CODEWITHSUNDEEP
pythonctypespython-2.5
sorin
sorin

Reputation: 170698

How to access a block of memory as a file object in Python 2.5?

I would like to get a file object for a block of memory in Python 2.5 under Windows.(for some reasons I cannot use newer versions for this tasks.)

So as input I do have a pointer and a size and let's assume that I need only read only access.

In case you'll wonder, I got these by using ctypes and I need to make them available to a function that expects a filehandler (read-only).

I considered using cStringIO but in order to create such an object I need a string object.

Upvotes: 2

Views: 597

Answers (2)

jsbueno
jsbueno

Reputation: 110591

You should use ctypes in there. As of Python 2.5 ctypes where already on the standard library, so a "win" situation for you.

With ctypes you can construct a python object representing a higher level pointe doing this:

import ctypes 
integer_pointer_type = ctypes.POINTER(ctypes.c_int)
my_pointer = integer_pointer_type.from_address(your_address)

You can then address the memory contents as a Python indexed object, like print my_pointer[0]

This won't give you a "file like interface" - although it would be trivial to wrap a class with a "read" and "seek" methods around such an object:

class MyMemoryFile(object):
    def __init__(self, pointer, size=None):
         integer_pointer_type = ctypes.POINTER(ctypes.c_uchar)
         self.pointer = integer_pointer_type.from_address(your_address)
         self.cursor = 0
         self.size = size

    def seek(self, position, whence=0):
         if whence == 0:
              self.cursor = position
         raise NotImplementedError
    def read(size=None):
         if size is None:
             res =  str(self.pointer[cursor:self.size])
             self.cursor = self.size
         else:
             res = str(self.pointer[self.cursor:self.cursor + size]
             self.cursor += size
         return res

(not tested - write me if it does not work - can be fixed)

Please note that attempts to read the memory beyond the space allocated for your data structure will have the exact same effects as doing so in C: in most cases,a segmentation fault.

Upvotes: 6

Adrien Plisson
Adrien Plisson

Reputation: 23303

from the ctypes documentation, it appears that you can get a string from an address in memory, using the ctypes.string_at() function.

the problem is that a string is not mutable, that means that you won't be able to modify the resulting string from python. to have a mutable buffer in python, you will need to call the ctypes.create_string_buffer() function from python.

Upvotes: 1

Related Questions