Graph

class langchain_core.runnables.graph.Graph(nodes: dict[str, ~langchain_core.runnables.graph.Node] = <factory>, edges: list[~langchain_core.runnables.graph.Edge] = <factory>)

Graph of nodes and edges.

Parameters:

• nodes (dict[str, Node]) – Dictionary of nodes in the graph. Defaults to an empty dictionary.

• edges (list[Edge]) – List of edges in the graph. Defaults to an empty list.

Attributes

Methods

__init__([nodes, edges])
add_edge(source, target[, data, conditional])	Add an edge to the graph and return it.
add_node(data[, id, metadata])	Add a node to the graph and return it.
draw_ascii()	Draw the graph as an ASCII art string.
draw_mermaid(*[, with_styles, curve_style, ...])	Draw the graph as a Mermaid syntax string.
draw_mermaid_png(*[, curve_style, ...])	Draw the graph as a PNG image using Mermaid.
draw_png()	Draw the graph as a PNG image.
extend(graph, *[, prefix])	Add all nodes and edges from another graph.
first_node()	Find the single node that is not a target of any edge.
last_node()	Find the single node that is not a source of any edge.
next_id()	Return a new unique node identifier that can be used to add a node to the graph.
print_ascii()	Print the graph as an ASCII art string.
reid()	Return a new graph with all nodes re-identified, using their unique, readable names where possible.
remove_node(node)	Remove a node from the graph and all edges connected to it.
to_json(*[, with_schemas])	Convert the graph to a JSON-serializable format.
trim_first_node()	Remove the first node if it exists and has a single outgoing edge, i.e., if removing it would not leave the graph without a "first" node.
trim_last_node()	Remove the last node if it exists and has a single incoming edge, i.e., if removing it would not leave the graph without a "last" node.

__init__(nodes: dict[str, ~langchain_core.runnables.graph.Node] = <factory>, edges: list[~langchain_core.runnables.graph.Edge] = <factory>) → None

Parameters:

• nodes (dict[str, Node])

• edges (list[Edge])

Return type:

None

add_edge(source: Node, target: Node, data: Stringifiable | None = None, conditional: bool = False) → Edge

Add an edge to the graph and return it.

Parameters:

• source (Node) – The source node of the edge.

• target (Node) – The target node of the edge.

• data (Stringifiable | None) – Optional data associated with the edge. Defaults to None.

• conditional (bool) – Whether the edge is conditional. Defaults to False.

Returns:

The edge that was added to the graph.

Raises:

ValueError – If the source or target node is not in the graph.

Return type:

Edge

add_node(data: type[BaseModel] | RunnableType, id: str | None = None, *, metadata: dict[str, Any] | None = None) → Node

Add a node to the graph and return it.

Parameters:

• data (Union[type[BaseModel], RunnableType]) – The data of the node.

• id (Optional[str]) – The id of the node. Defaults to None.

• metadata (Optional[dict[str, Any]]) – Optional metadata for the node. Defaults to None.

Returns:

The node that was added to the graph.

Raises:

ValueError – If a node with the same id already exists.

Return type:

Node

draw_ascii() → str

Draw the graph as an ASCII art string.

Return type:

str

draw_mermaid(*, with_styles: bool = True, curve_style: CurveStyle = CurveStyle.LINEAR, node_colors: NodeStyles | None = None, wrap_label_n_words: int = 9) → str

Draw the graph as a Mermaid syntax string.

Parameters:

• with_styles (bool) – Whether to include styles in the syntax. Defaults to True.

• curve_style (CurveStyle) – The style of the edges. Defaults to CurveStyle.LINEAR.

• node_colors (NodeStyles | None) – The colors of the nodes. Defaults to NodeStyles().

• wrap_label_n_words (int) – The number of words to wrap the node labels at. Defaults to 9.

Returns:

The Mermaid syntax string.

Return type:

str

draw_mermaid_png(*, curve_style: CurveStyle = CurveStyle.LINEAR, node_colors: NodeStyles | None = None, wrap_label_n_words: int = 9, output_file_path: str | None = None, draw_method: MermaidDrawMethod = MermaidDrawMethod.API, background_color: str = 'white', padding: int = 10) → bytes

Draw the graph as a PNG image using Mermaid.

Parameters:

• curve_style (CurveStyle) – The style of the edges. Defaults to CurveStyle.LINEAR.

• node_colors (NodeStyles | None) – The colors of the nodes. Defaults to NodeStyles().

• wrap_label_n_words (int) – The number of words to wrap the node labels at. Defaults to 9.

• output_file_path (str | None) – The path to save the image to. If None, the image is not saved. Defaults to None.

• draw_method (MermaidDrawMethod) – The method to use to draw the graph. Defaults to MermaidDrawMethod.API.

• background_color (str) – The color of the background. Defaults to “white”.

• padding (int) – The padding around the graph. Defaults to 10.

Returns:

The PNG image as bytes.

Return type:

bytes

draw_png(output_file_path: str, fontname: str | None = None, labels: LabelsDict | None = None) → None

draw_png(output_file_path: None, fontname: str | None = None, labels: LabelsDict | None = None) → bytes

Draw the graph as a PNG image.

Parameters:

• output_file_path – The path to save the image to. If None, the image is not saved. Defaults to None.

• fontname – The name of the font to use. Defaults to None.

• labels – Optional labels for nodes and edges in the graph. Defaults to None.

Returns:

The PNG image as bytes if output_file_path is None, None otherwise.

extend(graph: Graph, *, prefix: str = '') → tuple[Node | None, Node | None]

Add all nodes and edges from another graph. Note this doesn’t check for duplicates, nor does it connect the graphs.

Parameters:

• graph (Graph) – The graph to add.

• prefix (str) – The prefix to add to the node ids. Defaults to “”.

Returns:

A tuple of the first and last nodes of the subgraph.

Return type:

tuple[Node | None, Node | None]

first_node() → Node | None

Find the single node that is not a target of any edge. If there is no such node, or there are multiple, return None. When drawing the graph, this node would be the origin.

Return type:

Node | None

last_node() → Node | None

Find the single node that is not a source of any edge. If there is no such node, or there are multiple, return None. When drawing the graph, this node would be the destination.

Return type:

Node | None

next_id() → str

Return a new unique node identifier that can be used to add a node to the graph.

Return type:

str

print_ascii() → None

Print the graph as an ASCII art string.

Return type:

None

reid() → Graph

Return a new graph with all nodes re-identified, using their unique, readable names where possible.

Return type:

Graph

remove_node(node: Node) → None

Remove a node from the graph and all edges connected to it.

Parameters:

node (Node) – The node to remove.

Return type:

None

to_json(*, with_schemas: bool = False) → dict[str, list[dict[str, Any]]]

Convert the graph to a JSON-serializable format.

Parameters:

with_schemas (bool) – Whether to include the schemas of the nodes if they are Pydantic models. Defaults to False.

Returns:

A dictionary with the nodes and edges of the graph.

Return type:

dict[str, list[dict[str, Any]]]

trim_first_node() → None

Remove the first node if it exists and has a single outgoing edge, i.e., if removing it would not leave the graph without a “first” node.

Return type:

None

trim_last_node() → None

Remove the last node if it exists and has a single incoming edge, i.e., if removing it would not leave the graph without a “last” node.

Return type:

None