get_readable_size.py

#!/usr/bin/python

import os
import re
import sys
import hashlib

# get_readable_size by Lipis
# http://stackoverflow.com/posts/14998888/revisions


def get_readable_size(bytes, precision=1, unit="IEC"):
    unit = unit.upper()
    if(unit != "IEC" and unit != "SI"):
        unit = "IEC"
    if(unit == "IEC"):
        units = [" B", " KiB", " MiB", " GiB", " TiB", " PiB", " EiB", " ZiB"]
        unitswos = ["B", "KiB", "MiB", "GiB", "TiB", "PiB", "EiB", "ZiB"]
        unitsize = 1024.0
    if(unit == "SI"):
        units = [" B", " kB", " MB", " GB", " TB", " PB", " EB", " ZB"]
        unitswos = ["B", "kB", "MB", "GB", "TB", "PB", "EB", "ZB"]
        unitsize = 1000.0
    return_val = {}
    orgbytes = bytes
    for unit in units:
        if abs(bytes) < unitsize:
            strformat = "%3."+str(precision)+"f%s"
            pre_return_val = (strformat % (bytes, unit))
            pre_return_val = re.sub(
                r"([0]+) ([A-Za-z]+)", r" \2", pre_return_val)
            pre_return_val = re.sub(r"\. ([A-Za-z]+)", r" \1", pre_return_val)
            alt_return_val = pre_return_val.split()
            return_val = {'Bytes': orgbytes, 'ReadableWithSuffix': pre_return_val,
                          'ReadableWithoutSuffix': alt_return_val[0], 'ReadableSuffix': alt_return_val[1]}
            return return_val
        bytes /= unitsize
    strformat = "%."+str(precision)+"f%s"
    pre_return_val = (strformat % (bytes, "YiB"))
    pre_return_val = re.sub(r"([0]+) ([A-Za-z]+)", r" \2", pre_return_val)
    pre_return_val = re.sub(r"\. ([A-Za-z]+)", r" \1", pre_return_val)
    alt_return_val = pre_return_val.split()
    return_val = {'Bytes': orgbytes, 'ReadableWithSuffix': pre_return_val,
                  'ReadableWithoutSuffix': alt_return_val[0], 'ReadableSuffix': alt_return_val[1]}
    return return_val


def get_readable_size_from_file(infile, precision=1, unit="IEC", usehashes=False, usehashtypes="md5,sha1"):
    unit = unit.upper()
    usehashtypes = usehashtypes.lower()
    getfilesize = os.path.getsize(infile)
    return_val = get_readable_size(getfilesize, precision, unit)
    if(usehashes == True):
        hashtypelist = usehashtypes.split(",")
        openfile = open(infile, "rb")
        filecontents = openfile.read()
        openfile.close()
        listnumcount = 0
        listnumend = len(hashtypelist)
        while(listnumcount < listnumend):
            hashtypelistlow = hashtypelist[listnumcount].strip()
            hashtypelistup = hashtypelistlow.upper()
            filehash = hashlib.new(hashtypelistup)
            filehash.update(filecontents)
            filegethash = filehash.hexdigest()
            return_val.update({hashtypelistup: filegethash})
            listnumcount += 1
    return return_val


def get_readable_size_from_string(instring, precision=1, unit="IEC", usehashes=False, usehashtypes="md5,sha1"):
    unit = unit.upper()
    usehashtypes = usehashtypes.lower()
    getfilesize = len(instring)
    return_val = get_readable_size(getfilesize, precision, unit)
    if(usehashes == True):
        hashtypelist = usehashtypes.split(",")
        listnumcount = 0
        listnumend = len(hashtypelist)
        while(listnumcount < listnumend):
            hashtypelistlow = hashtypelist[listnumcount].strip()
            hashtypelistup = hashtypelistlow.upper()
            filehash = hashlib.new(hashtypelistup)
            if(sys.version[0] == "2"):
                filehash.update(instring)
            if(sys.version[0] >= "3"):
                filehash.update(instring.encode('utf-8'))
            filegethash = filehash.hexdigest()
            return_val.update({hashtypelistup: filegethash})
            listnumcount += 1
    return return_val


testfile = "/bin/sh"
print(get_readable_size_from_file(testfile, 1, "IEC", True))
print(get_readable_size_from_file(testfile, 1, "SI", True))