tabtree-example.py

LOCATIONS = """
country
    UK
        city
            London
            Guildford
        county
            Surray
            Sussex
    Belgium
    Russia
        city
            Moscow
        federal district
            Central
            South
    Germany
"""
import io
from itertools import takewhile


def tabtree_nodes(fileobj):
    path = []
    value = None
    indent_step = None
    depth = 0
    for line_n, line in enumerate(fileobj):
        line = line.rstrip()
        if not line:
            continue
        ind = len(list(takewhile(lambda ch: ch == ' ' or ch == '\t', line)))
        val = line[ind:]
        # setup indentation step size
        if not indent_step and ind:
            indent_step = ind
        # calc depth and check indentation size is correct
        d, reminder = divmod(ind, indent_step or 1)
        if reminder:
            raise Exception('Indentation size error, line: %d' % line_n)
        if d > depth:
            if d - depth > 1:
                raise Exception('Indentation problem, line: %d' % line_n)
            path.append(value)
            yield (path + [val])

        elif d < depth:
            path = path[:-(depth - d)]
            yield (path + [val])
        else:
            yield (path + [val])

        depth = d
        value = val


def locations_tree(fileobj):
    for node in tabtree_nodes(fileobj):
        if len(node) % 2 == 0:
            yield [dict(level=level, location=location) for level, location in zip(node[::2], node[1::2])]


for i in locations_tree(io.StringIO(LOCATIONS)):
    print(i)