rzzzwilson.pymlac/pyasm/pyasm

#!/usr/bin/python
# -*- coding: utf-8 -*-

"""
An assembler for the Imlac simulator.

Usage: pyasm [ -h ] [ -l <listfile> ] [ -o <outputfile> ] <asmfile>

Where <asmfile>  is the file to assemble,
      <outputfile>  is the output PTP file
      <listfile>    is the optional listing file

If <outputfile> is not specified the output filename is the input
<asmfile> with any extension removed and a .ptp axtenstion added.
"""

"""
The basic structure of the assembler:
0.  Read all file lines into memory
1.  Create ORG blocks
2.  Create CODE blocks from ORG blocks (assemble() function)
3.  Check for undefined things in the symbol table
4.  Allocate addresses to literal CODE blocks
5.  Fix relative addresses in literal blocks
6.  Backpatch all code blocks
7.  Emit PTP data
"""

import sys
import os
import copy
import string
import getopt

######
# Globals
######

# the input assembler filename
AsmFile = None

# the output listing file
ListFile = None
ListFileHandle = None       # open listing file

# the output PTP file
OutputFile = None
OutputFileHandle = None     # open output file

# the program start address (optional)
StartAddr = None

# the symbol table(s)
# {<name>: <value>, ... }
SymTable = {}
# {<name>: <line#>, ... }
SymTableLine = {}

# the backpatch list
# [[symname, coderef, offset], [symname, coderef, offset], ... ]
BackpatchList = []

# current line number and the line
CurrentLine = None
CurrentLineNumber = None

# Any undefined label
Undefined = None

######
# Mostly constant stuff
######

# the output PTP filename extension
PTPExtension = '.ptp'

# the output listing filename extension
ListFileExtension = '.lst'


def usage(msg=None):
    """Print usage and optional error message."""
    if msg:
        print('*'*60)
        print(msg)
        print('*'*60)
    print(__doc__)

def error(msg):
    """Print error message and stop."""

    print(msg)
    sys.exit(10)

def write_list(code, addr, lnum, line):
    """Generate one line of listing file.

    code  is the word of generated code
    addr  is the address of the generated code
    lnum  file line number
    line  complete text of the line of assembler
    """

    code_str = '%06o' % code if code else ''
    addr_str = '%04o' % addr if addr else ''
    lnum_str = '%04d:' % lnum if lnum else ''
    line_str = '%s' % line if line else ''

    ListFileHandle.write('%6s %4s %s %s\n'
                         % (code_str, addr_str, lnum_str, line_str))
    ListFileHandle.flush()

def eval_expression(expr, dot):
    """Evaluate an expression.

    expr  string holding expression
    dot   current code address (the "." symbol value)

    Valid expressions are:
        <expr>               ::= <string> | <numeric_expression>
        <string>             ::= ('|") <characters> ('|")
        <numeric_expression> ::= (<numeric_const>|<name_const>|".") [ (+|-) <numeric_expression> ]
        <name_const>         ::= <name_characters>
    """

    global Undefined

    # if no expression, do nothing
    if expr is None:
        return None

    # if expression is a string, return an iterable of ASCII values
    if expr[0] in "'\"":
        delim = expr[0]
        if expr[-1] != delim:
            error("Badly formed string value: '%s'" % expr)
        return expr[1:-1]

    # replace any "." value with "dot" defined in the symbol table
    expr = string.replace(expr, '.', 'dot')
    expr = expr.upper()
    env = copy.deepcopy(SymTable)
    env['DOT'] = dot       # add in the "." address

    # evaluate the expression
    try:
        result = eval(expr, env)
    except TypeError as e:
        error('ORG pseudo-opcode expression contains unsatisfied references\n'
              '%d: %s' % (CurrentLineNumber, CurrentLine))
    except NameError as e:
        Undefined = e.message
        if 'is not defined' in e.message:
            Undefined = e.message[len("name '"):-len("' is not defined")]
        raise

    return result

def assemble_line(lnum, opcode, addr):
    """Assemble one line of code.

    lnum    source file line number
    opcode  opcode, uppercase
    addr    address expression, uppercase

    Returns the 16bit word value, with possibly the backpatch list
    updated.
    """

    return 1025

def define_label(label, address, lnum):
    """Put 'label' into the symbol tables."""

    if label in SymTable:
        prev_lnum = SymTableLine[label]
        error("Label '%s' define twice, lines %d and %d."
              % (label, prev_lnum, lnum))
    SymTable[label] = address
    SymTableLine[label] = lnum

def gen_ascii(string, address, lnum, line):
    """Generate data words for ASCII string.

    string   the string to generate data for
    address  'dot' value for first word generated
    lnum     line number of code
    line     actual line of code
    """

    # ensure even number of bytes in string
    if len(string) % 2:
        string += '\0'      # append zero byte as padding

    # get 2 bytes at a time and generate words
    while string:
        ch1 = string[0]
        ch2 = string[1]
        string = string[2:]
        value = (ord(ch1) << 8) + ord(ch2)
        write_list(value, address, lnum, line)
        address += 1
        lnum = None
        line = None

def assemble_oblock(oblock):
    """Assemble one org block to a code block.

    An org block: [(lnum, line, label, opcode, addr), ...]
    A code block: [address, [word, word, word, ...]]
    """

    global CurrentLineNumber, CurrentLine

    cblock = [None, []]

    # if we get an empty block, do nothing
    if len(oblock) == 0:
        return cblock

    # assemble the org block
    address = None          # '.' value
    for (lnum, line, label, opcode, addr) in oblock:
        CurrentLineNumber = lnum
        CurrentLine = line

        # if no code, just list it
        if label is None and opcode is None:
            write_list(None, None, lnum, line)
        else:
            # maybe we have some code to generate
            if opcode:
                if opcode == 'ORG':
                    # ORG just sets the code address and starts the code block
                    if label:
                        error('%d: %s\nORG opcode may not have a label'
                              % (lnum, line))
                    try:
                        address = eval_expression(addr, address)
                    except NameError as e:
                        error("%d: %s\nName '%s' is undefined"
                              % (lnum, line, Undefined))
                    cblock = [address, []]
                    write_list(None, None, lnum, line)
                    continue
                elif opcode == 'END':
                    # END pseudo-op, terminate assembly
                    start_addr = None
                    if addr:
                        try:
                            start_addr = eval_expression(addr, address)
                        except NameError as e:
                            error("%d: %s\nName '%s' is undefined"
                                  % (lnum, line, undefined))
                    write_list(None, start_addr, lnum, line)
                    return cblock
                elif opcode == 'EQU':
                    if label is None:
                        error('%d: %s\nEQU opcode missing a label'
                              % (lnum, line))
                    try:
                        value = eval_expression(addr, None)
                    except NameError as e:
                        error("%d: %s\nName '%s' is undefined"
                              % (lnum, line, Undefined))
                    define_label(label, value, lnum)
                    write_list(None, None, lnum, line)
                    continue
                elif opcode == 'DATA':
                    try:
                        value = eval_expression(addr, address)
                    except NameError as e:
                        error("%d: %s\nName '%s' is undefined"
                              % (lnum, line, Undefined))
                    ln = lnum
                    li = line
                    if isinstance(value, basestring):
                        gen_ascii(value, address, lnum, line)
                    else:
                        write_list(value, address, lnum, line)
                        address += 1
                else:
                    code = assemble_line(lnum, opcode, addr)
                    cblock[1].append((code, addr))
                    write_list(code, address, lnum, line)
                    if label:
                        define_label(label, address, lnum)
                    address += 1
            elif label:
                # we have a label and no opcode, just define the label and list it
                define_label(label, address, lnum)
                write_list(None, None, lnum, line)

    return cblock

def assemble_org_blocks(blocks):
    """Assemble org blocks producing a code blocks list."""

    code_block = []

    # for each org block, assemble to code block
    for oblock in blocks:
        code_block.append(assemble_oblock(oblock))

    return code_block

def next_symbol(line):
    """Return next symbol and line remainder."""

    fields = string.split(line, maxsplit=1)
    if len(fields) != 2:
        fields.append('')

    return fields

def split_fields(line):
    """Split one ASM line into fields: label, opcode, address.

    Returns a tuple: (label, opcode, address).
    If label and opcode ar not None, uppercase the result string.

    If a field is missing, return None for it.  If the line is empty, return
    (None, None, None).

    We take pains not to split the address field if it's something like
        ALPHA + 100
    """

    if not line:
        return (None, None, None)

    # check for the label
    label = None
    if line[0] not in ' \t;':
        (label, remainder) = next_symbol(line)
        label = label.upper()
    else:
        remainder = line.strip()

    # get opcode
    opcode = None
    if remainder and remainder[0] != ';':
        (opcode, remainder) = next_symbol(remainder)
        opcode = opcode.upper()

    # get address
    address = None
    if remainder and remainder[0] != ';':
        # first, check for a string
        if remainder[0] in "'\"":
            delim = remainder[0]
            remainder = remainder[1:]
            ndx = remainder.find(delim)
            if ndx == -1:
                error('Unbalanced string delimiter:\n'
                      '%d: %s' % (CurrentLineNumber, CurrentLine))
            address = '"' + remainder[:ndx].strip() + '"'
            remainder = remainder[ndx+1:].strip()
        else:
            # strip off any comment
            ndx = remainder.find(';')
            if ndx != -1:
                remainder = remainder[:ndx].strip()
            address = remainder.strip()
            remainder = None

    # check that remainder is empty or only a comment
    if remainder and remainder[0] != ';':
        error('Badly formed instruction:\n'
              '%d: %s' % (CurrentLineNumber, CurrentLine))

    return (label, opcode, address)

def split_orgs(asm_lines):
    """Split ASM lines into ORG blocks (one or more).

    Error if no ORG pseudo-opcode found.
    """

    global CurrentLineNumber, CurrentLine

    result = []
    block = []

    for (lnum, line) in enumerate(asm_lines):
        (label, opcode, addr) = split_fields(line)
        CurrentLineNumber = lnum
        CurrentLine = line
        if opcode:
            if opcode.lower() == 'org':
                if block:
                    # save previous ORG block
                    result.append(block)
                    block = []
                if addr is None:
                    error('Line %d: %s\nORG pseudo-opcode without an address?'
                          % (lnum, line))
        block.append((lnum+1, line, label, opcode, addr))

    if block:
        # save previous ORG block
        result.append(block)
        block = []

    return result

def check_org_overlap(code_blocks):
    """Check that we don't have overlapping ORG blocks."""

   # remember org block span: [(start, end), (start', end'), ...]
    block_span = []

    for cb in code_blocks:
        (address, code) = cb
        if address is None and not code:
            # empty code block, ignore
            continue
        (block_start, code) = cb
        block_end = block_start + len(code) - 1

        for (start, end) in block_span:
            if start <= block_start <= end:
                error('ORG block at address %06o overwrites previous block at [%06o,%06o]'
                      % (block_start, start, end))
            if start <= block_end <= end:
                error('ORG block at address %06o overwrites previous block at [%06o,%06o]'
                      % (block_start, start, end))

        block_span.append((block_start, block_end))

def allocate_literals(code_blocks):
    """Allocate space for literal blocks."""

    pass

def backpatch(code_blocks):
    """Generate final code values, evaluate address fields.

    Returns a simplified code block list: [[address, [code, code, ...]], ...].
    """

    result = []

    for cb in code_blocks:
        (address, codes) = cb
        if address is None and not codes:
            # empty code block, ignore
            continue

        # got a code_block that has code
        dot = address
        new_code = []
        for (code, addr) in codes:
            try:
                addr_value = eval_expression(addr, dot)
            except NameError as e:
                error("Name '%s' is undefined" % Undefined)
            if addr_value:
                code |= addr_value
            new_code.append(code)
            dot += 1

        result.append([address, new_code])

    return [address, new_code]

def assemble_file():
    """Assemble the file and produce listing, output files."""

    # read all of ASM file into memory, strip \n, etc
    with open(AsmFile, 'rb') as fd:
        asm_lines = fd.readlines()
    asm_lines = [line.rstrip() for line in asm_lines]
    print('asm_lines=\n%s' % '\n'.join(asm_lines))

    org_blocks = split_orgs(asm_lines)
    print('org_blocks=%s' % str(org_blocks))

    code_blocks = assemble_org_blocks(org_blocks)
    print('code_blocks=%s' % str(code_blocks))
    print('SymTable=%s' % str(SymTable))
    print('SymTableLine=%s' % str(SymTableLine))

# allow programmer to overlap ORG blocks
#    check_org_overlap(code_blocks)

    allocate_literals(code_blocks)

    patched_code_blocks = backpatch(code_blocks)
    print('patched_code_blocks=%s' % str(patched_code_blocks))

def main():
    """The assembler."""

    global AsmFile, ListFile, OutputFile
    global ListFileHandle, OutputFileHandle

    # handle the options
    try:
        (opts, args) = getopt.gnu_getopt(sys.argv, "hl:o:",
                                         ["help", "list=", "output="])
    except getopt.GetoptError:
        usage()
        sys.exit(10)

    ListFile = None
    OutputFile = None

    for opt, arg in opts:
        if opt in ('-h', '--help'):
            usage()
            sys.exit(0)
        elif opt in ('-l', '--list'):
            ListFile = arg
        elif opt in ('-o', '--output'):
            OutputFile = arg

    if len(args) != 2:
        usage()
        sys.exit(10)

    # get ASM filename and make sure it exists
    AsmFile = args[1]
    try:
        f = open(AsmFile, 'rb')
    except IOError:
        print("Sorry, can't find file '%s'" % AsmFile)
        sys.exit(10)
    f.close()

    if OutputFile is None:
        (path, ext) = os.path.splitext(AsmFile)
        OutputFile = path + PTPExtension
    OutputFileHandle = open(OutputFile, 'wb')

    if ListFile is None:
        (path, ext) = os.path.splitext(AsmFile)
        ListFile = path + ListFileExtension
    ListFileHandle = open(ListFile, 'wb')

    print('ListFile=%s, OutputFile=%s, AsmFile=%s'
          % (str(ListFile), str(OutputFile), str(AsmFile)))

    assemble_file()

if __name__ == '__main__':
    main()