compute osdb (opensubtitles.org) hash code with the file size value, allowing to not download the entire file

osdb_hash.py

import struct
import os


BLOCK_SIZE = 65536
MIN_SIZE = BLOCK_SIZE * 2

# based on https://trac.opensubtitles.org/projects/opensubtitles/wiki/HashSourceCodes#Python
def osdb_hash(filename, filesize=0):
    try:
        longlongformat = "<q"  # little-endian long long
        bytesize = struct.calcsize(longlongformat)

        f = open(filename, "rb")

        if not filesize:
            filesize = os.path.getsize(filename)
        hash = filesize

        if filesize < MIN_SIZE:
            return "FileSizeError, at least %d, but %d" % (MIN_SIZE, hash)

        for x in range(BLOCK_SIZE // bytesize):
            buffer = f.read(bytesize)
            (l_value,) = struct.unpack(longlongformat, buffer)
            hash += l_value
            hash = hash & 0xFFFFFFFFFFFFFFFF  # to remain as 64bit number

        f.seek(max(0, filesize - BLOCK_SIZE), 0)
        for x in range(BLOCK_SIZE // bytesize):
            buffer = f.read(bytesize)
            (l_value,) = struct.unpack(longlongformat, buffer)
            hash += l_value
            hash = hash & 0xFFFFFFFFFFFFFFFF

        f.close()
        returnedhash = "%016x" % hash
        return returnedhash

    except (IOError):
        return "IOError"


if __name__ == '__main__':
    import sys
    path = sys.argv[1]
    if len(sys.argv) > 2:
        filesize = sys.argv[2]
        print(osdb_hash(path, int(filesize)))
    else:
        print(osdb_hash(path))