memoryscope.core.service

class memoryscope.core.service.BaseMemoryService(memory_operations: ~typing.Dict[str, dict], context: ~memoryscope.core.utils.singleton.singleton.<locals>._singleton, assistant_name: str | None = None, human_name: str | None = None, **kwargs)

Bases: object

An abstract base class for managing memory operations within a multithreaded context. It sets up the infrastructure for operation handling, message storage, and synchronization, along with logging capabilities and customizable configurations.

__init__(memory_operations: ~typing.Dict[str, dict], context: ~memoryscope.core.utils.singleton.singleton.<locals>._singleton, assistant_name: str | None = None, human_name: str | None = None, **kwargs)

Initializes the BaseMemoryService with operation definitions, keys for memory access, and additional keyword arguments for flexibility.

Parameters:

• memory_operations (Dict[str, dict]) – A dictionary defining available memory operations.

• context (MemoryscopeContext) – runtime context.

• human_name (str) – human name.

• assistant_name (str) – assistant name.

• **kwargs – Additional parameters to customize service behavior.

property human_name: str

property assistant_name: str

get_chat_messages_scatter(recent_n_pair: int) → List[Message]

property op_description_dict: Dict[str, str]

Property to retrieve a dictionary mapping operation keys to their descriptions. :returns: A dictionary where keys are operation identifiers and values are their descriptions. :rtype: Dict[str, str]

abstract add_messages_pair(messages: List[Message])

abstract register_operation(name: str, operation_config: dict, **kwargs)

abstract init_service(**kwargs)

start_backend_service(name: str | None = None, **kwargs)

stop_backend_service(wait_service: bool = False)

abstract run_operation(name: str, role_name: str = '', **kwargs)

class memoryscope.core.service.MemoryScopeService(history_msg_count: int = 100, contextual_msg_max_count: int = 10, contextual_msg_min_count: int = 0, **kwargs)

Bases: BaseMemoryService

__init__(history_msg_count: int = 100, contextual_msg_max_count: int = 10, contextual_msg_min_count: int = 0, **kwargs)

init function. :param history_msg_count: The conversation history in memory, control the quantity, and reduce memory usage. :type history_msg_count: int :param contextual_msg_max_count: The maximum context length in a conversation. If it exceeds this length,

it will not be included in the context to prevent token overflow.

Parameters:

• contextual_msg_min_count (int) – The minimum context length in a conversation. If it is shorter than this length, no conversation summary will be made and no long-term memory will be generated.

• kwargs (dict) – Additional parameters to customize service behavior.

add_messages_pair(messages: List[Message])

Adds a list of messages to the chat history, it can be a pair [user_message, assistant_message]. Ensuring the message list remains sorted by creation time and does not exceed the maximum history message count.

Parameters:

• messages (List[Message] | Message) – A single message instance or a list of message instances

• history. (to be added to the chat)

register_operation(name: str, operation_config: dict, **kwargs)

init_service(**kwargs)

run_operation(name: str, role_name: str = '', **kwargs)

Executes a specific operation by its name with provided keyword arguments.

Parameters:

• name (str) – The name of the operation to execute.

• role_name (str) – The name of the operation to execute.

• **kwargs – Keyword arguments for the operation’s execution.

Returns:

The result of the operation execution, if any. Otherwise, None.

Raises:

Warning – If the operation name is not initialized in _operation_dict.

start_backend_service(name: str | None = None, **kwargs)

Start all backend operations.

stop_backend_service(wait_service: bool = False)

Stops all backend operations that are currently running.