melodypapa/autosar-pdf2 — reverse-engineered prompt

Build me a small Java tool that can read AUTOSAR specification PDFs and turn them into usable structured output.

I want two simple command line modes. One should extract AUTOSAR model info like packages, classes, enums, primitives, attributes, inheritance, defaults, multiplicity, and source page info, then save that as either Markdown or JSON. The other should do a more direct PDF to Markdown conversion for the whole document, with tables kept readable.

Please make it handle messy real PDFs, especially tables that continue across pages, broken words inside cells, and optional title or table only output. I also want basic verbose logging and a way to write logs to a file.

Keep it as a Maven based Java 17 project with tests, and make the commands easy to run on a folder of PDFs or a single PDF file. Include examples and sensible sample output so I can quickly see it working. If anything is unclear, look up current docs for the PDF parsing libraries and make practical choices.