translatebot-django 0.1.2

Automatic AI-powered translations for Django .po files.

translatebot-django

Automatic AI-powered translations for Django .po files using the power of LLMs.

Translate your Django application's gettext messages automatically using state-of-the-art language models from OpenAI, Anthropic, Google, and many other providers.

Features

• 🤖 Multiple AI Providers: Supports OpenAI, Anthropic Claude, Google Gemini, Azure OpenAI, and many more models via LiteLLM
• 🎯 Smart Translation: Preserves placeholders (%(name)s, {0}, %s) and HTML tags
• ⚙️ Flexible Configuration: Django settings, environment variables, or command-line arguments
• 🔄 Selective Translation: Only translate empty entries or force re-translate all, saving costs
• 🧪 Dry Run Mode: Preview translations before applying them
• ✅ Well Tested: Comprehensive test suite with 100% code coverage
• 🚀 Easy Integration: Simple Django management command

Installation

pip install translatebot-django

Or with poetry:

poetry add translatebot-django

Or with uv:

uv add translatebot-django

Quick Start

1. Add to your Django project:

# settings.py
INSTALLED_APPS = [
    # ...
    'translatebot_django',
]

# Configure API key and model
TRANSLATEBOT_API_KEY = os.getenv("your-api-key-here")
TRANSLATEBOT_MODEL = "gpt-4o-mini"  # Optional, defaults to gpt-4o-mini

2. Generate your .po files (if you haven't already):

python manage.py makemessages -l de  # German
python manage.py makemessages -l fr  # French
python manage.py makemessages -l nl  # Dutch

3. Translate automatically:

python manage.py translate --target-lang nl

4. Compile the translations:

python manage.py compilemessages

Configuration

Django Settings

# settings.py

# Required: Your API key
TRANSLATEBOT_API_KEY = "your-api-key-here"

# Optional: Model to use (default: gpt-4o-mini)
TRANSLATEBOT_MODEL = "gpt-4o-mini"

Usage

Basic Usage

# Translate to French
python manage.py translate --target-lang fr

# Translate to German
python manage.py translate --target-lang de

# Translate to Dutch
python manage.py translate --target-lang nl

Advanced Options

# Preview translations without saving (dry run)
python manage.py translate --target-lang nl --dry-run

# Re-translate entries that already have translations
python manage.py translate --target-lang nl --overwrite

# Use a specific model
python manage.py translate --target-lang nl

Command Options

Option	Description	Default
--target-lang	Target language code (required)	-
--dry-run	Preview without saving	False
--overwrite	Re-translate existing entries	False

Supported Models

Thanks to LiteLLM, you can use models from many providers:

OpenAI

TRANSLATEBOT_MODEL = "gpt-4o-mini"      # Recommended, fast and cheap
TRANSLATEBOT_MODEL = "gpt-4o"           # More capable
TRANSLATEBOT_MODEL = "gpt-4-turbo"      # Good balance

Anthropic Claude

TRANSLATEBOT_MODEL = "claude-3-5-sonnet-20241022"  # Best quality
TRANSLATEBOT_MODEL = "claude-3-5-haiku-20241022"   # Fast and cheap
TRANSLATEBOT_MODEL = "claude-3-opus-20240229"      # Most capable

Google

TRANSLATEBOT_MODEL = "gemini/gemini-1.5-pro"
TRANSLATEBOT_MODEL = "gemini/gemini-1.5-flash"

Azure OpenAI

TRANSLATEBOT_MODEL = "azure/gpt-4o-mini"

See the LiteLLM providers documentation for the complete list.

How It Works

1. Reads your .po files from the locale directory
2. Identifies untranslated entries (empty msgstr)
3. Sends each message to the configured AI model with instructions to:
   • Preserve all placeholders and formatting
   • Maintain HTML tags exactly
   • Return only the translation
4. Updates the .po file with translations
5. Skips obsolete entries automatically

Example Workflow

# 1. Create your translatable strings in code
# myapp/views.py
from django.utils.translation import gettext as _

def my_view(request):
    message = _("Welcome to %(site_name)s!")
    # ...

# 2. Generate .po files
python manage.py makemessages -l fr -l de -l nl

# 3. Translate automatically
python manage.py translate --target-lang fr
python manage.py translate --target-lang de
python manage.py translate --target-lang nl

# 4. Compile messages
python manage.py compilemessages

# 5. Your app is now multilingual! 🎉

Development

Setup

# Clone the repository
git clone https://github.com/gettranslatebot/translatebot-django.git
cd translatebot-django

# Install with dev dependencies using uv
uv sync --extra dev

# Or with pip
pip install -e ".[dev]"

Running Tests

# Run all tests
uv run pytest

# Run with coverage
uv run pytest --cov=translatebot_django --cov-report=html

# Run specific test
uv run pytest tests/test_translate_command.py::TestTranslateText -v

Testing with Different Django Versions

# Test with Django 4.2
uv pip install "Django~=4.2.0"
uv run pytest

# Test with Django 5.2
uv pip install "Django~=5.2.0"
uv run pytest

Requirements

• Python 3.9+
• Django 4.2+
• LiteLLM 1.80.0+

FAQ

How much does it cost?

Costs depend on your chosen model and the amount of text to translate:

• GPT-4o-mini: ~$0.15 per million input tokens (~750k words)
• Claude 3.5 Haiku: ~$0.80 per million input tokens
• Claude 3.5 Sonnet: ~$3 per million input tokens

For a typical small-to-medium app with 500 translatable strings (~10,000 words), costs are usually under $0.01 per language with GPT-4o-mini.

Can I review translations before they're saved?

Yes! Use the --dry-run flag:

python manage.py translate --target-lang nl --dry-run

This shows you all translations without saving them to the .po file.

What if I want to update existing translations?

Use the --overwrite flag:

python manage.py translate --target-lang nl --overwrite

By default, only empty entries are translated.

Does it preserve Django template placeholders?

Yes! The AI is specifically instructed to preserve:

• Named placeholders: %(name)s, %(count)d
• Positional placeholders: %s, %d
• Format strings: {0}, {name}
• HTML tags: <strong>, <a href="...">, etc.

Can I use my own prompts?

Currently, the prompt is built-in to ensure reliable placeholder preservation. If you need custom prompts, please open an issue to discuss your use case.

Which models work best for translation?

Based on testing:

• Best quality: Claude 3.5 Sonnet, GPT-4o
• Best value: GPT-4o-mini, Claude 3.5 Haiku
• Best for Asian languages: GPT-4o, Claude 3.5 Sonnet

Start with GPT-4o-mini for most use cases.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Make your changes
4. Run tests (uv run pytest)
5. Commit your changes (git commit -m 'Add amazing feature')
6. Push to the branch (git push origin feature/amazing-feature)
7. Open a Pull Request

License

This project is licensed under the Mozilla Public License 2.0 - see the LICENSE file for details.

Credits

• Built with LiteLLM for universal LLM provider support
• Uses polib for .po file manipulation
• Inspired by the need to make Django internationalization easier

Support

• 🐛 Issue Tracker

---

Made with ❤️ for the Django community