def run_all(infile, outfile):
"Transforms the source infile to a binary outfile."
Err.count = 0
z = Ophis.Frontend.parse(infile)
env = Ophis.Environment.Environment()
m = Ophis.Passes.ExpandMacros()
i = Ophis.Passes.InitLabels()
l_basic = Ophis.Passes.UpdateLabels()
l = Ophis.Passes.FixPoint("label update", [l_basic], lambda: l_basic.changed == 0)
c = Ophis.Passes.Collapse()
a = Ophis.Passes.Assembler()
passes = []
passes.append(Ophis.Passes.DefineMacros())
passes.append(Ophis.Passes.FixPoint("macro expansion", [m], lambda: m.changed == 0))
passes.append(Ophis.Passes.FixPoint("label initialization", [i], lambda: i.changed == 0))
passes.extend([Ophis.Passes.CircularityCheck(), Ophis.Passes.CheckExprs(), Ophis.Passes.EasyModes()])
passes.append(Ophis.Passes.FixPoint("instruction selection", [l, c], lambda: c.collapsed == 0))
passes.extend([Ophis.Passes.NormalizeModes(), Ophis.Passes.UpdateLabels(), a])
for p in passes: p.go(z, env)
if Err.count == 0:
try:
output = file(outfile, 'wb')
output.write("".join(map(chr, a.output)))
except IOError:
print "Could not write to "+outfile
else:
Err.report()
def assemble(self, node, mode, env):
"A generic instruction called by the visitor methods themselves"
(opcode, expr, expr2) = node.data
bin_op = Ops.opcodes[opcode][mode]
if bin_op is None:
Err.log('%s does not have mode "%s"' % (opcode.upper(),
Ops.modes[mode]))
return
inst_bytes = []
self.outputbyte(IR.ConstantExpr(bin_op), env, inst_bytes)
arglen = Ops.lengths[mode]
val1 = None
val2 = None
if expr is not None:
val1 = expr.value(env)
if expr2 is not None:
val2 = expr2.value(env)
if mode == 15: # ZP Relative mode is wildly nonstandard
expr2 = self.relativize(expr2, env, arglen)
self.outputbyte(expr, env, inst_bytes)
self.outputbyte(expr2, env, inst_bytes)
else:
if mode == 14:
expr = self.relativize(expr, env, arglen)
if arglen == 1:
self.outputbyte(expr, env, inst_bytes)
elif arglen == 2:
self.outputword(expr, env, inst_bytes)
self.listing.listInstruction(self.listing_string(env.getPC(),
inst_bytes,
mode, opcode,
val1, val2))
env.incPC(1 + arglen)
self.code += 1 + arglen
def assemble(self, node, mode, env):
"A generic instruction called by the visitor methods themselves"
(opcode, expr) = node.data
bin_op = Ops.opcodes[opcode][mode]
if bin_op is None:
Err.log('%s does not have mode "%s"' % (opcode.upper(),
Ops.modes[mode]))
return
self.outputbyte(IR.ConstantExpr(bin_op), env)
arglen = Ops.lengths[mode]
if mode == 14: # Special handling for relative mode
arg = expr.value(env)
arg = arg - (env.getPC() + 2)
if arg < -128 or arg > 127:
Err.log("Branch target out of bounds")
arg = 0
if arg < 0:
arg += 256
expr = IR.ConstantExpr(arg)
if arglen == 1:
self.outputbyte(expr, env)
if arglen == 2:
self.outputword(expr, env)
env.incPC(1 + arglen)
self.code += 1 + arglen
def pragmaCbmfloat(ppt, line, result):
"Parses a string into a CBM BASIC format floating point number"
data = []
while True:
try:
v_str = line.expect("STRING").value
v = float(v_str)
if v == 0.0:
data.extend([0,0,0,0,0])
else:
if v < 0.0:
sign = 128
v = -v
else:
sign = 0
expt = math.floor(math.log(v, 2))
if expt >= -128 and expt <= 126:
mantissa = v / (2**expt)
m1 = (mantissa - 1.0) * 128 + sign
m2 = m1 * 256
m3 = m2 * 256
m4 = m3 * 256
data.extend([int(x) % 256 for x in [expt+129,m1,m2,m3,m4]])
else:
Err.log("Floating point constant out of range")
except ValueError:
Err.log("Expected: floating point")
next = line.expect(',', 'EOL').type
if next == 'EOL':
break
bytes = [IR.ConstantExpr(x) for x in data]
result.append(IR.Node(ppt, "Byte", *bytes))
def expect(self, *tokens):
"""Reads a token from the ParseLine line and returns it if it's of a type
in the sequence tokens. Otherwise, it logs an error."""
token = self.pop()
if token.type not in tokens:
Err.log('Expected: "'+'", "'.join(tokens)+'"')
return token
def pragmaIncbin(ppt, line, result):
"Includes a binary file"
filename = line.expect("STRING").value
offset = IR.ConstantExpr(0)
size = None
if str(line.lookahead(0)) == ",":
line.pop()
offset = FE.parse_expr(line)
if str(line.lookahead(0)) == ",":
line.pop()
size = FE.parse_expr(line)
line.expect("EOL")
if type(filename) == str:
try:
f = file(os.path.join(FE.context_directory, filename), "rb")
if offset.hardcoded and (size is None or size.hardcoded):
# We know how big it will be, we can just use the values.
# First check to make sure they're sane
if offset.value() < 0:
Err.log("Offset may not be negative")
f.close()
return
f.seek(0, 2) # Seek to end of file
if offset.value() > f.tell():
Err.log("Offset runs past end of file")
f.close()
return
if size is not None:
if size.value() < 0:
Err.log("Length may not be negative")
f.close()
return
if offset.value() + size.value() > f.tell():
Err.log(".incbin length too long")
f.close()
return
if size is None:
size = IR.ConstantExpr(-1)
f.seek(offset.value())
bytes = f.read(size.value())
bytes = [IR.ConstantExpr(ord(x)) for x in bytes]
result.append(IR.Node(ppt, "Byte", *bytes))
else:
# offset or length could change based on label placement.
# This seems like an unbelievably bad idea, but since we
# don't have constant prop it will happen for any symbolic
# alias. Don't use symbolic aliases when extracting tiny
# pieces out of humongous files, I guess.
bytes = f.read()
bytes = [IR.ConstantExpr(ord(x)) for x in bytes]
if size is None:
size = IR.SequenceExpr([IR.ConstantExpr(len(bytes)),
"-",
offset])
result.append(IR.Node(ppt, "ByteRange", offset, size, *bytes))
f.close()
except IOError:
Err.log("Could not read " + filename)
return
def __getitem__(self, item):
if item[0] == '_':
for dict in [self.dicts[i] for i in self.stack]:
if item in dict: return dict[item]
else:
if item in self.dicts[0]: return self.dicts[0][item]
Err.log("Unknown label '%s'" % item)
return 0
def visitMacroEnd(self, node, env):
if self.inDef:
Macro.endMacro()
node.nodetype = "None"
node.data = []
self.inDef = False
elif not self.nestedError:
Err.log("Unmatched .macend")
def visitMacroBegin(self, node, env):
if self.inDef:
Err.log("Nested macro definition")
self.nestedError = True
else:
Macro.newMacro(node.data[0])
node.nodetype = "None"
node.data = []
self.inDef = True
def reset(self):
"Clears out program counter, segment, and scoping information"
self.pc = 0
self.segmentdict = {}
self.segment = "*text-default*"
self.scopecount = 0
if len(self.stack) > 1:
Err.log("Unmatched .scope")
self.stack = [0]
def add_token(token):
"Converts a substring into a single lexeme"
if token == "":
return
if token == "0":
result.append(Lexeme("NUM", 0))
return
firstchar = token[0]
rest = token[1:]
if firstchar == '"':
result.append(Lexeme("STRING", rest))
return
elif firstchar in bases:
try:
result.append(Lexeme("NUM", long(rest, bases[firstchar][1])))
return
except ValueError:
Err.log('Invalid ' + bases[firstchar][0] + ' constant: ' +
rest)
result.append(Lexeme("NUM", 0))
return
elif firstchar.isdigit():
try:
result.append(Lexeme("NUM", long(token)))
except ValueError:
Err.log('Identifiers may not begin with a number')
result.append(Lexeme("LABEL", "ERROR"))
return
elif firstchar == "'":
if len(rest) == 1:
result.append(Lexeme("NUM", ord(rest)))
else:
Err.log("Invalid character constant '" + rest + "'")
result.append(Lexeme("NUM", 0))
return
elif firstchar in punctuation:
if rest != "":
Err.log("Internal lexer error! '" + token + "' can't happen!")
result.append(Lexeme(firstchar))
return
else: # Label, opcode, or index register
id = token.lower()
if is_opcode(id):
result.append(Lexeme("OPCODE", id))
elif id == "x":
result.append(Lexeme("X"))
elif id == "y":
result.append(Lexeme("Y"))
elif id == "z":
result.append(Lexeme("Z"))
elif id == "sp":
result.append(Lexeme("SP"))
else:
result.append(Lexeme("LABEL", id))
return
# should never reach here
Err.log("Internal lexer error: add_token fall-through")
def expandMacro(ppt, name, arglist):
global macros
if name not in macros:
Err.log("Undefined macro '%s'" % name)
return IR.NullNode
argexprs = [IR.Node(ppt, "Label", "_*%d" % i, arg) for (i, arg) in zip(xrange(1, sys.maxint), arglist)]
bindexprs = [IR.Node(ppt, "Label", "_%d" % i, IR.LabelExpr("_*%d" % i)) for i in range(1, len(arglist)+1)]
body = [IR.Node("%s->%s" % (ppt, node.ppt), node.nodetype, *node.data) for node in macros[name]]
invocation = [IR.Node(ppt, "ScopeBegin")] + argexprs + [IR.Node(ppt, "ScopeBegin")] + bindexprs + body + [IR.Node(ppt, "ScopeEnd"), IR.Node(ppt, "ScopeEnd")]
return IR.SequenceNode(ppt, invocation)
def endMacro():
global currentname
global currentbody
global macros
if currentname is None:
Err.log("Internal error! Ended a non-existent macro!")
else:
macros[currentname] = currentbody
currentname = None
currentbody = None
def outputbyte(self, expr, env):
'Outputs a byte, with range checking'
if self.writeOK:
val = expr.value(env)
if val < 0x00 or val > 0xff:
Err.log("Byte constant " + str(expr) + " out of range")
val = 0
self.output.append(int(val))
else:
Err.log("Attempt to write to data segment")
def visitAdvance(self, node, env):
pc = env.getPC()
target = node.data[0].value(env)
if (pc > target):
Err.log("Attempted to .advance backwards: $%x to $%x" % (pc, target))
else:
zero = IR.ConstantExpr(0)
for i in xrange(target-pc): self.outputbyte(zero, env)
self.filler += target-pc
env.setPC(target)
def visitLabel(self, node, env):
(label, val) = node.data
fulllabel = "%d:%s" % (env.stack[0], label)
if fulllabel in self.labelmap and self.labelmap[fulllabel] is not node:
Err.log("Duplicate label definition '%s'" % label)
if fulllabel not in self.labelmap:
self.labelmap[fulllabel] = node
if val.valid(env, self.PCvalid) and label not in env:
env[label]=0
self.changed=1
def relativize(self, expr, env, arglen):
"Convert an expression into one for use in relative addressing"
arg = expr.value(env)
arg = arg - (env.getPC() + arglen + 1)
if arg < -128 or arg > 127:
Err.log("Branch target out of bounds")
arg = 0
if arg < 0:
arg += 256
return IR.ConstantExpr(arg)
def outputword_be(self, expr, env):
'Outputs a big-endian word, with range checking'
if self.writeOK:
val = expr.value(env)
if val < 0x0000 or val > 0xFFFF:
Err.log("Word constant " + str(expr) + " out of range")
val = 0
self.output.append(int((val >> 8) & 0xFF))
self.output.append(int(val & 0xFF))
else:
Err.log("Attempt to write to data segment")
def go(self, node, env):
"""Runs this FixPoint's passes, in order, until the fixpoint
is true. Always runs the passes at least once."""
for i in xrange(100):
if Err.count != 0: break
for p in self.passes:
p.go(node, env)
if Err.count != 0: break
if self.fixpoint(): break
if Cmd.verbose > 1: print "Fixpoint failed, looping back"
else:
Err.log("Can't make %s converge! Maybe there's a recursive dependency somewhere?" % self.name)
def newMacro(name):
"Start creating a new macro with the specified name."
global currentname
global currentbody
global macros
if currentname is not None:
Err.log("Internal error! Nested macro attempt!")
else:
if name in macros:
Err.log("Duplicate macro definition '%s'" % name)
currentname = name
currentbody = []
def pragmaCharmapbin(ppt, line, result):
"Load a new character map from a file"
global currentcharmap
filename = line.expect("STRING").value
line.expect("EOL")
if type(filename)==str:
f = file(filename, "rb")
bytes = f.read()
f.close()
if len(bytes)==256:
currentcharmap = bytes
else:
Err.log("Character map "+filename+" not 256 bytes long")
def outputword(self, expr, env, tee=None):
'Outputs a little-endian word, with range checking'
if self.writeOK:
val = expr.value(env)
if val < 0x0000 or val > 0xFFFF:
Err.log("Word constant " + str(expr) + " out of range")
val = 0
self.output.append(int(val & 0xFF))
self.output.append(int((val >> 8) & 0xFF))
if tee is not None:
tee.extend(self.output[-2:])
else:
Err.log("Attempt to write to data segment")
def visitAdvance(self, node, env):
pc = env.getPC()
target = node.data[0].value(env)
if (pc > target):
Err.log("Attempted to .advance backwards: $%x to $%x" %
(pc, target))
else:
created = []
for i in xrange(target - pc):
self.outputbyte(node.data[1], env, created)
self.filler += target - pc
self.registerData(created, env.getPC())
env.setPC(target)
def expandMacro(ppt, name, arglist):
global macros
if name not in macros:
Err.log("Undefined macro '%s'" % name)
return IR.NullNode
argexprs = [IR.Node(ppt, "Label", "_*%d" % i, arg)
for (i, arg) in zip(xrange(1, sys.maxint), arglist)]
bindexprs = [IR.Node(ppt, "Label", "_%d" % i, IR.LabelExpr("_*%d" % i))
for i in range(1, len(arglist) + 1)]
body = [IR.Node("%s->%s" % (ppt, node.ppt), node.nodetype, *node.data)
for node in macros[name]]
invocation = [IR.Node(ppt, "ScopeBegin")] + argexprs + \
[IR.Node(ppt, "ScopeBegin")] + bindexprs + body + \
[IR.Node(ppt, "ScopeEnd"), IR.Node(ppt, "ScopeEnd")]
return IR.SequenceNode(ppt, invocation)
def outputdword_be(self, expr, env, tee=None):
'Outputs a big-endian dword, with range checking'
if self.writeOK:
val = expr.value(env)
if val < 0x00000000 or val > 0xFFFFFFFFL:
Err.log("DWord constant " + str(expr) + " out of range")
val = 0
self.output.append(int((val >> 24) & 0xFF))
self.output.append(int((val >> 16) & 0xFF))
self.output.append(int((val >> 8) & 0xFF))
self.output.append(int(val & 0xFF))
if tee is not None:
tee.extend(self.output[-4:])
else:
Err.log("Attempt to write to data segment")
def expect(self, *tokens):
"""Reads a token from the ParseLine line and returns it if it's of a
type in the sequence tokens. Otherwise, it logs an error."""
token = self.pop()
if token.type in tokens:
return token
if 'LABEL' in tokens:
if token.type in ['X', 'Y', 'Z', 'SP']:
token.value = token.type.lower()
token.type = 'LABEL'
return token
elif token.type == 'OPCODE':
token.type = 'LABEL'
return token
Err.log('Expected: "' + '", "'.join(tokens) + '"')
return token
def pragmaCharmap(ppt, line, result):
"Modify the character map."
global currentcharmap, basecharmap
bytes = readData(line)
if len(bytes) == 0:
currentcharmap = basecharmap
else:
try:
base = bytes[0].data
newsubstr = "".join([chr(x.data) for x in bytes[1:]])
currentcharmap = currentcharmap[:base] + newsubstr + currentcharmap[base+len(newsubstr):]
if len(currentcharmap) != 256 or base < 0 or base > 255:
Err.log("Charmap replacement out of range")
currentcharmap = currentcharmap[:256]
except ValueError:
Err.log("Illegal character in .charmap directive")
def visitLabel(self, node, env):
(label, val) = node.data
fulllabel = "%d:%s" % (env.stack[0], label)
if fulllabel in self.labelmap and self.labelmap[fulllabel] is not node:
Err.log("Duplicate label definition '%s'" % label)
if fulllabel not in self.labelmap:
self.labelmap[fulllabel] = node
if val.valid(env, self.PCvalid) and label not in env:
env[label] = 0
self.changed = True
if label in ['a', 'x', 'y'] and self.runcount == 1:
print>>sys.stderr, str(node.ppt) + ": WARNING: " \
"using register name as label"
if label in Ops.opcodes and self.runcount == 1:
print>>sys.stderr, str(node.ppt) + ": WARNING: " \
"using opcode name as label"
def parse_file(ppt, filename):
"Loads a .P65 source file, and returns an IR list."
Err.currentpoint = ppt
if Cmd.verbose > 0: print("Loading " + filename)
try:
f = open(filename, 'r', encoding='utf-8')
linelist = f.readlines()
f.close()
pptlist = ["%s:%d" % (filename, i + 1) for i in range(len(linelist))]
lexlist = map(lex, pptlist, linelist)
IRlist = map(parse_line, pptlist, lexlist)
IRlist = [node for node in IRlist if node is not IR.NullNode]
return IR.SequenceNode(ppt, IRlist)
except IOError:
Err.log("Could not read " + filename)
return IR.NullNode
def parse_file(ppt, filename):
"Loads a .P65 source file, and returns an IR list."
Err.currentpoint = ppt
if Cmd.verbose > 0: print "Loading "+filename
try:
f = file(filename)
linelist = f.readlines()
f.close()
pptlist = ["%s:%d" % (filename, i+1) for i in range(len(linelist))]
lexlist = map(lex, pptlist, linelist)
IRlist = map(parse_line, pptlist, lexlist)
IRlist = [node for node in IRlist if node is not IR.NullNode]
return IR.SequenceNode(ppt, IRlist)
except IOError:
Err.log ("Could not read "+filename)
return IR.NullNode
def expect(self, *tokens):
"""Reads a token from the ParseLine line and returns it if it's of a
type in the sequence tokens. Otherwise, it logs an error."""
token = self.pop()
if token.type in tokens:
return token
if 'LABEL' in tokens:
if token.type in ['X', 'Y', 'Z', 'SP']:
token.value = token.type.lower()
token.type = 'LABEL'
return token
elif token.type == 'OPCODE':
token.type = 'LABEL'
return token
Err.log('Expected: "' + '", "'.join(tokens) + '"')
return token
def visitLabel(self, node, env):
(label, val) = node.data
fulllabel = "%d:%s" % (env.stack[0], label)
if fulllabel in self.labelmap and self.labelmap[fulllabel] is not node:
Err.log("Duplicate label definition '%s'" % label)
if fulllabel not in self.labelmap:
self.labelmap[fulllabel] = node
if val.valid(env, self.PCvalid) and label not in env:
env[label] = 0
self.changed = True
if label in ['a', 'x', 'y'] and self.runcount == 1:
print>>sys.stderr, str(node.ppt) + ": WARNING: " \
"using register name as label"
if label in Ops.opcodes and self.runcount == 1:
print>>sys.stderr, str(node.ppt) + ": WARNING: " \
"using opcode name as label"
def expect(self, *tokens):
"""Reads a token from the ParseLine line and returns it if it's of a
type in the sequence tokens. Otherwise, it logs an error."""
token = self.pop()
if token.type in tokens:
return token
if "LABEL" in tokens:
if token.type in ["X", "Y"]:
token.value = token.type.lower()
token.type = "LABEL"
return token
elif token.type == "OPCODE":
token.type = "LABEL"
return token
Err.log('Expected: "' + '", "'.join(tokens) + '"')
return token
def go(self, node, env):
"""Runs this FixPoint's passes, in order, until the fixpoint
is true. Always runs the passes at least once."""
for i in xrange(100):
if Err.count != 0:
break
for p in self.passes:
p.go(node, env)
if Err.count != 0:
break
if self.fixpoint():
break
if Cmd.print_pass:
print>>sys.stderr, "Fixpoint failed, looping back"
else:
Err.log("Can't make %s converge! Maybe there's a recursive "
"dependency somewhere?" % self.name)
def pragmaCharmap(ppt, line, result):
"Modify the character map."
global currentcharmap, basecharmap
if str(line.lookahead(0)) == "EOL":
currentcharmap = basecharmap
else:
bytes = readData(line)
try:
base = bytes[0].data
newsubstr = "".join([chr(x.data) for x in bytes[1:]])
currentcharmap = currentcharmap[:base] + newsubstr + \
currentcharmap[base + len(newsubstr):]
if len(currentcharmap) != 256 or base < 0 or base > 255:
Err.log("Charmap replacement out of range")
currentcharmap = currentcharmap[:256]
except ValueError:
Err.log("Illegal character in .charmap directive")
def pragmaCharmapbin(ppt, line, result):
"Load a new character map from a file"
global currentcharmap
filename = line.expect("STRING").value
line.expect("EOL")
if type(filename) == str:
try:
f = open(os.path.join(FE.context_directory, filename), "rb")
bytes = f.read()
f.close()
except IOError:
Err.log("Could not read " + filename)
return
if len(bytes) == 256:
currentcharmap = bytes
else:
Err.log("Character map " + filename + " not 256 bytes long")
def pragmaCharmapbin(ppt, line, result):
"Load a new character map from a file"
global currentcharmap
filename = line.expect("STRING").value
line.expect("EOL")
if type(filename) == str:
try:
f = file(os.path.join(FE.context_directory, filename), "rb")
bytes = f.read()
f.close()
except IOError:
Err.log("Could not read " + filename)
return
if len(bytes) == 256:
currentcharmap = bytes
else:
Err.log("Character map " + filename + " not 256 bytes long")
def run_all(infile, outfile):
"Transforms the source infile to a binary outfile."
Err.count = 0
z = Ophis.Frontend.parse(infile)
env = Ophis.Environment.Environment()
m = Ophis.Passes.ExpandMacros()
i = Ophis.Passes.InitLabels()
l_basic = Ophis.Passes.UpdateLabels()
l = Ophis.Passes.FixPoint("label update", [l_basic],
lambda: l_basic.changed == 0)
c = Ophis.Passes.Collapse()
a = Ophis.Passes.Assembler()
passes = []
passes.append(Ophis.Passes.DefineMacros())
passes.append(
Ophis.Passes.FixPoint("macro expansion", [m], lambda: m.changed == 0))
passes.append(
Ophis.Passes.FixPoint("label initialization", [i],
lambda: i.changed == 0))
passes.extend([
Ophis.Passes.CircularityCheck(),
Ophis.Passes.CheckExprs(),
Ophis.Passes.EasyModes()
])
passes.append(
Ophis.Passes.FixPoint("instruction selection", [l, c],
lambda: c.collapsed == 0))
passes.extend(
[Ophis.Passes.NormalizeModes(),
Ophis.Passes.UpdateLabels(), a])
for p in passes:
p.go(z, env)
if Err.count == 0:
try:
output = open(outfile, 'wb')
output.write(bytes(a.output))
output.close()
except IOError:
print("Could not write to " + outfile)
else:
Err.report()
def parse_file(ppt, filename, load_once=False):
"Loads an Ophis source file, and returns an IR list."
global context_directory, loadedfiles
Err.currentpoint = ppt
old_context = context_directory
if filename != '-':
if context_directory is not None:
filename = os.path.abspath(
os.path.join(context_directory, filename))
if load_once and filename in loadedfiles:
if Cmd.print_loaded_files:
print("Skipping " + filename, file=sys.stderr)
return IR.NullNode
loadedfiles[filename] = True
if Cmd.print_loaded_files:
if filename != '-':
print("Loading " + filename, file=sys.stderr)
else:
print("Loading from standard input", file=sys.stderr)
try:
if filename != '-':
if context_directory is not None:
filename = os.path.join(context_directory, filename)
f = open(filename, "rt")
linelist = f.readlines()
f.close()
context_directory = os.path.abspath(os.path.dirname(filename))
else:
context_directory = os.getcwd()
linelist = sys.stdin.readlines()
pptlist = ["%s:%d" % (filename, i + 1) for i in range(len(linelist))]
lexlist = list(map(lex, pptlist, linelist))
IRlist = list(map(parse_line, pptlist, lexlist))
IRlist = [node for node in IRlist if node is not IR.NullNode]
context_directory = old_context
return IR.SequenceNode(ppt, IRlist)
except IOError:
Err.log("Could not read " + filename)
context_directory = old_context
return IR.NullNode
def parse_file(ppt, filename, load_once=False):
"Loads an Ophis source file, and returns an IR list."
global context_directory, loadedfiles
Err.currentpoint = ppt
old_context = context_directory
if filename != '-':
if context_directory is not None:
filename = os.path.abspath(os.path.join(context_directory,
filename))
if load_once and filename in loadedfiles:
if Cmd.print_loaded_files:
print>>sys.stderr, "Skipping " + filename
return IR.NullNode
loadedfiles[filename] = True
if Cmd.print_loaded_files:
if filename != '-':
print>>sys.stderr, "Loading " + filename
else:
print>>sys.stderr, "Loading from standard input"
try:
if filename != '-':
if context_directory is not None:
filename = os.path.join(context_directory, filename)
f = file(filename)
linelist = f.readlines()
f.close()
context_directory = os.path.abspath(os.path.dirname(filename))
else:
context_directory = os.getcwd()
linelist = sys.stdin.readlines()
pptlist = ["%s:%d" % (filename, i + 1) for i in range(len(linelist))]
lexlist = map(lex, pptlist, linelist)
IRlist = map(parse_line, pptlist, lexlist)
IRlist = [node for node in IRlist if node is not IR.NullNode]
context_directory = old_context
return IR.SequenceNode(ppt, IRlist)
except IOError:
Err.log("Could not read " + filename)
context_directory = old_context
return IR.NullNode
def assemble(self, node, mode, env):
"A generic instruction called by the visitor methods themselves"
(opcode, expr) = node.data
bin_op = Ops.opcodes[opcode][mode]
if bin_op is None:
Err.log('%s does not have mode "%s"' %
(opcode.upper(), Ops.modes[mode]))
return
self.outputbyte(IR.ConstantExpr(bin_op), env)
arglen = Ops.lengths[mode]
if mode == 11: # Special handling for relative mode
arg = expr.value(env)
arg = arg - (env.getPC() + 2)
if arg < -128 or arg > 127:
Err.log("Branch target out of bounds")
arg = 0
if arg < 0: arg += 256
expr = IR.ConstantExpr(arg)
if arglen == 1: self.outputbyte(expr, env)
if arglen == 2: self.outputword(expr, env)
env.incPC(1 + arglen)
self.code += 1 + arglen
def atom():
"Parses lowest-priority expression components."
global templabelcount
next = line.lookahead(0).type
if next == "NUM":
return IR.ConstantExpr(line.expect("NUM").value)
elif next in ["LABEL", "X", "Y", "Z", "SP", "OPCODE"]:
return IR.LabelExpr(line.expect("LABEL").value)
elif next == "^":
line.expect("^")
return IR.PCExpr()
elif next == "[":
line.expect("[")
result = parse_expr(line)
line.expect("]")
return result
elif next == "+":
offset = 0
while next == "+":
offset += 1
line.expect("+")
next = line.lookahead(0).type
return IR.LabelExpr("*" + str(templabelcount + offset))
elif next == "-":
offset = 1
while next == "-":
offset -= 1
line.expect("-")
next = line.lookahead(0).type
return IR.LabelExpr("*" + str(templabelcount + offset))
elif next == ">":
line.expect(">")
return IR.HighByteExpr(atom())
elif next == "<":
line.expect("<")
return IR.LowByteExpr(atom())
else:
Err.log('Expected: expression')
def assemble(self, node, mode, env):
"A generic instruction called by the visitor methods themselves"
(opcode, expr, expr2) = node.data
bin_op = Ops.opcodes[opcode][mode]
if bin_op is None:
Err.log('%s does not have mode "%s"' % (opcode.upper(),
Ops.modes[mode]))
return
inst_bytes = []
self.outputbyte(IR.ConstantExpr(bin_op), env, inst_bytes)
arglen = Ops.lengths[mode]
val1 = None
val2 = None
if expr is not None:
val1 = expr.value(env)
if expr2 is not None:
val2 = expr2.value(env)
if mode == Ops.modes.index("Zero Page, Relative"):
expr2 = self.relativize(expr2, env, arglen)
self.outputbyte(expr, env, inst_bytes)
self.outputbyte(expr2, env, inst_bytes)
else:
if mode == Ops.modes.index("Relative"):
expr = self.relativize(expr, env, arglen)
elif mode == Ops.modes.index("RelativeLong"):
expr = self.relativizelong(expr, env, arglen)
if arglen == 1:
self.outputbyte(expr, env, inst_bytes)
elif arglen == 2:
self.outputword(expr, env, inst_bytes)
self.listing.listInstruction(self.listing_string(env.getPC(),
inst_bytes,
mode, opcode,
val1, val2))
env.incPC(1 + arglen)
self.code += 1 + arglen
def visitByteRange(self, node, env):
offset = node.data[0].value(env) + 2
length = node.data[1].value(env)
if offset < 2:
Err.log("Negative offset in .incbin")
elif offset > len(node.data):
Err.log("Offset extends past end of file")
elif length < 0:
Err.log("Negative length")
elif offset + length > len(node.data):
Err.log("File too small for .incbin subrange")
else:
created = []
for expr in node.data[offset:(offset + length)]:
self.outputbyte(expr, env, created)
self.registerData(created, env.getPC())
env.incPC(length)
self.data += length
def aux():
"Accumulates all IR nodes defined by this line."
if line.lookahead(0).type == "EOL":
pass
elif line.lookahead(1).type == ":":
newlabel = line.expect("LABEL").value
line.expect(":")
result.append(IR.Node(ppt, "Label", newlabel, IR.PCExpr()))
aux()
elif line.lookahead(0).type == "*":
global templabelcount
templabelcount = templabelcount + 1
result.append(
IR.Node(ppt, "Label", "*" + str(templabelcount), IR.PCExpr()))
line.expect("*")
aux()
elif line.lookahead(0).type == "." or line.lookahead(0).type == "`":
which = line.expect(".", "`").type
if (which == "."): pragma = line.expect("LABEL").value
else: pragma = "invoke"
pragmaFunction = "pragma" + pragma.title()
for mod in pragma_modules:
if hasattr(mod, pragmaFunction):
getattr(mod, pragmaFunction)(ppt, line, result)
break
else:
Err.log("Unknown pragma " + pragma)
else: # Instruction
opcode = line.expect("OPCODE").value
if line.lookahead(0).type == "#":
mode = "Immediate"
line.expect("#")
arg = parse_expr(line)
line.expect("EOL")
elif line.lookahead(0).type == "(":
line.expect("(")
arg = parse_expr(line)
if line.lookahead(0).type == ",":
mode = "IndirectX"
line.expect(",")
line.expect("X")
line.expect(")")
line.expect("EOL")
else:
line.expect(")")
tok = line.expect(",", "EOL").type
if tok == "EOL":
mode = "Indirect"
else:
mode = "IndirectY"
line.expect("Y")
line.expect("EOL")
elif line.lookahead(0).type == "EOL":
mode = "Implied"
arg = None
else:
arg = parse_expr(line)
tok = line.expect("EOL", ",").type
if tok == ",":
tok = line.expect("X", "Y").type
if tok == "X": mode = "MemoryX"
else: mode = "MemoryY"
line.expect("EOL")
else:
mode = "Memory"
result.append(IR.Node(ppt, mode, opcode, arg))
def visitAdvance(self, node, env):
PCvalid = self.PCvalid
self.PCvalid = node.data[0].valid(env, self.PCvalid)
if not node.data[0].valid(env, PCvalid):
Err.log("Undefined or circular reference on .advance")
def postPass(self):
if self.inDef:
Err.log("Unmatched .macro")
elif Cmd.print_ir:
print>>sys.stderr, "Macro definitions:"
Macro.dump()
def run_all():
"""Transforms the source infiles to a binary outfile.
Returns a shell-style exit code: 1 if there were errors, 0 if there
were no errors.
"""
Err.count = 0
z = Ophis.Frontend.parse(Ophis.CmdLine.infiles)
env = Ophis.Environment.Environment()
m = Ophis.Passes.ExpandMacros()
i = Ophis.Passes.InitLabels()
l_basic = Ophis.Passes.UpdateLabels()
l = Ophis.Passes.FixPoint("label update", [l_basic],
lambda: not l_basic.changed)
# The instruction selector is a bunch of fixpoints, and which
# passes run depends on the command line options a bit.
c_basic = Ophis.Passes.Collapse()
c = Ophis.Passes.FixPoint("instruction selection 1", [l, c_basic],
lambda: not c_basic.changed)
if Ophis.CmdLine.enable_branch_extend:
b = Ophis.Passes.ExtendBranches()
instruction_select = Ophis.Passes.FixPoint("instruction selection 2",
[c, b],
lambda: not b.changed)
else:
instruction_select = c
a = Ophis.Passes.Assembler()
passes = []
passes.append(Ophis.Passes.DefineMacros())
passes.append(
Ophis.Passes.FixPoint("macro expansion", [m], lambda: not m.changed))
passes.append(
Ophis.Passes.FixPoint("label initialization", [i],
lambda: not i.changed))
passes.extend([
Ophis.Passes.CircularityCheck(),
Ophis.Passes.CheckExprs(),
Ophis.Passes.EasyModes()
])
passes.append(instruction_select)
passes.extend(
[Ophis.Passes.NormalizeModes(),
Ophis.Passes.UpdateLabels(), a])
for p in passes:
p.go(z, env)
if Err.count == 0:
try:
outfile = Ophis.CmdLine.outfile
if outfile == '-':
output = sys.stdout
if sys.platform == "win32":
# We can't dump our binary in text mode; that would be
# disastrous. So, we'll do some platform-specific
# things here to force our stdout to binary mode.
import msvcrt
msvcrt.setmode(sys.stdout.fileno(), os.O_BINARY)
elif outfile is None:
output = file('ophis.bin', 'wb')
else:
output = file(outfile, 'wb')
output.write("".join(map(chr, a.output)))
output.flush()
if outfile != '-':
output.close()
return 0
except IOError:
print >> sys.stderr, "Could not write to " + outfile
return 1
else:
Err.report()
return 1
def visitUnknown(self, node, env):
Err.log("Internal error! " + self.name +
" cannot understand node type " + node.nodetype)
def visitCheckPC(self, node, env):
pc = env.getPC()
target = node.data[0].value(env)
if (pc > target):
Err.log(".checkpc assertion failed: $%x > $%x" % (pc, target))
def aux():
"Accumulates all IR nodes defined by this line."
if line.lookahead(0).type == "EOL":
pass
elif line.lookahead(1).type == ":":
newlabel = line.expect("LABEL").value
line.expect(":")
result.append(IR.Node(ppt, "Label", newlabel, IR.PCExpr()))
aux()
elif line.lookahead(0).type == "*":
global templabelcount
templabelcount = templabelcount + 1
result.append(
IR.Node(ppt, "Label", "*" + str(templabelcount), IR.PCExpr()))
line.expect("*")
aux()
elif line.lookahead(0).type == "." or line.lookahead(0).type == "`":
which = line.expect(".", "`").type
if (which == "."):
pragma = line.expect("LABEL").value
else:
pragma = "invoke"
pragmaFunction = "pragma" + pragma.title()
for mod in pragma_modules:
if hasattr(mod, pragmaFunction):
getattr(mod, pragmaFunction)(ppt, line, result)
break
else:
Err.log("Unknown pragma " + pragma)
else: # Instruction
opcode = line.expect("OPCODE").value
arg2 = None
if line.lookahead(0).type == "#":
mode = "Immediate"
line.expect("#")
arg = parse_expr(line)
line.expect("EOL")
elif line.lookahead(0).type == "(":
line.expect("(")
arg = parse_expr(line)
if line.lookahead(0).type == ",":
line.expect(",")
if line.lookahead(0).type == "X":
mode = "PointerX"
line.expect("X")
line.expect(")")
line.expect("EOL")
else:
mode = "PointerSPY"
line.expect("SP")
line.expect(")")
line.expect(",")
line.expect("Y")
line.expect("EOL")
else:
line.expect(")")
tok = line.expect(",", "EOL").type
if tok == "EOL":
mode = "Pointer"
else:
if line.lookahead(0).type == "Y":
mode = "PointerY"
line.expect("Y")
line.expect("EOL")
else:
mode = "PointerZ"
line.expect("Z")
line.expect("EOL")
elif line.lookahead(0).type == "EOL":
mode = "Implied"
arg = None
else:
arg = parse_expr(line)
tok = line.expect("EOL", ",").type
if tok == ",":
# Parser has to special-case the BBXn instructions,
# Which uniquely take two addresses
if opcode[:3] in ["bbs", "bbr"]:
arg2 = parse_expr(line)
mode = "Memory2"
else:
tok = line.expect("X", "Y", "Z").type
if tok == "X":
mode = "MemoryX"
elif tok == "Y":
mode = "MemoryY"
else:
mode = "MemoryZ"
line.expect("EOL")
else:
mode = "Memory"
result.append(IR.Node(ppt, mode, opcode, arg, arg2))
def visitLabel(self, node, env):
(label, val) = node.data
if not val.valid(env, self.PCvalid):
Err.log("Undefined or circular dependency for label '%s'" % label)
def visitCheckPC(self, node, env):
if not node.data[0].valid(env, self.PCvalid):
Err.log("Undefined or circular reference on program counter check")
def endscope(self):
"Leaves a scope."
if len(self.stack) == 1:
Err.log("Unmatched .scend")
self.stack.pop(0)
def lex(point, line):
"""Turns a line of source into a sequence of lexemes."""
Err.currentpoint = point
result = []
def is_opcode(op):
"Tests whether a string is an opcode or an identifier"
return op in Ops.opcodes
def add_token(token):
"Converts a substring into a single lexeme"
if token == "":
return
if token == "0":
result.append(Lexeme("NUM", 0))
return
firstchar = token[0]
rest = token[1:]
if firstchar == '"':
result.append(Lexeme("STRING", rest))
return
elif firstchar in bases:
try:
result.append(Lexeme("NUM", int(rest, bases[firstchar][1])))
return
except ValueError:
Err.log('Invalid ' + bases[firstchar][0] + ' constant: ' +
rest)
result.append(Lexeme("NUM", 0))
return
elif firstchar.isdigit():
try:
result.append(Lexeme("NUM", int(token)))
except ValueError:
Err.log('Identifiers may not begin with a number')
result.append(Lexeme("LABEL", "ERROR"))
return
elif firstchar == "'":
if len(rest) == 1:
result.append(Lexeme("NUM", ord(rest)))
else:
Err.log("Invalid character constant '" + rest + "'")
result.append(Lexeme("NUM", 0))
return
elif firstchar in punctuation:
if rest != "":
Err.log("Internal lexer error! '" + token + "' can't happen!")
result.append(Lexeme(firstchar))
return
else: # Label, opcode, or index register
id = token.lower()
if is_opcode(id):
result.append(Lexeme("OPCODE", id))
elif id == "x":
result.append(Lexeme("X"))
elif id == "y":
result.append(Lexeme("Y"))
elif id == "z":
result.append(Lexeme("Z"))
elif id == "sp":
result.append(Lexeme("SP"))
else:
result.append(Lexeme("LABEL", id))
return
# should never reach here
Err.log("Internal lexer error: add_token fall-through")
def add_EOL():
"Adds an end-of-line lexeme"
result.append(Lexeme("EOL"))
# Actual routine begins here
value = ""
quotemode = False
backslashmode = False
for c in line.strip():
if backslashmode:
backslashmode = False
value = value + c
elif c == "\\":
backslashmode = True
elif quotemode:
if c == '"':
quotemode = False
else:
value = value + c
elif c == ';':
add_token(value)
value = ""
break
elif c == '.' and value != "":
value = value + c
elif c.isspace():
add_token(value)
value = ""
elif c in punctuation:
add_token(value)
add_token(c)
value = ""
elif c == '"':
add_token(value)
value = '"'
quotemode = True
else:
value = value + c
if backslashmode:
Err.log("Backslashed newline")
if quotemode:
Err.log("Unterminated string constant")
add_token(value)
add_EOL()
return result
def postPass(self):
if self.inDef:
Err.log("Unmatched .macro")
elif Cmd.verbose > 2:
print("Macro definitions:")
Macro.dump()
def visitSetPC(self, node, env):
PCvalid = self.PCvalid
self.PCvalid = node.data[0].valid(env, self.PCvalid)
if not node.data[0].valid(env, PCvalid):
Err.log("Undefined or circular reference on program counter set")