bubblebabble.py

#!/usr/bin/python3

"""
Python implementation of "The Bubble Babble Binary Data Encoding" (draft-huima-01)

Please note that this is not a cryptographic hash,
it can be reversed.
"""

class BubbleBabble:
    consonants = 'bcdfghklmnprstvz'
    vowels = 'aeiouy'
    
    def __init__(self):
        self.seed = 1
        self.out = ["x"]
        self.even = True

    def update(self, buf):
        for c in buf:
            if self.even:
                self.out.append(self.vowels[(((c >> 6) & 3) + self.seed) % 6])
                self.out.append(self.consonants[(c >> 2) & 15])
                self.out.append(self.vowels[((c & 3) + (self.seed // 6)) % 6])
                self.seed = (self.seed * 5) + (c * 7)
            else:
                self.seed = (self.seed + c) % 36
                self.out.append(self.consonants[(c >> 4) & 15])
                self.out.append('-')
                self.out.append(self.consonants[(c & 15)])
            self.even = not self.even
        return self

    def digest(self):
        out = self.out[:]
        if self.even:
            out.append(self.vowels[self.seed % 6])
            out.append('x')
            out.append(self.vowels[self.seed // 6])
        out.append('x')
        return ''.join(out)
        
def digest(buf):
    return BubbleBabble().update(buf).digest()

def test():
    assert digest(b'') == 'xexax'
    assert digest(b'1234567890') == 'xesef-disof-gytuf-katof-movif-baxux'
    assert digest(b'Pineapple') == 'xigak-nyryk-humil-bosek-sonax'

if __name__ == '__main__':
    import sys

    test()
    buf = sys.stdin.buffer.read()
    print(digest(buf))