# kcode-doc-extraction
Standalone synchronous extraction of searchable PDF, Word 97-2003 DOC, DOCX,
spreadsheet, and text-family files into normalized bounded text.
The library accepts complete in-memory bytes plus a filename and media type. It
performs no network access, filesystem access, subprocess invocation, OCR, or
persistence.
Legacy DOC support extracts readable body and textbox text through the
crates.io `unword` parser. Parser errors and ordinary unwinding panics are
contained as sanitized extraction failures, but strongly hostile DOC files
still require a separately bounded worker process for memory, CPU, timeout,
and crash isolation.
See `Documentation.md` for the Rust integration contract and
`Specification.md` for the complete scope and non-goals.