Naive brainfuck-to-C compiler

bf2c.py

#!/usr/bin/env python
import argparse
import sys


def tokenize(source):
    return list(source)


def parse(tokens):
    depth = 0
    ast = []
    stack = [ast]
    node = ast
    for token in tokens:
        if token == '[':
            node.append([])
            stack.append(node)
            node = node[-1]
            depth += 1
        elif token == ']':
            node = stack.pop()
            depth -= 1
        elif token in "+-<>,.":
            node.append(token)
    if not depth == 0:
        raise SyntaxError("Unmatched opening bracket")
    return ast


def compile_to_c(tree):
    buf = ""
    for node in tree:
        if isinstance(node, list):
            buf += "while (*p) { "
            buf += compile_to_c(node)
            buf += "} "
        elif node == '+':
            buf += "++*p; "
        elif node == '-':
            buf += "--*p; "
        elif node == '>':
            buf += "++p; "
        elif node == '<':
            buf += "--p; "
        elif node == ",":
            buf += "*p = getchar(); "
        elif node == ".":
            buf += "putchar(*p); "
    return buf

if __name__ == '__main__':
    ap = argparse.ArgumentParser(description="Transpile a Brainfuck program to a different host language")
    ap.add_argument('infile', nargs='?', type=argparse.FileType('r'), \
        default='-', help="File to read source from, defaults to stdin")
    ap.add_argument('-t', '--target', action='store', type=str, \
        choices=set(['c']), default='c', help="Output language to target")
    args = ap.parse_args()

    code = args.infile.read()
    tokens = tokenize(code)
    ast = parse(tokens)

    if args.target == 'c':
        print "#include <stdio.h>"
        print "#include <stdlib.h>"
        print "int main(int argc, char **argv) {"
        print "  int *p = (int *)malloc(30000 * sizeof(int));"
        print " ", compile_to_c(ast)
        print "  return 0;"
        print "}"

helloworld.bf

>+++++++++
[<++++++++>-]
<.>+++++++
[<++++>-]
<+.+++++++..+++.>>>++++++++
[<++++>-]
<.>>>++++++++++
[<+++++++++>-]
<---.<<<<.+++.------.--------.>>+.

helloworld.c

#include <stdio.h>
#include <stdlib.h>
int main(int argc, char **argv) {
  int *p = (int *)malloc(30000 * sizeof(int));
  ++p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  while (*p) {
    --p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++p;
    --*p;
  }
  --p;
  putchar(*p);
  ++p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  while (*p) {
    --p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++p;
    --*p;
  }
  --p;
  ++*p;
  putchar(*p);
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  putchar(*p);
  putchar(*p);
  ++*p;
  ++*p;
  ++*p;
  putchar(*p);
  ++p;
  ++p;
  ++p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  while (*p) {
    --p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++p;
    --*p;
  }
  --p;
  putchar(*p);
  ++p;
  ++p;
  ++p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  ++*p;
  while (*p) {
    --p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++*p;
    ++p;
    --*p;
  }
  --p;
  --*p;
  --*p;
  --*p;
  putchar(*p);
  --p;
  --p;
  --p;
  --p;
  putchar(*p);
  ++*p;
  ++*p;
  ++*p;
  putchar(*p);
  --*p;
  --*p;
  --*p;
  --*p;
  --*p;
  --*p;
  putchar(*p);
  --*p;
  --*p;
  --*p;
  --*p;
  --*p;
  --*p;
  --*p;
  --*p;
  putchar(*p);
  ++p;
  ++p;
  ++*p;
  putchar(*p);

  return 0;
}

Thanks for your article Justin.