Domain Genflow

verifia.generation.DomainGenFlow

A class to generate a domain graph flow based on provided dataset, PDFs or vector database, and a model card.

This class encapsulates the process of loading the necessary context (data, agents, and model card) to build and run a domain graph. It allows the context to be loaded from file paths or directly from provided objects. It orchestrates the domain graph generation, human interruptions, and final domain export.

__init__()

Initialize the DomainGenFlow instance.

launch(server_name=DEFAULT_HOST, server_port=DEFAULT_PORT)

Launch the interactive Gradio UI tied to this flow.

Parameters:

Name	Type	Description	Default
server_name	str	Host/IP for the Gradio server.	DEFAULT_HOST
server_port	int	Port for the Gradio server.	DEFAULT_PORT

load_ctx(data_fpath=None, dataframe=None, pdfs_dirpath=None, db_str_content=None, vectordb=None, model_card_fpath=None, model_card=None)

Load and initialize the domain context required to build the domain graph.

At least one of 'data_fpath' or 'dataframe' must be provided, as well as one of 'pdfs_dirpath', 'vectordb', or 'db_str_content'. Similarly, either 'model_card' or 'model_card_fpath' must be provided.

Parameters:

Name	Type	Description	Default
data_fpath	Optional[Union[str, Path]]	Path to the CSV or data file.	None
dataframe	Optional[DataFrame]	DataFrame containing the data.	None
pdfs_dirpath	Optional[Union[str, Path]]	Directory path containing PDF files.	None
db_str_content	Optional[str]	String content to build a vector database.	None
vectordb	Optional[Chroma]	Pre-initialized vector database.	None
model_card_fpath	Optional[Union[str, Path]]	Path to the model card YAML file.	None
model_card	Optional[Dict[str, Any]]	Model card as a dictionary.	None

Returns:

Name	Type	Description
DomainGenFlow	DomainGenFlow	The instance with the graph_context loaded.

Raises:

Type	Description
ValueError	If required context arguments are missing.

next_action(gen, action, yaml_str, instructions='')

Resume the flow based on a UI action (validate/regenerate/finish).

Parameters:

Name	Type	Description	Default
gen	Generator[Dict[str, Any], Any, None]	Active generator from start or previous next_action.	required
action	str	Action keyword.	required
yaml_str	str	Current YAML shown to the user.	required
instructions	str	User instructions for regeneration.	''

Returns:

Type	Description
Tuple[str, str, Generator[Dict[str, Any], Any, None]]	Tuple of updated (yaml_text, validation_output, generator).

start()

Kick off the flow until first human interrupt, returning initial YAML and validation.

Returns:

Name	Type	Description
init_yaml	str	Initial generated YAML spec.
init_val	str	Initial validation output.
gen	Generator[Dict[str, Any], Any, None]	Generator to drive subsequent actions.