ZDoc
ZDoc is a parser-driven documentation pipeline built around optimization — parsing large, legacy, mixed-language codebases (PL/X, PLAS, C, C++, and Java) in milliseconds so juniors don't need weeks to understand unfamiliar code. It's built as a set of standalone C components: each language parser reads source files and emits JSON on stdout, and downstream renderers turn that JSON into Markdown or HTML docs, similar to Doxygen or JavaDoc but purpose-built for mainframe codebases. Developed during a 2026 summer internship by a team of 6.
- ROLE
- Parser design, C implementation
- TIMELINE
- 2026 — team of 6
- LANGUAGES PARSED
- PL/X · PLAS C/C++ · Java
- STACK
- C
Optimization-first doc generator — parses legacy, mixed-language codebases in milliseconds.
Manual documentation for large, legacy mainframe codebases is slow, error-prone, and inconsistent — junior developers can spend weeks just understanding unfamiliar code. ZDoc solves this with an optimization-first pipeline that parses and documents large codebases in milliseconds, offline by default.
- 01
Optimized C parsers built for speed — documents large, legacy codebases in milliseconds
- 02
Dedicated parser per language (PL/X, PLAS, C/C++, Java), chosen by file extension
- 03
Extracts signatures, descriptions, parameters, and return values from comments into a shared JSON contract
- 04
Builds an expandable module tree rendered as Markdown or a standalone HTML site
- 05
Offline by default — local parsing and extraction, no external calls
- 06
Optional AI mode calls Bob for every symbol, producing a focused prompt and a Mermaid flowchart per function
- 07
Built during a 2026 summer internship by a team of 6