Python-сообщество

# coding: utf-8

import os as _os
import hashlib as _hashlib
from subprocess import Popen as _Popen

def file_md5(file, use_system = False):
    """file_md5(file, use_system = False) -> md5sum of "file" as hexdigest string.
       "file" may be a file name or file object, opened for read.
       If "use_system" is True, if possible use system specific program. This ignore, if file object given."""

    if isinstance(file, basestring):
        if use_system:
            sysname = _os.uname()[0]
            if sysname in ('Darwin', 'FreeBSD'):
                po = _Popen('md5 -q "%s"' % file, shell=True, stdout=-1, stderr=-1)
                po.wait()
                m = po.stdout.readline().strip()
                if len(m) == 32:
                    return m
#            elif sysname == ...

        file = open(file, 'rb')

    h = _hashlib.md5()
    block = file.read(h.block_size)
    while block:
        h.update(block)
        block = file.read(h.block_size)
    return h.hexdigest()

"""
fn = "/Users/Shared/Фильмы/Аниме/Evangelion/Neon Genesis Evangelion - 13.avi"
print file_md5(fn)
print file_md5(open(fn, 'rb'))
print file_md5(fn, True)
import os
os.system('md5 -q "%s"' % fn)
"""

Печатает или проверяет контрольные суммы MD5 (128-битные).
  -b, --binary            читать в двоичном режиме
  -t, --text              читать в текстовом режиме (по умолчанию)

По умолчанию печатает строку с контрольной суммой, знак, показывающий
тип файла (`*' для двоичных, ` ' для текстовых), и имя каждого ФАЙЛА.

svartalf@svartalf:~$ md5sum -b ~/amarok2.jpeg
c08cce704ad1877a6d62b47e1bf593f2 *~/amarok2.jpeg

svartalf@svartalf:~$ md5sum -t rfc3920.txt
f102c3f9f6e2e45601a083a03d0c4c4c  rfc3920.txt

svartalf@svartalf:~$ uname -a
Linux svartalf 2.6.27-11-generic #1 SMP Thu Jan 29 19:24:39 UTC 2009 i686 GNU/Linux

#!/usr/bin/python

import sys
import os

from md5 import md5
from subprocess import Popen
from timeit import Timer

def test1():
        po = Popen('md5sum -b "%s"' % sys.argv[1], shell=True, stdout=-1, stderr=-1)
        po.wait()
        return po.stdout.readline().strip()

def test2():
        #h = md5()
        #file = open(sys.argv[1], 'rb')
    #   block = file.read(h.block_size)
    #   while block:
    #           h.update(block)
    #           block = file.read(h.block_size)
    #   return h.hexdigest()
        return md5(open(sys.argv[1], 'rb').read()).hexdigest()

print os.uname()[0]
print sys.argv[1], os.path.getsize(sys.argv[1])
print test1()
print test2()
print Timer("test1()", "from __main__ import test1").timeit(3)
print Timer("test2()", "from __main__ import test2").timeit(3)

def file_md5(file, use_system=False, block_size=1024):
    """file_md5(file, use_system = False) -> md5sum of "file" as hexdigest string.
       "file" may be a file name or file object, opened for read.
       If "use_system" is True, if possible use system specific program. This ignore, if file object given."""

    if isinstance(file, basestring):
        if use_system:
            sysname = _os.uname()[0]
            if sysname in ('Darwin', 'FreeBSD'):
                po = _Popen('md5 -q "%s"' % file, shell=True, stdout=-1, stderr=-1)
                po.wait()
                m = po.stdout.readline().strip()
                if len(m) == 32:
                    return m
            elif sysname == "Linux2":
                po = _Popen('md5sum -b "%s"' % file, shell=True, stdout=-1, stderr=-1)
                po.wait()
                m = po.stdout.readline().strip()
                if len(m) > 32:
                    return m[:32]
#            elif sysname == ...

        file = open(file, 'rb')
    h = _hashlib.md5()
    block = file.read(block_size)
    while block:
        h.update(block)
        block = file.read(block_size)
    return h.hexdigest()

# coding: utf-8

import os as _os
import hashlib as _hashlib
from subprocess import Popen as _Popen


def file_md5(file, use_system=False, block_size=1024*14):
    """file_md5(file, use_system = False) -> md5sum of "file" as hexdigest string.
       "file" may be a file name or file object, opened for read.
       If "use_system" is True, if possible use system specific program. This ignore, if file object given.
       "block_size" -- size in bytes buffer for calc md5. Used with "use_system=False"."""

    if isinstance(file, basestring):
        if use_system:
            sysname = _os.uname()[0]
            if sysname in ('Darwin', 'FreeBSD'):
                po = _Popen('md5 -q "%s"' % file, shell=True, stdout=-1, stderr=-1)
                po.wait()
                m = po.stdout.readline().strip()
                if len(m) == 32:
                    return m
                else:
                    print "Warning: md5 on %s is error" % sysname
            elif sysname == "Linux":
                po = _Popen('md5sum -b "%s"' % file, shell=True, stdout=-1, stderr=-1)
                po.wait()
                m = po.stdout.readline().strip()
#                if len(m) > 32:
                if m.find(' ') == 32:
                    return m[:32]
                else:
                    print "Warning: md5sum on Linux is error"
#            elif sysname == ...

        file = open(file, 'rb')

    h = _hashlib.md5()
    block = file.read(block_size)
    while block:
        h.update(block)
        block = file.read(block_size)
    return h.hexdigest()