Enhanced Content Management and AI Integration

[zhuojg]

Key Features

ContentBase Framework

Introducing a new series of crates to replace file-handler:

GenDAM-specific:

• content-base: Core functionality for GenDAM
• content-base-pool: Task pool implementation for GenDAM

General-purpose:

• content-base-context: Task execution context
• content-base-task: File-related task definitions
• content-handler: File processing logic
• content-metadata: File metadata definitions

Expanded File Type Support

• Added support for video, audio, image, raw text, and web pages (direct URL import)
• Implementation: content-base (Rust-side)
• Enhanced TS types for improved Developer Experience
• Refined asset metadata handling:
  • Primary check: File extension
  • Secondary check: File magic number (if type is indeterminate)
• Streamlined asset metadata storage in database:
  • Removed MediaData model
  • Direct string storage of metadata in AssetObject

RAG-based Copilot

• Leveraging Large Language Models (LLMs) for answer generation using retrieved assets

Additional Improvements

• Enhanced support for LLM serving via Ollama
• Qdrant client upgrade
  • Note: Recommend upgrading Qdrant server to 1.10+ for development

Breaking Changes

• Asset metadata storage structure has been modified
• All artifacts should be regenerated

[zhuojg]

#72 fixed.

Use image crate instead of ffmpeg to handle image files

[zhuojg]

Related to #59.

• artifacts.json is more robust.
• Thumbnail information is saved into artifacts.json.