def __init__(self,
snapshot,
ctl_parser,
config=None,
final=False,
defb_size=8,
defb_mod=1,
zfill=False,
defm_width=66,
asm_hex=False,
asm_lower=False):
ctl_parser.apply_asm_data_directives(snapshot)
self.disassembler = Disassembler(snapshot, defb_size, defb_mod, zfill,
defm_width, asm_hex, asm_lower)
self.ctl_parser = ctl_parser
if asm_hex:
if asm_lower:
self.address_fmt = '{0:04x}'
else:
self.address_fmt = '{0:04X}'
else:
self.address_fmt = '{0}'
self.entry_map = {}
self.config = config or {}
self.build(final)
def _get_code_blocks(snapshot, start, end, fname):
if os.path.isdir(fname):
raise SkoolKitError('{0} is a directory'.format(fname))
try:
size = os.path.getsize(fname)
except OSError as e:
if e.errno == 2:
raise SkoolKitError('{0}: file not found'.format(fname))
raise SkoolKitError('Failed to get size of {}: {}'.format(
fname, e.strerror))
if size == 8192:
# Assume this is a Z80 map file
sys.stderr.write('Reading {0}'.format(fname))
sys.stderr.flush()
addresses = []
data = read_bin_file(fname)
address = start & 65528
for b in data[start // 8:end // 8 + 1]:
for i in range(8):
if b & 1 and start <= address < end:
addresses.append(address)
b >>= 1
address += 1
elif size == 65536:
# Assume this is a SpecEmu map file
sys.stderr.write('Reading {}'.format(fname))
sys.stderr.flush()
addresses = []
data = read_bin_file(fname)
for address in range(start, end):
if data[address] & 1:
addresses.append(address)
else:
sys.stderr.write('Reading {0}: '.format(fname))
sys.stderr.flush()
with open_file(fname) as f:
addresses = _get_addresses(f, fname, size, start, end)
sys.stderr.write('\n')
code_blocks = []
disassembler = Disassembler(snapshot)
for address in addresses:
size = disassembler.disassemble(address, address + 1)[0].size()
if code_blocks and address <= sum(code_blocks[-1]):
if address == sum(code_blocks[-1]):
code_blocks[-1][1] += size
else:
code_blocks.append([address, size])
return code_blocks
def _get_code_blocks(snapshot, start, end, fname):
if os.path.isdir(fname):
raise SkoolKitError('{0} is a directory'.format(fname))
try:
size = os.path.getsize(fname)
except OSError as e:
if e.errno == 2:
raise SkoolKitError('{0}: file not found'.format(fname))
raise # pragma: no cover
if size == 8192:
# Assume this is a Z80 map file
sys.stderr.write('Reading {0}'.format(fname))
sys.stderr.flush()
addresses = []
data = read_bin_file(fname)
address = start & 65528
for b in data[start // 8:end // 8 + 1]:
for i in range(8):
if b & 1 and start <= address < end:
addresses.append(address)
b >>= 1
address += 1
elif size == 65536:
# Assume this is a SpecEmu map file
sys.stderr.write('Reading {}'.format(fname))
sys.stderr.flush()
addresses = []
data = read_bin_file(fname)
for address in range(start, end):
if data[address] & 1:
addresses.append(address)
else:
sys.stderr.write('Reading {0}: '.format(fname))
sys.stderr.flush()
with open_file(fname) as f:
addresses = _get_addresses(f, fname, size, start, end)
sys.stderr.write('\n')
code_blocks = []
disassembler = Disassembler(snapshot)
for address in addresses:
size = disassembler.disassemble(address, address + 1)[0].size()
if code_blocks and address <= sum(code_blocks[-1]):
if address == sum(code_blocks[-1]):
code_blocks[-1][1] += size
else:
code_blocks.append([address, size])
return code_blocks
def __init__(self,
snapshot,
sftfile,
zfill=False,
asm_hex=False,
asm_lower=False):
self.snapshot = snapshot
self.disassembler = Disassembler(snapshot,
zfill=zfill,
asm_hex=asm_hex,
asm_lower=asm_lower)
self.sftfile = sftfile
self.address_fmt = get_address_format(asm_hex, asm_lower)
self.stack = []
self.disassemble = True
def __init__(self, snapshot, sftfile, zfill=False, asm_hex=False, asm_lower=False):
self.snapshot = snapshot
self.disassembler = Disassembler(snapshot, zfill=zfill, asm_hex=asm_hex, asm_lower=asm_lower)
self.sftfile = sftfile
self.address_fmt = get_address_format(asm_hex, asm_lower)
self.stack = []
self.disassemble = True
def __init__(self, snapshot, ctl_parser, final=False, defb_size=8, defb_mod=1, zfill=False, defm_width=66, asm_hex=False, asm_lower=False):
self.disassembler = Disassembler(snapshot, defb_size, defb_mod, zfill, defm_width, asm_hex, asm_lower)
self.ctl_parser = ctl_parser
if asm_hex:
if asm_lower:
self.address_fmt = '{0:04x}'
else:
self.address_fmt = '{0:04X}'
else:
self.address_fmt = '{0}'
self.entry_map = {}
self.build(final)
def __init__(self, snapshot, ctl_parser, config=None, final=False, defb_size=8, defb_mod=1,
zfill=False, defm_width=66, asm_hex=False, asm_lower=False):
ctl_parser.apply_asm_data_directives(snapshot)
self.disassembler = Disassembler(snapshot, defb_size, defb_mod, zfill, defm_width, asm_hex, asm_lower)
self.ctl_parser = ctl_parser
if asm_hex:
if asm_lower:
self.address_fmt = '{0:04x}'
else:
self.address_fmt = '{0:04X}'
else:
self.address_fmt = '{0}'
self.entry_map = {}
self.config = config or {}
self.build(final)
class SftParser:
def __init__(self,
snapshot,
sftfile,
zfill=False,
asm_hex=False,
asm_lower=False):
self.snapshot = snapshot
self.disassembler = Disassembler(snapshot,
zfill=zfill,
asm_hex=asm_hex,
asm_lower=asm_lower)
self.sftfile = sftfile
self.address_fmt = get_address_format(asm_hex, asm_lower)
self.stack = []
self.disassemble = True
def _parse_instruction(self, line):
ctl = line[0]
lengths = []
if line[1] == ';':
inst_ctl = None
start = None
i = line.find(' ', 2)
if i < 0:
i = len(line.rstrip())
comment_index = get_int_param(line[2:i])
else:
inst_ctl = line[1]
if inst_ctl == 'I':
i = find_unquoted(line.rstrip(), ' ', 2)
address_end = j = find_unquoted(line, ';', 2, i)
else:
address_end = line.index(',', 2)
i = find_unquoted(line.rstrip(), ' ', address_end + 1)
j = find_unquoted(line, ';', address_end + 1, i)
start = get_int_param(line[2:address_end])
if j == i:
comment_index = -1
else:
comment_index = get_int_param(line[j + 1:i])
if j > address_end + 1:
params = split_unquoted(line[address_end + 1:j], ',')
lengths = parse_params(inst_ctl, params, 0)
elif inst_ctl != 'I':
raise ValueError
comment = line[i:].strip()
return ctl, inst_ctl, start, lengths, comment_index, comment
def _parse_asm_directive(self, directive):
if parse_asm_block_directive(directive, self.stack):
self.disassemble = True
for _p, i in self.stack:
if i != '-':
self.disassemble = False
break
def _set_bytes(self, line):
address = parse_int(line[1:6])
if address is not None:
comment_index = find_unquoted(line, ';')
operation = line[7:comment_index].strip()
set_bytes(self.snapshot, address, operation)
def _parse_sft(self, min_address, max_address):
start_index = -1
lines = []
v_block_ctl = None
f = open_file(self.sftfile)
for line in f:
if line.startswith('#'):
# This line is a skool file template comment
continue
if not line.strip():
# This line is blank
lines.append(VerbatimLine(line))
v_block_ctl = None
continue
if line.startswith(';'):
# This line is an entry-level comment
lines.append(VerbatimLine(line))
continue
if line.startswith('@'):
lines.append(VerbatimLine(line))
self._parse_asm_directive(line[1:].rstrip())
continue
if not self.disassemble:
# This line is inside a '+' block, so include it as is
lines.append(VerbatimLine(line))
continue
# Check whether we're in a block that should be restored verbatim
if v_block_ctl is None and line.startswith(VERBATIM_BLOCKS):
v_block_ctl = line[0]
if v_block_ctl:
if v_block_ctl == 'd':
self._set_bytes(line)
lines.append(VerbatimLine(line))
continue
# Check whether the line starts with a valid character
if line[0] not in VALID_CTLS:
lines.append(VerbatimLine(line))
continue
try:
ctl, inst_ctl, start, lengths, comment_index, comment = self._parse_instruction(
line)
except (IndexError, ValueError):
raise SftParsingError("Invalid line: {0}".format(
line.split()[0]))
if start is not None:
# This line contains a control directive
if start >= min_address > 0 and start_index < 0:
start_index = len(lines)
instructions = []
for length, sublengths in lengths:
end = start + length
if inst_ctl == 'C':
base = sublengths[0][1]
instructions += self.disassembler.disassemble(
start, end, base)
elif inst_ctl == 'W':
instructions += self.disassembler.defw_range(
start, end, sublengths)
elif inst_ctl == 'T':
instructions += self.disassembler.defm_range(
start, end, sublengths)
elif inst_ctl == 'S':
instructions.append(
self.disassembler.defs(start, end, sublengths))
else:
instructions += self.disassembler.defb_range(
start, end, sublengths)
start += length
if instructions:
done = False
for instruction in instructions:
if instruction.address >= max_address:
while lines and lines[-1].is_trimmable():
lines.pop()
done = True
break
address = self.address_fmt.format(instruction.address)
lines.append(
InstructionLine(ctl, address,
instruction.operation,
comment_index, comment))
ctl = ' '
if done:
break
else:
lines.append(
InstructionLine(ctl, start, '', comment_index,
comment))
else:
# This line is an instruction-level comment continuation line
lines.append(
InstructionLine(comment_index=comment_index,
comment=comment))
f.close()
if start_index < 0:
return lines
if start_index < len(lines):
if str(lines[start_index])[0] in DIRECTIVES:
while start_index > 0 and not lines[start_index].is_blank():
start_index -= 1
else:
while start_index < len(
lines) and not lines[start_index].is_blank():
start_index += 1
return lines[start_index + 1:]
return []
def write_skool(self, min_address=0, max_address=65536):
for line in self._parse_sft(min_address, max_address):
write_line(str(line))
class Disassembly:
def __init__(self,
snapshot,
ctl_parser,
config=None,
final=False,
defb_size=8,
defb_mod=1,
zfill=False,
defm_width=66,
asm_hex=False,
asm_lower=False):
ctl_parser.apply_asm_data_directives(snapshot)
self.disassembler = Disassembler(snapshot, defb_size, defb_mod, zfill,
defm_width, asm_hex, asm_lower)
self.ctl_parser = ctl_parser
if asm_hex:
if asm_lower:
self.address_fmt = '{0:04x}'
else:
self.address_fmt = '{0:04X}'
else:
self.address_fmt = '{0}'
self.entry_map = {}
self.config = config or {}
self.build(final)
def build(self, final=False):
self.instructions = {}
self.entries = []
self._create_entries()
if self.entries:
self.org = self.entries[0].address
else:
self.org = None
if final:
self._calculate_references()
def _create_entries(self):
for block in self.ctl_parser.get_blocks():
if block.start in self.entry_map:
entry = self.entry_map[block.start]
self.entries.append(entry)
for instruction in entry.instructions:
self.instructions[instruction.address] = instruction
continue
title = block.title
if not title:
ctl = block.ctl
if ctl != 'i' or block.description or block.registers or block.blocks[
0].header:
name = 'Title-' + ctl
title = format_template(self.config.get(name, ''),
name,
address=self._address_str(
block.start))
for sub_block in block.blocks:
address = sub_block.start
if sub_block.ctl in 'cBT':
base = sub_block.sublengths[0][1]
instructions = self.disassembler.disassemble(
sub_block.start, sub_block.end, base)
elif sub_block.ctl in 'bgstuw':
sublengths = sub_block.sublengths
if sublengths[0][0]:
if sub_block.ctl == 's':
length = sublengths[0][0]
else:
length = sum([s[0] for s in sublengths])
else:
length = sub_block.end - sub_block.start
instructions = []
while address < sub_block.end:
end = min(address + length, sub_block.end)
if sub_block.ctl == 't':
instructions += self.disassembler.defm_range(
address, end, sublengths)
elif sub_block.ctl == 'w':
instructions += self.disassembler.defw_range(
address, end, sublengths)
elif sub_block.ctl == 's':
instructions.append(
self.disassembler.defs(address, end,
sublengths))
else:
instructions += self.disassembler.defb_range(
address, end, sublengths)
address += length
else:
instructions = self.disassembler.ignore(
sub_block.start, sub_block.end)
sub_block.instructions = instructions
for instruction in instructions:
self.instructions[instruction.address] = instruction
instruction.asm_directives = sub_block.asm_directives.get(
instruction.address, ())
sub_blocks = []
i = 0
while i < len(block.blocks):
sub_block = block.blocks[i]
i += 1
sub_blocks.append(sub_block)
if sub_block.multiline_comment is not None:
end, sub_block.comment = sub_block.multiline_comment
while i < len(
block.blocks) and block.blocks[i].start < end:
next_sub_block = block.blocks[i]
sub_block.instructions += next_sub_block.instructions
sub_block.end = next_sub_block.end
i += 1
entry = Entry(block.header, title, block.description, block.ctl,
sub_blocks, block.registers, block.end_comment,
block.footer, block.asm_directives,
block.ignoreua_directives)
self.entry_map[entry.address] = entry
self.entries.append(entry)
for i, entry in enumerate(self.entries[1:]):
self.entries[i].next = entry
def remove_entry(self, address):
if address in self.entry_map:
del self.entry_map[address]
def _calculate_references(self):
for entry in self.entries:
for instruction in entry.instructions:
instruction.referrers = []
for entry in self.entries:
for instruction in entry.instructions:
operation = instruction.operation
if operation.upper().startswith(
('DJ', 'JR', 'JP', 'CA', 'RS')):
addr_str = get_address(operation)
if addr_str:
callee = self.instructions.get(parse_int(addr_str))
if callee:
callee.add_referrer(entry)
def _address_str(self, address):
return self.address_fmt.format(address)
def _generate_ctls_with_code_map(snapshot, start, end, code_map):
# (1) Use the code map to create an initial set of 'c' ctls, and mark all
# unexecuted blocks as 'U' (unknown)
# (2) Where a 'c' block doesn't end with a RET/JP/JR, extend it up to the
# next RET/JP/JR in the following 'U' blocks, or up to the next 'c'
# block
# (3) Mark entry points in 'U' blocks that are CALLed or JPed to from 'c'
# blocks with 'c'
# (4) Split 'c' blocks on RET/JP/JR
# (5) Scan the disassembly for pairs of adjacent blocks where the start
# address of the second block is JRed or JPed to from the first block,
# and join such pairs
# (6) Examine the remaining 'U' blocks for text
# (7) Mark data blocks of all zeroes with 's'
# (1) Mark all executed blocks as 'c' and unexecuted blocks as 'U'
# (unknown)
ctls = {start: 'U', end: 'i'}
for address, length in _get_code_blocks(snapshot, start, end, code_map):
ctls[address] = 'c'
if address + length < end:
ctls[address + length] = 'U'
# (2) Where a 'c' block doesn't end with a RET/JP/JR, extend it up to the
# next RET/JP/JR in the following 'U' blocks, or up to the next 'c' block
disassembler = Disassembler(snapshot)
while 1:
done = True
for ctl, b_start, b_end in _get_blocks(ctls):
if ctl == 'c':
if _is_terminal_instruction(
disassembler.disassemble(b_start, b_end)[-1]):
continue
if _find_terminal_instruction(disassembler, ctls, b_end,
end) < end:
done = False
break
if done:
break
# (3) Mark entry points in 'U' blocks that are CALLed or JPed to from 'c'
# blocks with 'c'
ctl_parser = CtlParser(ctls)
disassembly = Disassembly(snapshot, ctl_parser)
while 1:
disassembly.build(True)
done = True
for entry in disassembly.entries:
if entry.ctl == 'U':
for instruction in entry.instructions:
for referrer in instruction.referrers:
if ctls[referrer.address] == 'c':
ctls[instruction.address] = 'c'
if entry.next:
e_end = entry.next.address
else:
e_end = 65536
_find_terminal_instruction(disassembler, ctls,
instruction.address,
e_end, entry.ctl)
disassembly.remove_entry(entry.address)
done = False
break
if not done:
break
if not done:
break
if done:
break
# (4) Split 'c' blocks on RET/JP/JR
for ctl, b_address, b_end in _get_blocks(ctls):
if ctl == 'c':
next_address = _find_terminal_instruction(disassembler, ctls,
b_address, b_end, 'c')
if next_address < b_end:
disassembly.remove_entry(b_address)
while next_address < b_end:
next_address = _find_terminal_instruction(
disassembler, ctls, next_address, b_end, 'c')
# (5) Scan the disassembly for pairs of adjacent blocks where the start
# address of the second block is JRed or JPed to from the first block, and
# join such pairs
while 1:
disassembly.build()
done = True
for entry in disassembly.entries[:-1]:
if entry.ctl == 'c':
for instruction in entry.instructions:
operation = instruction.operation
if operation[:2] in ('JR', 'JP') and operation[-5:] == str(
entry.next.address):
del ctls[entry.next.address]
disassembly.remove_entry(entry.address)
disassembly.remove_entry(entry.next.address)
done = False
break
if done:
break
# (6) Examine the 'U' blocks for text/data
for ctl, b_start, b_end in _get_blocks(ctls):
if ctl == 'U':
ctls[b_start] = 'b'
for t_start, t_end in _get_text_blocks(snapshot, b_start, b_end):
ctls[t_start] = 't'
if t_end < b_end:
ctls[t_end] = 'b'
# (7) Mark data blocks of all zeroes with 's'
for ctl, b_start, b_end in _get_blocks(ctls):
if ctl == 'b':
z_end = b_start
while z_end < b_end and snapshot[z_end] == 0:
z_end += 1
if z_end > b_start:
ctls[b_start] = 's'
if z_end < b_end:
ctls[z_end] = 'b'
return ctls
class SftParser:
def __init__(self, snapshot, sftfile, zfill=False, asm_hex=False, asm_lower=False):
self.snapshot = snapshot
self.disassembler = Disassembler(snapshot, zfill=zfill, asm_hex=asm_hex, asm_lower=asm_lower)
self.sftfile = sftfile
self.address_fmt = get_address_format(asm_hex, asm_lower)
self.stack = []
self.disassemble = True
def _parse_instruction(self, line):
ctl = line[0]
lengths = []
if line[1] == ';':
inst_ctl = None
start = None
i = line.find(' ', 2)
if i < 0:
i = len(line.rstrip())
comment_index = get_int_param(line[2:i])
else:
inst_ctl = line[1]
if inst_ctl == 'I':
i = find_unquoted(line.rstrip(), ' ', 2)
address_end = j = find_unquoted(line, ';', 2, i)
else:
address_end = line.index(',', 2)
i = find_unquoted(line.rstrip(), ' ', address_end + 1)
j = find_unquoted(line, ';', address_end + 1, i)
start = get_int_param(line[2:address_end])
if j == i:
comment_index = -1
else:
comment_index = get_int_param(line[j + 1:i])
if j > address_end + 1:
params = split_unquoted(line[address_end + 1:j], ',')
lengths = parse_params(inst_ctl, params, 0)
elif inst_ctl != 'I':
raise ValueError
comment = line[i:].strip()
return ctl, inst_ctl, start, lengths, comment_index, comment
def _parse_asm_directive(self, directive):
if parse_asm_block_directive(directive, self.stack):
self.disassemble = True
for _p, i in self.stack:
if i != '-':
self.disassemble = False
break
def _set_bytes(self, line):
address = parse_int(line[1:6])
if address is not None:
comment_index = find_unquoted(line, ';')
operation = line[7:comment_index].strip()
set_bytes(self.snapshot, address, operation)
def _parse_sft(self, min_address, max_address):
start_index = -1
lines = []
v_block_ctl = None
f = open_file(self.sftfile)
for line in f:
if line.startswith('#'):
# This line is a skool file template comment
continue
if not line.strip():
# This line is blank
lines.append(VerbatimLine(line))
v_block_ctl = None
continue
if line.startswith(';'):
# This line is an entry-level comment
lines.append(VerbatimLine(line))
continue
if line.startswith('@'):
lines.append(VerbatimLine(line))
self._parse_asm_directive(line[1:].rstrip())
continue
if not self.disassemble:
# This line is inside a '+' block, so include it as is
lines.append(VerbatimLine(line))
continue
# Check whether we're in a block that should be restored verbatim
if v_block_ctl is None and line[0] in 'dr' or (line[0] == 'i' and line[1] in '$0123456789'):
v_block_ctl = line[0]
if v_block_ctl:
if v_block_ctl == 'd':
self._set_bytes(line)
lines.append(VerbatimLine(line))
continue
# Check whether the line starts with a valid character
if line[0] not in VALID_CTLS:
lines.append(VerbatimLine(line))
continue
try:
ctl, inst_ctl, start, lengths, comment_index, comment = self._parse_instruction(line)
except (IndexError, ValueError):
raise SftParsingError("Invalid line: {0}".format(line.split()[0]))
if start is not None:
# This line contains a control directive
if start >= min_address > 0 and start_index < 0:
start_index = len(lines)
instructions = []
for length, sublengths in lengths:
end = start + length
if inst_ctl == 'C':
base = sublengths[0][1]
instructions += self.disassembler.disassemble(start, end, base)
elif inst_ctl == 'W':
instructions += self.disassembler.defw_range(start, end, sublengths)
elif inst_ctl == 'T':
instructions += self.disassembler.defm_range(start, end, sublengths)
elif inst_ctl == 'S':
instructions.append(self.disassembler.defs(start, end, sublengths))
else:
instructions += self.disassembler.defb_range(start, end, sublengths)
start += length
if instructions:
done = False
for instruction in instructions:
if instruction.address >= max_address:
while lines and lines[-1].is_trimmable():
lines.pop()
done = True
break
address = self.address_fmt.format(instruction.address)
lines.append(InstructionLine(ctl, address, instruction.operation, comment_index, comment))
ctl = ' '
if done:
break
else:
lines.append(InstructionLine(ctl, start, '', comment_index, comment))
else:
# This line is an instruction-level comment continuation line
lines.append(InstructionLine(comment_index=comment_index, comment=comment))
f.close()
if start_index < 0:
return lines
if start_index < len(lines):
if str(lines[start_index])[0] in DIRECTIVES:
while start_index > 0 and not lines[start_index].is_blank():
start_index -= 1
else:
while start_index < len(lines) and not lines[start_index].is_blank():
start_index += 1
return lines[start_index + 1:]
return []
def write_skool(self, min_address=0, max_address=65536):
for line in self._parse_sft(min_address, max_address):
write_line(str(line))
class Disassembly:
def __init__(self, snapshot, ctl_parser, final=False, defb_size=8, defb_mod=1, zfill=False, defm_width=66, asm_hex=False, asm_lower=False):
self.disassembler = Disassembler(snapshot, defb_size, defb_mod, zfill, defm_width, asm_hex, asm_lower)
self.ctl_parser = ctl_parser
if asm_hex:
if asm_lower:
self.address_fmt = '{0:04x}'
else:
self.address_fmt = '{0:04X}'
else:
self.address_fmt = '{0}'
self.entry_map = {}
self.build(final)
def build(self, final=False):
self.instructions = {}
self.entries = []
self._create_entries()
self.org = self.entries[0].address
if final:
self._calculate_references()
def _create_entries(self):
for block in self.ctl_parser.get_blocks():
if block.start in self.entry_map:
entry = self.entry_map[block.start]
self.entries.append(entry)
for instruction in entry.instructions:
self.instructions[instruction.address] = instruction
continue
title = block.title
if block.ctl == 'c':
title = title or 'Routine at {}'.format(self._address_str(block.start))
elif block.ctl in 'bw':
title = title or 'Data block at {}'.format(self._address_str(block.start))
elif block.ctl == 't':
title = title or 'Message at {}'.format(self._address_str(block.start))
elif block.ctl == 'g':
title = title or 'Game status buffer entry at {}'.format(self._address_str(block.start))
elif block.ctl in 'us':
title = title or 'Unused'
elif block.ctl == 'i' and (block.description or block.registers or block.blocks[0].header):
title = title or 'Ignored'
for sub_block in block.blocks:
address = sub_block.start
if sub_block.ctl in 'cBT':
base = sub_block.sublengths[0][1]
instructions = self.disassembler.disassemble(sub_block.start, sub_block.end, base)
elif sub_block.ctl in 'bgstuw':
sublengths = sub_block.sublengths
if sublengths[0][0]:
if sub_block.ctl == 's':
length = sublengths[0][0]
else:
length = sum([s[0] for s in sublengths])
else:
length = sub_block.end - sub_block.start
instructions = []
while address < sub_block.end:
end = min(address + length, sub_block.end)
if sub_block.ctl == 't':
instructions += self.disassembler.defm_range(address, end, sublengths)
elif sub_block.ctl == 'w':
instructions += self.disassembler.defw_range(address, end, sublengths)
elif sub_block.ctl == 's':
instructions.append(self.disassembler.defs(address, end, sublengths))
else:
instructions += self.disassembler.defb_range(address, end, sublengths)
address += length
else:
instructions = self.disassembler.ignore(sub_block.start, sub_block.end)
sub_block.instructions = instructions
for instruction in instructions:
self.instructions[instruction.address] = instruction
instruction.asm_directives = sub_block.asm_directives.get(instruction.address, ())
sub_blocks = []
i = 0
while i < len(block.blocks):
sub_block = block.blocks[i]
i += 1
sub_blocks.append(sub_block)
if sub_block.multiline_comment is not None:
end, sub_block.comment = sub_block.multiline_comment
while i < len(block.blocks) and block.blocks[i].start < end:
next_sub_block = block.blocks[i]
sub_block.instructions += next_sub_block.instructions
sub_block.end = next_sub_block.end
i += 1
entry = Entry(title, block.description, block.ctl, sub_blocks,
block.registers, block.end_comment, block.asm_directives,
block.ignoreua_directives)
self.entry_map[entry.address] = entry
self.entries.append(entry)
for i, entry in enumerate(self.entries[1:]):
self.entries[i].next = entry
def remove_entry(self, address):
if address in self.entry_map:
del self.entry_map[address]
def contains_entry_asm_directive(self, asm_dir):
for entry in self.entries:
for directive, value in entry.asm_directives:
if directive == asm_dir:
return True
def _calculate_references(self):
for entry in self.entries:
for instruction in entry.instructions:
instruction.referrers = []
for entry in self.entries:
for instruction in entry.instructions:
operation = instruction.operation
if operation.upper().startswith(('DJ', 'JR', 'JP', 'CA', 'RS')):
addr_str = get_address(operation)
if addr_str:
callee = self.instructions.get(parse_int(addr_str))
if callee:
callee.add_referrer(entry)
def _address_str(self, address):
return self.address_fmt.format(address)
def _generate_ctls_with_code_map(snapshot, start, end, code_map):
# (1) Use the code map to create an initial set of 'c' ctls, and mark all
# unexecuted blocks as 'U' (unknown)
# (2) Where a 'c' block doesn't end with a RET/JP/JR, extend it up to the
# next RET/JP/JR in the following 'U' blocks, or up to the next 'c'
# block
# (3) Mark entry points in 'U' blocks that are CALLed or JPed to from 'c'
# blocks with 'c'
# (4) Split 'c' blocks on RET/JP/JR
# (5) Scan the disassembly for pairs of adjacent blocks where the start
# address of the second block is JRed or JPed to from the first block,
# and join such pairs
# (6) Examine the remaining 'U' blocks for text
# (7) Mark data blocks of all zeroes with 's'
# (1) Mark all executed blocks as 'c' and unexecuted blocks as 'U'
# (unknown)
ctls = {start: 'U', end: 'i'}
for address, length in _get_code_blocks(snapshot, start, end, code_map):
ctls[address] = 'c'
if address + length < end:
ctls[address + length] = 'U'
# (2) Where a 'c' block doesn't end with a RET/JP/JR, extend it up to the
# next RET/JP/JR in the following 'U' blocks, or up to the next 'c' block
disassembler = Disassembler(snapshot)
while 1:
done = True
for ctl, b_start, b_end in _get_blocks(ctls):
if ctl == 'c':
if _is_terminal_instruction(disassembler.disassemble(b_start, b_end)[-1]):
continue
if _find_terminal_instruction(disassembler, ctls, b_end, end) < end:
done = False
break
if done:
break
# (3) Mark entry points in 'U' blocks that are CALLed or JPed to from 'c'
# blocks with 'c'
ctl_parser = CtlParser(ctls)
disassembly = Disassembly(snapshot, ctl_parser)
while 1:
disassembly.build(True)
done = True
for entry in disassembly.entries:
if entry.ctl == 'U':
for instruction in entry.instructions:
for referrer in instruction.referrers:
if ctls[referrer.address] == 'c':
ctls[instruction.address] = 'c'
if entry.next:
e_end = entry.next.address
else:
e_end = 65536
_find_terminal_instruction(disassembler, ctls, instruction.address, e_end, entry.ctl)
disassembly.remove_entry(entry.address)
done = False
break
if not done:
break
if not done:
break
if done:
break
# (4) Split 'c' blocks on RET/JP/JR
for ctl, b_address, b_end in _get_blocks(ctls):
if ctl == 'c':
next_address = _find_terminal_instruction(disassembler, ctls, b_address, b_end, 'c')
if next_address < b_end:
disassembly.remove_entry(b_address)
while next_address < b_end:
next_address = _find_terminal_instruction(disassembler, ctls, next_address, b_end, 'c')
# (5) Scan the disassembly for pairs of adjacent blocks where the start
# address of the second block is JRed or JPed to from the first block, and
# join such pairs
while 1:
disassembly.build()
done = True
for entry in disassembly.entries[:-1]:
if entry.ctl == 'c':
for instruction in entry.instructions:
operation = instruction.operation
if operation[:2] in ('JR', 'JP') and operation[-5:] == str(entry.next.address):
del ctls[entry.next.address]
disassembly.remove_entry(entry.address)
disassembly.remove_entry(entry.next.address)
done = False
break
if done:
break
# (6) Examine the 'U' blocks for text/data
for ctl, b_start, b_end in _get_blocks(ctls):
if ctl == 'U':
ctls[b_start] = 'b'
for t_start, t_end in _get_text_blocks(snapshot, b_start, b_end):
ctls[t_start] = 't'
if t_end < b_end:
ctls[t_end] = 'b'
# (7) Mark data blocks of all zeroes with 's'
for ctl, b_start, b_end in _get_blocks(ctls):
if ctl == 'b':
z_end = b_start
while z_end < b_end and snapshot[z_end] == 0:
z_end += 1
if z_end > b_start:
ctls[b_start] = 's'
if z_end < b_end:
ctls[z_end] = 'b'
return ctls
class Disassembly:
def __init__(self, snapshot, ctl_parser, config=None, final=False, defb_size=8, defb_mod=1,
zfill=False, defm_width=66, asm_hex=False, asm_lower=False):
ctl_parser.apply_asm_data_directives(snapshot)
self.disassembler = Disassembler(snapshot, defb_size, defb_mod, zfill, defm_width, asm_hex, asm_lower)
self.ctl_parser = ctl_parser
if asm_hex:
if asm_lower:
self.address_fmt = '{0:04x}'
else:
self.address_fmt = '{0:04X}'
else:
self.address_fmt = '{0}'
self.entry_map = {}
self.config = config or {}
self.build(final)
def build(self, final=False):
self.instructions = {}
self.entries = []
self._create_entries()
if self.entries:
self.org = self.entries[0].address
else:
self.org = None
if final:
self._calculate_references()
def _create_entries(self):
for block in self.ctl_parser.get_blocks():
if block.start in self.entry_map:
entry = self.entry_map[block.start]
self.entries.append(entry)
for instruction in entry.instructions:
self.instructions[instruction.address] = instruction
continue
title = block.title
if not any(title):
ctl = block.ctl
if ctl != 'i' or block.description or block.registers or block.blocks[0].header:
name = 'Title-' + ctl
title = [format_template(self.config.get(name, ''), name, address=self._address_str(block.start))]
for sub_block in block.blocks:
address = sub_block.start
if sub_block.ctl in 'cBT':
base = sub_block.sublengths[0][1]
instructions = self.disassembler.disassemble(sub_block.start, sub_block.end, base)
elif sub_block.ctl in 'bgstuw':
sublengths = sub_block.sublengths
if sublengths[0][0]:
if sub_block.ctl == 's':
length = sublengths[0][0]
else:
length = sum([s[0] for s in sublengths])
else:
length = sub_block.end - sub_block.start
instructions = []
while address < sub_block.end:
end = min(address + length, sub_block.end)
if sub_block.ctl == 't':
instructions += self.disassembler.defm_range(address, end, sublengths)
elif sub_block.ctl == 'w':
instructions += self.disassembler.defw_range(address, end, sublengths)
elif sub_block.ctl == 's':
instructions += self.disassembler.defs(address, end, sublengths)
else:
instructions += self.disassembler.defb_range(address, end, sublengths)
address += length
else:
instructions = self.disassembler.ignore(sub_block.start, sub_block.end)
self._add_instructions(sub_block, instructions)
sub_blocks = []
i = 0
while i < len(block.blocks):
sub_block = block.blocks[i]
i += 1
sub_blocks.append(sub_block)
if sub_block.multiline_comment is not None:
end, sub_block.comment = sub_block.multiline_comment
while i < len(block.blocks) and block.blocks[i].start < end:
next_sub_block = block.blocks[i]
sub_block.instructions += next_sub_block.instructions
sub_block.end = next_sub_block.end
i += 1
entry = Entry(block.header, title, block.description, block.ctl, sub_blocks,
block.registers, block.end_comment, block.footer, block.asm_directives,
block.ignoreua_directives)
self.entry_map[entry.address] = entry
self.entries.append(entry)
for i, entry in enumerate(self.entries[1:]):
self.entries[i].next = entry
def remove_entry(self, address):
if address in self.entry_map:
del self.entry_map[address]
def _add_instructions(self, sub_block, instructions):
sub_block.instructions = instructions
for instruction in instructions:
self.instructions[instruction.address] = instruction
instruction.asm_directives = sub_block.asm_directives.get(instruction.address, ())
instruction.label = None
for asm_dir in instruction.asm_directives:
if asm_dir.startswith(AD_LABEL + '='):
instruction.label = asm_dir[6:]
if instruction.label.startswith('*'):
instruction.ctl = '*'
break
def _calculate_references(self):
for entry in self.entries:
for instruction in entry.instructions:
instruction.referrers = []
for entry in self.entries:
for instruction in entry.instructions:
operation = instruction.operation
if operation.upper().startswith(('DJ', 'JR', 'JP', 'CA', 'RS')):
addr_str = get_address(operation)
if addr_str:
callee = self.instructions.get(parse_int(addr_str))
if callee and (entry.ctl != 'u' or callee.entry == entry) and callee.label != '':
callee.add_referrer(entry)
def _address_str(self, address):
return self.address_fmt.format(address)