Metadata-Version: 2.5
Name: jevex
Version: 0.0.1
Summary: Extract typed records from web pages and PDFs using Jev, with LLM fallbacks that learn declarative generators.
Project-URL: Homepage, https://github.com/davidpurkiss/jevex
Project-URL: Repository, https://github.com/davidpurkiss/jevex
Project-URL: Issues, https://github.com/davidpurkiss/jevex/issues
Author: David Purkiss
License-Expression: Apache-2.0
License-File: LICENSE
Keywords: extraction,html,jev,llm,pdf,pydantic,scraping
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Topic :: Text Processing :: Markup :: HTML
Classifier: Typing :: Typed
Requires-Python: >=3.12
Description-Content-Type: text/markdown

# jevex

Extract typed records from web pages and PDFs using [Jev](https://docs.typesafe.ai/introduction), TypeSafe AI's System One model.

jevex narrows each document step by step (document → component → statement → value), asking Jev small atomic questions at every level. Every LLM fallback is turned into a declarative generator, so each run needs fewer LLM calls than the last.

> **Status:** early planning. This release only reserves the package name; there is no usable API yet.

## License

Apache 2.0
