hare-nix/pkgs/hare-as.py

#!python3

import re
import sys
from subprocess import Popen, PIPE

def header(filename):
    return f"""
# ------------------
.file "{filename}"
# ------------------
"""

REPLACEMENTS = [
    (r'^\.section "\.text([^"]*)"(,"([^"]*))?"$', r'.text ; \1, \3'),
    (r'^\.section "\.data([^"]*)"$', r'.data ; \1'),
    (r'^\.section "\.bss([^"]*)"$', r'.bss ; \1'),
    (r'^\.section "\.init_array"(,"([^"]*)")?', r'.section __DATA, .init_array ; \2'),
    (r'^\.section "\.fini_array"(,"([^"]*)")?', r'.section __DATA, .fini_array ; \2'),
    (r'^\.section "\.test_array"(,"([^"]*)")?', r'.section __DATA, .test_array ; \2'),
    (r'^\.section \.abort "([^"]+)"', r'.section __DATA, .abort $COMMENT \1'),
    (r'^(\.type .+)', '; \1'),
    (r'^(\.global) (rt\..+)', r'\1 _\2'),
    (r'^(rt\..+:)', r'_\1'),
    (r'^(crypto\.aes\.x86ni_.+:)', r'_\1'),
    (r'^(debug\.getfp:)', r'_\1'),
    (r'adrp	x0, _environ@page', r'adrp	x0, _environ@GOTPAGE'),
    (r'add	x0, x0, _environ@pageoff', r'ldr	x0, [x0, _environ@GOTPAGEOFF]'),
]

def fix_asm(code):
    for (pattern, repl) in REPLACEMENTS:
        code = re.sub(pattern, repl, code, flags=re.MULTILINE)
    return code

args = ['as']
files = []
argv = sys.argv[1:]
while len(argv):
    if argv[0] == "--":
        args.append("--")
    elif argv[0] in ["-o", "-I", "-arch"]:
        args.append(argv[0])
        args.append(argv[1])
        argv.pop(0)
    elif argv[0][0] == "-":
        argv.append(argv[0])
    else:
        files.append(argv[0])

    argv.pop(0)

args.append("-")

code = ""
for file in files:
    with open(file) as fp:
        content = fp.read()
        code += header(file)
        code += "\n"
        code += fix_asm(content)
        code += "\n"

p = Popen(args, stdout=PIPE, stdin=PIPE, stderr=PIPE, text=True)
data = p.communicate(input=code)
print(data[1], end='', file=sys.stderr)
print(data[0], end='')