thundra.core package¶

Submodules¶

thundra.core.graph module¶

class thundra.core.graph.Graph[source]¶

Bases: ABC

Abstract base class representing a graph with nodes and edges.

classmethod combine_graph(*graphs)[source]¶

Combines multiple graph representations into a single graph.

Parameters:: graphs (Graph) – Multiple graph objects to combine.
Returns:: A string representation of the combined graph.
Return type:: str

abstract get_all_names()[source]¶

Abstract method to get all node names in the graph.

Yield:: A generator yielding node names.
Return type:: Generator[str, None, None]

graph()[source]¶

Generates a formatted string representing the graph, including its nodes and edges.

Returns:: A string representation of the graph.
Return type:: str

thundra.core.llm module¶

class thundra.core.llm.LLM[source]¶

Bases: object

A class to manage the lifecycle and availability of a language model (LLM) for chat.

chat_models¶

The chat model to be used.

Type:: BaseChatModel

llm_available¶

An event to signal the availability of the LLM.

Type:: Event

property available: bool¶

Check if the language model is available.

Returns:: True if the language model is available, False otherwise.
Return type:: bool

chat_models: BaseChatModel¶

property llm: BaseChatModel¶

Get the currently set language model. Waits if the model is not yet set.

Returns:: The currently set language model.
Return type:: BaseChatModel

remove_llm()[source]¶: Remove the currently set language model and reset its availability.

thundra.core.memory module¶

class thundra.core.memory.AIMemory(k=13)[source]¶

Bases: object

Manages the memory for multiple users, each with their own conversation buffer.

Parameters:: k (int, optional, default=13) – The default size of the conversation buffer.

clear_history(id)[source]¶

Clears the chat history for a given user while retaining their initial system message.

Parameters:: id (str) – The identifier for the user.

delete_memory(id)[source]¶

Deletes the memory for a given user.

Parameters:: id (str) – The identifier for the user.

get_memory(id)[source]¶

Retrieves the memory for a given user. If the user does not have memory initialized, it creates a new instance.

Parameters:: id (str) – The identifier for the user.
Returns:: The user’s conversation buffer memory.
Return type:: ConversationBufferWindowMemory

class thundra.core.memory.UserMemory(system_message, k, memory)[source]¶

Bases: object

Represents the memory for a user, including the system message and a conversation buffer.

Parameters:

system_message (SystemMessage) – The initial system message.
k (int) – The size of the conversation buffer.
memory (ConversationBufferWindowMemory) – The conversation buffer memory.

clear_history()[source]¶: Clears the chat history while retaining the initial system message.

classmethod create_ai_instance(k)[source]¶

Creates an instance of UserMemory with the specified buffer size.

Parameters:: k (int) – The size of the conversation buffer.
Returns:: A new instance of UserMemory.
Return type:: UserMemory

get_memory()[source]¶

Retrieves the current conversation buffer memory, ensuring its size does not exceed 2*k messages.

Returns:: The conversation buffer memory.
Return type:: ConversationBufferWindowMemory

k: int¶

memory: ConversationBufferWindowMemory¶

system_message: SystemMessage¶

thundra.core.memory.build_system_message()[source]¶

Builds the system message by replacing placeholders in the base system message with corresponding values from the configuration.

Returns:: The formatted system message.
Return type:: str

Module contents¶

class thundra.core.Graph[source]¶

Bases: ABC

Abstract base class representing a graph with nodes and edges.

classmethod combine_graph(*graphs)[source]¶

Combines multiple graph representations into a single graph.

Parameters:: graphs (Graph) – Multiple graph objects to combine.
Returns:: A string representation of the combined graph.
Return type:: str

abstract get_all_names()[source]¶

Abstract method to get all node names in the graph.

Yield:: A generator yielding node names.
Return type:: Generator[str, None, None]

graph()[source]¶

Generates a formatted string representing the graph, including its nodes and edges.

Returns:: A string representation of the graph.
Return type:: str