#!/usr/bin/env python3
"""DOSBox vs dosemu on the real workload: GSPA assembling TMS34010 source.

Workload A: gspa t1 t1.obj t1.lst -l     (4 KB region, listing)
Workload B: gspa -q rt16 rt16.obj        (16 KB region, no listing)

Both headless: DOSBox with SDL dummy drivers and core=dynamic/cycles=max;
dosemu in -t (terminal/S-Lang) mode.  Three measured runs each, plus one
warm-up.  Raw COFF md5 differs only in the 4-byte time stamp, so a
timestamp-normalised md5 is also reported.
"""
import hashlib
import os
import statistics
import subprocess
import time
from pathlib import Path

ROOT = Path(__file__).resolve().parent
BENCH = ROOT / "bench"
DHOME = ROOT / "dosemu-home"
DBENCH = DHOME / ".dosemu" / "drive_c" / "bench"
RUNS = 3


def norm_md5(p):
    if not p.exists():
        return "-"
    b = bytearray(p.read_bytes())
    if len(b) >= 8:
        b[4:8] = b"\0\0\0\0"
    return hashlib.md5(bytes(b)).hexdigest()


def raw_md5(p):
    return hashlib.md5(p.read_bytes()).hexdigest() if p.exists() else "-"


def clean(d, base):
    for name in (base + ".OBJ", base + ".LST", "MACLIB.$$$",
                 base.lower() + ".obj", base.lower() + ".lst", "maclib.$$$"):
        f = d / name
        if f.exists():
            f.unlink()


def run_dosbox(base, args):
    clean(BENCH, base)
    cmds = []
    for _ in range(1):
        cmds += ["-c", "gspa " + args]
    env = dict(os.environ, SDL_VIDEODRIVER="dummy", SDL_AUDIODRIVER="dummy")
    t0 = time.perf_counter()
    rc = subprocess.run(
        ["timeout", "120", "dosbox", "-conf", str(ROOT / "dosbox-bench.conf"),
         "-noprimaryconf", "-nolocalconf",
         "-c", "mount c " + str(BENCH), "-c", "c:"] + cmds +
        ["-c", "exit", "-exit"],
        env=env, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL).returncode
    dt = time.perf_counter() - t0
    # DOSBox uppercases 8.3 names
    obj = BENCH / (base.upper() + ".OBJ")
    return dt, rc, raw_md5(obj), norm_md5(obj), obj.stat().st_size if obj.exists() else 0


def run_dosemu(base, args, bat):
    clean(DBENCH, base)
    script = "@echo off\r\nc:\r\ncd \\bench\r\ngspa %s\r\nexitemu\r\n" % args
    (DBENCH / bat).write_bytes(script.encode())
    env = dict(os.environ, HOME=str(DHOME))
    t0 = time.perf_counter()
    rc = subprocess.run(
        ["timeout", "120", "dosemu", "-t", "-quiet", "-E", r"c:\bench" + "\\" + bat],
        env=env, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL).returncode
    dt = time.perf_counter() - t0
    # dosemu/lredir keeps the lowercase names written by FreeCom
    obj = DBENCH / (base.lower() + ".obj")
    if not obj.exists():
        obj = DBENCH / (base.upper() + ".OBJ")
    return dt, rc, raw_md5(obj), norm_md5(obj), obj.stat().st_size if obj.exists() else 0


def bench(name, base, args, bat):
    print("\n== %s ==  gspa %s" % (name, args))
    run_dosbox(base, args)            # warm-up
    run_dosemu(base, args, bat)
    results = {}
    for label, fn in (("dosbox", lambda: run_dosbox(base, args)),
                      ("dosemu", lambda: run_dosemu(base, args, bat))):
        rows = [fn() for _ in range(RUNS)]
        results[label] = rows
        for dt, rc, raw, norm, size in rows:
            print("  %-7s %6.2fs rc=%d size=%d raw=%s norm=%s" %
                  (label, dt, rc, size, raw[:8], norm[:8]))
        print("  %-7s median %.2fs" % (label, statistics.median(r[0] for r in rows)))
    n0 = {r[3] for r in results["dosbox"]}
    n1 = {r[3] for r in results["dosemu"]}
    print("  normalised object identical: %s (%s)" % (n0 == n1 and len(n0) == 1, n0))


if __name__ == "__main__":
    bench("A: 4 KB with listing", "t1", "t1 t1.obj t1.lst -l", "bench.bat")
    bench("B: 16 KB, no listing", "rt16", "-q rt16 rt16.obj", "bench16.bat")