v0.4.0: Add GUI for Configuring API Key, Model, Temperature, and Custom Prompt

- Implemented a GUI for setting the API Key, model, temperature, and custom prompt in the OmegaT Preferences dialog, replacing the previous system property-based configuration.
- Updated README.md to reflect the new configuration steps using the OmegaT GUI.
- Added screenshots to README to demonstrate the updated configuration process.

v0.3.4

Refactor: Ensure plugin's glossary search logic aligns with OmegaT's …

v0.3.3 - Custom Prompts and Performance Improvements

This release introduces new customization options and performance enhancements to the OpenAI plugin for OmegaT.

Key Features and Improvements:

• Custom User-Defined Prompts: Users can now add their own custom instructions to the translation process by defining a custom prompt via system properties. This allows greater control and flexibility over the translation output.
• Simplified Prompt Structure: The prompt generation logic has been streamlined to reduce complexity, improving performance and ensuring more accurate translations.
• Glossary and Tag Preservation: Continued support for glossary integration and tag preservation, ensuring that translations remain accurate and well-structured.

Performance Optimizations:

• Improved Prompt Efficiency: Addressed the complexity of prompts in previous versions, leading to faster and more reliable translations.
• Enhanced System Prompt Handling: Refined the handling of system prompts to maintain text structure and avoid unnecessary repetitions, improving overall translation quality.

To upgrade, download the latest version and follow the instructions in the updated README to take advantage of the new features.

Retaining v0.3.2 Prompt Behavior:

If you prefer the prompt from v0.3.2, you can achieve this by setting the following custom prompt in your configuration:

-Dcustom.prompt="Translate the following text exactly as it is, even if it seems incomplete or segmented. Handle line breaks appropriately and do not change the structure of the text. Do not repeat the original text. Ensure all tags are preserved and correctly placed in the translation. Do not ask for additional text or clarification."

v0.3.2: feat: improve prompt for better translation accuracy

- Enhanced system prompt to handle line breaks appropriately and maintain text structure.
- Added instructions to ensure all tags are preserved and correctly placed in the translation.
- Included directive to avoid repeating the original text.

v0.3.1: feat: Improve system prompt for handling segmented text

- Updated the system prompt to clarify the context of being integrated in a CAT tool.
- Added instructions to preserve tags and keep any segmentations intact.
- Included guidance to translate the text exactly as it is, even if it seems incomplete or segmented.

v0.3.0: feat: Update README and enhance OpenAITranslate functionality

- Add Key Features section to README
- Improve OpenAITranslate with glossary integration and tag preservation
- Add stub classes for compilation and testing
- Update .gitignore for generated files

v0.2.1

feat: Update system prompt to include tag preservation

• Modified system prompt to instruct the translation to preserve tags while translating from source language to target language

This change ensures that tags are preserved during the translation process, improving the accuracy and quality of the translated content.

v0.2.0: feat: Enhance OpenAITranslate plugin

- Added BaseCachedTranslate.java to utilize built-in caching mechanism
- Replaced custom caching implementation in OpenAITranslate with BaseCachedTranslate
- Updated OpenAITranslate from legacy completions endpoint and gpt-3.5-turbo-instruct model to the latest chat completions endpoint with gpt-4o model
- Removed max_tokens configuration as it is unnecessary; the model operates effectively without explicitly limiting the token count, making the setting redundant
- Configured model name and temperature to be retrieved from system properties
- Separated system and user prompts in the messages array
- Added exception handling to return appropriate error messages for network issues

These changes improve translation request handling by leveraging built-in caching, updating to the latest API and model, and providing clear error messages in case of network failures.

v0.1.2

Fix: Reduced max_tokens from 4000 to 2000 to prevent input and output

token limits from being exceeded.

v0.1.1

Full Changelog: v0.1.0...v0.1.1