isco.py

# filename: isco.py
from collections import defaultdict
import csv


def create_hierarchy_dict(filename):
    """
    Creates a dictionary where keys are nodes and values are sets of parent nodes representing the hierarchy of the ISCO-08 codes from the "unit" column of the isco_structure.csv file.

    Args:
    - filename: A string representing the path to the CSV file containing the ISCO-08 codes and their hierarchy.

    Returns:
    - A dictionary where keys are ISCO-08 unit codes and values are sets of their parent codes.
    """
    isco_hierarchy = {}

    with open(filename, newline="") as csvfile:
        reader = csv.DictReader(csvfile)
        for row in reader:
            # Extract unit group level code (4 digits)
            unit_code = row["unit"].zfill(4)
            # Extract the parent code for the unit group level, which is the minor group level (3 digits)
            parent_code = unit_code[:3]

            # Add the unit code to the hierarchy with its parent code
            isco_hierarchy[unit_code] = {parent_code}

            # Additionally, we can add the parent's parent codes if needed
            # For example, the major group level (1 digit) and sub-major group level (2 digits)
            major_code = unit_code[0]
            sub_major_code = unit_code[:2]
            isco_hierarchy[unit_code].update({major_code, sub_major_code})

    return isco_hierarchy


def create_hierarchy_tree(hierarchy_dict: dict) -> tuple:
    """
    Builds the hierarchy tree and a mapping from name to ISCO code.

    Args:
    - hierarchy_dict: A dictionary representing the hierarchical structure.

    Returns:
    - tree: A dictionary representing the hierarchical structure.
    - code_to_node: A dictionary mapping from ISCO code to node name.
    """

    tree = defaultdict(lambda: {"children": [], "parent": None})
    code_to_node = {}

    def add_node(parent_code, node):
        code = node["name"].split("=")[0].strip()
        code_to_node[code] = node["name"]
        tree[code]["parent"] = parent_code
        if parent_code:
            tree[parent_code]["children"].append(code)
        for child in node.get("children", []):
            add_node(code, child)

    add_node(None, hierarchy_dict)  # Root node has no parent
    return tree, code_to_node


# Example usage:
# hierarchy_dict = create_hierarchy("ISCO_structure.csv")
# tree, code_to_node = create_hierarchy_tree(hierarchy_dict)
# print(hierarchy)
# print(code_to_node)
# print(tree)