hasher.py

import hashlib
import sys
import os

def generate_hashes(filename):
    outfile = open(filename+'.md5s','w')
    infile = open(filename,'rb')
    length = os.stat(filename).st_size
    outfile.write(filename+'\n')
    outfile.write(str(length)+'\n')
    while True:
        block = infile.read(2**20)
        outfile.write(hashlib.md5(block).hexdigest())
        outfile.write('\n')
        if len(block) < (2**20):
            break

def check_hashes(filename):
    infile = open(filename,'rb')
    sums = open(filename+'.md5s','r')
    name = sums.readline()
    mylength = os.stat(filename).st_size
    reallength = int(sums.readline()[:-1])
    if mylength != reallength:
        print('length mismatch')
        return
    blockno = 0
    while True:
        goodhash = sums.readline()[:-1]
        block = infile.read(2**20)
        myhash = hashlib.md5(block).hexdigest()
        if myhash != goodhash:
            print('hash mismatch '+str(blockno))
        blockno+=1
        if len(block) < (2**20):
            break
    

if __name__ == '__main__':
    cmd = sys.argv[1]
    filename = sys.argv[2]
    if cmd == 'g':
        generate_hashes(filename)
    if cmd == 'v':
        check_hashes(filename)