GitHub ⭐

Sessions

Julep Python SDK Index / Julep / Managers / Session

Auto-generated documentation for julep.managers.session module.

AsyncSessionsManager

Show source in session.py:782

A class for managing asynchronous sessions.

This class handles operations related to creating, retrieving, updating, deleting, and interacting with sessions asynchronously. It extends the functionality of BaseSessionsManager with asynchronous behavior.

Attributes

Inherits attributes from the BaseSessionsManager class.

Methods

  • async get(id - Union[UUID, str]) -> Session: Retrieves a session by its ID.

async create(*, user_id: Union[str, UUID], agent_id: Union[str, UUID], situation: Optional[str]=None) -> ResourceCreatedResponse: Creates a new session with the specified user and agent IDs, and an optional situation.

async list(*, limit: Optional[int]=None, offset: Optional[int]=None) -> List[Session]: Lists sessions with an optional limit and offset for pagination.

  • async delete(session_id - Union[str, UUID]): Deletes a session by its ID.

async update(*, session_id: Union[str, UUID], situation: str) -> ResourceUpdatedResponse: Updates the situation for a session by its ID.

async chat(*, session_id: str, messages: List[InputChatMlMessage], tools: Optional[List[Tool]]=None, tool_choice: Optional[ToolChoiceOption]=None, frequency_penalty: Optional[float]=None, length_penalty: Optional[float]=None, logit_bias: Optional[Dict[str, Optional[int]]]=None, max_tokens: Optional[int]=None, presence_penalty: Optional[float]=None, repetition_penalty: Optional[float]=None, response_format: Optional[ChatSettingsResponseFormat]=None, seed: Optional[int]=None, stop: Optional[ChatSettingsStop]=None, stream: Optional[bool]=None, temperature: Optional[float]=None, top_p: Optional[float]=None, recall: Optional[bool]=None, remember: Optional[bool]=None) -> ChatResponse: Initiates a chat session with given messages and optional parameters for the chat behavior and output.

async suggestions(*, session_id: Union[str, UUID], limit: Optional[int]=None, offset: Optional[int]=None) -> List[Suggestion]: Retrieves suggestions related to a session optionally limited and paginated.

async history(*, session_id: Union[str, UUID], limit: Optional[int]=None, offset: Optional[int]=None) -> List[ChatMlMessage]: Retrieves the history of messages in a session, optionally limited and paginated.

  • async delete_history(session_id - Union[str, UUID]) -> None:

Notes

The @beartype decorator is used for runtime type checking of the arguments.

Additional methods may be provided by the BaseSessionsManager.

Signature

class AsyncSessionsManager(BaseSessionsManager): ...

See also

AsyncSessionsManager().chat

Show source in session.py:957

Sends a message in an asynchronous chat session and retrieves the response.

This method leverages the messaging interface with various options to adjust the behavior of the chat bot.

Arguments

  • session_id str - The unique identifier for the chat session.

  • messages List[InputChatMlMessage] - A list of chat messages in the session's context.

  • tools Optional[List[Tool]] - A list of tools, if provided, to enhance the chat capabilities.

  • tool_choice Optional[ToolChoiceOption] - A preference for tool selection during the chat.

  • frequency_penalty Optional[float] - Adjusts how much the model should avoid repeating the same line of thought.

  • length_penalty Optional[float] - Penalizes longer responses. logit_bias (Optional[Dict[str, Optional[int]]]): Biases the model's prediction towards or away from certain tokens.

  • max_tokens Optional[int] - The maximum length of the generated response.

  • presence_penalty Optional[float] - Adjusts how much the model should consider new concepts.

  • repetition_penalty Optional[float] - Adjusts how much the model should avoid repeating previous input.

  • response_format Optional[ChatSettingsResponseFormat] - The desired format for the response.

  • seed Optional[int] - A seed used to initialize the model's random number generator.

  • stop Optional[ChatSettingsStop] - Tokens that signify the end of the response.

  • stream Optional[bool] - Whether or not to stream the responses.

  • temperature Optional[float] - Controls randomness in the response generation.

  • top_p Optional[float] - Controls diversity via nucleus sampling.

  • recall Optional[bool] - If true, the model recalls previous messages within the same session.

  • remember Optional[bool] - If true, the model incorporates the context from the previous conversations in the session.

Returns

  • ChatResponse - The response from the chat bot, encapsulating the result of the chat action.

Notes

This function is decorated with @beartype, which enforces type annotations at runtime.

Examples

>>> response = await chat(...)
>>> print(response)

Signature

@beartype
async def chat(
    self,
    session_id: str,
    messages: List[Union[InputChatMlMessageDict, InputChatMlMessage]],
    tools: Optional[List[Union[ToolDict, Tool]]] = None,
    tool_choice: Optional[ToolChoiceOption] = None,
    frequency_penalty: Optional[float] = None,
    length_penalty: Optional[float] = None,
    logit_bias: Optional[Dict[str, Optional[int]]] = None,
    max_tokens: Optional[int] = None,
    presence_penalty: Optional[float] = None,
    repetition_penalty: Optional[float] = None,
    response_format: Optional[
        Union[ChatSettingsResponseFormatDict, ChatSettingsResponseFormat]
    ] = None,
    seed: Optional[int] = None,
    stop: Optional[ChatSettingsStop] = None,
    stream: Optional[bool] = None,
    temperature: Optional[float] = None,
    top_p: Optional[float] = None,
    recall: Optional[bool] = None,
    remember: Optional[bool] = None,
) -> ChatResponse: ...

AsyncSessionsManager().create

Show source in session.py:854

Asynchronously create a resource with the specified user and agent identifiers.

This function wraps an internal _create method and is decorated with beartype for run-time type checking.

Arguments

user_id (Union[str, UUID]): Unique identifier for the user. agent_id (Union[str, UUID]): Unique identifier for the agent.

  • situation Optional[str], optional - Description of the situation, defaults to None.

Returns

  • Session - The created Session object

Raises

  • BeartypeException - If any of the input arguments do not match their expected types. Any exception raised by the internal _create method.

Signature

@beartype
@rewrap_in_class(Session)
async def create(self, **kwargs: SessionCreateArgs) -> Session: ...

See also

AsyncSessionsManager().delete

Show source in session.py:907

Asynchronously delete a session given its ID.

Arguments

session_id (Union[str, UUID]): The unique identifier for the session, which can be either a string or a UUID.

Returns

Coroutine[Any, Any, Any]: A coroutine that, when awaited, completes the deletion process.

Raises

The decorators or the body of the '_delete' method may define specific exceptions that could be raised during the execution. Generally, include any exceptions that are raised by the '_delete' method or by the 'beartype' decorator in this section.

Signature

@beartype
async def delete(self, session_id: Union[str, UUID]): ...

AsyncSessionsManager().delete_history

Show source in session.py:1102

Delete the history of a session asynchronously.

Arguments

session_id (Union[str, UUID]): The unique identifier for the session.

Returns

  • None - The result of the delete operation.

Raises

  • AssertionError - If the session_id is not a valid UUID v4.

Signature

@beartype
async def delete_history(self, session_id: Union[str, UUID]) -> None: ...

AsyncSessionsManager().get

Show source in session.py:826

Asynchronously get a Session object by its identifier.

This method retrieves a Session based on the provided id. It uses an underlying asynchronous '_get' method to perform the operation.

Arguments

id (Union[UUID, str]): The unique identifier of the Session to retrieve. It can be either a string representation or a UUID object.

Returns

  • Session - The retrieved Session object associated with the given id.

Raises

  • TypeError - If the id is not of type UUID or str.

  • ValueError - If the id is not a valid UUID or an invalid string is provided.

  • AnyExceptionRaisedBy_get - Descriptive name of specific exceptions that '_get' might raise, if any. Replace this with the actual exceptions.

Notes

The @beartype decorator is being used to enforce type checking at runtime. This ensures that the argument id is of the correct type (UUID or str) and that the return value is a Session object.

Signature

@beartype
async def get(self, id: Union[UUID, str]) -> Session: ...

AsyncSessionsManager().history

Show source in session.py:1070

Retrieve a history of chat messages based on the session ID, with optional limit and offset.

This function is decorated with 'beartype' for runtime type checking.

Arguments

session_id (Union[str, UUID]): The unique identifier for the chat session.

  • limit Optional[int], optional - The maximum number of chat messages to return. Defaults to None.

  • offset Optional[int], optional - The number of chat messages to skip before starting to collect the history slice. Defaults to None.

Returns

  • List[ChatMlMessage] - A list of chat messages from the history that match the criteria.

Raises

Any exceptions that may be raised by the underlying '_history' method or 'beartype' decorator.

Signature

@beartype
async def history(
    self,
    session_id: Union[str, UUID],
    limit: Optional[int] = None,
    offset: Optional[int] = None,
) -> List[ChatMlMessage]: ...

AsyncSessionsManager().list

Show source in session.py:877

Asynchronously retrieves a list of sessions with optional pagination.

This method utilizes _list_items internally to obtain session data with support for limit and offset parameters. The beartype decorator is used to ensure that the function parameters match the expected types.

Arguments

  • limit Optional[int], optional - The maximum number of sessions to retrieve. Default is None, which retrieves all available sessions.

  • offset Optional[int], optional - The number to skip before starting to collect the response set. Default is None.

Returns

  • List[Session] - A list of Session objects containing session data.

Signature

@beartype
async def list(
    self,
    limit: Optional[int] = None,
    offset: Optional[int] = None,
    metadata_filter: Dict[str, Any] = {},
) -> List[Session]: ...

AsyncSessionsManager().suggestions

Show source in session.py:1038

Retrieve a list of suggestions asynchronously.

This function asynchronously fetches suggestions based on the provided session ID, with optional limit and offset parameters for pagination.

Arguments

session_id (Union[str, UUID]): The session identifier for which suggestions are to be retrieved.

  • limit Optional[int] - The maximum number of suggestions to return. Defaults to None, which means no limit.

  • offset Optional[int] - The number of suggestions to skip before starting to return results. Defaults to None, which means no offset.

Returns

  • List[Suggestion] - A list of Suggestion objects.

Raises

  • Exception - Raises an exception if the underlying _suggestions call fails.

Signature

@beartype
async def suggestions(
    self,
    session_id: Union[str, UUID],
    limit: Optional[int] = None,
    offset: Optional[int] = None,
) -> List[Suggestion]: ...

AsyncSessionsManager().update

Show source in session.py:926

Asynchronously update a resource with the given situation.

This method wraps the private _update method which performs the actual update operation asynchronously.

Arguments

session_id (Union[str, UUID]): The session ID of the resource to update. It can be either a str or a UUID object.

  • situation str - Description of the situation to update the resource with.

Returns

  • Session - The updated Session object

Notes

This function is decorated with @beartype, which will perform runtime type checking on the arguments.

Raises

  • BeartypeCallHintParamViolation - If the session_id or situation arguments do not match their annotated types.

Signature

@beartype
@rewrap_in_class(Session)
async def update(self, **kwargs: SessionUpdateArgs) -> Session: ...

See also

BaseSessionsManager

Show source in session.py:53

A class to manage sessions using base API client methods.

This manager handles CRUD operations and additional actions on the session data, such as chatting and retrieving history or suggestions.

Attributes

  • api_client - The client used for communicating with an API.

Methods

_get(id): Retrieve a specific session by its identifier.

Arguments

id (Union[str, UUID]): The unique identifier for the session.

agent_id (Union[str, UUID]): The unique identifier for the agent. user_id (Optional[Union[str, UUID]]): The unique identifier for the user.

  • situation Optional[str] - An optional description of the situation for the session.

  • limit Optional[int] - The limit on the number of items to be retrieved.

  • offset Optional[int] - The number of items to be skipped before starting to collect the result set.

session_id (Union[str, UUID]): The unique identifier for the session to be deleted.

session_id (Union[str, UUID]): The unique identifier for the session to be updated.

  • situation str - The new situation description for the session.

  • session_id str - The unique identifier for the session.

  • messages List[InputChatMlMessage] - The list of input chat messages to be sent.

  • tools Optional[List[Tool]] - ...

  • tool_choice Optional[ToolChoiceOption] - ...

  • ... - Other optional parameters for chat settings and modifiers.

session_id (Union[str, UUID]): The unique identifier for the session.

  • limit Optional[int] - The limit on the number of suggestions to be retrieved.

  • offset Optional[int] - The number of suggestions to be skipped before starting to collect the result set.

session_id (Union[str, UUID]): The unique identifier for the session.

  • limit Optional[int] - The limit on the number of history entries to be retrieved.

  • offset Optional[int] - The number of history entries to be skipped before starting to collect the result set.

session_id (Union[str, UUID]): The unique identifier for the session.

Returns

  • Union[Session, Awaitable[Session]] - The session object or an awaitable yielding it.

  • Union[ResourceCreatedResponse, Awaitable[ResourceCreatedResponse]] - The response for the created session or an awaitable yielding it.

_list_items(limit, offset): List multiple session items with optional pagination.

  • Union[ListSessionsResponse, Awaitable[ListSessionsResponse]] - The list of sessions or an awaitable yielding it.

_delete(session_id): Delete a session by its identifier.

  • Union[None, Awaitable[None]] - None or an awaitable yielding None if the operation is successful.

_update(session_id, situation): Update the situation for an existing session.

  • Union[ResourceUpdatedResponse, Awaitable[ResourceUpdatedResponse]] - The response for the updated session or an awaitable yielding it.

_chat(session_id, messages, ...): Send chat messages and get responses during a session.

  • Union[ChatResponse, Awaitable[ChatResponse]] - The chat response for the session or an awaitable yielding it.

_suggestions(session_id, limit, offset): Get suggestions for a session.

  • Union[GetSuggestionsResponse, Awaitable[GetSuggestionsResponse]] - The suggestions response for the session or an awaitable yielding it.

_history(session_id, limit, offset): Get the history for a session.

  • Union[GetHistoryResponse, Awaitable[GetHistoryResponse]] - The history response for the session or an awaitable yielding it.

_delete_history(session_id): Delete the history of a session.

  • Union[None, Awaitable[None]] - None or an awaitable yielding None if the operation is successful.

Raises

  • ValueError - If the id is not a valid UUID.

  • NetworkError - If there is an issue communicating with the API.

_create(user_id, agent_id, situation): Create a new session with specified user and agent identifiers.

Signature

class BaseSessionsManager(BaseManager): ...

BaseSessionsManager()._chat

Show source in session.py:298

Conducts a chat conversation with an AI model using specific parameters.

Arguments

  • session_id str - A unique identifier for the chat session.

  • messages List[InputChatMlMessage] - A list of input messages for the AI to respond to.

  • tools Optional[List[Tool]] - A list of tools to be used during the chat session.

  • tool_choice Optional[ToolChoiceOption] - A method for choosing which tools to apply.

  • frequency_penalty Optional[float] - A modifier to decrease the likelihood of frequency-based repetitions.

  • length_penalty Optional[float] - A modifier to control the length of the generated responses. logit_bias (Optional[Dict[str, Optional[int]]]): Adjustments to the likelihood of specific tokens appearing.

  • max_tokens Optional[int] - The maximum number of tokens to generate in the output.

  • presence_penalty Optional[float] - A modifier to control for new concepts' appearance.

  • repetition_penalty Optional[float] - A modifier to discourage repetitive responses.

  • response_format Optional[ChatSettingsResponseFormat] - The format in which the response is to be delivered.

  • seed Optional[int] - An integer to seed the random number generator for reproducibility.

  • stop Optional[ChatSettingsStop] - Tokens at which to stop generating further tokens.

  • stream Optional[bool] - Whether to stream the response or deliver it when it's complete.

  • temperature Optional[float] - A value to control the randomness of the output.

  • top_p Optional[float] - A value to control the nucleus sampling, i.e., the cumulative probability cutoff.

  • recall Optional[bool] - A flag to control the recall capability of the AI model.

  • remember Optional[bool] - A flag to control the persistence of the chat history in the AI's memory.

Returns

  • Union[ChatResponse, Awaitable[ChatResponse]] - The response from the AI given the input messages and parameters. This could be a synchronous ChatResponse object or an asynchronous Awaitable[ChatResponse] if the stream parameter is True.

Notes

The precise types of some arguments, like Tool, ToolChoiceOption, ChatSettingsResponseFormat, and ChatSettingsStop, are not defined within the given context. It's assumed that these types have been defined elsewhere in the code base.

Raises

It is not specified what exceptions this function might raise. Typically, one would expect potential exceptions to be associated with the underlying API client's chat method failure modes, such as network issues, invalid parameters, etc.

Signature

def _chat(
    self,
    session_id: str,
    messages: List[Union[InputChatMlMessageDict, InputChatMlMessage]],
    tools: Optional[List[Union[ToolDict, Tool]]] = None,
    tool_choice: Optional[ToolChoiceOption] = None,
    frequency_penalty: Optional[float] = None,
    length_penalty: Optional[float] = None,
    logit_bias: Optional[Dict[str, Optional[int]]] = None,
    max_tokens: Optional[int] = None,
    presence_penalty: Optional[float] = None,
    repetition_penalty: Optional[float] = None,
    response_format: Optional[
        Union[ChatSettingsResponseFormatDict, ChatSettingsResponseFormat]
    ] = None,
    seed: Optional[int] = None,
    stop: Optional[ChatSettingsStop] = None,
    stream: Optional[bool] = None,
    temperature: Optional[float] = None,
    top_p: Optional[float] = None,
    recall: Optional[bool] = None,
    remember: Optional[bool] = None,
) -> Union[ChatResponse, Awaitable[ChatResponse]]: ...

BaseSessionsManager()._create

Show source in session.py:178

Creates a session for a specified user and agent.

This internal method is responsible for creating a session using the API client. It validates that both the user and agent IDs are valid UUID v4 strings before proceeding with session creation.

Arguments

agent_id (Union[str, UUID]): The agent's identifier which could be a string or a UUID object. user_id (Optional[Union[str, UUID]]): The user's identifier which could be a string or a UUID object.

  • situation Optional[str], optional - An optional description of the situation. metadata (Dict[str, Any])

  • render_templates bool, optional - Whether to render templates in the metadata. Defaults to False.

Returns

  • Union[ResourceCreatedResponse, Awaitable[ResourceCreatedResponse]] - The response from the API client upon successful session creation, which can be a synchronous ResourceCreatedResponse or an asynchronous Awaitable of it.

Raises

  • AssertionError - If either user_id or agent_id is not a valid UUID v4.

Signature

def _create(
    self,
    agent_id: Union[str, UUID],
    user_id: Optional[Union[str, UUID]] = None,
    situation: Optional[str] = None,
    metadata: Dict[str, Any] = {},
    render_templates: bool = False,
) -> Union[ResourceCreatedResponse, Awaitable[ResourceCreatedResponse]]: ...

BaseSessionsManager()._delete

Show source in session.py:243

Delete a session given its session ID.

This is an internal method that asserts the provided session_id is a valid UUID v4 before making the delete request through the API client.

Args: session_id (Union[str, UUID]): The session identifier, which should be a valid UUID v4.

Returns: Union[None, Awaitable[None]]: The result of the delete operation, which can be either None or an Awaitable that resolves to None, depending on whether this is a synchronous or asynchronous call.

Raises: AssertionError: If the session_id is not a valid UUID v4.

Signature

def _delete(self, session_id: Union[str, UUID]) -> Union[None, Awaitable[None]]: ...

BaseSessionsManager()._delete_history

Show source in session.py:426

Delete the history of a session.

Arguments

session_id (Union[str, UUID]): The unique identifier for the session.

Returns

  • Union[None, Awaitable[None]] - The result of the delete operation, which can be either None or an Awaitable that resolves to None, depending on whether this is a synchronous or asynchronous call.

Raises

  • AssertionError - If the session_id is not a valid UUID v4.

Signature

def _delete_history(
    self, session_id: Union[str, UUID]
) -> Union[None, Awaitable[None]]: ...

BaseSessionsManager()._get

Show source in session.py:162

Get a session by its ID.

Arguments

id (Union[str, UUID]): A string or UUID representing the session ID.

Returns

  • Union[Session, Awaitable[Session]] - The session object associated with the given ID, which can be either a Session instance or an Awaitable that resolves to a Session.

Raises

  • AssertionError - If the id is not a valid UUID v4.

Signature

def _get(self, id: Union[str, UUID]) -> Union[Session, Awaitable[Session]]: ...

BaseSessionsManager()._history

Show source in session.py:398

Retrieve a session's history with optional pagination controls.

Args: session_id (Union[str, UUID]): Unique identifier for the session whose history is being queried. Can be a string or a UUID object. limit (Optional[int], optional): The maximum number of history entries to retrieve. Defaults to None, which uses the API's default setting. offset (Optional[int], optional): The number of initial history entries to skip. Defaults to None, which means no offset is applied.

Returns: Union[GetHistoryResponse, Awaitable[GetHistoryResponse]]: The history response object, which may be either synchronous or asynchronous (awaitable), depending on the API client configuration.

Signature

def _history(
    self,
    session_id: Union[str, UUID],
    limit: Optional[int] = None,
    offset: Optional[int] = None,
) -> Union[GetHistoryResponse, Awaitable[GetHistoryResponse]]: ...

BaseSessionsManager()._list_items

Show source in session.py:218

List items with optional pagination.

Arguments

  • limit Optional[int] - The maximum number of items to return. Defaults to None.

  • offset Optional[int] - The number of items to skip before starting to collect the result set. Defaults to None.

Returns

  • Union[ListSessionsResponse, Awaitable[ListSessionsResponse]] - The response object containing the list of items or an awaitable response object if called asynchronously.

Notes

The '_list_items' function is assumed to be a method of a class that has an 'api_client' attribute capable of listing sessions.

Signature

def _list_items(
    self,
    limit: Optional[int] = None,
    offset: Optional[int] = None,
    metadata_filter: str = "{}",
) -> Union[ListSessionsResponse, Awaitable[ListSessionsResponse]]: ...

BaseSessionsManager()._suggestions

Show source in session.py:375

Retrieve a list of suggestions for a given session.

Args: session_id (Union[str, UUID]): The ID of the session for which to get suggestions. limit (Optional[int], optional): The maximum number of suggestions to retrieve. Defaults to None. offset (Optional[int], optional): The offset from where to start retrieving suggestions. Defaults to None.

Returns: Union[GetSuggestionsResponse, Awaitable[GetSuggestionsResponse]]: The response containing the list of suggestions synchronously or asynchronously, depending on the API client.

Signature

def _suggestions(
    self,
    session_id: Union[str, UUID],
    limit: Optional[int] = None,
    offset: Optional[int] = None,
) -> Union[GetSuggestionsResponse, Awaitable[GetSuggestionsResponse]]: ...

BaseSessionsManager()._update

Show source in session.py:264

Update a session with a given situation.

Arguments

session_id (Union[str, UUID]): The session identifier, which can be a string-formatted UUID or an actual UUID object.

  • situation str - A string describing the current situation.

  • overwrite bool, optional - Whether to overwrite the existing situation. Defaults to False.

Returns

  • Union[ResourceUpdatedResponse, Awaitable[ResourceUpdatedResponse]] - The response from the update operation, which can be either synchronous or asynchronous.

Raises

  • AssertionError - If session_id is not a valid UUID v4.

Signature

def _update(
    self,
    session_id: Union[str, UUID],
    situation: Optional[str] = None,
    metadata: Optional[Dict[str, Any]] = None,
    overwrite: bool = False,
) -> Union[ResourceUpdatedResponse, Awaitable[ResourceUpdatedResponse]]: ...

SessionCreateArgs

Show source in session.py:37

Signature

class SessionCreateArgs(TypedDict): ...

SessionUpdateArgs

Show source in session.py:45

Signature

class SessionUpdateArgs(TypedDict): ...

SessionsManager

Show source in session.py:448

A class responsible for managing session interactions.

This class extends BaseSessionsManager and provides methods to get, create, list, delete, and update sessions, as well as to initiate a chat within a session, request suggestions, and access session history.

Methods

  • get (id - Union[str, UUID]) -> Session: Retrieves a session by its identifier.

create ( *, - user_id - Union[str, UUID], - agent_id - Union[str, UUID], - situation - Optional[str]=None ) -> ResourceCreatedResponse: Creates a new session given a user ID and an agent ID, and optionally a description of the situation.

list ( *, - limit - Optional[int]=None, - offset - Optional[int]=None ) -> List[Session]: Lists sessions with optional pagination via limit and offset.

  • delete (session_id - Union[str, UUID]): Deletes a session identified by the given session ID.

update ( *, - session_id - Union[str, UUID], - situation - str ) -> ResourceUpdatedResponse: Updates the situation of a specific session by its ID.

chat ( *args see full method signature for detailed arguments ) -> ChatResponse: Initiates a chat in the given session with messages and various settings, including tools, penalties, biases, tokens, response format, etc.

suggestions ( *, - session_id - Union[str, UUID], - limit - Optional[int]=None, - offset - Optional[int]=None ) -> List[Suggestion]: Retrieves a list of suggestions for a given session, supported by optional pagination parameters.

history ( *, - session_id - Union[str, UUID], - limit - Optional[int]=None, - offset - Optional[int]=None ) -> List[ChatMlMessage]: Retrieves the chat history for a given session, supported by optional pagination parameters.

  • delete_history (session_id - Union[str, UUID]) -> None:

Each method is decorated with @beartype for runtime type enforcement.

Signature

class SessionsManager(BaseSessionsManager): ...

See also

SessionsManager().chat

Show source in session.py:629

Initiate a chat session with the provided inputs and configurations.

Arguments

  • session_id str - Unique identifier for the chat session.

  • messages List[InputChatMlMessage] - List of messages to send in the chat session.

  • tools Optional[List[Tool]], optional - List of tools to be used in the session. Defaults to None.

  • tool_choice Optional[ToolChoiceOption], optional - The choice of tool to optimize response for. Defaults to None.

  • frequency_penalty Optional[float], optional - Penalty for frequent tokens to control repetition. Defaults to None.

  • length_penalty Optional[float], optional - Penalty for longer responses to control verbosity. Defaults to None. logit_bias (Optional[Dict[str, Optional[int]]], optional): Bias for or against specific tokens. Defaults to None.

  • max_tokens Optional[int], optional - Maximum number of tokens to generate in the response. Defaults to None.

  • presence_penalty Optional[float], optional - Penalty for new tokens to control topic introduction. Defaults to None.

  • repetition_penalty Optional[float], optional - Penalty to discourage repetition. Defaults to None.

  • response_format Optional[ChatSettingsResponseFormat], optional - Format of the response. Defaults to None.

  • seed Optional[int], optional - Random seed for deterministic responses. Defaults to None.

  • stop Optional[ChatSettingsStop], optional - Sequence at which to stop generating further tokens. Defaults to None.

  • stream Optional[bool], optional - Whether to stream responses or not. Defaults to None.

  • temperature Optional[float], optional - Sampling temperature for randomness in the response. Defaults to None.

  • top_p Optional[float], optional - Nucleus sampling parameter to control diversity. Defaults to None.

  • recall Optional[bool], optional - Whether to allow recalling previous parts of the chat. Defaults to None.

  • remember Optional[bool], optional - Whether to allow the model to remember previous chats. Defaults to None.

Returns

  • ChatResponse - The response object after processing chat messages.

Notes

The 'beartype' decorator is used for runtime type checking.

Signature

@beartype
def chat(
    self,
    session_id: str,
    messages: List[Union[InputChatMlMessageDict, InputChatMlMessage]],
    tools: Optional[List[Union[ToolDict, Tool]]] = None,
    tool_choice: Optional[ToolChoiceOption] = None,
    frequency_penalty: Optional[float] = None,
    length_penalty: Optional[float] = None,
    logit_bias: Optional[Dict[str, Optional[int]]] = None,
    max_tokens: Optional[int] = None,
    presence_penalty: Optional[float] = None,
    repetition_penalty: Optional[float] = None,
    response_format: Optional[
        Union[ChatSettingsResponseFormatDict, ChatSettingsResponseFormat]
    ] = None,
    seed: Optional[int] = None,
    stop: Optional[ChatSettingsStop] = None,
    stream: Optional[bool] = None,
    temperature: Optional[float] = None,
    top_p: Optional[float] = None,
    recall: Optional[bool] = None,
    remember: Optional[bool] = None,
) -> ChatResponse: ...

SessionsManager().create

Show source in session.py:533

Create a new resource with a user ID and an agent ID, optionally including a situation description.

Arguments

user_id (Union[str, UUID]): The unique identifier for the user. agent_id (Union[str, UUID]): The unique identifier for the agent.

  • situation Optional[str] - An optional description of the situation.

Returns

  • Session - The created Session object.

Raises

  • BeartypeException - If the provided user_id or agent_id do not match the required type. Any other exception that _create might raise.

Signature

@beartype
@rewrap_in_class(Session)
def create(self, **kwargs: SessionCreateArgs) -> Session: ...

See also

SessionsManager().delete

Show source in session.py:585

Deletes a session based on its session ID.

Arguments

session_id (Union[str, UUID]): The session ID to be deleted, which can be a string or a UUID object.

Returns

The result from the internal _delete method call.

Raises

The specific exceptions that self._delete might raise, which should be documented in its docstring.

Signature

@beartype
def delete(self, session_id: Union[str, UUID]): ...

SessionsManager().delete_history

Show source in session.py:765

Delete the history of a session.

Arguments

session_id (Union[str, UUID]): The unique identifier for the session.

Returns

  • None - The result of the delete operation.

Raises

  • AssertionError - If the session_id is not a valid UUID v4.

Signature

@beartype
def delete_history(self, session_id: Union[str, UUID]) -> None: ...

SessionsManager().get

Show source in session.py:516

Retrieve a Session object based on a given identifier.

Args: id (Union[str, UUID]): The identifier of the session, which can be either a string or a UUID.

Returns: Session: The session object associated with the given id.

Raises: TypeError: If the id is neither a string nor a UUID. KeyError: If the session with the given id does not exist.

Signature

@beartype
def get(self, id: Union[str, UUID]) -> Session: ...

SessionsManager().history

Show source in session.py:738

Retrieve a history of ChatMl messages for a given session.

This method uses the private method _history to fetch the message history and returns a list of ChatMlMessage objects.

Args: session_id (Union[str, UUID]): The session identifier to fetch the chat history for. limit (Optional[int], optional): The maximum number of messages to return. If None, no limit is applied. Defaults to None. offset (Optional[int], optional): The offset from where to start fetching messages. If None, no offset is applied. Defaults to None.

Returns: List[ChatMlMessage]: A list of ChatMlMessage objects representing the history of messages for the session.

Signature

@beartype
def history(
    self,
    session_id: Union[str, UUID],
    limit: Optional[int] = None,
    offset: Optional[int] = None,
) -> List[ChatMlMessage]: ...

SessionsManager().list

Show source in session.py:554

Retrieve a list of Session objects with optional pagination.

Args: limit (Optional[int]): The maximum number of Session objects to return. Defaults to None, which indicates no limit. offset (Optional[int]): The number of items to skip before starting to return the results. Defaults to None, which indicates no offset.

Returns: List[Session]: A list of Session objects meeting the criteria.

Raises: BeartypeException: If the input arguments do not match their annotated types.

Signature

@beartype
def list(
    self,
    limit: Optional[int] = None,
    offset: Optional[int] = None,
    metadata_filter: Dict[str, Any] = {},
) -> List[Session]: ...

SessionsManager().suggestions

Show source in session.py:704

Provides a list of suggestion objects based on the given session ID.

This method retrieves suggestions and is decorated with beartype for runtime type checking of the passed arguments.

Arguments

session_id (Union[str, UUID]): The ID of the session to retrieve suggestions for.

  • limit Optional[int], optional - The maximum number of suggestions to return. Defaults to None, which means no limit.

  • offset Optional[int], optional - The number to offset the list of returned suggestions by. Defaults to None, which means no offset.

Returns

  • List[Suggestion] - A list of suggestion objects.

Raises

Any exceptions that _suggestions might raise, for example, if the method is unable to retrieve suggestions based on the provided session ID.

Signature

@beartype
def suggestions(
    self,
    session_id: Union[str, UUID],
    limit: Optional[int] = None,
    offset: Optional[int] = None,
) -> List[Suggestion]: ...

SessionsManager().update

Show source in session.py:601

Updates the state of a resource based on a given situation.

This function is type-checked using beartype to ensure that the session_id parameter is either a string or a UUID and that the situation parameter is a string. It delegates the actual update process to an internal method '_update'.

Arguments

session_id (Union[str, UUID]): The session identifier, which can be a UUID or a string that uniquely identifies the session.

  • situation str - A string that represents the new situation for the resource update.

  • overwrite bool, optional - A flag to indicate whether to overwrite the existing

Returns

  • Session - The updated Session object.

Notes

The @beartype decorator is used for runtime type checking of the function arguments.

Signature

@beartype
@rewrap_in_class(Session)
def update(self, **kwargs: SessionUpdateArgs) -> Session: ...

See also

PreviousUsersNextMemories

Last updated