multranslate

English | Русский

Cross-platform terminal user interface based on the Blessed library for simultaneous text translation using several popular translation sources and LLM. All sources do not require an API access token (with the exception of official OpenAI). Supports automatic source and target language definition at code level between English and any of the supported languages, as well as access to translation history via SQLite (up to 500 requests, after which old records from the history are automatically cleared).

Translation providers

• Google - free and unlimited API using serverless hosted on the Vercel platform. Available for translation more than 5000 characters.
• DeepL - free API via DeepLX using serverless hosted on Vercel platform. There are limits on the number of translation requests that can be made frequently, and there may also be a limit on the number of characters that can be used (the official limit is 5000 characters per request).
• Reverso - the most stable, free and without any limitation on the number of characters (version on the site is limited to 2000 characters and 900 in the application, through the API can get up to 8000). Does not contain official documentation, request was received from official site via DevTools.
• MyMemory - free and open API (limit of 500 characters per request). Supports up to 3 response options for short queries.
• OpenAI - use LLM with a preset system prompt through official OpenAI (need to transfer the API key through the parameter that is stored in the file for further use) or LM Studio for the use of local models in offline mode (scheme of API requests and response correspond to OpenAI). It is recommended to choose a model previously trained in the right language.

Install

Use the NPM package manager to install a stable version:

npm install -g multranslate

Or install from the GitHub repository:

npm install -g ttps://github.com/Lifailon/multranslate

Run the application:

multranslate

Get help:

multranslate --help

Usage: multranslate [options]

Cross-platform TUI for translating text in multiple translators simultaneously and LLM, with support for
translation history and automatic language detection.

Options:
  -V, --version            output the version number
  -l, --language <name>    Select the language: ru, ja, zh, ko, ar, tr, uk, sk, pl, de, fr, it, es, el, hu, nl,
  sv, ro, cs, da, pt, vi (default: "ru")
  -t, --translator <name>  Select the translator: all, Google, DeepL, Reverso, MyMemory, OpenAI (default: "all")
  -k, --key <value>        API key for OpenAI (will be saved for future use)
  -s, --server <address>   Server address for OpenAI API or local LLM (default: "https://api.openai.com")
  -m, --model <name>       Select the LLM model (default: "gpt-4o-mini")
  -e, --temp <number>      Select the temperature for LLM (default: 0.7)
  -h, --help               display help for command

Build

Clone the repository:

git clone https://github.com/Lifailon/multranslate
cd multranslate

Install dependencies and run the application:

npm install
npm start

Supported languages

You can change the language for automatic definition between English and any of those presented in the table below:

Parameter	Language
ru	Russian (default)
ja	Japanese
zh	Chinese
ko	Korean
ar	Arabic
tr	Turkish
uk	Ukrainian
sk	Slovak
pl	Polish
de	German
fr	French
it	Italian
es	Spanish
el	Greek
hu	Hungarian
nl	Dutch
sv	Swedish
ro	Romanian
cs	Czech
da	Danish
pt	Portuguese (#1)
vi	Vietnam (#2)

All passed letters are analyzed to compare them between the English alphabet and the language specified in the --language parameter.

You can also use any of the translators individually by specifying the appropriate option at startup:

multranslate --translator Google --language tr	multranslate --translator DeepL --language de
multranslate --translator Reverso --language it	multranslate --translator MyMemory --language es

Hotkeys

• Ctrl+<Enter/S> - translation of text without breaking to a new line.
• Ctrl+V - paste text from the clipboard (defined at the code level).
• Alt+C - copy text from the input field to clipboard.
• Alt+<1/2/3/4/5> - copying translation results from output window to the clipboard (for each translator, the key combination is indicated in brackets), and the selected form will change its color to green.
• Ctrl+<N/Z> - move to the previous entry in the translation history.
• Ctrl+<P/X> - move to the next entry in the translation history.
• Shift+<Up/Down> - simultaneous scrolling of all output panels.
• Ctrl+<Up/Down> - scrolling the text input panel without changing the cursor position.
• Ctrl+<Left/Right> - quick cursor navigation through phrases.
• Ctrl+<A/E> - move the cursor to the ahead or end of text input.
• Ctrl+<C/U/L> - clear the text input field.
• Ctrl+W/Alt+Back - delete the word before the cursor.
• Del/Ctrl+K - deletes one letter or character after the cursor.
• F2 - switch to OpenAI with a preset translation prompt.
• F3 - switch to OpenAI in chat mode.
• Escape - exit the program.

Use the F1 key to get help on available keyboard shortcuts:

Contributing

If your language is not listed or you have problems with translation, please open an issue in the Issues.

You can also offer another source for translating the text through API, which does not require an access key.

If you like to use this interface, you can make a contribution, just translate this readme file to your own language and pass it through Pull Request.

Backlog

• Rewrite code to TypeScript.
• Implement native cursor support (developments in multranslate-native-cursor).
• Check texts for style and grammar (spelling) via LanguageTool.

Text buffer

The Blessed library is great and has no analogues in its functionality for JavaScript or TypeScript (and even surpasses some libraries in other languages), but it is outdated (not supported since 2015) and has a number of technical limitations, for example, there is no ability to navigate the cursor in the text input field. For this reason, a mechanism for managing the input content through a text buffer was implemented, which allows you to use a custom cursor for navigation using keyboard arrows and automatic scrolling.

If you plan to use this library for similar tasks where text input is required, then add class TextBuffer and control of hotkeys (keypress) via inputBox.on() to your code. It took me several weeks to achieve full functionality for working with text, the key problem is the built-in line break (autowrap) and the shift of the custom cursor.

Over the course of daily use for 6 months, I can confirm that it works stably.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Copyright (C) 2024 Lifailon (Alex Kup)