someml-indent

#!/usr/bin/env python3

import os, sys, io, re


p_err = lambda *a,**k: print(*a, file=sys.stderr, **k) or 1


def parse_tags(src):
	'Returns "content" list of strings or (tag, attrs, content) tuples.'
	dst = list()

	while True:
		m = re.search(
			r'^(?P<pre>.*?)<\s*(?P<tag>[^\s>]+)\s*(?:\s(?P<attrs>[^>]+))?>'
			r'(?P<content>.*?)' r'</\s*(?P=tag)\s*>', src, re.DOTALL )
		if not m:
			if src.strip(): dst.append(src)
			break
		src = src[m.end():]

		pre, tag, attrs, c = m.groups()
		if c:
			if re.search(fr'<\s*{tag}\s*(?:\s[^>]+)?>', c):
				p_err(f'WARNING: nested <{tag}> tags were matched incorrectly')
			c = parse_tags(c)
		if pre.strip(): dst.append(pre)
		dst.append((tag, attrs, c))

	return dst


def print_tags(src, dst, indent, pre=None, path=()):
	if not src: return
	inline = len(src) == 1
	if pre is None: pre2 = ''
	else: pre2 = (pre + indent) if not inline else pre
	for n, v in enumerate(src):
		if not inline: dst.write('\n' + pre2)
		if isinstance(v, str):
			dst.write(v.strip())
			continue
		tag, attrs, c = v
		attrs = f' {attrs}' if attrs else ''
		dst.write(f'<{tag}{attrs}>')
		if c: print_tags(c, dst, indent, pre2, path=path+(f'{tag}.{n}',))
		dst.write(f'</{tag}>')
	if not inline: dst.write('\n' + (pre or ''))
	if pre is None: dst.write('\n')


def main(args=None):
	import argparse
	parser = argparse.ArgumentParser(
		description='Insert newlines into XML-ish thing, without really'
				' validating or messing it up otherwise, except for added/stripped spaces.'
			' Input - stdin, output - stdout.')
	parser.add_argument('-i', '--indent-spaces',
		type=int, default=2, metavar='num',
		help='Number of spaces to use for a level indent, -1 = use tabs (default: %(default)s).')
	opts = parser.parse_args(sys.argv[1:] if args is None else args)

	src = sys.stdin.read()
	src, dst = parse_tags(src), io.StringIO()
	# exit(print(src))

	indent = opts.indent_spaces
	indent = (' '*indent) if indent >= 0 else '\t'
	print_tags(src, dst, indent)
	print(dst.getvalue().strip())

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