#!/usr/bin/python
'generate EQU statements for CALL, JMP, LOOP statements'
import sys, os, re
instruction_pattern = '^\( [0-9A-F]{4},[0-9A-F]{4} \)\s+' + \
 '(J|CALL|LOOP|\S+).* ([0-9A-F-]+) \(R[BWL],\).*$'
long_instruction_pattern = '^\( [0-9A-F]{4},[0-9A-F]{4} \)\s+' + \
 '(J|CALL|LOOP|\S+).* 0020,([0-9A-F-]+) I?L,.*$'
address_pattern = '^\( ([0-9A-F]{4}),([0-9A-F]{4}) \)\s+'
equ_pattern = '^([0-9A-F]+) EQU '
names = {'J': 'skip', 'CALL': 'subr', 'LOOP': 'loop'}
def readall(filename):
 file = open(filename)
 lines = map(str.strip, file.readlines())
 file.close()
 return lines
def append(filename, lines):
 newfile = readall(filename)
 for line in lines:
  if line not in newfile:
   insert(newfile, line)
 file = open(filename, 'w')
 file.writelines(map(lambda l: l + os.linesep, newfile))
 file.close()
def insert(lines, new):
 '''insert new EQU line in numeric order

    for this to work, you must have manually inserted at least one EQU'''
 newvalue = int(re.compile(equ_pattern).match(new).groups()[0], 16)
 for index in range(len(lines)):
  first = re.compile(equ_pattern).match(lines[index])
  second = re.compile(equ_pattern).match(lines[index + 1])
  if not first:
   continue
  else:
   first = int(first.groups()[0], 16)
  if second:
   second = int(second.groups()[0], 16)
  else:
   second = 0xffffffffL
  if newvalue < first:
   lines.insert(index, new)
   break
  elif newvalue < second:
   lines.insert(index + 1, new)
   break
def genequ(asmfile, culfile = None, equates = []):
 lines = readall(asmfile)
 for index in range(len(lines)):
  line = lines[index]
  instruction_match = re.compile(instruction_pattern).match(line)
  long_instruction_match = None
  #long_instruction_match = re.compile(long_instruction_pattern).match(line)
  if instruction_match:
   if index + 1 < len(lines) and \
    re.compile(address_pattern).match(lines[index + 1]):
    instr, offset = instruction_match.groups()
    address = re.compile(address_pattern).match(lines[index + 1]).groups()
    final_address = int(address[0], 16) * 0x10000 + \
     int(address[1], 16) + int(offset, 16)
   addr = '%X' % final_address
  elif long_instruction_match:
   instr, address = long_instruction_match.groups()
   final_address = 0x200000 + int(address, 16)
   addr = '0020,%04X' % (final_address & 0xffff)
  if instruction_match or long_instruction_match:
   try:
    label = '%s%X' % (names[instr], final_address)
   except:
    label = '%s%X' % ('data', final_address)
   equ = '%s EQU %s' % (addr, label)
   if re.compile('^[ABCDEF]').match(equ):
    equ = '0' + equ
   equates.append(equ)
 if culfile:
  append(culfile, equates)
 else:
  print repr(equates)
if __name__ == '__main__':
 sys.argv += ['', '']
 genequ(sys.argv[1], sys.argv[2])