jan/mem_edit
1
1
Fork 2
Code Issues 1 Pull Requests Releases Wiki Activity

Compare commits

base: jan:master
Branches Tags
jan:master
jan:test
shokerplz:master
shokerplz:test
jan:v0.8
jan:release
jan:v0.7
jan:v0.6
jan:v0.5
jan:v0.4
jan:v0.3
jan:0.2
jan:0.1
..
compare: shokerplz:master
Branches Tags
shokerplz:master
shokerplz:test
jan:master
jan:test
jan:v0.8
jan:release
jan:v0.7
jan:v0.6
jan:v0.5
jan:v0.4
jan:v0.3
jan:0.2
jan:0.1

No commits in common. "master" and "master" have entirely different histories.
master ... master

14 changed files with 327 additions and 497 deletions
Show Stats Expand all files Collapse all files

29
.flake8
View File

@ -1,29 +0,0 @@
[flake8]
ignore =
# E501 line too long
E501,
# W391 newlines at EOF
W391,
# E241 multiple spaces after comma
E241,
# E302 expected 2 newlines
E302,
# W503 line break before binary operator (to be deprecated)
W503,
# E265 block comment should start with '# '
E265,
# E123 closing bracket does not match indentation of opening bracket's line
E123,
# E124 closing bracket does not match visual indentation
E124,
# E221 multiple spaces before operator
E221,
# E201 whitespace after '['
E201,
# E741 ambiguous variable name 'I'
E741,
per-file-ignores =
# F401 import without use
*/__init__.py: F401,

7
.gitignore vendored
View File

@ -1,13 +1,8 @@
.idea/
__pycache__/
__pycache__
*.pyc
*.egg-info/
build/
dist/
.pytest_cache/
.mypy_cache/
*.pickle

3
MANIFEST.in Normal file
View File

@ -0,0 +1,3 @@
include README.md
include LICENSE.md
include mem_edit/VERSION

4
README.md
View File

@ -3,8 +3,6 @@
**mem_edit** is a multi-platform memory editing library written in Python.
**Homepage:** https://mpxd.net/code/jan/mem_edit
* PyPI: https://pypi.org/project/mem-edit/
* Github mirror: https://github.com/anewusername/mem_edit
**Capabilities:**
* Scan all readable memory used by a process.
@ -20,7 +18,7 @@
## Installation
**Dependencies:**
* python >=3.11
* python 3 (written and tested with 3.7)
* ctypes
* typing (for type annotations)

1
mem_edit/LICENSE.md
View File

@ -1 +0,0 @@
../LICENSE.md

1
mem_edit/README.md
View File

@ -1 +0,0 @@
../README.md

1
mem_edit/VERSION Normal file
View File

@ -0,0 +1 @@
0.3

11
mem_edit/__init__.py
View File

@ -12,19 +12,22 @@ To get started, try:
"""
import platform
import pathlib
from .utils import MemEditError
__author__ = 'Jan Petykiewicz'
__version__ = '0.8'
version = __version__ # legacy compatibility
with open(pathlib.Path(__file__).parent / 'VERSION', 'r') as f:
__version__ = f.read().strip()
version = __version__
system = platform.system()
if system == 'Windows':
from .windows import Process as Process
from .windows import Process
elif system == 'Linux':
from .linux import Process as Process
from .linux import Process
else:
raise MemEditError('Only Linux and Windows are currently supported.')

218
mem_edit/abstract.py
View File

@ -1,8 +1,8 @@
"""
Abstract class for cross-platform memory editing.
"""
from typing import Self
from collections.abc import Generator
from typing import List, Tuple
from abc import ABCMeta, abstractmethod
from contextlib import contextmanager
import copy
@ -13,6 +13,7 @@ from . import utils
from .utils import ctypes_buffer_t
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
@ -22,8 +23,8 @@ class Process(metaclass=ABCMeta):
(i.e., by reading from or writing to the memory used by a given process).
The static methods
`Process.list_available_pids()`
`Process.get_pid_by_name(executable_filename)`
Process.list_available_pids()
Process.get_pid_by_name(executable_filename)
can be used to help find the process id (pid) of the target process. They are
provided for convenience only; it is probably better to use the tools built
in to your operating system to discover the pid of the specific process you
@ -31,19 +32,18 @@ class Process(metaclass=ABCMeta):
Once you have found the pid, you are ready to construct an instance of Process
and use it to read and write to memory. Once you are done with the process,
use `.close()` to free up the process for access by other debuggers etc.
```
use .close() to free up the process for access by other debuggers etc.
p = Process(1239)
p.close()
```
To read/write to memory, first create a buffer using ctypes:
```
buffer0 = (ctypes.c_byte * 5)(39, 50, 03, 40, 30)
buffer1 = ctypes.c_ulong()
```
and then use
```
p.write_memory(0x2fe, buffer0)
val0 = p.read_memory(0x220, buffer0)[:]
@ -51,52 +51,52 @@ class Process(metaclass=ABCMeta):
val1a = p.read_memory(0x149, buffer1).value
val2b = buffer1.value
assert(val1a == val2b)
```
Searching for a value can be done in a number of ways:
Search a list of addresses:
`found_addresses = p.search_addresses([0x1020, 0x1030], buffer0)`
found_addresses = p.search_addresses([0x1020, 0x1030], buffer0)
Search the entire memory space:
`found_addresses = p.search_all_memory(buffer0, writeable_only=False)`
found_addresses = p.search_all_memory(buffer0, writeable_only=False)
You can also get a list of which regions in memory are mapped (readable):
`regions = p.list_mapped_regions(writeable_only=False)`
regions = p.list_mapped_regions(writeable_only=False)
which can be used along with search_buffer(...) to re-create .search_all_memory(...):
```
found = []
for region_start, region_stop in regions:
region_buffer = (ctypes.c_byte * (region_stop - region_start))()
p.read_memory(region_start, region_buffer)
found += utils.search_buffer(ctypes.c_ulong(123456790), region_buffer)
```
Other useful methods include the context manager, implemented as a static method:
```
with Process.open_process(pid) as p:
# use p here, no need to call p.close()
```
.get_path(), which reports the path of the executable file which was used
to start the process:
```
executable_path = p.get_path()
```
and deref_struct_pointer, which takes a pointer to a struct and reads out the struct members:
```
# struct is a list of (offset, buffer) pairs
struct_defintion = [(0x0, ctypes.c_ulong()),
(0x20, ctypes.c_byte())]
values = p.deref_struct_pointer(0x0feab4, struct_defintion)
```
which is shorthand for
```
struct_addr = p.read_memory(0x0feab4, ctypes.c_void_p())
values = [p.read_memory(struct_addr + 0x0, ctypes.c_ulong()),
p.read_memory(struct_addr + 0x20, ctypes.c_byte())]
```
=================
Putting all this together, a simple program which alters a magic number in the only running
instance of 'magic.exe' might look like this:
```
import ctypes
from mem_edit import Process
@ -107,9 +107,9 @@ class Process(metaclass=ABCMeta):
addrs = p.search_all_memory(magic_number)
assert(len(addrs) == 1)
p.write_memory(addrs[0], ctypes.c_ulong(42))
```
Searching for a value which changes:
```
pid = Process.get_pid_by_name('monitor_me.exe')
with Process.open_process(pid) as p:
addrs = p.search_all_memory(ctypes.c_int(40))
@ -118,111 +118,93 @@ class Process(metaclass=ABCMeta):
print('Found addresses:')
for addr in filtered_addrs:
print(hex(addr))
```
"""
@abstractmethod
def __init__(self, process_id: int) -> None:
def __init__(self, process_id: int):
"""
Constructing a Process object prepares the process with specified process_id for
memory editing. Finding the `process_id` for the process you want to edit is often
memory editing. Finding the process_id for the process you want to edit is often
easiest using os-specific tools (or by launching the process yourself, e.g. with
`subprocess.Popen(...)`).
subprocess.Popen(...)).
Args:
process_id: Process id (pid) of the target process
:param process_id: Process id (pid) of the target process
"""
pass
@abstractmethod
def close(self) -> None:
def close(self):
"""
Detach from the process, removing our ability to edit it and
letting other debuggers attach to it instead.
This function should be called after you are done working with the process
and will no longer need it. See the `Process.open_process(...)` context
and will no longer need it. See the Process.open_process(...) context
manager to avoid having to call this function yourself.
"""
pass
@abstractmethod
def write_memory(
self,
base_address: int,
write_buffer: ctypes_buffer_t,
) -> None:
def write_memory(self, base_address: int, write_buffer: ctypes_buffer_t):
"""
Write the given buffer to the process's address space, starting at `base_address`.
Write the given buffer to the process's address space, starting at base_address.
Args:
base_address: The address to write at, in the process's address space.
write_buffer: A ctypes object, for example, `ctypes.c_ulong(48)`,
`(ctypes.c_byte * 3)(43, 21, 0xff)`, or a subclass of `ctypes.Structure`,
which will be written into memory starting at `base_address`.
:param base_address: The address to write at, in the process's address space.
:param write_buffer: A ctypes object, for example, ctypes.c_ulong(48),
(ctypes.c_byte * 3)(43, 21, 0xff), or a subclass of ctypes.Structure,
which will be written into memory starting at base_address.
"""
pass
@abstractmethod
def read_memory(
self,
base_address: int,
read_buffer: ctypes_buffer_t,
) -> ctypes_buffer_t:
def read_memory(self, base_address: int, read_buffer: ctypes_buffer_t) -> ctypes_buffer_t:
"""
Read into the given buffer from the process's address space, starting at `base_address`.
Read into the given buffer from the process's address space, starting at base_address.
Args:
base_address: The address to read from, in the process's address space.
read_buffer: A `ctypes` object, for example. `ctypes.c_ulong()`,
`(ctypes.c_byte * 3)()`, or a subclass of `ctypes.Structure`, which will be
overwritten with the contents of the process's memory starting at `base_address`.
Returns:
`read_buffer` is returned as well as being overwritten.
:param base_address: The address to read from, in the process's address space.
:param read_buffer: A ctypes object, for example. ctypes.c_ulong(),
(ctypes.c_byte * 3)(), or a subclass of ctypes.Structure, which will be
overwritten with the contents of the process's memory starting at base_address.
:returns: read_buffer is returned as well as being overwritten.
"""
pass
@abstractmethod
def list_mapped_regions(self, writeable_only: bool = True) -> list[tuple[int, int]]:
def list_mapped_regions(self, writeable_only=True) -> List[Tuple[int, int]]:
"""
Return a list of `(start_address, stop_address)` for the regions of the address space
Return a list of (start_address, stop_address) for the regions of the address space
accessible to (readable and possibly writable by) the process.
By default, this function does not return non-writeable regions.
Args:
writeable_only: If `True`, only return regions which are also writeable.
Default `True`.
Returns:
List of `(start_address, stop_address)` for each accessible memory region.
:param writeable_only: If True, only return regions which are also writeable.
Default true.
:return: List of (start_address, stop_address) for each accessible memory region.
"""
pass
@abstractmethod
def get_path(self) -> str | None:
def get_path(self) -> str:
"""
Return the path to the executable file which was run to start this process.
Returns:
A string containing the path, or None if no path was found.
:return: A string containing the path.
"""
pass
@staticmethod
@abstractmethod
def list_available_pids() -> list[int]:
def list_available_pids() -> List[int]:
"""
Return a list of all process ids (pids) accessible on this system.
Returns:
List of running process ids.
:return: List of running process ids.
"""
pass
@staticmethod
@abstractmethod
def get_pid_by_name(target_name: str) -> int | None:
def get_pid_by_name(target_name: str) -> int or None:
"""
Attempt to return the process id (pid) of a process which was run with an executable
file with the provided name. If no process is found, return None.
@ -233,65 +215,44 @@ class Process(metaclass=ABCMeta):
Don't rely on this method if you can possibly avoid it, since it makes no
attempt to confirm that it found a unique process and breaks trivially (e.g. if the
executable file is renamed).
Args:
target_name: Name of the process to find the PID for
Returns:
Process id (pid) of a process with the provided name, or `None`.
:return: Process id (pid) of a process with the provided name, or None.
"""
pass
def deref_struct_pointer(
self,
def deref_struct_pointer(self,
base_address: int,
targets: list[tuple[int, ctypes_buffer_t]],
) -> list[ctypes_buffer_t]:
targets: List[Tuple[int, ctypes_buffer_t]],
) -> List[ctypes_buffer_t]:
"""
Take a pointer to a struct and read out the struct members:
```
struct_defintion = [(0x0, ctypes.c_ulong()),
(0x20, ctypes.c_byte())]
values = p.deref_struct_pointer(0x0feab4, struct_defintion)
```
which is shorthand for
```
struct_addr = p.read_memory(0x0feab4, ctypes.c_void_p())
values = [p.read_memory(struct_addr + 0x0, ctypes.c_ulong()),
p.read_memory(struct_addr + 0x20, ctypes.c_byte())]
```
Args:
base_address: Address at which the struct pointer is located.
targets: List of `(offset, read_buffer)` pairs which will be read from the struct.
Return:
List of read values corresponding to the provided targets.
:param base_address: Address at which the struct pointer is located.
:param targets: List of (offset, read_buffer) pairs which will be read from the struct.
:return: List of read values corresponding to the provided targets.
"""
base = self.read_memory(base_address, ctypes.c_void_p()).value
values = [self.read_memory(base + offset, buffer) for offset, buffer in targets]
return values
def search_addresses(
self,
addresses: list[int],
needle_buffer: ctypes_buffer_t,
verbatim: bool = True,
) -> list[int]:
def search_addresses(self, addresses: List[int], needle_buffer: ctypes_buffer_t, verbatim: bool=True) -> List[int]:
"""
Search for the provided value at each of the provided addresses, and return the addresses
where it is found.
Args:
addresses: List of addresses which should be probed.
needle_buffer: The value to search for. This should be a `ctypes` object of the same
sorts as used by `.read_memory(...)`, which will be compared to the contents of
:param addresses: List of addresses which should be probed.
:param needle_buffer: The value to search for. This should be a ctypes object of the same
sorts as used by .read_memory(...), which will be compared to the contents of
memory at each of the given addresses.
verbatim: If `True`, perform bitwise comparison when searching for `needle_buffer`.
If `False`, perform `utils.ctypes_equal`-based comparison. Default `True`.
Returns:
List of addresses where the `needle_buffer` was found.
:param verbatim: If True, perform bitwise comparison when searching for needle_buffer.
If False, perform utils.ctypes_equal-based comparison. Default True.
:return: List of addresses where the needle_buffer was found.
"""
found = []
read_buffer = copy.copy(needle_buffer)
@ -308,26 +269,18 @@ class Process(metaclass=ABCMeta):
found.append(address)
return found
def search_all_memory(
self,
needle_buffer: ctypes_buffer_t,
writeable_only: bool = True,
verbatim: bool = True,
) -> list[int]:
def search_all_memory(self, needle_buffer: ctypes_buffer_t, writeable_only: bool=True, verbatim: bool=True) -> List[int]:
"""
Search the entire memory space accessible to the process for the provided value.
Args:
needle_buffer: The value to search for. This should be a ctypes object of the same
sorts as used by `.read_memory(...)`, which will be compared to the contents of
:param needle_buffer: The value to search for. This should be a ctypes object of the same
sorts as used by .read_memory(...), which will be compared to the contents of
memory at each accessible address.
writeable_only: If `True`, only search regions where the process has write access.
Default `True`.
verbatim: If `True`, perform bitwise comparison when searching for `needle_buffer`.
If `False`, perform `utils.ctypes_equal-based` comparison. Default `True`.
Returns:
List of addresses where the `needle_buffer` was found.
:param writeable_only: If True, only search regions where the process has write access.
Default True.
:param verbatim: If True, perform bitwise comparison when searching for needle_buffer.
If False, perform utils.ctypes_equal-based comparison. Default True.
:return: List of addresses where the needle_buffer was found.
"""
found = []
if verbatim:
@ -341,25 +294,20 @@ class Process(metaclass=ABCMeta):
self.read_memory(start, region_buffer)
found += [offset + start for offset in search(needle_buffer, region_buffer)]
except OSError:
logger.exception(f'Failed to read in range 0x{start} - 0x{stop}')
logger.error('Failed to read in range 0x{} - 0x{}'.format(start, stop))
return found
@classmethod
@contextmanager
def open_process(cls: type[Self], process_id: int) -> Generator[Self, None, None]:
def open_process(cls, process_id: int) -> 'Process':
"""
Context manager which automatically closes the constructed Process:
```
with Process.open_process(2394) as p:
# use p here
# no need to run p.close()
```
Args:
process_id: Process id (pid), passed to the Process constructor.
Returns:
Constructed Process object.
:param process_id: Process id (pid), passed to the Process constructor.
:return: Constructed Process object.
"""
process = cls(process_id)
yield process

60
mem_edit/linux.py
View File

@ -2,6 +2,7 @@
Implementation of Process class for Linux
"""
from typing import List, Tuple
from os import strerror
import os
import os.path
@ -9,12 +10,12 @@ import signal
import ctypes
import ctypes.util
import logging
from pathlib import Path
from .abstract import Process as AbstractProcess
from .utils import ctypes_buffer_t, MemEditError
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
@ -35,61 +36,52 @@ _ptrace.argtypes = (ctypes.c_ulong,) * 4
_ptrace.restype = ctypes.c_long
def ptrace(
command: int,
pid: int = 0,
arg1: int = 0,
arg2: int = 0,
) -> int:
def ptrace(command: int, pid: int = 0, arg1: int = 0, arg2: int = 0) -> int:
"""
Call ptrace() with the provided pid and arguments. See `man ptrace`.
Call ptrace() with the provided pid and arguments. See the ```man ptrace```.
"""
logger.debug(f'ptrace({command}, {pid}, {arg1}, {arg2})')
logger.debug('ptrace({}, {}, {}, {})'.format(command, pid, arg1, arg2))
result = _ptrace(command, pid, arg1, arg2)
if result == -1:
err_no = ctypes.get_errno()
if err_no:
raise MemEditError(f'ptrace({command}, {pid}, {arg1}, {arg2})'
+ f' failed with error {err_no}: {strerror(err_no)}')
raise MemEditError('ptrace({}, {}, {}, {})'.format(command, pid, arg1, arg2) +
' failed with error {}: {}'.format(err_no, strerror(err_no)))
return result
class Process(AbstractProcess):
pid: int | None
pid = None
def __init__(self, process_id: int) -> None:
def __init__(self, process_id: int):
ptrace(ptrace_commands['PTRACE_SEIZE'], process_id)
self.pid = process_id
def close(self) -> None:
if self.pid is None:
return
def close(self):
os.kill(self.pid, signal.SIGSTOP)
os.waitpid(self.pid, 0)
ptrace(ptrace_commands['PTRACE_DETACH'], self.pid, 0, 0)
os.kill(self.pid, signal.SIGCONT)
self.pid = None
def write_memory(self, base_address: int, write_buffer: ctypes_buffer_t) -> None:
with Path(f'/proc/{self.pid}/mem').open('rb+') as mem:
def write_memory(self, base_address: int, write_buffer: ctypes_buffer_t):
with open('/proc/{}/mem'.format(self.pid), 'rb+') as mem:
mem.seek(base_address)
mem.write(write_buffer)
def read_memory(self, base_address: int, read_buffer: ctypes_buffer_t) -> ctypes_buffer_t:
with Path(f'/proc/{self.pid}/mem').open('rb+') as mem:
with open('/proc/{}/mem'.format(self.pid), 'rb+') as mem:
mem.seek(base_address)
mem.readinto(read_buffer)
return read_buffer
def get_path(self) -> str | None:
def get_path(self) -> str:
try:
with Path(f'/proc/{self.pid}/cmdline').open('rb') as ff:
return ff.read().decode().split('\x00')[0]
with open('/proc/{}/cmdline', 'rb') as f:
return f.read().decode().split('\x00')[0]
except FileNotFoundError:
return None
return ''
@staticmethod
def list_available_pids() -> list[int]:
def list_available_pids() -> List[int]:
pids = []
for pid_str in os.listdir('/proc'):
try:
@ -99,26 +91,26 @@ class Process(AbstractProcess):
return pids
@staticmethod
def get_pid_by_name(target_name: str) -> int | None:
def get_pid_by_name(target_name: str) -> int or None:
for pid in Process.list_available_pids():
try:
logger.debug(f'Checking name for pid {pid}')
with Path(f'/proc/{pid}/cmdline').open('rb') as cmdline:
logger.info('Checking name for pid {}'.format(pid))
with open('/proc/{}/cmdline'.format(pid), 'rb') as cmdline:
path = cmdline.read().decode().split('\x00')[0]
except FileNotFoundError:
continue
name = Path(path).name
logger.debug(f'Name was "{name}"')
name = os.path.basename(path)
logger.info('Name was "{}"'.format(name))
if path is not None and name == target_name:
return pid
logger.info(f'Found no process with name {target_name}')
logger.info('Found no process with name {}'.format(target_name))
return None
def list_mapped_regions(self, writeable_only: bool = True) -> list[tuple[int, int]]:
def list_mapped_regions(self, writeable_only: bool = True) -> List[Tuple[int, int]]:
regions = []
with Path(f'/proc/{self.pid}/maps').open('r') as maps:
with open('/proc/{}/maps'.format(self.pid), 'r') as maps:
for line in maps:
bounds, privileges = line.split()[0:2]

52
mem_edit/utils.py
View File

@ -11,34 +11,25 @@ Utility functions and types:
Check if two buffers (ctypes objects) store equal values:
ctypes_equal(a, b)
"""
from typing import List
import ctypes
ctypes_buffer_t = (
ctypes._SimpleCData
| ctypes.Array
| ctypes.Structure
| ctypes.Union
)
ctypes_buffer_t = ctypes._SimpleCData or ctypes.Array or ctypes.Structure or ctypes.Union
class MemEditError(Exception):
pass
def search_buffer_verbatim(
needle_buffer: ctypes_buffer_t,
haystack_buffer: ctypes_buffer_t,
) -> list[int]:
def search_buffer_verbatim(needle_buffer: ctypes_buffer_t, haystack_buffer: ctypes_buffer_t) -> List[int]:
"""
Search for a buffer inside another buffer, using a direct (bitwise) comparison
Args:
needle_buffer: Buffer to search for.
haystack_buffer: Buffer to search in.
Returns:
List of offsets where the `needle_buffer` was found.
:param needle_buffer: Buffer to search for.
:param haystack_buffer: Buffer to search in.
:return: List of offsets where the needle_buffer was found.
"""
found = []
@ -54,20 +45,14 @@ def search_buffer_verbatim(
return found
def search_buffer(
needle_buffer: ctypes_buffer_t,
haystack_buffer: ctypes_buffer_t,
) -> list[int]:
def search_buffer(needle_buffer: ctypes_buffer_t, haystack_buffer: ctypes_buffer_t) -> List[int]:
"""
Search for a buffer inside another buffer, using `ctypes_equal` for comparison.
Much slower than `search_buffer_verbatim`.
Search for a buffer inside another buffer, using ctypes_equal for comparison.
Much slower than search_buffer_verbatim.
Args:
needle_buffer: Buffer to search for.
haystack_buffer: Buffer to search in.
Returns:
List of offsets where the needle_buffer was found.
:param needle_buffer: Buffer to search for.
:param haystack_buffer: Buffer to search in.
:return: List of offsets where the needle_buffer was found.
"""
found = []
read_type = type(needle_buffer)
@ -78,14 +63,11 @@ def search_buffer(
return found
def ctypes_equal(
a: ctypes_buffer_t,
b: ctypes_buffer_t,
) -> bool:
def ctypes_equal(a: ctypes_buffer_t, b: ctypes_buffer_t) -> bool:
"""
Check if the values stored inside two ctypes buffers are equal.
"""
if not type(a) == type(b): # noqa: E721
if not type(a) == type(b):
return False
if isinstance(a, ctypes.Array):
@ -93,10 +75,10 @@ def ctypes_equal(
elif isinstance(a, ctypes.Structure) or isinstance(a, ctypes.Union):
for attr_name, attr_type in a._fields_:
a_attr, b_attr = (getattr(x, attr_name) for x in (a, b))
if isinstance(a, (ctypes.Array, ctypes.Structure, ctypes.Union, ctypes._SimpleCData)):
if isinstance(a, ctypes_buffer_t):
if not ctypes_equal(a_attr, b_attr):
return False
elif a_attr != b_attr:
elif not a_attr == b_attr:
return False
return True

149
mem_edit/windows.py
View File

@ -2,9 +2,10 @@
Implementation of Process class for Windows
"""
from typing import List, Tuple
from math import floor
from os import strerror
from pathlib import Path
import os.path
import ctypes
import ctypes.wintypes
import logging
@ -13,6 +14,7 @@ from .abstract import Process as AbstractProcess
from .utils import ctypes_buffer_t, MemEditError
logging.basicConfig(level=logging.INFO)
logger = logging.getLogger(__name__)
@ -24,10 +26,10 @@ privileges = {
'PROCESS_VM_WRITE': 0x0020,
}
privileges['PROCESS_RW'] = (
privileges['PROCESS_QUERY_INFORMATION']
| privileges['PROCESS_VM_OPERATION']
| privileges['PROCESS_VM_READ']
| privileges['PROCESS_VM_WRITE']
privileges['PROCESS_QUERY_INFORMATION'] |
privileges['PROCESS_VM_OPERATION'] |
privileges['PROCESS_VM_READ'] |
privileges['PROCESS_VM_WRITE']
)
# Memory region states
@ -49,13 +51,13 @@ page_protections = {
}
# Custom (combined) permissions
page_protections['PAGE_READABLE'] = (
page_protections['PAGE_EXECUTE_READ']
| page_protections['PAGE_EXECUTE_READWRITE']
| page_protections['PAGE_READWRITE']
page_protections['PAGE_EXECUTE_READ'] |
page_protections['PAGE_EXECUTE_READWRITE'] |
page_protections['PAGE_READWRITE']
)
page_protections['PAGE_READWRITEABLE'] = (
page_protections['PAGE_EXECUTE_READWRITE']
| page_protections['PAGE_READWRITE']
page_protections['PAGE_EXECUTE_READWRITE'] |
page_protections['PAGE_READWRITE']
)
# Memory types
@ -65,56 +67,19 @@ mem_types = {
'MEM_PRIVATE': 0x20000,
}
# C struct for VirtualQueryEx
class MEMORY_BASIC_INFORMATION32(ctypes.Structure):
class MEMORY_BASIC_INFORMATION(ctypes.Structure):
_fields_ = [
('BaseAddress', ctypes.wintypes.DWORD),
('AllocationBase', ctypes.wintypes.DWORD),
('BaseAddress', ctypes.c_void_p),
('AllocationBase', ctypes.c_void_p),
('AllocationProtect', ctypes.wintypes.DWORD),
('RegionSize', ctypes.wintypes.DWORD),
('RegionSize', ctypes.wintypes.UINT),
('State', ctypes.wintypes.DWORD),
('Protect', ctypes.wintypes.DWORD),
('Type', ctypes.wintypes.DWORD),
]
class MEMORY_BASIC_INFORMATION64(ctypes.Structure):
_fields_ = [
('BaseAddress', ctypes.c_ulonglong),
('AllocationBase', ctypes.c_ulonglong),
('AllocationProtect', ctypes.wintypes.DWORD),
('__alignment1', ctypes.wintypes.DWORD),
('RegionSize', ctypes.c_ulonglong),
('State', ctypes.wintypes.DWORD),
('Protect', ctypes.wintypes.DWORD),
('Type', ctypes.wintypes.DWORD),
('__alignment2', ctypes.wintypes.DWORD),
]
PTR_SIZE = ctypes.sizeof(ctypes.c_void_p)
MEMORY_BASIC_INFORMATION: type[ctypes.Structure]
if PTR_SIZE == 8: # 64-bit python
MEMORY_BASIC_INFORMATION = MEMORY_BASIC_INFORMATION64
elif PTR_SIZE == 4: # 32-bit python
MEMORY_BASIC_INFORMATION = MEMORY_BASIC_INFORMATION32
ctypes.windll.kernel32.VirtualQueryEx.argtypes = [
ctypes.wintypes.HANDLE,
ctypes.wintypes.LPCVOID,
ctypes.c_void_p,
ctypes.c_size_t]
ctypes.windll.kernel32.ReadProcessMemory.argtypes = [
ctypes.wintypes.HANDLE,
ctypes.wintypes.LPCVOID,
ctypes.c_void_p,
ctypes.c_size_t,
ctypes.c_void_p]
ctypes.windll.kernel32.WriteProcessMemory.argtypes = [
ctypes.wintypes.HANDLE,
ctypes.wintypes.LPCVOID,
ctypes.c_void_p,
ctypes.c_size_t,
ctypes.c_void_p]
# C struct for GetSystemInfo
class SYSTEM_INFO(ctypes.Structure):
@ -124,7 +89,7 @@ class SYSTEM_INFO(ctypes.Structure):
('dwPageSize', ctypes.wintypes.DWORD),
('lpMinimumApplicationAddress', ctypes.c_void_p),
('lpMaximumApplicationAddress', ctypes.c_void_p),
('dwActiveProcessorMask', ctypes.c_void_p),
('dwActiveProcessorMask', ctypes.wintypes.DWORD),
('dwNumberOfProcessors', ctypes.wintypes.DWORD),
('dwProcessorType', ctypes.wintypes.DWORD),
('dwAllocationGranularity', ctypes.wintypes.DWORD),
@ -134,9 +99,9 @@ class SYSTEM_INFO(ctypes.Structure):
class Process(AbstractProcess):
process_handle: int | None
process_handle = None
def __init__(self, process_id: int) -> None:
def __init__(self, process_id: int):
process_handle = ctypes.windll.kernel32.OpenProcess(
privileges['PROCESS_RW'],
False,
@ -144,15 +109,15 @@ class Process(AbstractProcess):
)
if not process_handle:
raise MemEditError(f'Couldn\'t open process {process_id}')
raise MemEditError('Couldn\'t open process {}'.format(process_id))
self.process_handle = process_handle
def close(self) -> None:
def close(self):
ctypes.windll.kernel32.CloseHandle(self.process_handle)
self.process_handle = None
def write_memory(self, base_address: int, write_buffer: ctypes_buffer_t) -> None:
def write_memory(self, base_address: int, write_buffer: ctypes_buffer_t):
try:
ctypes.windll.kernel32.WriteProcessMemory(
self.process_handle,
@ -161,8 +126,8 @@ class Process(AbstractProcess):
ctypes.sizeof(write_buffer),
None
)
except (BufferError, ValueError, TypeError) as err:
raise MemEditError(f'Error with handle {self.process_handle}: {self._get_last_error()}') from err
except (BufferError, ValueError, TypeError):
raise MemEditError('Error with handle {}: {}'.format(self.process_handle, self._get_last_error()))
def read_memory(self, base_address: int, read_buffer: ctypes_buffer_t) -> ctypes_buffer_t:
try:
@ -173,79 +138,78 @@ class Process(AbstractProcess):
ctypes.sizeof(read_buffer),
None
)
except (BufferError, ValueError, TypeError) as err:
raise MemEditError(f'Error with handle {self.process_handle}: {self._get_last_error()}') from err
except (BufferError, ValueError, TypeError):
raise MemEditError('Error with handle {}: {}'.format(self.process_handle, self._get_last_error()))
return read_buffer
@staticmethod
def _get_last_error() -> tuple[int, str]:
def _get_last_error() -> Tuple[int, str]:
err = ctypes.windll.kernel32.GetLastError()
return err, strerror(err)
def get_path(self) -> str | None:
def get_path(self) -> str:
max_path_len = 260
name_buffer = (ctypes.c_char * max_path_len)()
rval = ctypes.windll.psapi.GetProcessImageFileNameA(
self.process_handle,
name_buffer,
max_path_len,
max_path_len
)
if rval <= 0:
return None
if rval > 0:
return name_buffer.value.decode()
else:
return None
@staticmethod
def list_available_pids() -> list[int]:
def list_available_pids() -> List[int]:
# According to EnumProcesses docs, you can't find out how many processes there are before
# fetching the list. As a result, we grab 100 on the first try, and if we get a full list
# of 100, repeatedly double the number until we get fewer than we asked for.
nn = 100
n = 100
returned_size = ctypes.wintypes.DWORD()
returned_size_ptr = ctypes.byref(returned_size)
while True:
pids = (ctypes.wintypes.DWORD * nn)()
pids = (ctypes.wintypes.DWORD * n)()
size = ctypes.sizeof(pids)
pids_ptr = ctypes.byref(pids)
success = ctypes.windll.Psapi.EnumProcesses(pids_ptr, size, returned_size_ptr)
if not success:
raise MemEditError(f'Failed to enumerate processes: nn={nn}')
raise MemEditError('Failed to enumerate processes: n={}'.format(n))
num_returned = floor(returned_size.value / ctypes.sizeof(ctypes.wintypes.DWORD))
if nn != num_returned:
if n == num_returned:
n *= 2
continue
else:
break
nn *= 2
return pids[:num_returned]
@staticmethod
def get_pid_by_name(target_name: str) -> int | None:
def get_pid_by_name(target_name: str) -> int or None:
for pid in Process.list_available_pids():
try:
logger.debug(f'Checking name for pid {pid}')
logger.info('Checking name for pid {}'.format(pid))
with Process.open_process(pid) as process:
path = process.get_path()
if path is None:
continue
name = Path(path).name
logger.debug(f'Name was "{name}"')
name = os.path.basename(path)
logger.info('Name was "{}"'.format(name))
if path is not None and name == target_name:
return pid
except ValueError:
pass
except MemEditError as err:
logger.debug(repr(err))
logger.info(f'Found no process with name {target_name}')
logger.info('Found no process with name {}'.format(target_name))
return None
def list_mapped_regions(self, writeable_only: bool = True) -> list[tuple[int, int]]:
def list_mapped_regions(self, writeable_only: bool = True) -> List[Tuple[int, int]]:
sys_info = SYSTEM_INFO()
sys_info_ptr = ctypes.byref(sys_info)
ctypes.windll.kernel32.GetSystemInfo(sys_info_ptr)
@ -253,7 +217,7 @@ class Process(AbstractProcess):
start = sys_info.lpMinimumApplicationAddress
stop = sys_info.lpMaximumApplicationAddress
def get_mem_info(address: int) -> MEMORY_BASIC_INFORMATION:
def get_mem_info(address):
"""
Query the memory region starting at or before 'address' to get its size/type/state/permissions.
"""
@ -265,12 +229,14 @@ class Process(AbstractProcess):
self.process_handle,
address,
mbi_ptr,
mbi_size)
mbi_size,
)
if success != mbi_size:
if success == 0:
raise MemEditError('Failed VirtualQueryEx with handle '
+ f'{self.process_handle}: {self._get_last_error()}')
raise MemEditError('Failed VirtualQueryEx with handle ' +
'{}: {}'.format(self.process_handle, self._get_last_error()))
else:
raise MemEditError('VirtualQueryEx output too short!')
return mbi
@ -279,11 +245,10 @@ class Process(AbstractProcess):
page_ptr = start
while page_ptr < stop:
page_info = get_mem_info(page_ptr)
if (page_info.Type == mem_types['MEM_PRIVATE']
and page_info.State == mem_states['MEM_COMMIT']
and page_info.Protect & page_protections['PAGE_READABLE'] != 0
and (page_info.Protect & page_protections['PAGE_READWRITEABLE'] != 0
or not writeable_only)):
if page_info.Type == mem_types['MEM_PRIVATE'] and \
page_info.State == mem_states['MEM_COMMIT'] and \
page_info.Protect & page_protections['PAGE_READABLE'] != 0 and \
(page_info.Protect & page_protections['PAGE_READWRITEABLE'] != 0 or not writeable_only):
regions.append((page_ptr, page_ptr + page_info.RegionSize))
page_ptr += page_info.RegionSize

87
pyproject.toml
View File

@ -1,87 +0,0 @@
[build-system]
requires = ["hatchling"]
build-backend = "hatchling.build"
[project]
name = "mem_edit"
description = "Multi-platform library for memory editing"
readme = "README.md"
license = { file = "LICENSE.md" }
authors = [
{ name="Jan Petykiewicz", email="jan@mpxd.net" },
]
homepage = "https://mpxd.net/code/jan/mem_edit"
repository = "https://mpxd.net/code/jan/mem_edit"
keywords = [
"memory",
"edit",
"editing",
"ReadProcessMemory",
"WriteProcessMemory",
"proc",
"mem",
"ptrace",
"multiplatform",
"scan",
"scanner",
"search",
"debug",
"cheat",
"trainer",
]
classifiers = [
"Programming Language :: Python :: 3",
"Development Status :: 5 - Production/Stable",
"Environment :: Other Environment",
"Intended Audience :: Developers",
"License :: OSI Approved :: GNU Affero General Public License v3",
"Operating System :: POSIX :: Linux",
"Operating System :: Microsoft :: Windows",
"Topic :: Software Development",
"Topic :: Software Development :: Debuggers",
"Topic :: Software Development :: Testing",
"Topic :: System",
"Topic :: Games/Entertainment",
"Topic :: Utilities",
]
requires-python = ">=3.11"
dynamic = ["version"]
dependencies = [
]
[tool.hatch.version]
path = "mem_edit/__init__.py"
[tool.ruff]
exclude = [
".git",
"dist",
]
line-length = 145
indent-width = 4
lint.dummy-variable-rgx = "^(_+|(_+[a-zA-Z0-9_]*[a-zA-Z0-9]+?))$"
lint.select = [
"NPY", "E", "F", "W", "B", "ANN", "UP", "SLOT", "SIM", "LOG",
"C4", "ISC", "PIE", "PT", "RET", "TCH", "PTH", "INT",
"ARG", "PL", "R", "TRY",
"G010", "G101", "G201", "G202",
"Q002", "Q003", "Q004",
]
lint.ignore = [
#"ANN001", # No annotation
"ANN002", # *args
"ANN003", # **kwargs
"ANN401", # Any
"ANN101", # self: Self
"SIM108", # single-line if / else assignment
"RET504", # x=y+z; return x
"PIE790", # unnecessary pass
"ISC003", # non-implicit string concatenation
"C408", # dict(x=y) instead of {'x': y}
"PLR09", # Too many xxx
"PLR2004", # magic number
"PLC0414", # import x as x
"TRY003", # Long exception message
]

61
setup.py Normal file
View File

@ -0,0 +1,61 @@
#!/usr/bin/env python3
from setuptools import setup, find_packages
with open('README.md', 'r') as f:
long_description = f.read()
with open('mem_edit/VERSION', 'r') as f:
version = f.read().strip()
setup(name='mem_edit',
version=version,
description='Multi-platform library for memory editing',
long_description=long_description,
long_description_content_type='text/markdown',
author='Jan Petykiewicz',
author_email='anewusername@gmail.com',
url='https://mpxd.net/code/jan/mem_edit',
keywords=[
'memory',
'edit',
'editing',
'ReadProcessMemory',
'WriteProcessMemory',
'proc',
'mem',
'ptrace',
'multiplatform',
'scan',
'scanner',
'search',
'debug',
'cheat',
'trainer',
],
classifiers=[
'Programming Language :: Python :: 3',
'Development Status :: 4 - Beta',
'Environment :: Other Environment',
'Intended Audience :: Developers',
'License :: OSI Approved :: GNU Affero General Public License v3',
'Operating System :: POSIX :: Linux',
'Operating System :: Microsoft :: Windows',
'Topic :: Software Development',
'Topic :: Software Development :: Debuggers',
'Topic :: Software Development :: Testing',
'Topic :: System',
'Topic :: Games/Entertainment',
'Topic :: Utilities',
],
packages=find_packages(),
package_data={
'mem_edit': ['VERSION']
},
install_requires=[
'typing',
],
extras_require={
},
)