vault-engine 0.1.0__tar.gz

vault_engine-0.1.0/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2026 fishonbike
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

vault_engine-0.1.0/NOTICE

@@ -0,0 +1,6 @@
+vault-engine
+Copyright 2026 fishonbike
+
+Licensed under the Apache License, Version 2.0 (see LICENSE).
+
+This product includes software developed by fishonbike.

vault_engine-0.1.0/PKG-INFO

@@ -0,0 +1,270 @@
+Metadata-Version: 2.4
+Name: vault-engine
+Version: 0.1.0
+Summary: Identity de-identification for cloud-LLM hand-off: local detection, consistent pseudonyms, a reversible local map, and a swappable model backend.
+Author: fishonbike
+License: Apache-2.0
+Project-URL: Source, https://github.com/fishonbike/vault-engine
+Project-URL: Issues, https://github.com/fishonbike/vault-engine/issues
+Keywords: privacy,pii,de-identification,anonymization,redaction,pseudonymization,llm,ollama,qwen,egress
+Classifier: Development Status :: 4 - Beta
+Classifier: Intended Audience :: Developers
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Security
+Classifier: Topic :: Text Processing :: Filters
+Requires-Python: >=3.9
+Description-Content-Type: text/markdown
+License-File: LICENSE
+License-File: NOTICE
+Provides-Extra: dev
+Requires-Dist: pytest; extra == "dev"
+Dynamic: license-file
+
+# vault-engine
+
+**A local-LLM privacy layer for anything you paste into a cloud model.**
+
+Strip the identities out of your text *before* it reaches ChatGPT / Claude /
+Gemini — a model running **on your own machine** finds the names, orgs, places
+and quasi-identifiers, replaces them with stable tokens, and keeps the only
+key-back-to-reality on disk. When the cloud answers in tokens, you put the real
+identities back locally.
+
+> 出云前做身份脱敏：本地模型检测 → 代号化 → 云端用代号分析 → 本地还原真身。
+> 检测**不出本机**，身份映射**只存本地**，大模型**一行换**。零依赖。
+
+<p align="center">
+  <img src="docs/demo.svg" alt="vault-engine scrubs English and Chinese names and PII into tokens before the cloud sees them, then restores them locally" width="760">
+</p>
+
+![CI](https://github.com/fishonbike/vault-engine/actions/workflows/ci.yml/badge.svg)
+&nbsp;·&nbsp; Python ≥3.9 &nbsp;·&nbsp; stdlib-only &nbsp;·&nbsp; Apache-2.0
+
+```text
+# notes.txt  ── private, on your machine
+林若曦是星澜资本的合伙人，在深圳见了字节跳动的陈大壮，邮箱 lin@xinglan.example
+
+        ▼  vault-engine scrub  (local qwen3.6:27b)
+
+# safe.txt  ── what the cloud sees: identities swapped for tokens
+P-n1 是 ORG_1 的合伙人，在 LOC_1 见了 ORG_2 的 P-n2，邮箱 EMAIL_1
+```
+
+---
+
+## Why
+
+You want a frontier cloud model to analyze sensitive notes — but you don't want
+the cloud to learn *who* they're about. Masking only the names you already know
+leaks everything you don't: an unregistered name, an employer, a city + a rare
+title, a project codename. Pattern-based redaction never sees those at all.
+
+`vault-engine` puts a **local model** in front as the detector, so the semantic
+identifiers get caught too — and nothing but the sanitized text ever leaves.
+
+## How it works
+
+```
+ private text                                    cloud model
+      │                                          (sees only tokens)
+      ▼                                                ▲
+┌─────────────────────────── vault-engine ────────────┼───────────┐
+│  ① regex PII detectors  (offline floor)              │           │
+│  ② LLM detector         (local model finds names,    │           │
+│                          orgs, places, quasi-IDs)    │           │
+│  ③ consistent pseudonyms (张三→P-n1, 同名同号)        │           │
+│  ④ residual-risk critic  (re-scan: anything left?)   │  ① send   │
+│        │                                             │           │
+│   sanitized text ────────────────────────────────────┘           │
+│        ▲                                                          │
+│   reverse map (token → real identity) ── stays LOCAL ──┐  ② reply │
+│        └───────────────────── ⑤ rehydrate ◀────────────┘          │
+└──────────────────────────────────────────────────────────────────┘
+      ▼
+ real identities restored locally → use in your own system
+```
+
+## Benchmark
+
+How much identity each detector actually catches, on a labelled bilingual
+dataset (reproduce with `python eval/run_eval.py`; methodology in
+[`eval/`](eval/README.md)):
+
+<!-- BENCHMARK:START -->
+77 gold identities across 15 bilingual documents — easy PII plus hard cases
+(ambiguous common-word names, abbreviations, transliterations, @handles, a badge
+number, a license plate). Reproduce:
+`python eval/run_eval.py --provider ollama --with-presidio`.
+
+> ⚠️ A **small synthetic set** for regression testing and rough comparison —
+> **not** evidence of legal anonymization or complete privacy. "Recall" means
+> flagged-for-redaction; LLM detection is non-deterministic. See the
+> [threat model](#threat-model--limitations-honest).
+
+| detector | person | org | location | project | contact | id | **overall** | over-redaction |
+|---|---|---|---|---|---|---|---|---|
+| regex only | 0% | 0% | 0% | 0% | 69% | 33% | **13%** | 0% |
+| Microsoft Presidio (en/zh `lg`) | 78% | 59% | 80% | 33% | 38% | 0% | **61%** | 4% |
+| **vault-engine (qwen3.6:27b)** | 100% | 100% | 100% | 100% | 100% | 100% | **100%** | 0% |
+
+Same set where Presidio's NER scores 61%, the local LLM clears 100% — gap widest
+on codenames, @handles, IDs, and Chinese names/orgs. Trade-off is speed: Presidio
+~6s, the LLM ~25s/doc.
+<!-- BENCHMARK:END -->
+
+The point isn't a leaderboard — it's the **shape**: pattern-only redaction can't
+see names, organizations, locations, or codenames at all; a local LLM can.
+
+## Install
+
+```bash
+pip install vault-engine
+```
+
+Or get the latest straight from source:
+
+```bash
+pip install git+https://github.com/fishonbike/vault-engine
+```
+
+For the default local backend, install [Ollama](https://ollama.com) and pull a
+model:
+
+```bash
+ollama pull qwen3.6:27b
+```
+
+No model yet? The deterministic floor (emails, phones, IDs, cards, URLs) works
+with zero setup via `--no-llm`.
+
+## Quickstart
+
+```bash
+vault-engine scrub notes.txt -o notes.safe.txt
+```
+
+That writes `notes.safe.txt` (send this to the cloud) and
+`notes.safe.txt.map.json` (**local only** — the identities). Paste the sanitized
+text into your model, save its reply, then restore the real identities:
+
+```bash
+vault-engine rehydrate reply.json --map notes.safe.txt.map.json -o reply.real.json
+```
+
+### The clipboard one-liner
+
+The fastest path — scrub whatever you're about to paste into a chatbot, in place:
+
+```bash
+vault-engine clip               # de-identifies the clipboard
+#   …paste into ChatGPT/Claude, copy its reply, then:
+vault-engine clip --rehydrate   # restores the real identities in the clipboard
+```
+
+Works on macOS, Windows, and Linux (with `xclip`/`xsel`/`wl-clipboard`).
+
+Library:
+
+```python
+from vaultengine import deidentify, rehydrate, Config
+
+result = deidentify(open("notes.txt").read(), Config(model="qwen3.6:27b"))
+send_to_cloud(result.text)                  # tokens only
+restored = rehydrate(get_cloud_reply(), result.vault)   # real identities, locally
+result.vault.save("notes.map.json")         # the reverse map — keep it local
+```
+
+## Use cases
+
+- **Pseudonymize before pasting into ChatGPT/Claude** — analyze private notes,
+  contracts, or chats with direct identifiers stripped.
+- **Redact logs & support tickets** before sharing them or feeding an LLM.
+- **Anonymize a dataset** for LLM-assisted analysis, then map results back.
+- **Air-gapped review loops** — a model on a locked-down box only ever sees
+  tokens.
+
+## How it compares
+
+Presidio and LLM Guard are excellent, mature tools. vault-engine's bet is
+different: a **local LLM** as the detector catches semantic/quasi-identifiers
+that label-based NER misses, with **zero runtime deps** and first-class Chinese.
+
+| | **vault-engine** | Presidio | LLM Guard (Anonymize) | regex / scrubadub |
+|---|---|---|---|---|
+| Detection | local LLM + regex | NER (spaCy) + regex | NER / transformers | patterns only |
+| Unregistered names / orgs / quasi-IDs | ✅ LLM | ⚠️ NER labels only | ⚠️ NER-limited | ❌ |
+| Reversible round-trip | ✅ local map | ✅ deanonymizer | ✅ Vault | ❌ |
+| Fully local / offline | ✅ Ollama | ✅ | ⚠️ varies | ✅ |
+| Runtime dependencies | **none (stdlib)** | spaCy + models | several | varies |
+| Chinese (中文) | ✅ strong | ⚠️ needs model | ⚠️ | ❌ |
+| Swap the model | ✅ one line | — | partial | — |
+| Fail-loud if detector errors | ✅ degrades + non-zero exit | — | — | — |
+
+## Redaction policy (privacy ↔ utility)
+
+| `--policy`  | Persons | Orgs / places / roles | Dates | Token shape |
+|-------------|---------|-----------------------|-------|-------------|
+| `balanced` *(default)* | ✅ | ✅ typed (`ORG_1`, `LOC_2`) | kept | typed |
+| `max`       | ✅ | ✅ opaque `R_1` (type hidden) | coarsened | opaque |
+| `light`     | ✅ | left in place | kept | typed |
+
+`balanced` keeps coarse structure — the cloud still reads "`ORG_1` hired `P-n2`
+as `ROLE_1` in `LOC_1`" and can reason about it, while no real identity ships.
+**Persons are tokenized in every policy.**
+
+## Swap the model
+
+```bash
+vault-engine models                                   # list local Ollama tags
+vault-engine scrub notes.txt --model qwen3.6:35b-a3b  # any local model
+vault-engine scrub notes.txt --provider null          # offline, regex only
+```
+
+Built-in providers: `ollama` (default), `openai-compat` (any OpenAI-style
+endpoint — opt-in; ⚠️ sends raw text to that endpoint), `null` (offline). Add
+your own by implementing one method (`complete`) and registering it.
+
+## ⚠️ Security model — read this
+
+- **The reverse map (`*.map.json`) *is* the identity.** It's the only thing that
+  links tokens back to real people. Keep it local. Never send it to a cloud
+  model, never commit it — `.gitignore` excludes `*.map.json` and the CLI warns
+  every run. Use `--one-way` to produce no map (irreversible publish).
+- **Detection stays local by default.** Only the sanitized text is meant to
+  leave, and only when you send it.
+
+## Threat model & limitations (honest)
+
+- LLM detection is **best-effort, not a guarantee** of non-identifiability — a
+  model can miss a name or a rare quasi-identifier. It is **not** k-anonymity or
+  differential privacy.
+- The critic pass and the risk report reduce and surface residual risk; they
+  don't certify its absence. Writing style and domain-unique facts can still
+  identify with names removed — use `max` for higher-stakes material.
+- If the model backend is unreachable, the run **degrades to regex-only and
+  exits non-zero** (`--allow-degraded` to override) — it will never silently ship
+  under-redacted text.
+
+## Protecting code & schemas (`--format markdown`)
+
+With `--format markdown` (or `auto`, which switches on at a fenced block),
+anything inside fenced code blocks is preserved verbatim — a JSON reply-schema or
+code sample you include for the model survives untouched while the prose around
+it is scrubbed. Pre-existing placeholder tokens (e.g. `P-7`) pass through
+unchanged.
+
+## Development
+
+```bash
+python -m unittest discover -t . -s tests -v   # 59 tests, offline, no model
+python eval/run_eval.py --provider ollama       # reproduce the benchmark
+```
+
+Fully offline and deterministic (null/fake providers); every fixture is
+synthetic — no real data lives in this repo.
+
+## License
+
+Apache-2.0 © 2026 fishonbike. See [LICENSE](LICENSE) and [NOTICE](NOTICE).