Skip to main content

​
mellea.stdlib.base

Basic stdlib data structures.

​
Functions

​
get_images_from_component

get_images_from_component(c: Component) -> None | list[ImageBlock]
Gets images from a Component if they are present and a non-empty list, otherwise returns None.

​
blockify

blockify(s: str | CBlock | Component) -> CBlock | Component
blockify is a helper function that turns raw strings into CBlocks.

​
Classes

​
CBlock

A CBlock is a block of content that can serve as input to or output from an LLM. Methods:

​
value

value(self) -> str | None
Gets the value of the block.

​
value

value(self, v: str)
Sets the value of the block.

​
ImageBlock

A ImageBlock represents an image (as base64 PNG). Methods:

​
is_valid_base64_png

is_valid_base64_png(s: str) -> bool
Checks if a string is a valid base64 string [AIA PAI Nc Hin R v1.0].

​
pil_to_base64

pil_to_base64(image: PILImage.Image) -> str
Converts a PIL image to a base64 string representation.

​
from_pil_image

from_pil_image(cls, image: PILImage.Image, meta: dict[str, Any] | None = None) -> ImageBlock
Converts a PIL image to a base64 string representation.

​
Component

A Component is a composite data structure that is intended to be represented to an LLM. Methods:

​
parts

parts(self) -> list[Component | CBlock]
The set of all the constituent parts of the Component.

​
format_for_llm

format_for_llm(self) -> TemplateRepresentation | str
Formats the Component into a TemplateRepresentation or string. Returns: a TemplateRepresentation or string

​
GenerateType

Used to track what functions can be used to extract a value from a ModelOutputThunk.

​
ModelOutputThunk

A ModelOutputThunk is a special type of CBlock that we know came from a model’s output. It is possible to instantiate one without the output being computed yet. Methods:

​
is_computed

is_computed(self)
Returns true only if this Thunk has already been filled.

​
value

value(self) -> str | None
Gets the value of the block.

​
value

value(self, v: str)
Sets the value of the block.

​
avalue

avalue(self) -> str
Returns the value of the ModelOutputThunk. Can be used for both async streaming and async non-streaming. Raises:
  • Exception: Propagates any errors from the underlying inference engine api request.
  • RuntimeError: If called when the ModelOutputThunk’s generate function is not async compatible.

​
astream

astream(self) -> str
Returns the ModelOutputThunk’s partial value including the next chunk(s). Can be used for both async streaming and async non-streaming. Returns the value of the ModelOutputThunk if streaming is done. Note: Be careful with calling this function. Only call it from one location at a time. This means you shouldn’t pass a ModelOutputThunk to multiple coroutines/tasks and call astream from those coroutines/tasks simultaneously. We have considered solutions to this but are waiting until we see this error happen in a real use case. Raises:
  • Exception: Propagates any errors from the underlying inference engine api request.
  • RuntimeError: If called when the ModelOutputThunk’s generate function is not async compatible.

​
ContextTurn

A turn of model input and model output.

​
Context

A Context is used to track the state of a MelleaSession. A context is immutable. Every alteration leads to a new context. Methods:

​
from_previous

from_previous(cls: type[ContextT], previous: Context, data: Component | CBlock) -> ContextT
Constructs a new context from an existing context.

​
reset_to_new

reset_to_new(cls: type[ContextT]) -> ContextT
Returns an empty context for convenience.

​
is_root_node

is_root_node(self) -> bool
Returns whether this context is the root context node.

​
previous_node

previous_node(self) -> Context | None
Returns the context node from which this context node was created. Internal use: Users should not need to use this property.

​
node_data

node_data(self) -> Component | CBlock | None
Returns the data associated with this context node. Internal use: Users should not need to use this property.

​
is_chat_context

is_chat_context(self) -> bool
Returns whether this context is a chat context.

​
as_list

as_list(self, last_n_components: int | None = None) -> list[Component | CBlock]
Returns a list of the last n components in the context sorted from FIRST TO LAST. If last_n_components is None, then all components are returned.

​
actions_for_available_tools

actions_for_available_tools(self) -> list[Component | CBlock] | None
Provides a list of actions to extract tools from for use with during generation, or None if that’s not possible. Can be used to make the available tools differ from the tools of all the actions in the context. Can be overwritten by subclasses.

​
last_output

last_output(self, check_last_n_components: int = 3) -> ModelOutputThunk | None
The last output thunk of the context.

​
last_turn

last_turn(self)
The last input/output turn of the context. This can be partial. If the last event is an input, then the output is None.

​
add

add(self, c: Component | CBlock) -> Context
Returns a new context obtained by adding c to this context.

​
view_for_generation

view_for_generation(self) -> list[Component | CBlock] | None
Provides a linear list of context components to use for generation, or None if that is not possible to construct.

​
ChatContext

Initializes a chat context with unbounded window_size and is_chat=True by default. Methods:

​
add

add(self, c: Component | CBlock) -> ChatContext
Add a new component/cblock to the context. Returns the new context.

​
view_for_generation

view_for_generation(self) -> list[Component | CBlock] | None
Returns the context in a linearized form. Uses the window_size set during initialization.

​
SimpleContext

A SimpleContext is a context in which each interaction is a separate and independent turn. The history of all previous turns is NOT saved.. Methods:

​
add

add(self, c: Component | CBlock) -> SimpleContext
Add a new component/cblock to the context. Returns the new context.

​
view_for_generation

view_for_generation(self) -> list[Component | CBlock] | None
Returns an empty list.

​
TemplateRepresentation

Representing a component as a set of important attributes that can be consumed by the formatter.

​
GenerateLog

A dataclass for capturing log entries. GenerateLog provides a structured way to include various details in log entries, making it useful for maintaining detailed records of events or operations where context and additional data are significant.

​
ModelToolCall

A dataclass for capturing the tool calls a model wants to make. Provides a unified way to call tools post generation. Methods:

​
call_func

call_func(self) -> Any
A helper function for calling the function/tool represented by this object.