Summary

This PR enhances the PreloadMemoryTool class to properly handle multiple text parts within memory events. The implementation now correctly joins text parts from each event to provide comprehensive context to the LLM.

Changes

• Improved text extraction from memory event parts with robust attribute checking
• Enhanced the memory formatting logic to join multiple text parts with spaces
• Added comprehensive error handling for various edge cases (missing attributes, empty collections)
• Updated tests to validate correct handling of multi-part text events
• Maintained clean output by properly filtering non-text content

Implementation Details

The implementation now:

• Properly iterates through all events in each memory entry
• Collects and joins all text parts from each event with proper spacing
• Maintains author attribution and timestamp information
• Handles edge cases gracefully (missing content, empty parts, non-text parts)

Testing

All tests are now passing with the updated implementation. The test suite covers:

• Single text part events
• Multiple text part events
• Mixed content events (text and non-text)
• Events without text content
• Edge cases (missing or empty attributes)

This implementation focuses specifically on improving text part handling as requested in the review feedback.
#129