jb302@1: #!/usr/bin/env python2
jb302@1: # assembler.py
jb302@1: import struct
jb302@1: import sys
jb302@1: from language import *
jb302@1: 
jb302@1: # take source file and return preprocessed assembly code
jb302@1: # for each non-empty line in the file:
jb302@1: #   remove comments from source
jb302@25: #   replace equated strings
jb302@25: #   store label definitions and remove label from source
jb302@25: #   store new equates
jb302@25: #   make hashable format symbol from arguments
jb302@25: #   identify and save constant data
jb302@25: #   save instruction, arguments, symbol and data to list
jb302@25: # also prepares org and db instructions for second_pass()
jb302@1: def first_pass(f):
jb302@25:     asm = []
jb302@25:     labels = {}
jb302@25:     equates = {}
jb302@25:     pc = 0
jb302@25:     
jb302@1:     # read file into list, remove blank line
jb302@1:     f.seek(0)
jb302@25:     source_code = filter(lambda l: l != '\n', f.readlines())  
jb302@14:     
jb302@1:     # <line> ::= [<statement>] [";"<comment>] <EOL>
jb302@22:     for line in source_code:  
jb302@18:         try:
jb302@22:             # remove trailing whitespace and comments
jb302@18:             line = line.strip()
jb302@18:             for i in range(len(line)):
jb302@18:                 if line[i] == ';':
jb302@18:                     line = line[:i]
jb302@18:                     break
jb302@18:             
jb302@18:             # <statement> ::= [ <label> ":"] <mnemonic> [<arguments>]
jb302@18:             #                 | <label> ":"
jb302@18:             #                 | "EOF"
jb302@18:             statement = line.split()
jb302@22:             if not statement:
jb302@22:                 continue
jb302@22: 
jb302@22:             # replace equated strings
jb302@22:             # because this happens on the first pass
jb302@22:             # equates must be assigned before they are used
jb302@25:             i = 1
jb302@25:             for s in statement[1:]:
jb302@22:                 # replace any equates already stored
jb302@22:                 # remove prefixes and suffixes before attempting to replace
jb302@25:                 prefix = suffix = ''
jb302@25:                 # prefixes
jb302@25:                 if s[0] in ids:
jb302@25:                     prefix = prefix + s[0]
jb302@25:                     s = s[1:]
jb302@25:                 if s[0] == '(':
jb302@25:                     prefix = prefix + s[0]
jb302@25:                     s = s[1:]
jb302@25:                 # suffixes
jb302@25:                 if s and (s[-1] == ','):
jb302@25:                     suffix = suffix + s[-1]
jb302@25:                     s = s[:-1]
jb302@25:                 if s and (s[-1] == ')'):
jb302@25:                     suffix = s[-1] + suffix
jb302@25:                     s = s[:-1]
jb302@22:                 # replace and put removed characters back
jb302@25:                 if s in equates:
jb302@25:                     statement[i] = prefix + equates[s] + suffix
jb302@23:                 # labels can be used in equates but they have
jb302@25:                 # to be assigned before they are used as well
jb302@25:                 elif s in labels:
jb302@34:                     if statement[0].lower() in rinst:
jb302@26:                         statement[i] = prefix + str(labels[s] - pc) + suffix
jb302@26:                     else:
jb302@26:                         statement[i] = prefix + str(labels[s]) + suffix
jb302@25:                 i = i + 1                   
jb302@18:             
jb302@23:             # deal with org
jb302@22:             if statement[0].lower() == 'org':
jb302@23:                 asm.append(['org', statement[1:], ('',), ''])
jb302@25:                 pc = stoi(statement[1])
jb302@18:                 continue
jb302@23:             # if needed update index and remove label 
jb302@23:             elif statement[0][-1] == ':':
jb302@25:                 labels[statement[0][:-1]] = pc;
jb302@23:                 del statement[0]
jb302@18:             # store equates
jb302@22:             # these are case sensative
jb302@26:             elif (len(statement) >= 3) and (statement[1].lower() == 'equ'):
jb302@25:                 equates[statement[0]] = ' '.join(statement[2:])
jb302@18:                 continue
jb302@18:                 
jb302@22:             if not statement:
jb302@22:                 continue
jb302@18:             
jb302@18:             # <statement> ::= <mnemonic> [<arguments>]
jb302@25:             mne = statement[0].lower()
jb302@25:             args = ''.join(statement[1:]).split(',')
jb302@25:         
jb302@22:             # deal with db
jb302@25:             if mne == 'db':
jb302@25:                 const = ''
jb302@25:                 for a in args:
jb302@25:                     data = tokenize(mne, ['#' + a])[1]
jb302@25:                     # deal with leading zeros
jb302@25:                     # skip zeros unless zero is the 
jb302@25:                     # only number
jb302@25:                     if data == '\x00\x00':
jb302@25:                         const = const + '\x00'
jb302@25:                         continue
jb302@25:                     i = 0
jb302@25:                     for c in data:
jb302@25:                         if c == '\x00':
jb302@25:                             i = i + 1
jb302@25:                         else:
jb302@25:                             pass
jb302@25:                     const = const + data[i:]
jb302@25:                 asm.append([mne, args, ('',), const])
jb302@25:                 pc = pc + len(const)
jb302@22:                 continue
jb302@40:             elif mne == 'ds':
jb302@40:                 asm.append([mne, args, ('',), ''])
jb302@40:                 pc = pc + stoi(args[0])
jb302@40:                 continue
jb302@40: 
jb302@40: 
jb302@22:             
jb302@25:             # tokenize
jb302@25:             sym, const = tokenize(mne, args)
jb302@25:             asm.append([mne, args, sym, const])
jb302@25:             # increase pc
jb302@25:             width = iset[mne][sym][1]
jb302@18:             pc = pc + width
jb302@12:         
jb302@18:         except:
jb302@18:             print ' ** first pass error **\nline:\n', line
jb302@18:             raise
jb302@19:     
jb302@25:     return asm, labels
jb302@1: 
jb302@25: # take a preprocessed object asm and write machine code to binary file
jb302@25: # for each line of asm:
jb302@19: #   check if it's an org or db command deal with it accordingly
jb302@19: #   check if arguments are labels and replace with value
jb302@19: #   write instruction to file
jb302@34: def second_pass(f, asm, labels, d=None):
jb302@1:     pc = 0
jb302@14: 
jb302@14:     for line in asm:
jb302@18:         f.seek(pc)
jb302@14:         mne, args, sym, const = line
jb302@1:     
jb302@1:         try:
jb302@25:             # deal with org and db
jb302@18:             if mne == 'org':
jb302@25:                 pc = stoi(args[0])
jb302@18:                 continue
jb302@18:             elif mne == 'db':
jb302@23:                 f.write(const)
jb302@23:                 pc = pc + len(const)
jb302@18:                 continue
jb302@40:             elif mne == 'ds':
jb302@40:                 pc = pc + stoi(args[0])
jb302@40:                 continue
jb302@18:             
jb302@18:             # replace labels with addresses
jb302@18:             i = 0
jb302@18:             for a in args:
jb302@22:                 if not a:
jb302@22:                     continue
jb302@19:                 elif (sym[i] == 'label') or (sym[i] == '@label'):
jb302@19:                     # labeled pointer uglyness
jb302@25:                     if (a[0] == '@') and (a[1:] in labels):
jb302@25:                         args[i] = '@' + str(labels[a[1:]])
jb302@25:                         const = const + tokenize(mne, [args[i]])[1]
jb302@25:                     else:
jb302@19:                         # check if constant needs to be a relative address
jb302@25:                         if mne in rinst:
jb302@25:                             args[i] = str(labels[a] - pc)
jb302@19:                         else:
jb302@25:                             args[i] = str(labels[a])
jb302@25:                         const = const + tokenize(mne, [args[i]])[1]  
jb302@18:                 i = i + 1
jb302@18:             
jb302@18:             # assemble to file
jb302@18:             op, width = iset[mne][sym]
jb302@22:             # theres gotta be a better way do deal with paged addresses
jb302@22:             if mne in ['pcall', 'pjmp']:
jb302@25:                 op = op | ((stoi(args[0]) &  0x7FF) >> 8)
jb302@22:                 const = const[-1]
jb302@17:             f.write(struct.pack('>B', op))
jb302@25:             
jb302@17:             # pad if needed
jb302@25:             # i don't think this ever happens
jb302@25:             #for i in range(width - len(const) - 1):
jb302@25:             #    f.write(struct.pack('>B', 0))
jb302@25:             
jb302@25:             # check length and write constant or throw error
jb302@25:             of = len(const) - width + 1
jb302@25:             if of > 0:
jb302@25:                 if const[0] == ('\x00'):
jb302@25:                     const = const[of:]
jb302@25:                 else:
jb302@25:                     raise ValueError
jb302@1:             f.write(const)
jb302@34:             
jb302@34:             # write debug file
jb302@34:             if d != None:
jb302@34:                 if len(const) == 0:
jb302@34:                     uconst = ' '
jb302@34:                 elif len(const) == 2:
jb302@34:                     uconst = hex(struct.unpack('>H', const)[0])
jb302@34:                 else:
jb302@34:                     if mne in rinst:
jb302@34:                         fmt = '>b'
jb302@34:                     else: 
jb302@34:                         fmt = '>B'
jb302@34:                     uconst = hex(struct.unpack(fmt, const)[0])
jb302@34:                 argstr = ', '.join(args)
jb302@34:                 d.write(hex(pc) + '\t' + hex(op)  + '\t' + uconst + '\t'  + mne + '\t' + argstr + '\n');
jb302@34:             
jb302@18:             pc = pc + width
jb302@18:         
jb302@1:         except:
jb302@18:             print '** second pass error **\nline:\n', line
jb302@18:             raise
jb302@18: 
jb302@34:     return f, d
jb302@1:     
jb302@1: if __name__ == '__main__':
jb302@1:     f = open(sys.argv[1], 'r')
jb302@25:     try:
jb302@25:         b = open(sys.argv[2], 'wb')
jb302@34:         d = open(sys.argv[2] + '.dsm', 'w')
jb302@25:     except IndexError:
jb302@25:         b = open('a.out', 'wb')
jb302@34:         d = open('a.dsm', 'wb')
jb302@25:     asm, labels = first_pass(f)
jb302@34:     b, d= second_pass(b, asm, labels, d)
jb302@1:     f.close()
jb302@1:     b.close()
jb302@34:     d.close()
jb302@1: