ai_deepseek_text_editor-1.0.3

Release Notes: Version 1.0.3

This release includes improvements to error handling, logging, and API request verification in source/operators/autocomplete.py.

Enhancements to Error Handling and Logging

• Added logging for the start of API requests and successful stream connections in the stream_generation method. [1] [2]
• Improved error handling with specific messages for JSON decoding errors, unexpected response structures, and request exceptions.
• Added a print statement for errors encountered during generation in the modal method.

API Request Verification

• Implemented verification for API key and URL settings before making requests.
• Ensured the presence of the OpenAI API key in environment variables before initiating requests.

ai_deepseek_text_editor-1.0.2

DeepSeek Blender Addon v1.0.2 Release Notes

New Features 🚀

• Real-Time Streaming Support
  Experience live code generation with streaming responses directly in Blender's text editor
• Dual-Model Configuration
  Added separate model selection for error fixing (model_name_fix_errors) to complement deepseek-reasoner
• AI Reasoning Visualization
  Integrated deepThink process showing AI's thought process as Python comments

Improvements ✨

• Enhanced Context Awareness
  Automatic scene analysis now includes:
  • Blender version & render engine
  • Object inventory (meshes, cameras, lights)
  • Selection context & render settings
  • Detailed technical specs for all scene elements
• Smart Response Cleaning
  New regex filters for cleaner code output:

  response = re.sub(r'\bNone\b', '', response)
  response = re.sub(r'```\w*\s*', '', response, flags=re.MULTILINE)

• Threaded Operations
  Background processing with Queue system for smooth UI interaction

Configuration Changes ⚙️

• Added model_name_fix_errors preference for error correction workflows
• Expanded prompt templating with new placeholders:

  "{code_context}"  # Current script state
  "{scene_context}" # Full scene analysis

ai_deepseek_text_editor-1.0.1

v1.0.1

• Code structure improvements for better maintainability (no impact on functionality)
• Increased maximum output tokens from 4000 to 8000 for longer AI completions

This update focuses on internal code organization and allows for more extensive code generation. No action required from users - the addon will continue to work as before.

ai_deepseek_text_editor-1.0.0

An experimental add-on, AI-powered code assistant for Blender's Text Editor, integrating with DeepSeek's API. Provides intelligent code completion and error fixing capabilities for Python scripting in Blender.