -0
  25. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/extraction/geometry.py +647 -0
  26. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/extraction/liteparse_backend.py +152 -0
  27. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/extraction/pdf_inventory.py +166 -0
  28. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/extraction/worker.py +50 -0
  29. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/ids.py +35 -0
  30. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/models.py +195 -0
  31. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/opencode.py +134 -0
  32. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/audit-region-v1.schema.json +53 -0
  33. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/diagnostic-v1.schema.json +45 -0
  34. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/ingest-config-v1.schema.json +89 -0
  35. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/page-v1.schema.json +271 -0
  36. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/page-visual-audit-v1.schema.json +127 -0
  37. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/search-hit-v1.schema.json +73 -0
  38. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/table-v1.schema.json +55 -0
  39. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/text-block-v1.schema.json +78 -0
  40. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/verification-result-v1.schema.json +44 -0
  41. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/visual-description-v1.schema.json +99 -0
  42. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/visual-occurrence-v1.schema.json +65 -0
  43. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/visual-task-v1.schema.json +153 -0
  44. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/schemas/workspace-v1.schema.json +111 -0
  45. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/search.py +287 -0
  46. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/skill/agent-pdf-workspace/SKILL.md +97 -0
  47. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/skill/agent-pdf-workspace/agents/openai.yaml +4 -0
  48. agent_pdf_workspace-0.1.1/src/agent_pdf_workspace/workspace.py +1352 -0
  49. agent_pdf_workspace-0.1.1/tests/conftest.py +233 -0
  50. agent_pdf_workspace-0.1.1/tests/test_bundled_contracts.py +72 -0
  51. agent_pdf_workspace-0.1.1/tests/test_cli.py +220 -0
  52. agent_pdf_workspace-0.1.1/tests/test_golden_fixtures.py +140 -0
  53. agent_pdf_workspace-0.1.1/tests/test_models_and_ids.py +152 -0
  54. agent_pdf_workspace-0.1.1/tests/test_opencode.py +86 -0
  55. agent_pdf_workspace-0.1.1/tests/test_package.py +14 -0
  56. agent_pdf_workspace-0.1.1/tests/test_security.py +326 -0
  57. agent_pdf_workspace-0.1.1/tests/test_semantic_markdown.py +75 -0
  58. agent_pdf_workspace-0.1.1/tests/test_visual_candidates.py +21 -0
  59. agent_pdf_workspace-0.1.1/tests/test_workspace_exploration.py +393 -0
  60. agent_pdf_workspace-0.1.1/tests/test_workspace_ingest.py +437 -0
  61. agent_pdf_workspace-0.1.1/uv.lock +1266 -0
@@ -0,0 +1,76 @@
1
+ name: CI
2
+
3
+ on:
4
+ push:
5
+ pull_request:
6
+
7
+ jobs:
8
+ test:
9
+ strategy:
10
+ fail-fast: false
11
+ matrix:
12
+ os: [ubuntu-latest, macos-latest, windows-latest]
13
+ python: ["3.11", "3.12", "3.13", "3.14"]
14
+ runs-on: ${{ matrix.os }}
15
+ steps:
16
+ - uses: actions/checkout@v4
17
+ - uses: astral-sh/setup-uv@v6
18
+ with:
19
+ python-version: ${{ matrix.python }}
20
+ enable-cache: true
21
+ - if: runner.os == 'Linux'
22
+ run: |
23
+ sudo apt-get update
24
+ sudo apt-get install -y tesseract-ocr tesseract-ocr-deu tesseract-ocr-eng
25
+ tesseract --list-langs | grep -qx deu
26
+ - if: runner.os == 'macOS'
27
+ run: |
28
+ brew install tesseract tesseract-lang
29
+ tesseract --list-langs | grep -qx deu
30
+ - if: runner.os == 'Windows'
31
+ run: |
32
+ choco install tesseract --no-progress -y
33
+ $language = "C:\Program Files\Tesseract-OCR\tessdata\deu.traineddata"
34
+ Invoke-WebRequest -Uri "https://raw.githubusercontent.com/tesseract-ocr/tessdata_fast/87416418657359cb625c412a48b6e1d6d41c29bd/deu.traineddata" -OutFile $language
35
+ if ((Get-FileHash -Algorithm SHA256 $language).Hash -ne "19D219BBB6672C869D20A9636C6816A81EB9A71796CB93EBE0CB1530E2CDB22D") { throw "German OCR data checksum mismatch" }
36
+ echo "C:\Program Files\Tesseract-OCR" >> $env:GITHUB_PATH
37
+ & "C:\Program Files\Tesseract-OCR\tesseract.exe" --list-langs | Select-String -Pattern '^deu$'
38
+ - run: uv sync --frozen
39
+ - run: uv run pytest
40
+ - run: uv run ruff check .
41
+ - run: uv run ruff format --check .
42
+ - run: uv run mypy src/agent_pdf_workspace
43
+ - run: uv run python -m build
44
+
45
+ coverage:
46
+ runs-on: ubuntu-latest
47
+ steps:
48
+ - uses: actions/checkout@v4
49
+ - uses: astral-sh/setup-uv@v6
50
+ with:
51
+ python-version: "3.12"
52
+ enable-cache: true
53
+ - run: sudo apt-get update && sudo apt-get install -y tesseract-ocr tesseract-ocr-deu tesseract-ocr-eng
54
+ - run: uv sync --frozen
55
+ - run: uv run pytest --cov=agent_pdf_workspace --cov-fail-under=80
56
+
57
+ release-smoke:
58
+ runs-on: ubuntu-latest
59
+ strategy:
60
+ fail-fast: false
61
+ matrix:
62
+ python: ["3.11", "3.12", "3.13", "3.14"]
63
+ steps:
64
+ - uses: actions/checkout@v4
65
+ - uses: astral-sh/setup-uv@v6
66
+ with:
67
+ python-version: ${{ matrix.python }}
68
+ enable-cache: true
69
+ - run: uv build
70
+ - name: Install wheel in an empty environment
71
+ run: |
72
+ uv venv --python ${{ matrix.python }} .smoke-venv
73
+ wheel=$(find dist -name '*.whl' -type f -print -quit)
74
+ uv pip install --python .smoke-venv/bin/python "$wheel"
75
+ - name: Run installed CLI without network access
76
+ run: .smoke-venv/bin/python scripts/smoke_wheel.py
@@ -0,0 +1,11 @@
1
+ .DS_Store
2
+ .coverage
3
+ .coverage.*
4
+ .mypy_cache/
5
+ .pytest_cache/
6
+ .ruff_cache/
7
+ .venv/
8
+ __pycache__/
9
+ build/
10
+ dist/
11
+ *.egg-info/
@@ -0,0 +1,11 @@
1
+ # Contributing
2
+
3
+ Use Python 3.11–3.14 and `uv`. Every behavior change starts with a failing test and ends with the
4
+ full test, lint, type, and build gates from the README.
5
+
6
+ Keep the runtime offline and model-free. Do not add URL ingestion, telemetry, automatic link
7
+ following, attachment execution, LLM clients, embedding clients, or hidden network fallbacks.
8
+ Changes to public models require matching JSON Schema and compatibility tests. Changes to the
9
+ bundled skill require validation with the skill creator's `quick_validate.py`.
10
+
11
+ Synthetic fixtures belong in `tests/`; never commit private or copyrighted source PDFs.
@@ -0,0 +1,201 @@
1
+ Apache License
2
+ Version 2.0, January 2004
3
+ http://www.apache.org/licenses/
4
+
5
+ TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
6
+
7
+ 1. Definitions.
8
+
9
+ "License" shall mean the terms and conditions for use, reproduction,
10
+ and distribution as defined by Sections 1 through 9 of this document.
11
+
12
+ "Licensor" shall mean the copyright owner or entity authorized by
13
+ the copyright owner that is granting the License.
14
+
15
+ "Legal Entity" shall mean the union of the acting entity and all
16
+ other entities that control, are controlled by, or are under common
17
+ control with that entity. For the purposes of this definition,
18
+ "control" means (i) the power, direct or indirect, to cause the
19
+ direction or management of such entity, whether by contract or
20
+ otherwise, or (ii) ownership of fifty percent (50%) or more of the
21
+ outstanding shares, or (iii) beneficial ownership of such entity.
22
+
23
+ "You" (or "Your") shall mean an individual or Legal Entity
24
+ exercising permissions granted by this License.
25
+
26
+ "Source" form shall mean the preferred form for making modifications,
27
+ including but not limited to software source code, documentation
28
+ source, and configuration files.
29
+
30
+ "Object" form shall mean any form resulting from mechanical
31
+ transformation or translation of a Source form, including but
32
+ not limited to compiled object code, generated documentation,
33
+ and conversions to other media types.
34
+
35
+ "Work" shall mean the work of authorship, whether in Source or
36
+ Object form, made available under the License, as indicated by a
37
+ copyright notice that is included in or attached to the work
38
+ (an example is provided in the Appendix below).
39
+
40
+ "Derivative Works" shall mean any work, whether in Source or Object
41
+ form, that is based on (or derived from) the Work and for which the
42
+ editorial revisions, annotations, elaborations, or other modifications
43
+ represent, as a whole, an original work of authorship. For the purposes
44
+ of this License, Derivative Works shall not include works that remain
45
+ separable from, or merely link (or bind by name) to the interfaces of,
46
+ the Work and Derivative Works thereof.
47
+
48
+ "Contribution" shall mean any work of authorship, including
49
+ the original version of the Work and any modifications or additions
50
+ to that Work or Derivative Works thereof, that is intentionally
51
+ submitted to Licensor for inclusion in the Work by the copyright owner
52
+ or by an individual or Legal Entity authorized to submit on behalf of
53
+ the copyright owner. For the purposes of this definition, "submitted"
54
+ means any form of electronic, verbal, or written communication sent
55
+ to the Licensor or its representatives, including but not limited to
56
+ communication on electronic mailing lists, source code control systems,
57
+ and issue tracking systems that are managed by, or on behalf of, the
58
+ Licensor for the purpose of discussing and improving the Work, but
59
+ excluding communication that is conspicuously marked or otherwise
60
+ designated in writing by the copyright owner as "Not a Contribution."
61
+
62
+ "Contributor" shall mean Licensor and any individual or Legal Entity
63
+ on behalf of whom a Contribution has been received by Licensor and
64
+ subsequently incorporated within the Work.
65
+
66
+ 2. Grant of Copyright License. Subject to the terms and conditions of
67
+ this License, each Contributor hereby grants to You a perpetual,
68
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
69
+ copyright license to reproduce, prepare Derivative Works of,
70
+ publicly display, publicly perform, sublicense, and distribute the
71
+ Work and such Derivative Works in Source or Object form.
72
+
73
+ 3. Grant of Patent License. Subject to the terms and conditions of
74
+ this License, each Contributor hereby grants to You a perpetual,
75
+ worldwide, non-exclusive, no-charge, royalty-free, irrevocable
76
+ (except as stated in this section) patent license to make, have made,
77
+ use, offer to sell, sell, import, and otherwise transfer the Work,
78
+ where such license applies only to those patent claims licensable
79
+ by such Contributor that are necessarily infringed by their
80
+ Contribution(s) alone or by combination of their Contribution(s)
81
+ with the Work to which such Contribution(s) was submitted. If You
82
+ institute patent litigation against any entity (including a
83
+ cross-claim or counterclaim in a lawsuit) alleging that the Work
84
+ or a Contribution incorporated within the Work constitutes direct
85
+ or contributory patent infringement, then any patent licenses
86
+ granted to You under this License for that Work shall terminate
87
+ as of the date such litigation is filed.
88
+
89
+ 4. Redistribution. You may reproduce and distribute copies of the
90
+ Work or Derivative Works thereof in any medium, with or without
91
+ modifications, and in Source or Object form, provided that You
92
+ meet the following conditions:
93
+
94
+ (a) You must give any other recipients of the Work or
95
+ Derivative Works a copy of this License; and
96
+
97
+ (b) You must cause any modified files to carry prominent notices
98
+ stating that You changed the files; and
99
+
100
+ (c) You must retain, in the Source form of any Derivative Works
101
+ that You distribute, all copyright, patent, trademark, and
102
+ attribution notices from the Source form of the Work,
103
+ excluding those notices that do not pertain to any part of
104
+ the Derivative Works; and
105
+
106
+ (d) If the Work includes a "NOTICE" text file as part of its
107
+ distribution, then any Derivative Works that You distribute must
108
+ include a readable copy of the attribution notices contained
109
+ within such NOTICE file, excluding those notices that do not
110
+ pertain to any part of the Derivative Works, in at least one
111
+ of the following places: within a NOTICE text file distributed
112
+ as part of the Derivative Works; within the Source form or
113
+ documentation, if provided along with the Derivative Works; or,
114
+ within a display generated by the Derivative Works, if and
115
+ wherever such third-party notices normally appear. The contents
116
+ of the NOTICE file are for informational purposes only and
117
+ do not modify the License. You may add Your own attribution
118
+ notices within Derivative Works that You distribute, alongside
119
+ or as an addendum to the NOTICE text from the Work, provided
120
+ that such additional attribution notices cannot be construed
121
+ as modifying the License.
122
+
123
+ You may add Your own copyright statement to Your modifications and
124
+ may provide additional or different license terms and conditions
125
+ for use, reproduction, or distribution of Your modifications, or
126
+ for any such Derivative Works as a whole, provided Your use,
127
+ reproduction, and distribution of the Work otherwise complies with
128
+ the conditions stated in this License.
129
+
130
+ 5. Submission of Contributions. Unless You explicitly state otherwise,
131
+ any Contribution intentionally submitted for inclusion in the Work
132
+ by You to the Licensor shall be under the terms and conditions of
133
+ this License, without any additional terms or conditions.
134
+ Notwithstanding the above, nothing herein shall supersede or modify
135
+ the terms of any separate license agreement you may have executed
136
+ with Licensor regarding such Contributions.
137
+
138
+ 6. Trademarks. This License does not grant permission to use the trade
139
+ names, trademarks, service marks, or product names of the Licensor,
140
+ except as required for reasonable and customary use in describing the
141
+ origin of the Work and reproducing the content of the NOTICE file.
142
+
143
+ 7. Disclaimer of Warranty. Unless required by applicable law or
144
+ agreed to in writing, Licensor provides the Work (and each
145
+ Contributor provides its Contributions) on an "AS IS" BASIS,
146
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
147
+ implied, including, without limitation, any warranties or conditions
148
+ of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
149
+ PARTICULAR PURPOSE. You are solely responsible for determining the
150
+ appropriateness of using or redistributing the Work and assume any
151
+ risks associated with Your exercise of permissions under this License.
152
+
153
+ 8. Limitation of Liability. In no event and under no legal theory,
154
+ whether in tort (including negligence), contract, or otherwise,
155
+ unless required by applicable law (such as deliberate and grossly
156
+ negligent acts) or agreed to in writing, shall any Contributor be
157
+ liable to You for damages, including any direct, indirect, special,
158
+ incidental, or consequential damages of any character arising as a
159
+ result of this License or out of the use or inability to use the
160
+ Work (including but not limited to damages for loss of goodwill,
161
+ work stoppage, computer failure or malfunction, or any and all
162
+ other commercial damages or losses), even if such Contributor
163
+ has been advised of the possibility of such damages.
164
+
165
+ 9. Accepting Warranty or Additional Liability. While redistributing
166
+ the Work or Derivative Works thereof, You may choose to offer,
167
+ and charge a fee for, acceptance of support, warranty, indemnity,
168
+ or other liability obligations and/or rights consistent with this
169
+ License. However, in accepting such obligations, You may act only
170
+ on Your own behalf and on Your sole responsibility, not on behalf
171
+ of any other Contributor, and only if You agree to indemnify,
172
+ defend, and hold each Contributor harmless for any liability
173
+ incurred by, or claims asserted against, such Contributor by reason
174
+ of your accepting any such warranty or additional liability.
175
+
176
+ END OF TERMS AND CONDITIONS
177
+
178
+ APPENDIX: How to apply the Apache License to your work.
179
+
180
+ To apply the Apache License to your work, attach the following
181
+ boilerplate notice, with the fields enclosed by brackets "[]"
182
+ replaced with your own identifying information. (Don't include
183
+ the brackets!) The text should be enclosed in the appropriate
184
+ comment syntax for the file format. We also recommend that a
185
+ file or class name and description of purpose be included on the
186
+ same "printed page" as the copyright notice for easier
187
+ identification within third-party archives.
188
+
189
+ Copyright 2026 Dark Light
190
+
191
+ Licensed under the Apache License, Version 2.0 (the "License");
192
+ you may not use this file except in compliance with the License.
193
+ You may obtain a copy of the License at
194
+
195
+ http://www.apache.org/licenses/LICENSE-2.0
196
+
197
+ Unless required by applicable law or agreed to in writing, software
198
+ distributed under the License is distributed on an "AS IS" BASIS,
199
+ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
200
+ See the License for the specific language governing permissions and
201
+ limitations under the License.
@@ -0,0 +1,173 @@
1
+ Metadata-Version: 2.4
2
+ Name: agent-pdf-workspace
3
+ Version: 0.1.1
4
+ Summary: Offline, agent-oriented PDF exploration workspaces
5
+ Author-email: Dark Light <darklight@noreply.com>
6
+ License-Expression: Apache-2.0
7
+ License-File: LICENSE
8
+ Keywords: agents,document-processing,ocr,offline,pdf
9
+ Classifier: Development Status :: 3 - Alpha
10
+ Classifier: License :: OSI Approved :: Apache Software License
11
+ Classifier: Programming Language :: Python :: 3
12
+ Classifier: Programming Language :: Python :: 3.11
13
+ Classifier: Programming Language :: Python :: 3.12
14
+ Classifier: Programming Language :: Python :: 3.13
15
+ Classifier: Programming Language :: Python :: 3.14
16
+ Requires-Python: <3.15,>=3.11
17
+ Requires-Dist: liteparse<2.7,>=2.6
18
+ Requires-Dist: pdfplumber<0.12,>=0.11
19
+ Requires-Dist: pydantic<3,>=2.11
20
+ Requires-Dist: pypdf<7,>=5.7
21
+ Requires-Dist: pyyaml<7,>=6
22
+ Requires-Dist: regex<2027,>=2024.11
23
+ Requires-Dist: typer<1,>=0.16
24
+ Description-Content-Type: text/markdown
25
+
26
+ # agent-pdf-workspace
27
+
28
+ `agent-pdf-workspace` converts one local PDF into a persistent, agent-oriented workspace. Text,
29
+ layout, tables, OCR, metadata, exact/regex/full-text search, rendering, and integrity checks run
30
+ locally. The package contains no LLM, vision-model, embedding, or network client.
31
+
32
+ An external agent can inspect queued image crops with its own vision capability and return short,
33
+ schema-validated descriptions. PDF content is always treated as untrusted data.
34
+
35
+ ## Install
36
+
37
+ Python 3.11–3.14 is supported.
38
+
39
+ ```bash
40
+ uv tool install .
41
+ # or, for development from this checkout
42
+ uv sync
43
+ uv run pdfws --version
44
+ ```
45
+
46
+ LiteParse supplies local PDF parsing and OCR. Check the OCR languages available on the machine when
47
+ using non-English scans:
48
+
49
+ ```bash
50
+ tesseract --list-langs
51
+ ```
52
+
53
+ In `auto` mode the package first inspects native extraction and sends only image-dominated or
54
+ text-empty complex pages through local OCR; both layers retain their provenance when merged.
55
+
56
+ ## Quick start
57
+
58
+ ```bash
59
+ pdfws ingest report.pdf --target ./report-workspace --ocr auto --language deu+eng --json
60
+ pdfws inspect ./report-workspace --json
61
+ pdfws search ./report-workspace "operating income" --json
62
+ pdfws read ./report-workspace --page 12 --json
63
+ pdfws visuals next ./report-workspace --json
64
+ pdfws verify ./report-workspace --json
65
+ ```
66
+
67
+ For encrypted input, provide the password through standard input so it is not exposed in process
68
+ arguments or persisted:
69
+
70
+ ```bash
71
+ printf '%s\n' "$PDF_PASSWORD" | pdfws ingest protected.pdf --target ./workspace --password-stdin
72
+ ```
73
+
74
+ ### Visual descriptions
75
+
76
+ `pdfws visuals next` returns either a `describe_region` task or an annotated `audit_page` task.
77
+ For a region, have the calling agent inspect `asset_path` and submit JSON like:
78
+
79
+ ```json
80
+ {
81
+ "visual_id": "visual-p0001-…",
82
+ "crop_sha256": "…",
83
+ "kind": "chart",
84
+ "decorative": false,
85
+ "short_description": "Revenue rises through Q3 and dips slightly in Q4.",
86
+ "text_in_visual": "| Quarter | Revenue (EUR m) |\n|---|---:|\n| Q1 | 12 |\n| Q2 | 15 |",
87
+ "agent_id": "document-agent",
88
+ "model_id": "vision-model"
89
+ }
90
+ ```
91
+
92
+ ```bash
93
+ pdfws visuals submit ./workspace --task-id TASK --input description.json --json
94
+ ```
95
+
96
+ For charts and plots, `text_in_visual` should be a Markdown table with one row per visible data
97
+ point and explicit series/category, value, and unit columns. Preserve signs, decimals, years, and
98
+ label/value associations. Mark estimates as estimates instead of inventing precision. To correct an
99
+ already submitted description, recheck the crop and existing JSON and pass `--replace`; a
100
+ formatting-only revision must not recompute or change verified values.
101
+
102
+ Page audits can add bboxes missed by deterministic raster/vector detection; every added region then
103
+ becomes its own description task. Accepted descriptions are persisted as both readable Markdown
104
+ and a versioned JSON sidecar under `visuals/descriptions/`.
105
+
106
+ ## Python API
107
+
108
+ ```python
109
+ from agent_pdf_workspace import Workspace
110
+
111
+ workspace = Workspace.ingest(
112
+ "report.pdf",
113
+ "report-workspace",
114
+ ocr="auto",
115
+ ocr_languages=("deu", "eng"),
116
+ )
117
+ hits = workspace.search("cash flow", mode="fts")
118
+ page = workspace.read_page(hits[0].page)
119
+ crop_path = workspace.render_region(hits[0].page, (72, 90, 520, 420))
120
+ ```
121
+
122
+ Search hits include a stable hit ID, page, page-coordinate bounding box, kind, and snippet. Custom
123
+ ingestion limits supplied with `IngestConfig` are persisted without secrets and remain effective
124
+ when the workspace is reopened.
125
+
126
+ The main public contracts are Pydantic models exported from `agent_pdf_workspace`. Versioned JSON
127
+ Schemas and the bundled agent skill are included in the wheel; export the latter with
128
+ `pdfws skill export TARGET`.
129
+
130
+ ### OpenCode installation
131
+
132
+ OpenCode calls its configurable home `OPENCODE_CONFIG_DIR`. The installer reads that variable
133
+ first and otherwise asks the installed CLI via `opencode debug paths`; it does not assume that
134
+ `~/.config/opencode` is the active directory.
135
+
136
+ ```bash
137
+ pdfws opencode path --json
138
+ pdfws opencode install --json
139
+ ```
140
+
141
+ The install command writes the skill to `<config>/skills/agent-pdf-workspace` and native tools to
142
+ `<config>/tools/agent_pdf_workspace.ts`. Use `--config-dir DIR` for an explicit directory or
143
+ `--force` to replace only those two package-owned artifacts. The OpenCode tools invoke `pdfws`
144
+ without a shell and expose bounded inspect, search, page-read, region-render, and visual-task
145
+ operations.
146
+
147
+ ## Workspace format
148
+
149
+ The source PDF is copied unchanged to `source/original.pdf`. Human-readable Markdown provides
150
+ navigation while JSON sidecars preserve coordinates and provenance. Tables are also emitted as CSV.
151
+ SQLite FTS and rendered query crops live under `cache/` and are regenerable.
152
+
153
+ The layout is OKF-inspired but does not claim compliance with a separate OKF standard. ODT/ODF
154
+ export is intentionally outside version 1.
155
+
156
+ The reproducible local comparison and its limitations are documented in
157
+ [benchmarks/parsebench/results/2026-07-15-test-suite.md](benchmarks/parsebench/results/2026-07-15-test-suite.md).
158
+
159
+ See [docs/format-v1.md](docs/format-v1.md) for the format contract and
160
+ [SECURITY.md](SECURITY.md) for the trust model.
161
+
162
+ ## Development
163
+
164
+ ```bash
165
+ uv sync --python 3.12
166
+ uv run pytest --cov=agent_pdf_workspace --cov-fail-under=80
167
+ uv run ruff check .
168
+ uv run ruff format --check .
169
+ uv run mypy src/agent_pdf_workspace
170
+ uv run python -m build
171
+ ```
172
+
173
+ No public PyPI release or remote repository is created automatically.
@@ -0,0 +1,148 @@
1
+ # agent-pdf-workspace
2
+
3
+ `agent-pdf-workspace` converts one local PDF into a persistent, agent-oriented workspace. Text,
4
+ layout, tables, OCR, metadata, exact/regex/full-text search, rendering, and integrity checks run
5
+ locally. The package contains no LLM, vision-model, embedding, or network client.
6
+
7
+ An external agent can inspect queued image crops with its own vision capability and return short,
8
+ schema-validated descriptions. PDF content is always treated as untrusted data.
9
+
10
+ ## Install
11
+
12
+ Python 3.11–3.14 is supported.
13
+
14
+ ```bash
15
+ uv tool install .
16
+ # or, for development from this checkout
17
+ uv sync
18
+ uv run pdfws --version
19
+ ```
20
+
21
+ LiteParse supplies local PDF parsing and OCR. Check the OCR languages available on the machine when
22
+ using non-English scans:
23
+
24
+ ```bash
25
+ tesseract --list-langs
26
+ ```
27
+
28
+ In `auto` mode the package first inspects native extraction and sends only image-dominated or
29
+ text-empty complex pages through local OCR; both layers retain their provenance when merged.
30
+
31
+ ## Quick start
32
+
33
+ ```bash
34
+ pdfws ingest report.pdf --target ./report-workspace --ocr auto --language deu+eng --json
35
+ pdfws inspect ./report-workspace --json
36
+ pdfws search ./report-workspace "operating income" --json
37
+ pdfws read ./report-workspace --page 12 --json
38
+ pdfws visuals next ./report-workspace --json
39
+ pdfws verify ./report-workspace --json
40
+ ```
41
+
42
+ For encrypted input, provide the password through standard input so it is not exposed in process
43
+ arguments or persisted:
44
+
45
+ ```bash
46
+ printf '%s\n' "$PDF_PASSWORD" | pdfws ingest protected.pdf --target ./workspace --password-stdin
47
+ ```
48
+
49
+ ### Visual descriptions
50
+
51
+ `pdfws visuals next` returns either a `describe_region` task or an annotated `audit_page` task.
52
+ For a region, have the calling agent inspect `asset_path` and submit JSON like:
53
+
54
+ ```json
55
+ {
56
+ "visual_id": "visual-p0001-…",
57
+ "crop_sha256": "…",
58
+ "kind": "chart",
59
+ "decorative": false,
60
+ "short_description": "Revenue rises through Q3 and dips slightly in Q4.",
61
+ "text_in_visual": "| Quarter | Revenue (EUR m) |\n|---|---:|\n| Q1 | 12 |\n| Q2 | 15 |",
62
+ "agent_id": "document-agent",
63
+ "model_id": "vision-model"
64
+ }
65
+ ```
66
+
67
+ ```bash
68
+ pdfws visuals submit ./workspace --task-id TASK --input description.json --json
69
+ ```
70
+
71
+ For charts and plots, `text_in_visual` should be a Markdown table with one row per visible data
72
+ point and explicit series/category, value, and unit columns. Preserve signs, decimals, years, and
73
+ label/value associations. Mark estimates as estimates instead of inventing precision. To correct an
74
+ already submitted description, recheck the crop and existing JSON and pass `--replace`; a
75
+ formatting-only revision must not recompute or change verified values.
76
+
77
+ Page audits can add bboxes missed by deterministic raster/vector detection; every added region then
78
+ becomes its own description task. Accepted descriptions are persisted as both readable Markdown
79
+ and a versioned JSON sidecar under `visuals/descriptions/`.
80
+
81
+ ## Python API
82
+
83
+ ```python
84
+ from agent_pdf_workspace import Workspace
85
+
86
+ workspace = Workspace.ingest(
87
+ "report.pdf",
88
+ "report-workspace",
89
+ ocr="auto",
90
+ ocr_languages=("deu", "eng"),
91
+ )
92
+ hits = workspace.search("cash flow", mode="fts")
93
+ page = workspace.read_page(hits[0].page)
94
+ crop_path = workspace.render_region(hits[0].page, (72, 90, 520, 420))
95
+ ```
96
+
97
+ Search hits include a stable hit ID, page, page-coordinate bounding box, kind, and snippet. Custom
98
+ ingestion limits supplied with `IngestConfig` are persisted without secrets and remain effective
99
+ when the workspace is reopened.
100
+
101
+ The main public contracts are Pydantic models exported from `agent_pdf_workspace`. Versioned JSON
102
+ Schemas and the bundled agent skill are included in the wheel; export the latter with
103
+ `pdfws skill export TARGET`.
104
+
105
+ ### OpenCode installation
106
+
107
+ OpenCode calls its configurable home `OPENCODE_CONFIG_DIR`. The installer reads that variable
108
+ first and otherwise asks the installed CLI via `opencode debug paths`; it does not assume that
109
+ `~/.config/opencode` is the active directory.
110
+
111
+ ```bash
112
+ pdfws opencode path --json
113
+ pdfws opencode install --json
114
+ ```
115
+
116
+ The install command writes the skill to `<config>/skills/agent-pdf-workspace` and native tools to
117
+ `<config>/tools/agent_pdf_workspace.ts`. Use `--config-dir DIR` for an explicit directory or
118
+ `--force` to replace only those two package-owned artifacts. The OpenCode tools invoke `pdfws`
119
+ without a shell and expose bounded inspect, search, page-read, region-render, and visual-task
120
+ operations.
121
+
122
+ ## Workspace format
123
+
124
+ The source PDF is copied unchanged to `source/original.pdf`. Human-readable Markdown provides
125
+ navigation while JSON sidecars preserve coordinates and provenance. Tables are also emitted as CSV.
126
+ SQLite FTS and rendered query crops live under `cache/` and are regenerable.
127
+
128
+ The layout is OKF-inspired but does not claim compliance with a separate OKF standard. ODT/ODF
129
+ export is intentionally outside version 1.
130
+
131
+ The reproducible local comparison and its limitations are documented in
132
+ [benchmarks/parsebench/results/2026-07-15-test-suite.md](benchmarks/parsebench/results/2026-07-15-test-suite.md).
133
+
134
+ See [docs/format-v1.md](docs/format-v1.md) for the format contract and
135
+ [SECURITY.md](SECURITY.md) for the trust model.
136
+
137
+ ## Development
138
+
139
+ ```bash
140
+ uv sync --python 3.12
141
+ uv run pytest --cov=agent_pdf_workspace --cov-fail-under=80
142
+ uv run ruff check .
143
+ uv run ruff format --check .
144
+ uv run mypy src/agent_pdf_workspace
145
+ uv run python -m build
146
+ ```
147
+
148
+ No public PyPI release or remote repository is created automatically.
@@ -0,0 +1,29 @@
1
+ # Security policy
2
+
3
+ ## Trust model
4
+
5
+ PDFs are hostile input. Text, metadata, links, annotations, forms, images, QR codes, JavaScript,
6
+ actions, attachments, and OCR results are data and never instructions.
7
+
8
+ The package:
9
+
10
+ - accepts local `.pdf` paths only and makes no network requests;
11
+ - validates file header, size, page count, target structure, bboxes, schemas, and search limits;
12
+ - parses through abortable LiteParse, pypdf, and pdfplumber workers with time and resource caps;
13
+ - blocks network access inside every parser and renderer worker process;
14
+ - inventories active actions and attachments without executing or extracting them;
15
+ - uses parameterized SQLite statements and bounded regex evaluation;
16
+ - keeps raw table cells in JSON/CSV as untrusted data; do not open CSV output in a mode that
17
+ evaluates spreadsheet formulas;
18
+ - never persists a PDF password;
19
+ - prevents concurrent writers and refuses symlink workspace targets;
20
+ - refuses symlinks anywhere inside a reopened workspace;
21
+ - hashes canonical output for later integrity verification.
22
+
23
+ The caller is responsible for deciding whether a visual crop may be sent to an external agent or
24
+ service. The bundled skill requires applying the caller's privacy and authorization policy first.
25
+
26
+ ## Reporting
27
+
28
+ Do not open a public issue containing a sensitive PDF. Report the package version, platform, a
29
+ minimal synthetic reproducer, and the output of `pdfws verify` without secrets or document content.
@@ -0,0 +1,13 @@
1
+ # ParseBench adapter
2
+
3
+ `agent_pdf_workspace.py` is the local ParseBench provider used to normalize a persistent workspace
4
+ to ParseBench Markdown and layout IR. Copy it into ParseBench's parse-provider package, register the
5
+ provider and a pipeline named `agent_pdf_workspace`, and install this checkout in ParseBench's
6
+ environment.
7
+
8
+ Set `AGENT_PDF_WORKSPACE_BENCHMARK_CACHE` to reuse completed workspaces. Visual descriptions are
9
+ read only when their `crop_sha256` still matches the visual occurrence. The package itself does not
10
+ call a vision model; the recorded chart experiment used an external agent through OpenCode.
11
+
12
+ See [results/2026-07-15-chart-test.md](results/2026-07-15-chart-test.md) for the first same-fixture
13
+ comparison with local Docling.