Skip to content

Interface ZIF_LLM_CLIENT

AI Generated documentation.

Overview

The zif_llm_client interface provides a structured way to interact with a Language Learning Model (LLM) client. It includes public constants defining various roles and public methods for obtaining a new client instance and executing chat requests.

Public Constants

The interface defines four public constants to represent different roles:

  • role_user: Represents a user role.
  • role_system: Represents a system role.
  • role_assistant: Represents an assistant role.
  • role_tool: Represents a tool role.

Public Methods

Class Methods

  • get_client
  • Purpose: Obtains a new client instance based on the provided configuration.
  • Parameters:
    • client_config: Configuration for the client.
    • provider_config: Configuration for the provider.
  • Returns: A reference to an instance of zif_llm_client.
  • Exceptions:
    • zcx_llm_validation: Raised for validation errors.
    • zcx_llm_authorization: Raised for authorization errors.

Instance Methods

  • chat
  • Purpose: Executes a chat request.
  • Parameters:
    • request: A reference to a zif_llm_chat_request object containing the chat request details.

  • Returns: A zllm_response object containing the response from the chat execution.

  • new_request

  • Purpose: Creates a new chat request.
  • Returns: A reference to a new zif_llm_chat_request object.

Dependencies

The interface depends on the following types and exceptions:

  • zllm_role: Type used for role constants.
  • zllm_clnt_config: Type for client configuration.
  • zllm_providers: Type for provider configuration.
  • zif_llm_client: The interface itself, used for returning client instances.
  • zif_llm_chat_request: Type for chat request objects.
  • zllm_response: Type for chat response objects.
  • zcx_llm_validation: Exception for validation errors.
  • zcx_llm_authorization: Exception for authorization errors.