patmos-hex2bin

#!/usr/bin/python

from __future__ import print_function
import argparse
import sys
import os.path
import re
from binascii import unhexlify

""" 
This script reads the hexadecimal instruction encoding in Patmos .s files
generated by patmos-clang or llc -show-mc-encoding, and adds them in
binary encoding to the comments.

Author: Stefan Hepp <hepp@complang.tuwien.ac.at>
"""

parser = argparse.ArgumentParser(description='Reformat assembler encoding comments.', 
                                 usage='%(prog)s [-b|-f] [-o <outfile.s>|-i] <asmfile>')
parser.add_argument('asmfile', help='assembler file containing instruction encoding comments')
parser.add_argument('-b', '--binary', action='store_true', help='Produce binary file containing all instructions (use with -o)')
parser.add_argument('-f', '--bytes', action='store_true',  help='Group binary representation per byte, not per field')
parser.add_argument('-i', '--inplace', action='store_true', help='Perform replace in place on the input file.')
parser.add_argument('-o', '--output', help='Write output to file instead of stdout')

args = parser.parse_args()

if args.inplace and args.output:
    print("Error: -i not allowed with -o")
    sys.exit(1)

if args.binary:
    if args.inplace or args.asmfile == args.output:
	print("You probably do not want to replace your .s file with the binary, cowardly backing out.")
	sys.exit(2)
    if args.output is None:
	print("Error: -b must be used with -o")
	sys.exit(1)

if not os.path.isfile(args.asmfile):
    print("Input " + args.asmfile + " not found or not a file.")
    sys.exit(1)

try:
    infile = open(args.asmfile, 'r')
    lines = infile.readlines()
    infile.close()
except IOError as e:
    print("Cannot read input file " + args.asmfile + ": " + e)
    sys.exit(1)

outfile = None
if args.inplace or args.output:
    try:
        outfile = open(args.asmfile if args.inplace else args.output, 'w')
    except IOError as e:
        print("Cannot open output file for writing: " + e)
        sys.exit(2)

pat = re.compile("# encoding: \[( ?([0-9a-fx,A]*) *(\| [01x ]*)?)\]")

for line in lines:
    
    m = pat.search(line)
    if m:
        hex = m.group(2)
        
        bny = ""
        pad = ""
        for val in hex.split(','):
            if val == "A":
                bny += "00" if args.binary else "xxxxxxxx"
                pad += "   "
                continue
            if args.binary:
                bny += val[2:4]
            else:
                bny += bin(int(val[2:4], 16))[2:].zfill(8) 
                bny += "" if args.bytes else " "
                 
        if args.binary:
            line = unhexlify(bin)
        else:
            if args.bytes:
                bny = (bny[0:1] + " " + bny[1:5] + " " + bny[5:10] + " " +
                       bny[10:15] + " " + bny[15:20] + " " + bny[20:25] + " " +
                       bny[25:32] + " " + bny[32:] + 
                       (" " if len(bny) >= 32 else ""))
            
            line = line.replace(m.group(1), " %s %s| %s" % (hex, pad, bny))
    
    if outfile:
        outfile.write(line)
    else:
        print(line, end='')

if outfile:
    outfile.close()