LLM-as-Translator

A simple and flexible Python-based translator using chat models to translate text between user-configured source and target languages. Supports both command-line interactive use and direct programmatic integration.

Features

• Translate text between any two languages by specifying source and target languages.
• Interactive CLI with commands to change languages on the fly and clean exit support.
• Keeps a configurable history of recent translations in memory.
• Uses chat models from langchain for flexible, high-quality translation.
• Environment-variable support for API key management.
• Graceful handling of user interrupts (Ctrl+C).
• Command-line arguments for instant translation or interactive mode.

Installation

git clone https://github.com/hnthap/llm-as-translator
cd llm-as-translator

# You are recommended to create an environment before installing anything
# to avoid conflicts.
pip install -r requirements.txt

Before running, set your Google Gemini API key in .env or input it at runtime if not preset. Read .env.example to know the expected format for .env.

Usage

Run interactively

python translator.py

Follow on-screen instructions to translate text or update source/target languages with commands:

• \source <language> — change the source language.
• \target <language> — change the target language.
• \exit — exit the CLI.

Translate via CLI arguments

python translator.py English French --text "One hundred years in the human realm"

Translate from English to French, returning the result immediately.

Configuration

• source_language and target_language (required): Set the languages for translation.
• model (optional): Model identifier (default: gemini-2.5-flash).
• model_provider (optional): Model provider (default: google_genai).
• max_history: Number of recent translations saved (default: 100).

Planned Improvements

Search for TODO inside translator.py to read planned improvements.

Testing

Unit and integration tests are in progress.

License

Licensed under MIT License. See LICENSE for more.

---

For detailed usage, examples, and development notes, see the code inline documentation.