Translator

Summary

The translation API is very simple, it's only a single Translator interface allowing you to translate text to other languages, using completable futures.

import arc.util.Log;
import com.xpdustry.flex.FlexAPI;
import com.xpdustry.flex.translator.RateLimitedException;
import com.xpdustry.flex.translator.UnsupportedLanguageException;
import mindustry.gen.Player;
import java.util.Locale;

final Player player = /* get player */;
final var text = "Hello";
final var sourceLocale = Locale.ENGLISH;
final var targetLocale = Locale.FRENCH;
FlexAPI.get().getTranslator().translate(text, sourceLocale, targetLocale).whenComplete((result, throwable) -> {
    // The translation succeeded, send the result to the player
    if (result != null) {
        player.sendMessage(result); // Bonjour
        return;
    }
    // The translation failed, if its not one of the expected exceptions from the translator, log it
    if (!(throwable instanceof RateLimitedException || throwable instanceof UnsupportedLanguageException)) {
        Log.err(throwable);
    }
    // Send the original text to the player
    player.sendMessage(text);
});

Configuration

The API is simple, but the back-end is very configurable.

Implementations

None

The None translator is the default translator, it will always fail.

"None"

DeepL

DeepL provides high quality translations and a free tier of 500 000 characters. Ideal for small to medium sized servers.

# Your DeepL API key
deepl-api-key: "xxx"

Google Basic

The basic Google translation API, has 500 000 free characters too and covers way more languages than DeepL.

# Your Google API key
google-basic-api-key: "xxx"

LibreTranslate

A self-hosted translator, recommended as a fallback. Provides medium quality translations, which can do the job. But has poor language auto detection capabilities.

# Your LibreTranslate API endpoint
lt-endpoint: "https://translation.example.com/"
# Your LibreTranslate API key (optional if you use a instance)
lt-api-key: "xxx"

Caching

A wrapper translator caching the results of its inner translator, heavily recommended to use.

# How much time a successful translation should be cached (optional)
success-retention: "10m"
# How much time a failed translation should be cached (optional)
failure-retention: "10s"
# The maximum size of cached translations (optional)
maximum-size: 1000
# The wrapped translator whose results are cached
caching-translator: "..."

Rolling

A wrapper translator switching between multiple translators, ideal if you want to abuse free tiers 😈. Has also a fallback translator if all primary translators fail.

# The list of sub-translators
translators:
  - "..."
  - "..."
# The fallback translator (optional)
fallback: "None"

Putting it all together

Here the configuration used by chaotic neutral, it uses all the above back-end in a way that maximizes efficiency.

caching-translator:
  translators:
    - deepl-api-key: "xxx"
    - google-basic-api-key: "xxx"
  fallback:
    lt-endpoint: "https://translate.xpdustry.com/"
    lt-api-key: "xxx"