arb_translate

A command-line tool for automatically generating missing translations to ARB files using Google Gemini or OpenAI ChatGPT by LeanCode

Installation

$ dart pub global activate arb_translate

Configuration

arb_translate has been designed to seamlessly integrate with Flutter apps using flutter_localizations for code generation from ARB files. Thanks to this integration, the setup process for arb_translate can be completed in just a few steps:

1. Generate your API key. You can create your Gemini key here or your OpenAI key here

2. Save your API token in the environment variable ARB_TRANSLATE_API_KEY or add arb-translate-api-key: {your-api-key} to l10n.yaml in your project.

3. (Optional) Select model used for translation. To do it add arb-translate-model to l10n.yaml or use command argument --model. The default model is gemini-2.5-flash. For all available models see arb_translate --help.

4. (Optional) Add context of your application arb-translate-context: {your-app-context} eg. "sporting goods store app"

All other required parameters match flutter_localizations parameters and will be read from l10n.yaml file. You can override them using command arguments if necessary. See arb_translate --help for more information.

Configuration without l10n.yaml

If you project doesn't include l10n.yaml configuration you have to provide configuration using environment variables and command arguments. You also have to provide:

1. --arb-dir The directory where the template and translated ARB files are located

2. --template-arb-file The template ARB file that will be used as the basis for translation

See arb_translate --help for more information.

Custom model configuration

You can use arb_translate with any model with an OpenAI-compatible API. To configure a custom model:

1. Add arb-translate-custom-model: {your-model-name} to l10n.yaml or specify as command argument custom-model: {your-model-name}
2. Add arb-translate-custom-model-provider-base-url: {your-model-url} to l10n.yaml or specify as command argument: --custom-model-provider-base-url: {your-model-url}
3. (Optional) Set target batch size appropriately to model token count limits by adding arb-translate-batch-size: {size} to your l10n.yaml or specify as command argument batch-size: {size}. Batch size is the number of characters of ARB messages in a single batch and does not include the prompt or app context.

Usage

To generate translations, simply call arb_translate. All messages included in the template ARB file but missing from other files will be translated. To add a new locale, simply add an empty ARB file.

$ arb_translate

Or without l10n.yaml file

$ arb_translate --arb-dir...

Read more

If you want to know how we made this tool and what challenges we had, read the story.

Built with ☕️ by LeanCode