What is translate-pdf-skill?

translate-pdf-skill is a Python script and accompanying Codex skill definition for translating text-based PDF files into Chinese and producing a new readable Chinese PDF. It is designed for PDFs that already contain a selectable text layer, not scanned images, and prioritises readability of the output over pixel-perfect reproduction of the original layout. The workflow has five steps: extract text from the source PDF using pypdf, clean up line-break artifacts and layout noise, protect product names and acronyms so they are not accidentally translated, translate the text blocks using a translation library with caching to avoid re-translating the same content, and finally rebuild the output as a new PDF using reportlab. Optional flags let you supply a custom cache file, a glossary file for terminology control, a custom title, and a custom font file. The repository also includes minimal unit tests for block parsing and heading detection, a terminology guidance file for protected terms, and example artifacts showing an English founders playbook alongside its generated Chinese equivalent. You would use this tool when you have an English PDF, such as a document, guide, or report, and need a Chinese version quickly for internal review. It is not suited for scanned PDFs without an OCR step, and complex tables or heavily designed pages may need manual cleanup after translation. The script auto-installs the deep-translator package locally if it is not already present.