import sys

def binarize(file_in, file_out):

    """ Create a binary file from an ASCII file. Everything 
        but '0' and '1' (alternatively '_' and 'X') is ignored 
        from the input file. 

        The first two lines are width and height and will be saved
        as the first two bytes of the resulting file"""

    print ("binarizing " + file_in + " to "+ file_out + ".")
    
    try:

        f=open(file_in,'r')

        try:

            o=open(file_out,'wb')

            try:

                w=int(f.readline())
                h=int(f.readline())
                o.write(bytes([w]))
                o.write(bytes([h]))

                while True:

                    b=readnext(f)
                    if b==-1:
                        break

                    o.write(bytes([b]))

            except IOError:
                print("IOError happened during processing.");
                print("I am closing the input and ouput files and aborting.");
                print("Warning: The output file is very likely incomplete.");

            o.close()

        except IOError:
            print("IOError opening output file: " + file_out);

        f.close()

    except IOError:
        print("IOError opening input file: " + file_in);


def readnext(f):

    """ Read next 'byte' from the given ASCII file (BufferedWriter)
        everything but '_','X','0','1' is ignored. 
        Returns the byte as integer or -1 on EOF """

    l=0 
    b=""

    while l<8:

        c=f.read(1)

        if not c:
            if l==0:
                return -1
            else:
                print("trailing only ",b," with ",l," bits");
                return int(b,2)

        if c=="0" or c=='_':
            l+=1
            b+="0"

        if c=="1" or c=='X':
            l+=1
            b+="1"

    return  int(b,2)

def usage():

    """ Prints usage information to sdtout """

    print ("python3.x binarize.py [file_in] [file_out]")

""" direct use """

if __name__ == "__main__":

    if(len(sys.argv)==3):
        binarize(sys.argv[1],sys.argv[2])
    else:
        usage()