TurboModula2/tools/score.py

"""MCD proc-compare scorer. Usage: score.py ORIG_TXT OURS_TXT MOD PAIRS...
PAIRS = OrigName:OursName. Masks absolute targets, proc refs, dep slots;
leave-truncates bodies; reports SAME/PREFIX/DIFF per pair."""
import re, sys

def procs(path, mod):
    txt = open(path, errors='replace').read()
    heads = [(m.group(1), int(m.group(2), 16))
             for m in re.finditer(r'(?m)^;? ?' + mod + r'\.(\w+)\s*: ([0-9a-f]{4})$', txt)]
    ops = []
    cur = None
    for line in txt.splitlines():
        m = re.match(r'^\s*([0-9a-f]{4})\s+((?:[0-9a-f]{2}\s*)+?)(?:\s{2,}|\t+|$)(.*)$', line)
        if m:
            cur = int(m.group(1), 16)
            ops.append((cur, m.group(2).split(), m.group(3).strip()))
            continue
        # case-table lines carry no address: attach to preceding cd op so
        # CASE bounds/labels participate in comparison (else invisible).
        # NOTE: unassemble prints labels +90 (subtracts 32678 not 32768);
        # the shift applies equally to both sides, so comparison stays valid.
        c = re.match(r'^\s*case (\d+) => ([0-9a-f]{4})$', line)
        if c and cur is not None:
            ops.append((cur, ['cd'], 'case %d => %s' % (int(c.group(1)), c.group(2))))
    out = {}
    addrs = sorted(set(a for _, a in heads))
    for name, a in heads:
        if name in out:
            continue
        nxt = min([x for x in addrs if x > a] or [0x10000])
        out[name] = [(o, b, m) for (o, b, m) in ops if a <= o < nxt and b != ['00'] and m]
    return out

def norm(m):
    m = re.sub(r'\b[0-9a-f]{4}\b', '@@@@', m)
    m = re.sub(r'call_with_frame proc\d+', 'call_with_frame proc##', m)
    m = re.sub(r'nested_call proc\d+', 'nested_call proc##', m)
    m = re.sub(r'(?<!_)proc\d+', 'proc##', m)
    return m

def trunc(ms):
    li = [i for i, m in enumerate(ms)
          if m.split()[0] in ('leave', 'leave0', 'leave2', 'leave4', 'leave6',
                              'fct_leave', 'longfct_leave', 'end_program', 'Raise')]
    return ms[:li[-1] + 1] if li else ms

def main():
    orig_txt, ours_txt, mod = sys.argv[1], sys.argv[2], sys.argv[3]
    A, B = procs(orig_txt, mod), procs(ours_txt, mod)
    same = tot = 0
    for pair in sys.argv[4:]:
        o, u = pair.split(':')
        tot += 1
        if o not in A:
            print(f'{u}/{o} ORIG-MISSING'); continue
        if u not in B:
            print(f'{u}/{o} OURS-MISSING'); continue
        xa, xb = trunc([norm(m) for (_, _, m) in A[o]]), trunc([norm(m) for (_, _, m) in B[u]])
        n = min(len(xa), len(xb))
        if xa == xb:
            same += 1
            print(f'{u}/{o} {len(xa)} vs {len(xb)} SAME')
        elif xa[:n] == xb[:n]:
            print(f'{u}/{o} {len(xa)} vs {len(xb)} PREFIX(+{abs(len(xa)-len(xb))})')
        else:
            print(f'{u}/{o} {len(xa)} vs {len(xb)} DIFF')
    print(f'{same}/{tot} exact')

main()