OpenAI-Compatiable Translation provider for XUnity.AutoTranslator

A translation Custom provider for XUnity.AutoTranslator.
This project is compatible with openai API.

Installation

Requirements

• Python 3.8 or later

Dependencies

• fastapi
• uvicorn
• openai

You can install the requirements with:

pip3 install -r requirements.txt

Configuration

CLI-configuration

```bash
usage: main.py [-h] [--base-url BASE_URL] [--api-key API_KEY] [--model-name MODEL_NAME] [--temperature TEMPERATURE] [--max-tokens MAX_TOKENS]
            [--frequency-penalty FREQUENCY_PENALTY] [--presence-penalty PRESENCE_PENALTY] [--host HOST] [--port PORT] [--use-history]
            [--max-history MAX_HISTORY] [--use-latest-history] [--db-type DB_TYPE] [--cache-translation] [--use-cached-translation]
            [--use-latest-records] [--init-latest-records INIT_LATEST_RECORDS] [--postgres-host POSTGRES_HOST] [--postgres-port POSTGRES_PORT]
            [--postgres-user POSTGRES_USER] [--postgres-password POSTGRES_PASSWORD] [--postgres-db POSTGRES_DB] [--sqlite-db-path SQLITE_DB_PATH]
            [--log-file LOG_FILE] [--log-level {DEBUG,INFO,WARNING,ERROR,CRITICAL}] [--task-template TASK_TEMPLATE] [--specify-language]
            [--language-template LANGUAGE_TEMPLATE] [--src-start SRC_START] [--src-end SRC_END] [--tgt-start TGT_START] [--tgt-end TGT_END]
            [--use-system-prompt] [--system-prompt SYSTEM_PROMPT] [--config CONFIG]

Application Configuration CLI

options:
-h, --help            show this help message and exit
--base-url BASE_URL   Base URL for OpenAI API
--api-key API_KEY     openai
--model-name MODEL_NAME
                        OpenAI model name
--temperature TEMPERATURE
                        Model temperature (randomness control)
--max-tokens MAX_TOKENS
                        Maximum number of tokens to generate
--frequency-penalty FREQUENCY_PENALTY
                        Penalty for repeated tokens
--presence-penalty PRESENCE_PENALTY
                        Penalty for new tokens
--host HOST           Server host address
--port PORT           Server port
--use-history         Enable history usage
--max-history MAX_HISTORY
                        Maximum number of history records
--use-latest-history  Use latest history records
--db-type DB_TYPE     Database type to use
--cache-translation   Enable translation caching
--use-cached-translation
                        Use cached translations if available
--use-latest-records  Use latest database records
--init-latest-records INIT_LATEST_RECORDS
                        Number of initial latest records
--postgres-host POSTGRES_HOST
                        PostgreSQL server host
--postgres-port POSTGRES_PORT
                        PostgreSQL server port
--postgres-user POSTGRES_USER
                        PostgreSQL username
--postgres-password POSTGRES_PASSWORD
                        PostgreSQL password
--postgres-db POSTGRES_DB
                        PostgreSQL database name
--sqlite-db-path SQLITE_DB_PATH
                        Path to the SQLite database file
--log-file LOG_FILE   Log file path
--log-level {DEBUG,INFO,WARNING,ERROR,CRITICAL}
                        Logging level
--task-template TASK_TEMPLATE
                        Template for the translation task
--specify-language    Specify source and target languages in the prompt
--language-template LANGUAGE_TEMPLATE
                        Template for specifying languages
--src-start SRC_START
                        Start tag for the source language
--src-end SRC_END     End tag for the source language
--tgt-start TGT_START
                        Start tag for the target language
--tgt-end TGT_END     End tag for the target language
--use-system-prompt   Enable system prompt
--system-prompt SYSTEM_PROMPT
                        System prompt to be used
--config CONFIG       Path to the TOML configuration file
```

For docker, you can change args using Environment variables. Environment variables are listed in Dockerfile.

pre-configuration file

For docker, set CONFIG environment variable to target path, and you mount configuration file.

Usage

Local Usage

1. Install Dependencies
2. Configure the provider in XUnity.AutoTranslator:
   1. Change Provider to 'CustomTranslate' in [Service] section.

          [Service]
          Endpoint=CustomTranslate
      

   2. Set server URL in [Custom] section.

      [Custom]
      Url=http://<server_url>:<server_port>/translate
      

   3. start main.py Recommand args is

   python main.py --model-name <model_name_to_use> --base-url <model_url_to_use> --api-key <api_key_to_use> --use-latest-history --cache-translation --use-cached-translation --use-latest-records --specify-language --use-system-prompt

Docker Usage

docker run

If you want to run it with docker run, use command below.

docker run -d \
  --name xunity \
  -e BASE_URL="https://api.openai.com/v1" \
  -e API_KEY="api_key_here" \
  -e MODEL_NAME="gpt-3.5-turbo" \
  -e SQLITE_DB_PATH="translation.db" \
  -v ./translation.db:translation.db \
  -p 5000:5000 \
  ghcr.io/returntofirst/xunity-autotranslator-openai:latest

This command will server on port 5000.

docker compose

If you want to run this container with docker compose, use docker-compose.yaml to configure container

Kubernetes

If you want to run this container with kubernetes, use k8s-example.yaml to configure container

Features

Translation Provider

• CustomTranslate Endpoint: Integrate with XUnity.AutoTranslator using a custom endpoint that leverages OpenAI's translation capabilities.

Advanced Translation Features

• Translation Caching: Store translated text in a local database to avoid redundant API calls and improve performance.
• Context-Aware Translation: Use previous and latest chat history to provide more contextually accurate translations.
• Configurable Prompts: Customize the system and task prompts used in translation requests via configuration files.
• Supported Languages: Support for multiple source and target languages that LLMs can handle.

API Endpoints

• Translation: Perform translation using OpenAI's API.

  • Endpoint: /translate
  • Method: GET
  • Parameters: to={tgt_lang}&from={src_lang}&text={src_text}
  • Description: Sends a text translation request to the OpenAI API.

• Reset: Reset the chat history to start fresh.

  • Endpoint: /reset
  • Method: GET
  • Description: Clears the current chat history.

TODO

• Docs
  • Installation
  • Requirements
  • Usage
  • Configuration
• Refactor
  • Server
  • Logger
  • Error handling
• UI
  • WebUI
    • Control configs
    • Reset
    • Custom prompts
    • System prompt
    • Task template
    • Translation history
      • Pre-translated dictonary
      • Remove bad translation
• Deployment
  • Github Actions
  • Docker-compose
  • Kubernetes