* refactor: Simplify message resend logic and enhance abort controller handling
- Updated MessageMenubar to streamline message resend functionality.
- Improved abort controller management in BaseProvider and related services.
- Adjusted sendMessage to handle both single and multiple assistant messages.
- Enhanced logging for better debugging and tracking of message flow.
* feat: Enhance message handling and queue management
- Updated Inputbar to include mentions in dispatched messages.
- Introduced appendMessage action to manage message insertion at specific positions in the state.
- Improved sendMessage logic to handle mentions and maintain message order.
- Refactored getTopicQueue to accept options for better queue configuration.
* refactor: Improve abort handling and message operations
- Refactored useMessageOperations to streamline message pausing logic.
- Enhanced abort controller in BaseProvider to handle abort events more effectively.
- Updated OpenAIProvider to utilize new abort handling mechanism.
- Adjusted fetchChatCompletion to set message status based on abort conditions.
- Improved message dispatching in sendMessage for better queue management.
* refactor: Enhance signal promise handling in BaseProvider and OpenAIProvider
- Updated signal handling in BaseProvider to use a structured signalPromise object for better clarity and management.
- Adjusted error handling in OpenAIProvider to correctly catch and throw errors from the signalPromise.
- Improved overall abort handling logic to ensure robust message operations.
* fix:lint
* feat(websearch): implement API key formatting and add WebSearchApiCheckPopup for multiple keys validation
- Introduced a new WebSearchApiCheckPopup component to validate multiple API keys.
- Added formatApiKeys function to standardize API key input.
- Updated WebSearchProviderSetting to utilize the new popup for checking multiple keys.
- Enhanced error handling and user feedback for API key validation.
* feat(settings): enhance API key validation for providers and web search
- Updated ApiCheckPopup to handle both provider and web search API key validation.
- Refactored key checking logic to differentiate between provider and web search types.
- Removed the obsolete WebSearchApiCheckPopup component and integrated its functionality into ApiCheckPopup.
- Adjusted WebSearchProviderSetting to utilize the updated ApiCheckPopup for checking multiple keys.
- Reformatted sections in README.md for better readability.
- Added new development documentation in docs/dev.md to guide setup and installation processes.
- Updated Japanese and Chinese README translations to reference the new development documentation.
- Reorganized scripts to include a new build check that runs tests, type checks, and i18n checks before publishing.
- Restored and structured type checking and linting scripts for better clarity and usability.
- Ensured post-installation dependencies are correctly handled.
- Modified Inputbar to include a new resizeTextArea function for better text area management.
- Adjusted MessageGroup and MessageStream components to remove unnecessary props and streamline message handling.
- Enhanced message dispatching logic in the messages store for improved state management.
- Cleaned up unused imports and code in Message components to enhance readability and maintainability.
* Enhance update error logging and fix duplicate type import
- Improve error logging in AppUpdater with more detailed error information and timestamps
- Remove duplicate MCPServer type import in Inputbar component
* refactor: optimize file loader with switch-case structure
- Added Gemma3 as a vision model.
- Improved functioncall model check logic.
- Introduced a new method to clean tool call arguments, ensuring proper formatting and extraction of parameters.
- Adjusted tool call handling in OpenAIProvider to accommodate new GLM-4-alltools model checks and argument processing.
Introduce isGemmaModel function to identify Gemma models and adjust system instruction handling in GeminiProvider based on model type. Ensure proper message formatting for Gemma models during chat initialization.
- Added NewContextButton component to manage new context creation with a dedicated button.
- Refactored Inputbar to utilize the new NewContextButton, improving code organization.
- Removed deprecated context shortcut handling from Inputbar.
- Enhanced MCPToolsButton visibility logic based on model type.
- Reordered parameters in completions methods across AiProvider, AnthropicProvider, GeminiProvider, and OpenAIProvider to improve consistency.
- Added detailed JSDoc comments for methods to clarify parameter usage and functionality.
- Ensured mcpTools parameter is consistently included in completions method signatures.
- Implemented a method to extract file content from messages, supporting text and document types.
- Updated message parameter handling to include file content when the model does not support files.
- Added detailed JSDoc comments for new methods and existing functionalities for better documentation.
- Implement token counting functionality in both AddAgentPopup and AssistantPromptSettings components.
- Display the token count dynamically as the user types in the prompt text area.
- Refactor text area components to include a styled token count indicator.
This commit enhances the chat navigation buttons with a more intelligent
visibility system that prevents interference with scrolling while maintaining
easy access to navigation controls.
Key improvements:
- Replace static trigger area with dynamic cursor position tracking to allow
unobstructed scrolling
- Show navigation buttons only when cursor is near the button area or when
actively interacting with them
- Add throttled mouse position detection (50ms) for better performance
- Use passive scroll event listeners for smoother scrolling
- Implement smarter auto-hide behavior with 1.5s timeout when cursor leaves
the button area
This change resolves the issue where navigation buttons would interfere with
scrolling when the cursor was in the detection area, creating a more seamless
user experience.
- Added result check after saving markdown files to ensure success messages are only shown when the save operation is successful.
- Standardized error message keys for consistency.
- Added loading state checks in MessageMenubar to prevent actions during loading.
- Updated useMessageOperations to await database updates for message operations.
- Improved error handling for message edits and resends to ensure proper state management.
