Script to convert retdec disassembly output into GNU as input

dsm2s.py

#!/usr/bin/env python3
import sys, argparse, re

#0x438e1d:   8b 45 84                            mov eax, dword ptr [ebp - 0x7c]
#0x402a4c:   cc cc cc cc                                        |....            |
re_line = re.compile(r'''^0x([0-9a-f]+):   ((?:[0-9a-f][0-9a-f]  ?){1,}) *\t(.*)$''')
re_hex  = re.compile(r'''^0x([0-9a-f]+):   ((?:[0-9a-f?][0-9a-f?]  ?){1,}) *\|(.{16})\|(.*)$''')

def run():
  ap = argparse.ArgumentParser()
  ap.add_argument('input-file')
  args = vars(ap.parse_args())

  with open(args['input-file'], 'r') as f:
    for L in f:
      L = L.rstrip('\r\n')
      if L == '':
        continue

      if L.startswith('; section: '):
        secName = L[11:]
        print('.section %s' % secName)
        continue

      if L[0] == ';':
        print('#'+L[1:])
        continue

      m = re_line.match(L)
      if m:
        addr    = m.group(1)
        mcode   = ['0x'+x for x in m.group(2).strip(' ').replace('  ',' ').split(' ')]
        disasm  = m.group(3)

        s = '.byte %s' % ','.join(mcode)

        s = s.ljust(62,' ')
        s += '#%s; %s' % (addr, disasm)
        print(s)
        continue

      m = re_hex.match(L)
      if m:
        addr    = m.group(1)
        mcode   = ['0x'+x for x in m.group(2).strip(' ').replace('  ',' ').split(' ')]
        hexdump = m.group(3)
        comment = m.group(4)

        s = '.byte %s' % ','.join(mcode)
        s = s.ljust(90,' ')
        s += '#%s;|%s|%s' % (addr, hexdump, comment)
        print(s)
        continue

      print('NOMATCH: %s' % L)

  return 0

if __name__ == '__main__':
  sys.exit(run())