Asynchronous version of generate_completion. Note: IBM Watson X SDK may not have native async support, so this is a

Generate embeddings using Watson X embedding models. Note: This requires using Watson X embedding models through the embeddings API.

Factory function to get an instance of the RAG agent based on the specified mode. This uses local imports to prevent circular dependencies.

Get all document content for a session (for LLM context)

Get all documents for a session

Get the global PDF processor instance

Processes PDF files: extracts text and stores it in the database. DEPRECATED: This is the old method. Use handle_file_upload instead.

Perform health check on a service.

Return up to *max_items* distinct document titles (or IDs). This is used only for prompt-routing, so we favour robustness over perfec

Custom log format

Emit a nicely-formatted block describing the incoming query and any decomposition.

Process a batch of chunk indices for contextual enrichment

Process a PDF file and store in database

Process items one at a time with minimal memory footprint Args: items: List of items to process proc

Performs a search on a single LanceDB table. If a reranker is provided, it performs a hybrid search. Otherwise, it performs a

Lazily exposes the main (dense) retriever so external components like the ReAct agent tools can call `.retrieve()` directly without re

Start the enhanced API server with a reusable TCP socket.

Generator that yields partial response strings as they arrive. Note: Watson X streaming support depends on the SDK version a