ebook-GPT-translator

jesselau76/ebook-GPT-translator is ranked 9,582 among these 8 peers; tom-doerr/zsh_codex is ranked 9,583.

ebook-GPT-translator Modernized ebook translation toolkit for TXT, EPUB, DOCX, PDF, and optional MOBI input. Supports Codex, Claude Code, and Gemini CLI as local translation providers alongside OpenAI/Azure APIs. This version upgrades the original single-file project into a package with a real CLI, modern OpenAI SDK support, multi-provider support, resume cache, glossary handling, tests, and release-ready packaging. gif What changed in v2 Replaced the legacy global OpenAI API usage with client-based modern SDK integration

Added provider abstraction for , , , and offline Added resume-safe SQLite translation cache and manifest files Added configurable chunk and token limits for long books Added , , and full legacy compatibility Added maintainable package layout under Added unit tests and GitHub Actions CI Preserved the old entrypoint as a compatibility wrapper Supported providers : local Codex CLI using your ChatGPT subscription login : Claude Code CLI using your Anthropic account (models: , , ) : Gemini CLI using your Google account (models: , , , , ) : official OpenAI API : Azure OpenAI deployment

: any OpenAI-compatible endpoint, including Venice. ai-style APIs : offline smoke testing without any API key Supported formats Input: , , , , Optional input: with Output: translated and Installation Optional MOBI support: Optional XLSX glossary support: Quick start Create config: Launch the desktop GUI: Or, after installation: GUI notes: is optional. Leave it empty to use the values currently selected in the GUI. Use a config file when you want to save a repeatable preset for provider, language, context window, and output options

. The field in the GUI is a dropdown that updates automatically when you switch providers (Codex, Claude Code, or Gemini models). The field in the GUI is a common-language dropdown, but you can still type a custom language. The GUI includes a box for style instructions such as The GUI now shows live progress for both blocks and chunked sub-steps, so long files no longer appear frozen during translation. If translation is interrupted, rerun the same file and

settings to continue from cached chunks while rebuilding the same consistency state. The GUI includes a panel showing install/auth status for Codex, Claude Code, and Gemini CLI with one-click login for each. Run a real translation: Run with Codex subscription login: Run with Claude Code: Run with Gemini CLI: Run an offline smoke test: Save OpenAI credentials locally: Default provider notes: The default provider is now The default model is The default reasoning effort is Long-form consistency

is enabled by default with Chapter memory, rolling translated context, and automatic term memory are enabled by default Cross-chapter memory is persisted to a sidecar memory file and reused on reruns You can choose cheaper/faster Codex runs with You can switch models with , for example , , or You can override the rolling context size with The GUI exposes the same provider/model/context settings without requiring long CLI commands Use Codex login status: List available models: Use the legacy

entrypoint: Azure OpenAI example OpenAI-compatible endpoint example Key features skips files when translated outputs already exist The SQLite cache