#!/usr/bin/env python3
"""Merge several Mesen CDL captures into one.

Mesen's CDL accumulates: a capture is the OR of everything the CPU did since
the logger was reset.  Two captures of different sessions therefore hold
different subsets of the execution, and the union of them covers more of the
game than either does alone.  The flags are merged with OR, so a byte that is
Code in one capture stays Code in the merge.

Usage:
    python3 _tools/merge_cdl.py <out.cdl> <in.cdl> [in.cdl ...] [--size 32768]
"""

import argparse
from pathlib import Path

CODE, DATA, EXEC, JUMP = 0x01, 0x02, 0x04, 0x08


def read(path, size):
    raw = Path(path).read_bytes()
    if raw[:5] == b"CDLv2":
        body = raw[9:9 + size]
    else:
        body = raw[-size:]
    if len(body) < size:
        raise SystemExit(f"{path}: expected {size} CDL bytes, got {len(body)}")
    return body


def main():
    ap = argparse.ArgumentParser(description=__doc__.splitlines()[0])
    ap.add_argument("out")
    ap.add_argument("ins", nargs="+")
    ap.add_argument("--size", type=int, default=32768)
    args = ap.parse_args()

    merged = bytearray(args.size)
    for p in args.ins:
        body = read(p, args.size)
        for i, b in enumerate(body):
            merged[i] |= b

    code = sum(1 for b in merged if b & CODE)
    data = sum(1 for b in merged if b & DATA)
    jump = sum(1 for b in merged if b & JUMP)
    print(f"merged {len(args.ins)} capture(s): {code} code bytes, "
          f"{data} data bytes, {jump} jump targets")

    out = bytearray(b"CDLv2")
    out += args.size.to_bytes(4, "little")
    out += merged
    Path(args.out).write_bytes(bytes(out))
    print(f"wrote {args.out}")


if __name__ == "__main__":
    main()