Archipelago/worlds/smz3/ips.py

255 lines
11 KiB
Python

import itertools
from .TotalSMZ3.Text.Texts import openFile
def range_union(ranges):
ret = []
for rg in sorted([[r.start, r.stop] for r in ranges]):
begin, end = rg[0], rg[-1]
if ret and ret[-1][1] > begin:
ret[-1][1] = max(ret[-1][1], end)
else:
ret.append([begin, end])
return [range(r[0], r[1]) for r in ret]
# adapted from ips-util for python 3.2 (https://pypi.org/project/ips-util/)
class IPS_Patch(object):
def __init__(self, patchDict=None):
self.records = []
self.truncate_length = None
self.max_size = 0
if patchDict is not None:
for addr, data in patchDict.items():
byteData = bytearray(data)
self.add_record(addr, byteData)
def toDict(self):
ret = {}
for record in self.records:
if 'rle_count' in record:
ret[record['address']] = [int.from_bytes(record['data'],'little')]*record['rle_count']
else:
ret[record['address']] = [int(b) for b in record['data']]
return ret
@staticmethod
def load(filename):
loaded_patch = IPS_Patch()
with openFile(filename, 'rb') as file:
header = file.read(5)
if header != b'PATCH':
raise Exception('Not a valid IPS patch file!')
while True:
address_bytes = file.read(3)
if address_bytes == b'EOF':
break
address = int.from_bytes(address_bytes, byteorder='big')
length = int.from_bytes(file.read(2), byteorder='big')
rle_count = 0
if length == 0:
rle_count = int.from_bytes(file.read(2), byteorder='big')
length = 1
data = file.read(length)
if rle_count > 0:
loaded_patch.add_rle_record(address, data, rle_count)
else:
loaded_patch.add_record(address, data)
truncate_bytes = file.read(3)
if len(truncate_bytes) == 3:
loaded_patch.set_truncate_length(int.from_bytes(truncate_bytes, byteorder='big'))
return loaded_patch
@staticmethod
def create(original_data, patched_data):
# The heuristics for optimizing a patch were chosen with reference to
# the source code of Flips: https://github.com/Alcaro/Flips
patch = IPS_Patch()
run_in_progress = False
current_run_start = 0
current_run_data = bytearray()
runs = []
if len(original_data) > len(patched_data):
patch.set_truncate_length(len(patched_data))
original_data = original_data[:len(patched_data)]
elif len(original_data) < len(patched_data):
original_data += bytes([0] * (len(patched_data) - len(original_data)))
if original_data[-1] == 0 and patched_data[-1] == 0:
patch.add_record(len(patched_data) - 1, bytes([0]))
for index, (original, patched) in enumerate(zip(original_data, patched_data)):
if not run_in_progress:
if original != patched:
run_in_progress = True
current_run_start = index
current_run_data = bytearray([patched])
else:
if original == patched:
runs.append((current_run_start, current_run_data))
run_in_progress = False
else:
current_run_data.append(patched)
if run_in_progress:
runs.append((current_run_start, current_run_data))
for start, data in runs:
if start == int.from_bytes(b'EOF', byteorder='big'):
start -= 1
data = bytes([patched_data[start - 1]]) + data
grouped_byte_data = list([
{'val': key, 'count': sum(1 for _ in group), 'is_last': False}
for key,group in itertools.groupby(data)
])
grouped_byte_data[-1]['is_last'] = True
record_in_progress = bytearray()
pos = start
for group in grouped_byte_data:
if len(record_in_progress) > 0:
# We don't want to interrupt a record in progress with a new header unless
# this group is longer than two complete headers.
if group['count'] > 13:
patch.add_record(pos, record_in_progress)
pos += len(record_in_progress)
record_in_progress = bytearray()
patch.add_rle_record(pos, bytes([group['val']]), group['count'])
pos += group['count']
else:
record_in_progress += bytes([group['val']] * group['count'])
elif (group['count'] > 3 and group['is_last']) or group['count'] > 8:
# We benefit from making this an RLE record if the length is at least 8,
# or the length is at least 3 and we know it to be the last part of this diff.
# Make sure not to overflow the maximum length. Split it up if necessary.
remaining_length = group['count']
while remaining_length > 0xffff:
patch.add_rle_record(pos, bytes([group['val']]), 0xffff)
remaining_length -= 0xffff
pos += 0xffff
patch.add_rle_record(pos, bytes([group['val']]), remaining_length)
pos += remaining_length
else:
# Just begin a new standard record.
record_in_progress += bytes([group['val']] * group['count'])
if len(record_in_progress) > 0xffff:
patch.add_record(pos, record_in_progress[:0xffff])
record_in_progress = record_in_progress[0xffff:]
pos += 0xffff
# Finalize any record still in progress.
if len(record_in_progress) > 0:
patch.add_record(pos, record_in_progress)
return patch
def add_record(self, address, data):
if address == int.from_bytes(b'EOF', byteorder='big'):
raise RuntimeError('Start address {0:x} is invalid in the IPS format. Please shift your starting address back by one byte to avoid it.'.format(address))
if address > 0xffffff:
raise RuntimeError('Start address {0:x} is too large for the IPS format. Addresses must fit into 3 bytes.'.format(address))
if len(data) > 0xffff:
raise RuntimeError('Record with length {0} is too large for the IPS format. Records must be less than 65536 bytes.'.format(len(data)))
if len(data) == 0: # ignore empty records
return
record = {'address': address, 'data': data, 'size':len(data)}
self.appendRecord(record)
def add_rle_record(self, address, data, count):
if address == int.from_bytes(b'EOF', byteorder='big'):
raise RuntimeError('Start address {0:x} is invalid in the IPS format. Please shift your starting address back by one byte to avoid it.'.format(address))
if address > 0xffffff:
raise RuntimeError('Start address {0:x} is too large for the IPS format. Addresses must fit into 3 bytes.'.format(address))
if count > 0xffff:
raise RuntimeError('RLE record with length {0} is too large for the IPS format. RLE records must be less than 65536 bytes.'.format(count))
if len(data) != 1:
raise RuntimeError('Data for RLE record must be exactly one byte! Received {0}.'.format(data))
record = {'address': address, 'data': data, 'rle_count': count, 'size': count}
self.appendRecord(record)
def appendRecord(self, record):
sz = record['address'] + record['size']
if sz > self.max_size:
self.max_size = sz
self.records.append(record)
def set_truncate_length(self, truncate_length):
self.truncate_length = truncate_length
def encode(self):
encoded_bytes = bytearray()
encoded_bytes += 'PATCH'.encode('ascii')
for record in self.records:
encoded_bytes += record['address'].to_bytes(3, byteorder='big')
if 'rle_count' in record:
encoded_bytes += (0).to_bytes(2, byteorder='big')
encoded_bytes += record['rle_count'].to_bytes(2, byteorder='big')
else:
encoded_bytes += len(record['data']).to_bytes(2, byteorder='big')
encoded_bytes += record['data']
encoded_bytes += 'EOF'.encode('ascii')
if self.truncate_length is not None:
encoded_bytes += self.truncate_length.to_bytes(3, byteorder='big')
return encoded_bytes
# save patch into IPS file
def save(self, path):
with open(path, 'wb') as ipsFile:
ipsFile.write(self.encode())
# applies patch on an existing bytearray
def apply(self, in_data):
out_data = bytearray(in_data)
for record in self.records:
if record['address'] >= len(out_data):
out_data += bytes([0] * (record['address'] - len(out_data) + 1))
if 'rle_count' in record:
out_data[record['address'] : record['address'] + record['rle_count']] = b''.join([record['data']] * record['rle_count'])
else:
out_data[record['address'] : record['address'] + len(record['data'])] = record['data']
if self.truncate_length is not None:
out_data = out_data[:self.truncate_length]
return out_data
# applies patch on an opened file
def applyFile(self, handle):
for record in self.records:
handle.seek(record['address'])
if 'rle_count' in record:
handle.write(bytearray(b'').join([record['data']]) * record['rle_count'])
else:
handle.write(record['data'])
# appends an IPS_Patch on top of this one
def append(self, patch):
if patch.truncate_length is not None and (self.truncate_length is None or patch.truncate_length > self.truncate_length):
self.set_truncate_length(patch.truncate_length)
for record in patch.records:
if record['size'] > 0: # ignore empty records
self.appendRecord(record)
# gets address ranges written to by this patch
def getRanges(self):
def getRange(record):
return range(record['address'], record['address']+record['size'])
return range_union([getRange(record) for record in self.records])