#!/usr/bin/env python3
"""Turn a false decode back into data.

A `jsr $XXXX` whose target carries no instruction is the tracer having
decoded a data byte as a call.  The instruction is not code, so it is
emitted as the explicit `.byte` row it always was, with the same bytes and
the same audit comment - the build cannot change - and the raw operand
disappears.

The tool only touches a call when the target really has no instruction
somewhere in the file; a call into a routine the tracer simply never reached
is left alone, because that is a coverage problem and not a data problem.

Usage:
    python3 _tools/fix_false_decodes.py <project dir> [--write]
"""

import argparse
import re
from pathlib import Path

INSN = re.compile(
    r"^    (?P<mn>[A-Za-z@][A-Za-z0-9_]*)\s+(?P<op>\S+)\s*; "
    r"CPU:\$(?P<cpu>[0-9A-F]{4}) PRG:(?P<prg>[0-9A-F]{4})\s+"
    r"(?P<bytes>(?:[0-9A-F]{2} ?)+)\s*$")
LABEL = re.compile(r"^([A-Za-z_][A-Za-z0-9_]*):\s*$")
RAW = re.compile(r"^\$(?:jsr|jmp)\s")
CALL = re.compile(r"^\$([0-9A-F]{4})$")


def main():
    ap = argparse.ArgumentParser(description=__doc__.splitlines()[0])
    ap.add_argument("project")
    ap.add_argument("--write", action="store_true")
    args = ap.parse_args()

    path = Path(args.project) / "prg.s"
    lines = path.read_text().split("\n")

    # every address that carries an instruction
    have = set()
    for line in lines:
        m = INSN.match(line)
        if m:
            have.add(int(m.group("cpu"), 16))

    out = []
    changed = 0
    for line in lines:
        m = INSN.match(line)
        if m:
            op = m.group("op")
            c = CALL.match(op)
            if m.group("mn") in ("jsr", "jmp") and c and \
                    int(c.group(1), 16) not in have:
                # a call into nowhere: the bytes are data
                raw = m.group("bytes").split()
                vals = ", ".join(f"${b}" for b in raw)
                out.append(f"    .byte {vals}  ; CPU:${m.group('cpu')} "
                           f"PRG:{m.group('prg')}  {m.group('bytes').strip()}"
                           f"  -- data, not a call")
                changed += 1
                continue
        out.append(line)

    print(f"{changed} false decode(s) turned back into data")
    if args.write and changed:
        path.write_text("\n".join(out))
        print("written")


if __name__ == "__main__":
    main()