asciitree-parse

#!/usr/bin/env python3

import itertools as it, operator as op, functools as ft
from collections import defaultdict, OrderedDict
import os, sys, re


def parse_tree(src):
	tree, node_pos = OrderedDict(), dict()

	for line in src:
		if line.lstrip().startswith('#') or not line.strip(): continue
		tokens = iter(re.finditer(r'(\|)|(\+-+)|(\S.*$)', line))
		node_parsed = False
		for m in tokens:
			assert not node_parsed, line # only one node per line
			pos, (t_next, t_leaf, val) = m.start(), m.groups()

			if t_next:
				assert pos in node_pos, [line, pos, node_pos]
				continue
			elif t_leaf:
				m = next(tokens)
				val, pos_sub = m.group(3), m.start()
			elif val: # root
				assert not tree, line
				pos_sub = 1

			parent = node_pos[pos] if pos else tree
			node = parent[val] = dict()
			node_parsed, node_pos[pos_sub] = True, node

	return tree


def main(args=None):
	import argparse
	parser = argparse.ArgumentParser(
		description='Parse vertical ASCII tree'
				' to nested OrderedDict structure and print/pickle it to stdout.'
			' Tree must be in asciitree.LeftAligned format, asciitree is a python module.')
	parser.add_argument('tree_file', nargs='?',
		help='File that ascitree is stored in.'
			' Any empty lines or lines starting with # will be ignored.'
			' If not specified or "-", stdin will be used.')
	parser.add_argument('-p', '--pickle', nargs='?', metavar='path', const='-',
		help='Pickle (python binary format) resulting structure to'
			' specified file or stdout (if no path or "-" is provided as arg).')
	opts = parser.parse_args(sys.argv[1:] if args is None else args)

	src = open(opts.tree_file) if opts.tree_file and opts.tree_file != '-' else sys.stdin
	with src as src: tree = parse_tree(src)

	if opts.pickle:
		import pickle
		dst = open(opts.pickle, 'wb') if opts.pickle and opts.pickle != '-' else sys.stdout
		with dst as dst: pickle.dump(tree, dst)
	else: print(tree)

if __name__ == '__main__': sys.exit(main())