prog8/il65/generateasm.py
Irmen de Jong ee9a5716b0 vars init
2018-01-14 02:32:03 +01:00

500 lines
25 KiB
Python

"""
Programming Language for 6502/6510 microprocessors, codename 'Sick'
This is the assembly code generator (from the parse tree)
Written by Irmen de Jong (irmen@razorvine.net) - license: GNU GPL 3.0
"""
import os
import re
import subprocess
import datetime
from collections import defaultdict
from typing import Dict, TextIO, List, Any
from .plylex import print_bold
from .plyparse import Module, ProgramFormat, Block, Directive, VarDef, Label, Subroutine, AstNode, ZpOptions, \
InlineAssembly, Return, Register, LiteralValue
from .datatypes import VarType, DataType, to_hex, to_mflpt5, STRING_DATATYPES
class CodeError(Exception):
pass
class AssemblyGenerator:
BREAKPOINT_COMMENT_SIGNATURE = "~~~BREAKPOINT~~~"
BREAKPOINT_COMMENT_DETECTOR = r".(?P<address>\w+)\s+ea\s+nop\s+;\s+{:s}.*".format(BREAKPOINT_COMMENT_SIGNATURE)
def __init__(self, module: Module) -> None:
self.module = module
self.cur_block = None
self.output = None # type: TextIO
def p(self, text, *args, **vargs):
# replace '\v' (vertical tab) char by the actual line indent (2 tabs) and write to the stringIo
print(text.replace("\v", "\t\t"), *args, file=self.output, **vargs)
def generate(self, filename: str) -> None:
with open(filename, "wt") as self.output:
try:
self._generate()
except Exception as x:
self.output.write(".error \"****** ABORTED DUE TO ERROR: " + str(x) + "\"\n")
raise
def _generate(self) -> None:
self.sanitycheck()
self.header()
self.init_vars_and_start()
self.blocks()
self.footer()
def sanitycheck(self):
start_found = False
for block, parent in self.module.all_scopes():
for label in block.nodes:
if isinstance(label, Label) and label.name == "start" and block.name == "main":
start_found = True
break
if start_found:
break
if not start_found:
print_bold("ERROR: program entry point is missing ('start' label in 'main' block)\n")
raise SystemExit(1)
all_blocknames = [b.name for b in self.module.scope.filter_nodes(Block)]
unique_blocknames = set(all_blocknames)
if len(all_blocknames) != len(unique_blocknames):
for name in unique_blocknames:
all_blocknames.remove(name)
raise CodeError("there are duplicate block names", all_blocknames)
zpblock = self.module.zeropage()
if zpblock:
# ZP block contains no code?
for stmt in zpblock.scope.nodes:
if not isinstance(stmt, (Directive, VarDef)):
raise CodeError("ZP block can only contain directive and var")
def header(self):
self.p("; code generated by il65.py - codename 'Sick'")
self.p("; source file:", self.module.sourceref.file)
self.p("; compiled on:", datetime.datetime.now())
self.p("; output options:", self.module.format, self.module.zp_options)
self.p("; assembler syntax is for the 64tasm cross-assembler")
self.p("\n.cpu '6502'\n.enc 'none'\n")
assert self.module.address is not None
if self.module.format in (ProgramFormat.PRG, ProgramFormat.BASIC):
if self.module.format == ProgramFormat.BASIC:
if self.module.address != 0x0801:
raise CodeError("BASIC output mode must have load address $0801")
self.p("; ---- basic program with sys call ----")
self.p("* = " + to_hex(self.module.address))
year = datetime.datetime.now().year
self.p("\v.word (+), {:d}".format(year))
self.p("\v.null $9e, format(' %d ', _il65_entrypoint), $3a, $8f, ' il65 by idj'")
self.p("+\v.word 0")
self.p("_il65_entrypoint\v; assembly code starts here\n")
else:
self.p("; ---- program without sys call ----")
self.p("* = " + to_hex(self.module.address) + "\n")
elif self.module.format == ProgramFormat.RAW:
self.p("; ---- raw assembler program ----")
self.p("* = " + to_hex(self.module.address) + "\n")
def init_vars_and_start(self) -> None:
if self.module.zp_options == ZpOptions.CLOBBER_RESTORE:
self.p("\vjsr _il65_save_zeropage")
self.p("\v; initialize all blocks (reset vars)")
if self.module.zeropage():
self.p("\vjsr ZP._il65_init_block")
for block in self.module.nodes:
if isinstance(block, Block) and block.name != "ZP":
self.p("\vjsr {}._il65_init_block".format(block.name))
self.p("\v; call user code")
if self.module.zp_options == ZpOptions.CLOBBER_RESTORE:
self.p("\vjsr {:s}.start".format(self.module.main().label))
self.p("\vcld")
self.p("\vjmp _il65_restore_zeropage\n")
# include the assembly code for the save/restore zeropage routines
zprestorefile = os.path.join(os.path.split(__file__)[0], "lib", "restorezp.asm")
with open(zprestorefile, "rU") as f:
for line in f.readlines():
self.p(line.rstrip("\n"))
else:
self.p("\vjmp {:s}.start".format(self.module.main().label))
self.p("")
def blocks(self) -> None:
zpblock = self.module.zeropage()
if zpblock:
# if there's a Zeropage block, it always goes first
self.cur_block = zpblock # type: ignore
self.p("\n; ---- zero page block: '{:s}' ----".format(zpblock.name))
self.p("; file: '{:s}' src l. {:d}\n".format(zpblock.sourceref.file, zpblock.sourceref.line))
self.p("{:s}\t.proc\n".format(zpblock.label))
self.generate_block_init(zpblock)
self.generate_block_vars(zpblock, True)
# there's no code in the zero page block.
self.p("\v.pend\n")
for block in sorted(self.module.scope.filter_nodes(Block), key=lambda b: b.address or 0):
if block.name == "ZP":
continue # already processed
self.cur_block = block
self.p("\n; ---- block: '{:s}' ----".format(block.name))
self.p("; file: '{:s}' src l. {:d}\n".format(block.sourceref.file, block.sourceref.line))
if block.address:
self.p(".cerror * > ${0:04x}, 'block address overlaps by ', *-${0:04x},' bytes'".format(block.address))
self.p("* = ${:04x}".format(block.address))
self.p("{:s}\t.proc\n".format(block.label))
self.generate_block_init(block)
self.generate_block_vars(block)
subroutines = list(sub for sub in block.scope.filter_nodes(Subroutine) if sub.address is not None)
if subroutines:
# these are (external) subroutines that are defined by address instead of a scope/code
self.p("; external subroutines")
for subdef in subroutines:
assert subdef.scope is None
self.p("\v{:s} = {:s}".format(subdef.name, to_hex(subdef.address)))
self.p("; end external subroutines\n")
for stmt in block.scope.nodes:
if isinstance(stmt, (VarDef, Directive, Subroutine)):
continue # should have been handled already or will be later
self.generate_statement(stmt)
if block.name == "main" and isinstance(stmt, Label) and stmt.name == "start":
# make sure the main.start routine clears the decimal and carry flags as first steps
self.p("\vcld\n\vclc\n\vclv")
subroutines = list(sub for sub in block.scope.filter_nodes(Subroutine) if sub.address is None)
if subroutines:
# these are subroutines that are defined by a scope/code
self.p("; -- block subroutines")
for subdef in subroutines:
assert subdef.scope is not None
self.p("{:s}\v; src l. {:d}".format(subdef.name, subdef.sourceref.line))
params = ", ".join("{:s} -> {:s}".format(name or "<unnamed>", registers) for name, registers in subdef.param_spec)
returns = ",".join(sorted(register for register in subdef.result_spec if register[-1] != '?'))
clobbers = ",".join(sorted(register for register in subdef.result_spec if register[-1] == '?'))
self.p("\v; params: {}\n\v; returns: {} clobbers: {}"
.format(params or "-", returns or "-", clobbers or "-"))
cur_block = self.cur_block
self.cur_block = subdef.scope
for stmt in subdef.scope.nodes:
if isinstance(stmt, (VarDef, Directive)):
continue # should have been handled already
self.generate_statement(stmt)
self.cur_block = cur_block
self.p("")
self.p("; -- end block subroutines")
self.p("\n\v.pend\n")
def footer(self) -> None:
self.p("\t.end")
def output_string(self, value: str, screencodes: bool = False) -> str:
if len(value) == 1 and screencodes:
if value[0].isprintable() and ord(value[0]) < 128:
return "'{:s}'".format(value[0])
else:
return str(ord(value[0]))
result = '"'
for char in value:
if char in "{}":
result += '", {:d}, "'.format(ord(char))
elif char.isprintable() and ord(char) < 128:
result += char
else:
if screencodes:
result += '", {:d}, "'.format(ord(char))
else:
if char == '\f':
result += "{clear}"
elif char == '\b':
result += "{delete}"
elif char == '\n':
result += "{cr}"
elif char == '\r':
result += "{down}"
elif char == '\t':
result += "{tab}"
else:
result += '", {:d}, "'.format(ord(char))
return result + '"'
def generate_block_init(self, block: Block) -> None:
# generate the block initializer
# @todo add a block initializer subroutine that can contain custom reset/init code? (static initializer)
def _memset(varname: str, value: int, size: int) -> None:
if size > 6:
self.p("\vlda #<" + varname)
self.p("\vsta il65_lib.SCRATCH_ZPWORD1")
self.p("\vlda #>" + varname)
self.p("\vsta il65_lib.SCRATCH_ZPWORD1+1")
self.p("\vlda #" + to_hex(value))
self.p("\vldx #<" + to_hex(size))
self.p("\vldy #>" + to_hex(size))
self.p("\vjsr il65_lib.memset")
else:
self.p("\vlda #" + to_hex(value))
for i in range(size):
self.p("\vsta {:s}+{:d}".format(varname, i))
def _memsetw(varname: str, value: int, size: int) -> None:
if size > 4:
self.p("\vlda #<" + varname)
self.p("\vsta il65_lib.SCRATCH_ZPWORD1")
self.p("\vlda #>" + varname)
self.p("\vsta il65_lib.SCRATCH_ZPWORD1+1")
self.p("\vlda #<" + to_hex(size))
self.p("\vsta il65_lib.SCRATCH_ZPWORD2")
self.p("\vlda #>" + to_hex(size))
self.p("\vsta il65_lib.SCRATCH_ZPWORD2+1")
self.p("\vlda #<" + to_hex(value))
self.p("\vldx #>" + to_hex(value))
self.p("\vjsr il65_lib.memsetw")
else:
self.p("\vlda #<" + to_hex(value))
self.p("\vldy #>" + to_hex(value))
for i in range(size):
self.p("\vsta {:s}+{:d}".format(varname, i*2))
self.p("\vsty {:s}+{:d}".format(varname, i*2+1))
self.p("_il65_init_block\v; (re)set vars to initial values")
float_inits = {}
prev_value_a, prev_value_x = None, None
vars_by_datatype = defaultdict(list) # type: Dict[DataType, List[VarDef]]
for vardef in block.scope.filter_nodes(VarDef):
if vardef.vartype == VarType.VAR:
vars_by_datatype[vardef.datatype].append(vardef)
for bytevar in sorted(vars_by_datatype[DataType.BYTE], key=lambda vd: vd.value):
assert isinstance(bytevar.value, int)
if bytevar.value != prev_value_a:
self.p("\vlda #${:02x}".format(bytevar.value))
prev_value_a = bytevar.value
self.p("\vsta {:s}".format(bytevar.name))
for wordvar in sorted(vars_by_datatype[DataType.WORD], key=lambda vd: vd.value):
assert isinstance(wordvar.value, int)
v_hi, v_lo = divmod(wordvar.value, 256)
if v_hi != prev_value_a:
self.p("\vlda #${:02x}".format(v_hi))
prev_value_a = v_hi
if v_lo != prev_value_x:
self.p("\vldx #${:02x}".format(v_lo))
prev_value_x = v_lo
self.p("\vsta {:s}".format(wordvar.name))
self.p("\vstx {:s}+1".format(wordvar.name))
for floatvar in vars_by_datatype[DataType.FLOAT]:
assert isinstance(floatvar.value, (int, float))
fpbytes = to_mflpt5(floatvar.value) # type: ignore
float_inits[floatvar.name] = (floatvar.name, fpbytes, floatvar.value)
for arrayvar in vars_by_datatype[DataType.BYTEARRAY]:
assert isinstance(arrayvar.value, int)
_memset(arrayvar.name, arrayvar.value, arrayvar.size[0])
for arrayvar in vars_by_datatype[DataType.WORDARRAY]:
assert isinstance(arrayvar.value, int)
_memsetw(arrayvar.name, arrayvar.value, arrayvar.size[0])
for arrayvar in vars_by_datatype[DataType.MATRIX]:
assert isinstance(arrayvar.value, int)
_memset(arrayvar.name, arrayvar.value, arrayvar.size[0] * arrayvar.size[1])
if float_inits:
self.p("\vldx #4")
self.p("-")
for varname, (vname, b, fv) in sorted(float_inits.items()):
self.p("\vlda _init_float_{:s},x".format(varname))
self.p("\vsta {:s},x".format(vname))
self.p("\vdex")
self.p("\vbpl -")
self.p("\vrts\n")
for varname, (vname, fpbytes, fpvalue) in sorted(float_inits.items()):
self.p("_init_float_{:s}\t\t.byte ${:02x}, ${:02x}, ${:02x}, ${:02x}, ${:02x}\t; {}".format(varname, *fpbytes, fpvalue))
all_string_vars = []
for svtype in STRING_DATATYPES:
all_string_vars.extend(vars_by_datatype[svtype])
for strvar in all_string_vars:
# string vars are considered to be a constant, and are statically initialized.
self._generate_string_var(strvar)
self.p("")
def _numeric_value_str(self, value: Any, as_hex: bool=False) -> str:
if isinstance(value, bool):
return "1" if value else "0"
if isinstance(value, int):
if as_hex:
return to_hex(value)
return str(value)
if isinstance(value, (int, float)):
if as_hex:
raise TypeError("cannot output float as hex")
return str(value)
raise TypeError("no numeric representation possible", value)
def generate_block_vars(self, block: Block, zeropage: bool=False) -> None:
# Generate the block variable storage.
# The memory bytes of the allocated variables is set to zero (so it compresses very well),
# their actual starting values are set by the block init code.
vars_by_vartype = defaultdict(list) # type: Dict[VarType, List[VarDef]]
for vardef in block.scope.filter_nodes(VarDef):
vars_by_vartype[vardef.vartype].append(vardef)
self.p("; constants")
for vardef in vars_by_vartype.get(VarType.CONST, []):
if vardef.datatype == DataType.FLOAT:
self.p("\v{:s} = {}".format(vardef.name, self._numeric_value_str(vardef.value)))
elif vardef.datatype in (DataType.BYTE, DataType.WORD):
self.p("\v{:s} = {:s}".format(vardef.name, self._numeric_value_str(vardef.value, True)))
elif vardef.datatype.isstring():
# a const string is just a string variable in the generated assembly
self._generate_string_var(vardef)
else:
raise CodeError("invalid const type", vardef)
self.p("; memory mapped variables")
for vardef in vars_by_vartype.get(VarType.MEMORY, []):
# create a definition for variables at a specific place in memory (memory-mapped)
if vardef.datatype.isnumeric():
assert vardef.size == [1]
self.p("\v{:s} = {:s}\t; {:s}".format(vardef.name, to_hex(vardef.value), vardef.datatype.name.lower()))
elif vardef.datatype == DataType.BYTEARRAY:
assert len(vardef.size) == 1
self.p("\v{:s} = {:s}\t; array of {:d} bytes".format(vardef.name, to_hex(vardef.value), vardef.size[0]))
elif vardef.datatype == DataType.WORDARRAY:
assert len(vardef.size) == 1
self.p("\v{:s} = {:s}\t; array of {:d} words".format(vardef.name, to_hex(vardef.value), vardef.size[0]))
elif vardef.datatype == DataType.MATRIX:
assert len(vardef.size) in (2, 3)
if len(vardef.size) == 2:
comment = "matrix of {:d} by {:d} = {:d} bytes".format(vardef.size[0], vardef.size[1], vardef.size[0]*vardef.size[1])
elif len(vardef.size) == 3:
comment = "matrix of {:d} by {:d}, interleave {:d}".format(vardef.size[0], vardef.size[1], vardef.size[2])
else:
raise CodeError("matrix size should be 2 or 3 numbers")
self.p("\v{:s} = {:s}\t; {:s}".format(vardef.name, to_hex(vardef.value), comment))
else:
raise CodeError("invalid var type")
self.p("; normal variables - initial values will be set by init code")
if zeropage:
# zeropage uses the zp_address we've allocated, instead of allocating memory here
for vardef in vars_by_vartype.get(VarType.VAR, []):
assert vardef.zp_address is not None
if vardef.datatype.isstring():
raise CodeError("cannot put strings in the zeropage", vardef.sourceref)
if vardef.datatype.isarray():
size_str = "size " + str(vardef.size)
else:
size_str = ""
self.p("\v{:s} = {:s}\t; {:s} {:s}".format(vardef.name, to_hex(vardef.zp_address),
vardef.datatype.name.lower(), size_str))
else:
# create definitions for the variables that takes up empty space and will be initialized at startup
string_vars = []
for vardef in vars_by_vartype.get(VarType.VAR, []):
if vardef.datatype.isnumeric():
assert vardef.size == [1]
if vardef.datatype == DataType.BYTE:
self.p("{:s}\v.byte ?".format(vardef.name))
elif vardef.datatype == DataType.WORD:
self.p("{:s}\v.word ?".format(vardef.name))
elif vardef.datatype == DataType.FLOAT:
self.p("{:s}\v.fill 5\t\t; float".format(vardef.name))
else:
raise CodeError("weird datatype")
elif vardef.datatype in (DataType.BYTEARRAY, DataType.WORDARRAY):
assert len(vardef.size) == 1
if vardef.datatype == DataType.BYTEARRAY:
self.p("{:s}\v.fill {:d}\t\t; bytearray".format(vardef.name, vardef.size[0]))
elif vardef.datatype == DataType.WORDARRAY:
self.p("{:s}\v.fill {:d}*2\t\t; wordarray".format(vardef.name, vardef.size[0]))
else:
raise CodeError("invalid datatype", vardef.datatype)
elif vardef.datatype == DataType.MATRIX:
assert len(vardef.size) == 2
self.p("{:s}\v.fill {:d}\t\t; matrix {:d}*{:d} bytes"
.format(vardef.name, vardef.size[0] * vardef.size[1], vardef.size[0], vardef.size[1]))
elif vardef.datatype.isstring():
string_vars.append(vardef)
else:
raise CodeError("unknown variable type " + str(vardef.datatype))
# string vars are considered to be a constant, and are not re-initialized.
self.p("")
def _generate_string_var(self, vardef: VarDef) -> None:
if vardef.datatype == DataType.STRING:
# 0-terminated string
self.p("{:s}\n\v.null {:s}".format(vardef.name, self.output_string(str(vardef.value))))
elif vardef.datatype == DataType.STRING_P:
# pascal string
self.p("{:s}\n\v.ptext {:s}".format(vardef.name, self.output_string(str(vardef.value))))
elif vardef.datatype == DataType.STRING_S:
# 0-terminated string in screencode encoding
self.p(".enc 'screen'")
self.p("{:s}\n\v.null {:s}".format(vardef.name, self.output_string(str(vardef.value), True)))
self.p(".enc 'none'")
elif vardef.datatype == DataType.STRING_PS:
# 0-terminated pascal string in screencode encoding
self.p(".enc 'screen'")
self.p("{:s}n\v.ptext {:s}".format(vardef.name, self.output_string(str(vardef.value), True)))
self.p(".enc 'none'")
def generate_statement(self, stmt: AstNode) -> None:
if isinstance(stmt, Label):
self.p("\n{:s}\v\t\t; {:s}".format(stmt.name, stmt.lineref))
elif isinstance(stmt, Return):
if stmt.value_A:
self.generate_assignment(Register(name="A", sourceref=stmt.sourceref), '=', stmt.value_A) # type: ignore
if stmt.value_X:
self.generate_assignment(Register(name="X", sourceref=stmt.sourceref), '=', stmt.value_X) # type: ignore
if stmt.value_Y:
self.generate_assignment(Register(name="Y", sourceref=stmt.sourceref), '=', stmt.value_Y) # type: ignore
self.p("\vrts")
elif isinstance(stmt, InlineAssembly):
self.p("\n\v; inline asm, " + stmt.lineref)
self.p(stmt.assembly)
self.p("\v; end inline asm, " + stmt.lineref + "\n")
else:
self.p("\vrts; " + str(stmt)) # @todo rest of the statement nodes
def generate_assignment(self, lvalue: AstNode, operator: str, rvalue: Any) -> None:
assert rvalue is not None
if isinstance(rvalue, LiteralValue):
rvalue = rvalue.value
print("ASSIGN", lvalue, lvalue.datatype, operator, rvalue) # @todo
class Assembler64Tass:
def __init__(self, format: ProgramFormat) -> None:
self.format = format
def assemble(self, inputfilename: str, outputfilename: str) -> None:
args = ["64tass", "--ascii", "--case-sensitive", "-Wall", "-Wno-strict-bool",
"--dump-labels", "--vice-labels", "-l", outputfilename+".vice-mon-list",
"-L", outputfilename+".final-asm", "--no-monitor", "--output", outputfilename, inputfilename]
if self.format in (ProgramFormat.PRG, ProgramFormat.BASIC):
args.append("--cbm-prg")
elif self.format == ProgramFormat.RAW:
args.append("--nostart")
else:
raise CodeError("don't know how to create format "+str(self.format))
try:
if self.format == ProgramFormat.PRG:
print("\nCreating C-64 prg.")
elif self.format == ProgramFormat.RAW:
print("\nCreating raw binary.")
try:
subprocess.check_call(args)
except FileNotFoundError as x:
raise SystemExit("ERROR: cannot run assembler program: "+str(x))
except subprocess.CalledProcessError as x:
raise SystemExit("assembler failed with returncode " + str(x.returncode))
def generate_breakpoint_list(self, program_filename: str) -> str:
breakpoints = []
with open(program_filename + ".final-asm", "rU") as f:
for line in f:
match = re.fullmatch(AssemblyGenerator.BREAKPOINT_COMMENT_DETECTOR, line, re.DOTALL)
if match:
breakpoints.append("$" + match.group("address"))
cmdfile = program_filename + ".vice-mon-list"
with open(cmdfile, "at") as f:
print("; vice monitor breakpoint list now follows", file=f)
print("; {:d} breakpoints have been defined here".format(len(breakpoints)), file=f)
print("del", file=f)
for b in breakpoints:
print("break", b, file=f)
return cmdfile