- Introduced a new function to extract the first state code and clean content by removing state tags.
- Updated the chat endpoint to utilize the new extraction function, improving state handling and content processing.
- Enhanced logging to provide clearer insights into extracted state codes and cleaned content.
- Revised the formUpdate section to clarify the conditions under which structured updates are pushed.
- Added specific examples for accident vehicle count and cause of the incident to enhance understanding.
- Adjusted text_delta examples for improved clarity in response handling.
- Replaced hardcoded URLs with a base URL variable for better configurability.
- Updated request payloads to use sessionId and timeStamp variables.
- Enhanced the chat request with new parameters: needFormUpdate and useTextChunk.
- Adjusted response content types and lengths for consistency.
- Updated timestamps in response headers for accuracy.
- Moved SentenceTextChunker and SentenceTextChunkerConfig to the utils module for better organization.
- Updated pytest.ini to include the current directory in the Python path.
- Added a new utils module with shared utility helpers.
- Adjusted import paths in the test files to reflect the new location of text chunking classes.
- Introduced SentenceTextChunker and SentenceTextChunkerConfig for improved text processing in chat responses.
- Updated chat endpoint to conditionally use text chunking based on the new 'useTextChunk' parameter from the request.
- Enhanced logging to include 'useTextChunk' status and adjusted text delta handling to support chunked responses.
- Modified ProcessRequest_chat model to include 'useTextChunk' field for request handling.
- Added unit tests for SentenceTextChunker to ensure correct chunking behavior and edge case handling.
- Added detailed logging for chat request processing, including session ID, input text length, and latency measurements.
- Implemented performance tracking for streaming events, capturing first event and text delta timings.
- Improved error handling logging to include session context and duration on failures.
- Updated non-stream response logging to include latency and output details for better debugging.
- Introduced a new function to parse JSON values in endpoints.py for improved data handling.
- Updated extract_form_update_from_flow_nodes to return structured data instead of strings.
- Changed formUpdate field in ProcessResponse_chat model to use Any type with a default empty dictionary for better flexibility in handling updates.
- Changed ANALYSIS_SERVICE_URL to localhost for local development.
- Updated ANALYSIS_AUTH_TOKEN and APP_ID for improved security.
- Added new functions in endpoints.py for form extraction and stage code normalization.
- Enhanced chat handling to support form updates and improved event streaming.
- Updated models to include new fields for form update handling.
- Added support for UserStartedSpeakingFrame to enhance user interaction tracking.
- Updated the pipeline to reset idle prompt count when a user starts speaking, improving responsiveness during conversations.
- Integrated new event handlers for better management of user turn events and upstream frame processing.
- Added idle prompt timeout, maximum count, and text to multiple voice configuration files to improve user interaction during idle periods.
- Updated greeting mode to 'fastgpt_opener' in relevant configurations for a more dynamic greeting experience.
- Introduced a new voice configuration file for xfyun TTS, including detailed service settings and parameters.
- Refactored the pipeline to handle idle prompts and user turn events, ensuring smoother interaction flow.
- Adjusted the VAD and turn configurations to accommodate new idle prompt features.
- Introduced a camera drawer for capturing images during the conversation flow.
- Added prompts for various camera states to guide users through the photo capture process.
- Updated HTML structure to include camera-related elements and integrated them with existing chat functionality.
- Enhanced JavaScript logic to manage camera state and button enabling/disabling based on connection status.
- Updated CSS for styling the camera drawer and its components, ensuring responsive design across devices.
- Adjusted README to reflect the new demo URL for voice functionality.
Parse leading <state> tags from LLM replies and emit response.state over the product websocket while stripping tags from TTS/text streams. Add FastGPT+Xfyun voice configs (including state-enabled preset), SuperTTS support, and context sync for interrupted turns. Refresh the voice demo with a state indicator and collapsible audio delta websocket log groups.
Co-authored-by: Cursor <cursoragent@cursor.com>
