Implementing Brainfuck loops in an interpreter

Question

I want to build a Brainfuck (Damn that name) interpreter in my freshly created programming language to prove it\'s turing-completeness.

Now, everything is clear so far (

Answer 1

Here is my "optimizing" version of interpreter, that pre-compiles the loop jumps.

def interpret2(code):
    data = [0] * 5000   # data memory
    cp = 0              # code pointer
    dp = 0              # data pointer
    # pre-compile a jump table
    stack = []
    jump = [None] * len(code)
    for i,o in enumerate(code):
        if o=='[':
            stack.append(i)
        elif o==']':
            jump[i] = stack.pop()
            jump[jump[i]] = i
    # execute
    while cp < len(code):
        cmd = code[cp]
        if   cmd == '>': dp += 1
        elif cmd == '<': dp -= 1
        elif cmd == '+': data[dp] += 1 
        elif cmd == '-': data[dp] -= 1 
        elif cmd == '.': stdout.write(chr(data[dp]))
        elif cmd == ',': data[dp] = ord(stdin.read(1))
        elif cmd == '[' and not data[dp]: # skip loop if ==0
            cp = jump[cp]
        elif cmd == ']' and data[dp]:     # loop back if !=0
            cp = jump[cp]
        cp += 1

It does a dry run of the code, keeping track of the brackets (in a stack) and marks the goto addresses in parallel jump array which is later consulted during execution.

I compared the execution speed on long-running BF program (calculate N digits of Pi) and this increased the speed 2x compared to an innocent implementation in which source is scanned forward to exit [ and scanned backwards to loop on ].