Analysis updated 2026-05-18
Translate a full EPUB novel into Chinese while keeping character names and terms consistent.
Convert a PDF novel into a bilingual EPUB with the original text alongside a Chinese translation.
Resume an interrupted book translation without redoing finished chapters.
| bigdawnghost/wenyi | wxyhgk/retain-pdf | karpathy/arxiv-sanity-lite | |
|---|---|---|---|
| Stars | 1,647 | 1,654 | 1,637 |
| Language | Python | Python | Python |
| Last pushed | — | — | 2023-06-19 |
| Maintenance | — | — | Dormant |
| Setup difficulty | moderate | easy | moderate |
| Complexity | 3/5 | 3/5 | 2/5 |
| Audience | developer | researcher | researcher |
Figures from each repo's GitHub metadata at analysis time.
Requires a DeepSeek API key, and PDF input needs a separate MinerU API key.
Wenyi is a command line tool that translates full length novels into Chinese. It works with EPUB, FB2, TXT, Markdown, HTML, and PDF files, and it can translate from multiple source languages. The project's goal is to get long form machine translation closer to something readable and consistent across an entire book, not just accurate sentence by sentence. The tool needs Python 3.10 or newer and a package manager called uv. After running a sync command and setting an API key for a translation model, a user can point the tool at a book file with a single translate command. By default it writes a Chinese only EPUB back into the book's own output folder, though a bilingual edition with both the source text and the translation side by side can be turned on if wanted. Wenyi tracks its progress in files under a state folder, and it can pick up a run again where it left off if it gets interrupted, rather than starting over. The pipeline behind a translation is more than one pass. It reads the whole book first to build a style guide and a glossary of names and recurring terms, then translates using that context, then optionally polishes and reviews the finished text before treating it as done. Reviewing the completed book is off by default and can be switched on or run separately once a translation is complete. For PDFs specifically, the first conversion uses an outside service called MinerU and needs its own API key, with the converted file cached afterward so later runs do not repeat that step. Wenyi also tries to keep an EPUB's original images, styling, table of contents, and internal links intact while replacing the text. This is described as an early personal project rather than a finished product. The author welcomes bug reports about naming inconsistencies, repeated phrasing, missing text, or formatting issues, and takes pull requests through GitHub. Community support runs through a Discord server and a QQ group. Do not publish copyrighted or private book text, or any state folder holding sensitive information, without permission.
A command line tool that translates full length EPUB, FB2, TXT, Markdown, HTML, and PDF novels into Chinese, using whole book context to keep names and terms consistent.
Mainly Python. The stack also includes Python, uv, DeepSeek API.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.