bayesline.api.equity.hierarchy_from_df

bayesline.api.equity.hierarchy_from_df(df: DataFrame) → tuple[Hierarchy, dict[str, str]]

Convert a hierarchy DataFrame back to a (Hierarchy, labels) pair.

Inverse of hierarchy_to_df(). The input must have columns level, code, label, parent_code. Roots are rows where parent_code is null. A node with no children in the table is a leaf; when every child of a node is a leaf, that node’s value is a list of leaf codes (matching the canonical Hierarchy = list[str] | Mapping[str, Hierarchy] shape).

Parameters

dfpl.DataFrame

Hierarchy in the long (level, code, label, parent_code) shape.

Returns

tuple[Hierarchy, dict[str, str]]

The reconstructed hierarchy and a code → label mapping.