sf2xed/sf2xed.py

#!/usr/bin/env python3
import enum
from collections import namedtuple

class writes(enum.Enum):
	zero, one, same = range(3)

State = namedtuple('State', ['write', 'move', 'transitions'])

class SFParseError(Exception): pass

def parse_smallfuck(s):
	def parse(s, index = 0, in_loop = False):
		parsed = []
		while True:
			assert index <= len(s)
			if index == len(s):
				if in_loop:
					raise SFParseError('Unexpected end of text (expected "]")')
				else:
					break

			c = s[index]
			index += 1

			if c == ']':
				if in_loop:
					break
				else:
					raise SFParseError('Unexpected "]"')

			elif c == '[':
				loop, index = parse(s, index, True)
				parsed.append(loop)

			elif c in ['<', '>', '*']:
				parsed.append(c)

		return parsed, index

	parsed, index = parse(s)
	return parsed

def chunk(parsed):
	def append_change_move():
		nonlocal change, move, chunked
		while change or move != 0:
			chunk = '*' if change else ''
			change = False

			if move < 0:
				chunk += '<'
				move += 1
			elif move > 0:
				chunk += '>'
				move -= 1

			if chunk != '':
				chunked.append(chunk)

	chunked = []
	change = False
	move = 0
	for c in parsed:
		if type(c) == list:
			append_change_move()
			chunked.append(chunk(c))

		elif c == '*':
			if move == 0:
				change = not change
			else:
				append_change_move()
				change = True

		elif c == '<':
			move -= 1

		elif c == '>':
			move += 1

		else:
			assert not "Unreachable"

	append_change_move()

	return chunked

def turingify(chunked):
	states = []
	def add_state(write, move, transitions):
		nonlocal states
		assert write in writes
		assert -1 <= move <= 1
		assert len(transitions) == 2

		index = len(states)
		states.append(State(write, move, transitions))

		return index

	def worker(chunked, end = None, in_loop = False):
		nonlocal states

		ifunset = ifset = end
		while len(chunked) > 0:
			c = chunked[-1]
			chunked = chunked[:-1]

			if type(c) == list:
				loop_test = add_state(writes.same, 0, [ifunset, None])
				loop_body = worker(c, loop_test, True)
				# We want the loop to repeat if the cell is set
				states[loop_test].transitions[1] = loop_body
				ifunset = ifset = loop_test

			else:
				change = '*' in c
				move = (1 if '>' in c else 0) - (1 if '<' in c else 0)

				assert change or move != 0

				if change:
					# Create a pair of states
					new_ifunset = add_state(writes.one, move, [ifunset, ifset])
					new_ifset = add_state(writes.zero, move, [ifunset, ifset])

					ifunset, ifset = new_ifunset, new_ifset

				else:
					# Create only one state
					ifunset = ifset = add_state(writes.same, move, [ifunset, ifset])

		if in_loop:
			# At the start of the loop body, the cell will always be set
			return ifset
		else:
			# At the start of the program, the cell will always be unset
			return ifunset

	return worker(chunked), states

def prettyprint_states(states):
	for i in range(len(states)):
		state = states[i]
		write = {writes.zero: '0 ', writes.one: '1 ', writes.same: ''}[state.write]
		move = {-1: '< ', 0: '', 1: '> '}[state.move]
		print('%i: %s%s%s' % (i, write, move, state.transitions))

def tree_shake(start, states):
	# Since we always end up with start at index 0, no need to create shaken_start
	shaken = [states[start]]
	translations = {start: 0, None: None}

	shaken_sweep = 0
	while shaken_sweep < len(shaken):
		for i in range(len(shaken[shaken_sweep].transitions)):
			index_states = shaken[shaken_sweep].transitions[i]

			if index_states not in translations:
				index_shaken = len(shaken)
				shaken.append(states[index_states])
				translations[index_states] = index_shaken

			shaken[shaken_sweep].transitions[i] = translations[index_states]

		shaken_sweep += 1

	return shaken

def main():
	program = input('program: ')
	start_nonshaken, states_nonshaken = turingify(chunk(parse_smallfuck(program)))
	shaken = tree_shake(start_nonshaken, states_nonshaken)
	prettyprint_states(shaken)

if __name__ == '__main__':
	main()