creole/asm/creole.py

# Copyright (c) 2023 Peter McGoron <code@mcgoron.com>
#
# Permission to use, copy, modify, and/or distribute this software for any
# purpose with or without fee is hereby granted, provided that the above
# copyright notice and this permission notice appear in all copies.
#
# THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
# WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
# MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
# ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
# WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
# ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
# OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.

from enum import Enum

class MalformedArgument(Exception):
	pass

def word_2c(w):
	""" Negate a non-negative integer using 32 bit two's compliment.

	:param w: An integer in two's compliment. A non-negative Python
	integer will work.
	:return: The negation of the integer stored as a two's compliment
	integer.
	"""
	return (~w + 1) & 0xFFFFFFFF
def ti(w):
	""" Explicitly transform integer into two's compliment representation.

	:param w: A python integer.
	:return: The integer in two's compliment.
	"""
	return w if w >= 0 else word_wc(-w)
def from_2c(w):
	""" Turn two's compliment word into Python integer.

	:param w: An integer in 32 bit twos compliment.
	:return: The integer as a proper Python string.
	"""
	if (w >> 31) & 1 == 0:
		return w
	return -word_2c(w)

class Argument:
	""" Class of arguments. Not used directly: It is used to store
	intermediate information during the assembly process. """

	def __init__(self, argtype, val, sign=False):
		""" Initialize an argument.

		:param argtype: Type of the argument (instance of ArgType).
		:param val: Python integer value of the argument.
		:param sign: If the argument should be treated as signed.
		Otherwise, the integer will be interpreted in execution
		as an unsigned integer.
		"""
		self.at = argtype
		self.sign = sign
		self.val = val
	def __str__(self):
		return f'({self.at}, {self.sign}, {self.val})'
	def high_bits(self):
		""" Returns the high bits that the argument would have
		    in the opcode. """
		return int(self.sign) << 1 | (self.at == ArgType.REG)
	def __call__(self):
		l = 2 if self.val < 0x80 else None
		return encode_pseudo_utf8(self.val, self.high_bits(), l)

class StringArgument(Argument):
	def __init__(self, *args, **kwargs):
		super().__init__(*args, **kwargs)
	def __call__(self):
		b = bytes()
		for v in self.val:
			b = b + Argument(ArgType.IMM, int(v, base=16))()
		return b

class LabelArgument(Argument):
	def __init__(self, *args, **kwargs):
		super().__init__(*args, **kwargs)
	def __call__(self):
		return self.val

class TypecheckException(Exception):
	""" Exception thrown when an argument to an instruction are of the
	    incorrect type. """
	def __init__(self, got, argtype, sarg, opcode, i):
		self.argtype = argtype
		self.sarg = sarg
		self.opcode = opcode
		self.got = got
		self.i = i
		self.message = f'''\
opcode {self.opcode.name} has invalid value {self.sarg} ({self.got})
at {self.i} (expected {self.argtype})\
'''

class ArgType(Enum):
	""" Class denoting the type of an argument to an instruction. """

	IMM = 1
	""" Immediate values are ones that must be numbers (positive or negative). """

	REG = 2
	""" Type of registers. """

	VAL = 3
	""" Type that denotes either immediate values or registers. """

	DAT = 4
	""" Type of data label. """

	STR = 5
	""" Type of a string of 32 bit integers. """

	LAB = 6
	""" Type of a label (.name). """

	def is_number(t):
		return t == ArgType.IMM or t == ArgType.REG

	def gettype(s):
		""" Parses the type of the argument represented as a string
		    and returns a tuple with the first the first element being
		    the type and the second element being the integer value of
		    the argument.

		    Valid parameters are:

		    * `r` followed by a nonnegative integer (register)
		    * `l` followed by a nonnegative integer (label)
		    * any integer (immediate value)

		    :param s: String representing the argument.
		    :return: The Argument object representing the argument.
		    :raises MalformedArgument:
		"""
		if type(s) is list:
			return StringArgument(ArgType.STR, s)
		elif s.isnumeric():
			return Argument(ArgType.IMM, int(s))
		elif s[0] == "-" and s[1:].isnumeric():
			return Argument(ArgType.IMM, word_2c(int(s[1:])), True)
		elif s[0] == 'r' and s[1:].isnumeric():
			return Argument(ArgType.REG, int(s[1:]))
		elif s[0] == 'd' and s[1:].isnumeric():
			return Argument(ArgType.DAT, int(s[1:]))
		elif s[0] == '.':
			return LabelArgument(ArgType.LAB, s[:])
		else:
			raise MalformedArgument(s)

	def typecheck(self, s, opcode, i):
		""" Parses the type of the string and returns it if it fits
		    the type of the enum value.

		    :param s: String argument representing an argument.
		    :param opcode: Opcode of the argument. Used for debugging.
		    :param i: Argument number. Used for debugging.
		    :return: The Argument class containing the object.
		    :raises TypecheckException: """
		t = ArgType.gettype(s)
		if self == ArgType.VAL:
			if t.at == ArgType.REG or t.at == ArgType.IMM:
				return t
			else:
				raise TypecheckException(t.at, self, s, opcode, i)
		elif t.at == self:
			return t
		else:
			raise TypecheckException(t.at, self, s, opcode, i)

class OpcodeException(Exception):
	pass
class TypecheckLenException(Exception):
	""" Exception thrown when arguments to an instruction are of the
	    incorrect length. """
	def __init__(self, opcode, insargs, argtypelen):
		self.opcode = opcode
		self.insargs = insargs
		self.argtypelen = argtypelen
	def __str__(self):
		return f'''\
arguments {self.insargs} to opcode {self.opcode.name} not of length {self.argtypelen}\
'''

class Instruction(Enum):
	""" Class of microcode instructions. The first number is the opcode
	    and the suceeding values are the types of each of the
	    arguments. The first argument is the opcode and the second
	    argument is what function is used to compile the instruction
	    (some instructions are actually versions of other instructions). """
	NOP = 0, "_render_default"
	PUSH = 1, "_render_default", ArgType.VAL
	POP = 2, "_render_default", ArgType.REG
	ADD = 3, "_render_default", ArgType.REG, ArgType.VAL, ArgType.VAL
	MOV = "ADD", "_render_mov", ArgType.REG, ArgType.VAL
	MUL = 4, "_render_default", ArgType.REG, ArgType.VAL, ArgType.VAL
	DIV = 5, "_render_default", ArgType.REG, ArgType.VAL, ArgType.VAL
	SDIV = "DIV", "_render_change_args", ArgType.REG, ArgType.VAL, ArgType.VAL
	SYS = 6, "_render_default", ArgType.VAL
	JL = 7, "_render_default", ArgType.LAB, ArgType.VAL, ArgType.VAL
	JLS = "JL", "_render_change_args", ArgType.LAB, ArgType.VAL, ArgType.VAL
	JLE = 8, "_render_default", ArgType.LAB, ArgType.VAL, ArgType.VAL
	JLES = "JLE", "_render_change_args", ArgType.LAB, ArgType.VAL, ArgType.VAL
	JE = 9, "_render_default", ArgType.LAB, ArgType.VAL, ArgType.VAL
	J = "JE", "_render_j", ArgType.LAB
	JNE = 10, "_render_default", ArgType.LAB, ArgType.VAL, ArgType.VAL
	DB = 11, "_render_default", ArgType.DAT, ArgType.STR

	def __int__(self):
		""" Returns the opcode associated with the Instruction.
		    If it is a virtual instruction, it will resolve the string
		    name of the opcode and return its opcode. """
		if type(self.opcode) is int:
			return self.opcode
		return int(Instruction[self.opcode])

	def __init__(self, opcode, renderfun, *args):
		""" Initialize an Instruction. Do not call this function: it is
		    used to make enum values. To add a new instruction, modify
		    the Instruction enum.
		    This function sometimes takes string arguments because
		    certain values may not be loaded until later.

		    :param opcode: Opcode of the instruction, or a string
		    containing the case-sensitive name of the instruction from
		    which this instruction derives from.

		    :param renderfun: a string with the name of a function
		    in the class that returns the instruction opcode.

		    :param *args: Type of each argument to the instruction.
		    The amount of arguments denotes the amount of instructions.
		"""

		if type(opcode) is int and (opcode > 0x7F or opcode < 0):
			raise OpcodeException(opcode)

		self.opcode = opcode
		self.argtypes = args
		self.render = getattr(self, renderfun)

	def typecheck(self, sargs):
		""" Pass arguments to the instruction and check if the
		    arguments are correct.

		    :param sargs: List of arguments to the instruction
		    as strings.
		    :return: List of arguments (as Argument objects).
		    :raises TypecheckLenException:
		"""
		rargs = []
		if len(sargs) != len(self.argtypes):
			raise TypecheckLenException(self, sargs,
			                            len(self.argtypes))
		for i in range(0, len(sargs)):
			t = self.argtypes[i].typecheck(sargs[i], self, i)
			rargs.append(t)
		return rargs

	def _render_mov(self, args):
		args = [args[0], args[1], Argument(ArgType.IMM, 0)]
		return Instruction[self.opcode].render(args)

	def _render_j(self, args):
		args = [args[0], Argument(ArgType.IMM, 0),
		        Argument(ArgType.IMM, 0)]
		return Instruction[self.opcode].render(args)

	def _render_change_args(self, args):
		for i in range(0,len(args)):
			if ArgType.is_number(args[i].at):
				args[i].sign = True
		return Instruction[self.opcode].render(args)

	def _render_default(self, args):
		comps = [bytes([self.opcode])]
		for a in args:
			comps.append(a())
		comps.append(b'\x00')
		return comps

encoding_types = {
#           start      mask   B
	2: (0x7F,       0xC0, 7),
	3: (0xFFF,      0xE0, 12),
	4: (0x1FFFF,    0xF0, 17),
	5: (0x3FFFFF,   0xF8, 22),
	6: (0x7FFFFFF,  0xFC, 27),
	7: (0xFFFFFFFF, 0xFE, 32),
#	B : Total number of bits excluding high bits
}

def pseudo_utf8_len(n):
	for k in sorted(encoding_types):
		if n <= encoding_types[k][0]:
			return k
	return None

class InvalidNumberException(Exception):
	pass
class InvalidLengthException(Exception):
	pass
def encode_pseudo_utf8(n, high_bits, to):
	if n < 0:
		raise InvalidNumberException(n)
	if to is None or to < 0:
		to = pseudo_utf8_len(n)
		if to is None:
			raise InvalidNumberException(n)
	if to > 8 or to < 0:
		raise InvalidLengthException(to)
	elif to == 1:
		if n < 0x80:
			return bytes([n])
		else:
			raise InvalidNumberException(n,to)

	(maxval, start_byte, n_tot) = encoding_types[to]
	if n > maxval or high_bits > 15:
		raise InvalidNumberException(n, high_bits)
	n = n | (high_bits << n_tot)
	all_bytes = []
	for i in range(0, to - 1):
		all_bytes.append(0x80 | (n & 0x3F))
		n >>= 6
	all_bytes.append(start_byte | n)
	return bytes(reversed(all_bytes))

class RangeCheckException(Exception):
	pass
class Line:
	def __init__(self, ins, args):
		self.ins = ins
		self.args = args

	def check_line(self, reglen, datlen):
		for a in self.args:
			if a.at == ArgType.REG:
				if a.val < 0 or a.val >= reglen:
					raise RangeCheckException(a.at,
					                          a.val,
					                          reglen)
			elif a.at == ArgType.DAT:
				if a.val < 0 or a.val >= datlen:
					raise RangeCheckException(a.at,
					                          a.val,
					                          reglen)
	def __call__(self):
		return self.ins.render(self.args)

class InstructionNotFoundException(Exception):
	pass

def _term_sep(s):
	""" Split up the arguments of an instruction.
	OP arg1 arg2 [data,data,data,...]
	"""

	s = s.strip()
	s_data = s.split('[')
	if len(s_data) == 2:
		return s_data[0].split() + [s_data[1].rstrip('] \t\n\r\v').split(',')]
	else:
		return s.split()

class Program:
	def _asm_push_line(self, ins, args):
		l = Line(ins, args)
		l.check_line(self.reglen, self.datlen)
		self.asm.append(l)

	def parse_asm_line(self, line):
		""" Parse and add a single assembly line to the program.
		:param line: String containing the line.
		:raises InstructionNotFoundException:
		"""
		line = _term_sep(line)
		line[0] = line[0].casefold()
		if line[0][0] == '.':
			self.asm.append(line[0])
			return None

		try:
			ins = Instruction[line[0].upper()]
		except Exception as e:
			raise InstructionNotFoundException(line[0])

		args_w_type = ins.typecheck(line[1:])
		self._asm_push_line(ins, args_w_type)

	def parse_lines(self, lines):
		""" Parse a list of lines. See parse_asm_line.
		    :param lines: List of assembly lines.
		"""
		for l in lines:
			self.parse_asm_line(l)

	def __call__(self):
		""" Generate bytecode. """

		# Labels may jump forward in the program, which means
		# multiple passes are required to properly calculate
		# jump locations.
		# This algorithm makes every jump destination the same
		# width in each operation, and calculates the smallest
		# width that will allow all labels to jump to any location
		# in the program.
		# The algorithm calculates the length of the program
		# with all jump arguments given a length of 0. Each label
		# is noted with its offset in the program (with all jump
		# arguments given zero length) and the amount of jump arguments
		# that occur prior to the label.
		# When the code is emitted, the label length is properly
		# calculated with the length of each label.
		# This method is not optimal, but will work well for small
		# programs.

		ins = []
		curlen = 0

		# This dictonary contains a tuple (len, refs)
		# that denotes that a label points to len + lablen*refs
		# where lablen is a to-be-determined number.
		labels = {}
		labelrefs = 0
		for line in self.asm:
			if type(line) is str:
				labels[line] = (curlen, labelrefs)
				continue

			next_ins = line()
			for v in next_ins:
				if type(v) is str:
					labelrefs += 1
				else:
					curlen += len(v)
			ins.append(next_ins)

		# Calculate a label length, such that the entire program
		# can be contained in this length.
		for i in encoding_types:
			if curlen + labelrefs*i < encoding_types[i][0]:
				lablen = i
				break

		# Emit bytecode.
		b = bytes()
		for line in ins:
			for arg in line:
				if type(arg) is str:
					off = labels[arg][0] + labels[arg][1]*lablen
					arg = encode_pseudo_utf8(off, 0, lablen)
				b = b + arg
		assert len(b) < encoding_types[lablen][0]
		return b

	def __init__(self, reglen=16, datlen=16):
		self.asm = []
		self.reglen = reglen
		self.datlen = datlen