mirror of
https://github.com/mnaberez/py65.git
synced 2024-09-06 10:54:32 +00:00
133 lines
4.6 KiB
Python
133 lines
4.6 KiB
Python
import re
|
|
from py65.utils.addressing import AddressParser
|
|
|
|
|
|
class Assembler:
|
|
Statement = re.compile(r'^([A-z]{3}[0-7]?\s+'
|
|
r'\(?\s*)([^,\s\)]+)(\s*[,xXyY\s]*\)?'
|
|
r'[,xXyY\s]*)$')
|
|
|
|
Addressing = (
|
|
('zpi', "($00FF)"),
|
|
('zpx', "$00FF,X"),
|
|
('zpy', "$00FF,Y"),
|
|
('zpg', "$00FF"),
|
|
('inx', "($00FF,X)"),
|
|
('iax', "($FFFF,X)"),
|
|
('iny', "($00FF),Y"),
|
|
('ind', "($FFFF)"),
|
|
('abx', "$FFFF,X"),
|
|
('aby', "$FFFF,Y"),
|
|
('abs', "$FFFF"),
|
|
('rel', "$FFFF"),
|
|
('imp', ""),
|
|
('acc', ""),
|
|
('acc', "A"),
|
|
('imm', "#$FF")
|
|
)
|
|
|
|
def __init__(self, mpu, address_parser=None):
|
|
""" If a configured AddressParser is passed, symbolic addresses
|
|
may be used in the assembly statements.
|
|
"""
|
|
self._mpu = mpu
|
|
|
|
if address_parser is None:
|
|
address_parser = AddressParser()
|
|
self._address_parser = address_parser
|
|
|
|
self._addressing = []
|
|
numchars = mpu.BYTE_WIDTH / 4 # 1 byte = 2 chars in hex
|
|
for mode, format in self.Addressing:
|
|
pat = "^" + re.escape(format) + "$"
|
|
pat = pat.replace('00', '0{%d}' % numchars)
|
|
pat = pat.replace('FF', '([0-9A-F]{%d})' % numchars)
|
|
self._addressing.append([mode, re.compile(pat)])
|
|
|
|
def assemble(self, statement, pc=0000):
|
|
""" Assemble the given assembly language statement. If the statement
|
|
uses relative addressing, the program counter (pc) must also be given.
|
|
The result is a list of bytes. Raises when assembly fails.
|
|
"""
|
|
opcode, operand = self.normalize_and_split(statement)
|
|
|
|
for mode, pattern in self._addressing:
|
|
match = pattern.match(operand)
|
|
|
|
if match:
|
|
# check if opcode supports this addressing mode
|
|
try:
|
|
bytes = [self._mpu.disassemble.index((opcode, mode))]
|
|
except ValueError:
|
|
continue
|
|
|
|
operands = match.groups()
|
|
|
|
if mode == 'rel':
|
|
# relative branch
|
|
absolute = int(''.join(operands), 16)
|
|
relative = (absolute - pc) - 2
|
|
relative = relative & self._mpu.byteMask
|
|
operands = [(self._mpu.BYTE_FORMAT % relative)]
|
|
|
|
elif len(operands) == 2:
|
|
# swap bytes
|
|
operands = (operands[1], operands[0])
|
|
|
|
operands = [int(hex, 16) for hex in operands]
|
|
bytes.extend(operands)
|
|
|
|
# raise if the assembled bytes would exceed top of memory
|
|
if (pc + len(bytes)) > (2 ** self._mpu.ADDR_WIDTH):
|
|
raise OverflowError
|
|
|
|
return bytes
|
|
|
|
# assembly failed
|
|
raise SyntaxError(statement)
|
|
|
|
def normalize_and_split(self, statement):
|
|
""" Given an assembly language statement like "lda $c12,x", normalize
|
|
the statement by uppercasing it, removing unnecessary whitespace,
|
|
and parsing the address part using AddressParser. The result of
|
|
the normalization is a tuple of two strings (opcode, operand).
|
|
"""
|
|
statement = ' '.join(str.split(statement))
|
|
|
|
# normalize target in operand
|
|
match = self.Statement.match(statement)
|
|
if match:
|
|
before, target, after = match.groups()
|
|
|
|
# target is an immediate value
|
|
if target.startswith('#'):
|
|
try:
|
|
if target[1] in ("'", '"'): # quoted ascii character
|
|
number = ord(target[2])
|
|
else:
|
|
number = self._address_parser.number(target[1:])
|
|
except IndexError:
|
|
raise SyntaxError(statement)
|
|
|
|
if (number < 0) or (number > self._mpu.byteMask):
|
|
raise OverflowError
|
|
statement = before + '#$' + self._mpu.BYTE_FORMAT % number
|
|
|
|
# target is the accumulator
|
|
elif target in ('a', 'A'):
|
|
pass
|
|
|
|
# target is an address or label
|
|
else:
|
|
address = self._address_parser.number(target)
|
|
statement = before + '$' + self._mpu.ADDR_FORMAT % address + after
|
|
|
|
# separate opcode and operand
|
|
splitted = statement.split(" ", 2)
|
|
opcode = splitted[0].strip().upper()
|
|
if len(splitted) > 1:
|
|
operand = splitted[1].strip().upper()
|
|
else:
|
|
operand = ''
|
|
return (opcode, operand)
|