femtolisp_dis.py

"disassemble femtoLisp bytecode (kind of)"

"""
a serialized function looks like this:

#fn(
    ":000r1|Mc0<17702|M]<6@0|N\x8550|M;c1|NK;|N\x85@0c2|Mi10~N31L3;|\x84c3\x82W0e4e5|31316A0c6qe7e5|313141;c8qc93041;c:|Mc1|NKi10~N31L4;"
    [
        else begin or => 1arg-lambda? caddr
        #fn("=000r1c0|~ML2L1c1|c2e3e4~3131Ki20i10N31L4L3;" [let if begin cddr caddr])
        caadr
        #fn("<000r1c0|~ML2L1c1|e2~31|L2i20i10N31L4L3;" [let if caddr])
        #fn(gensym) if
    ]
)

from:
    https://github.com/JeffBezanson/femtolisp/blob/dc621773236a55a441dc2dce6b1c5d08e7fa10a1/flisp.boot
"""

code = b':000r1|Mc0<17702|M]<6@0|N\x8550|M;c1|NK;|N\x85@0c2|Mi10~N31L3;|\x84c3\x82W0e4e5|31316A0c6qe7e5|313141;c8qc93041;c:|Mc1|NKi10~N31L4'

vals = [
    'else', 'begin', 'or', '=>', '1arg-lambda?', 'caddr',
    '#fn("=000r1c0|~ML2L1c1|c2e3e4~3131Ki20i10N31L4L3;" [let if begin cddr caddr])',
    'caadr',
    '#fn("<000r1c0|~ML2L1c1|e2~31|L2i20i10N31L4L3;" [let if caddr])',
    '#fn(gensym)', 'if'
]

def main():
    disassemble(code, vals)




# from:
#   https://github.com/JeffBezanson/femtolisp/blob/dc621773236a55a441dc2dce6b1c5d08e7fa10a1/compiler.lsp

_insts = """nop dup pop call tcall jmp brf brt jmp.l brf.l brt.l ret

eq? eqv? equal? atom? not null? boolean? symbol?
number? bound? pair? builtin? vector? fixnum? function?

cons list car cdr set-car! set-cdr!
apply

+ - * / div0 = < compare

vector aref aset!

loadt loadf loadnil load0 load1 loadi8
loadv loadv.l
loadg loadg.l
loada loada.l loadc loadc.l
setg setg.l
seta seta.l setc setc.l

closure argc vargc trycatch for tapply
add2 sub2 neg largc lvargc
loada0 loada1 loadc00 loadc01 call.l tcall.l
brne brne.l cadr brnn brnn.l brn brn.l
optargs brbound keyargs

dummy_t dummy_f dummy_nil"""

insts = dict(enumerate(_insts.split()))

import struct


def disassemble(code, func_vals=None):
    if func_vals is None: func_vals = {}

    # https://github.com/JeffBezanson/femtolisp/blob/ec7601076a976f845bc05ad6bd3ed5b8cde58a97/flisp.c#L2038
    if code[4] > len(insts):
        code = bytes(b-48 for b in code) 

    res = []
    max_stack = i32_le(code[0: 4])
    i = 4
    while i < len(code):
        try:
            byte = code[i]
            inst_loc = i
            inst = insts[byte]
            i += 1
            args = []
            for off in arg_offsets(inst):
                val = code[i] if off == 1 else {2:i16_le, 4:i32_le}[off](code[i:i+off])
                args.append(val)
                i += off
            if inst is None:
                inst = byte
            res.append((inst, args))
            out = ' '.join((
                '{:>4} '.format(inst_loc), '{:<7}'.format(inst), *('{:>3}'.format(a) for a in args),
            ))
            if is_jump(inst):
                out += '  | ' + str(args[0] + inst_loc)
            elif uses_vals(inst):
                out += '  | ' + str(list_get(func_vals, get_val_index(inst, args), ''))
            print(out)

        except Exception as e:
            msg = ' '.join(['error at i =', i, 'inst =', inst, 'code[i] = ', code[i], ' = ', repr(chr(code[i]))])
            raise Exception((msg, i, res)) from e
    return (max_stack, res)






def is_jump(inst):
    return inst in 'jmp brf brt brne brnn brn jmp.l brf.l brt.l brne.l brnn.l brn.l'.split(' ')

def uses_vals(inst):
    return (inst in 'loadv.l loadg.l setg.l loadv loadg setg load0 load1') 

def get_val_index(inst, args):
    if (inst in 'loadv.l loadg.l setg.l loadv loadg setg'): return args[0]
    if (inst in 'load0 load1'): return int(inst[-1])


def arg_offsets(inst):
    if   inst in ('loadv.l loadg.l setg.l'.split(' ')):
        offs = [4]
    elif inst in ('loadv loadg setg'.split(' ')):
        offs = [1]
    elif inst in ('loada seta call tcall list + - * / vector argc vargc loadi8 apply tapply'.split(' ')):
        offs = [1]
    elif inst in ('loada.l seta.l largc lvargc call.l tcall.l'.split(' ')):
        offs = [4]
    elif inst in ('loadc setc'.split(' ')):
        offs = [1, 1]
    elif inst in ('loadc.l setc.l optargs'.split(' ')):
        offs = [4, 4]
    elif inst in ('keyargs'.split(' ')):
        offs = [4, 4, 4]
    elif inst in ('brbound'.split(' ')):
        offs = [4]
    elif inst in ('jmp brf brt brne brnn brn'.split(' ')):
        offs = [2]
    elif inst in ('jmp.l brf.l brt.l brne.l brnn.l brn.l'.split(' ')):
        offs = [4]
    else:
        offs = []
    return offs


def list_get(xs, i, default=None): return xs[i] if i < len(xs) else default

def i32_le(bs): return struct.unpack('<i', bs)[0]
def i16_le(bs): return struct.unpack('<h', bs)[0]

if __name__ == '__main__':
    main()