KonText vs. Manatee - multi-dimensional frequency distribution

gistfile1.py

"""
KonText vs. Manatee - multi-dimensional frequency distribution

words = manatee.StrVector()
freqs = manatee.NumVector()
norms = manatee.NumVector()
crit = 'opus.rokvyd 0 opus.genre 0'
corpus.freq_dist(range_stream, crit, limit, words, freqs, norms)
"""

import re

# this is what Manatee returns
words = [u'2006\tTRV', u'2007\tSPO', u'2000\tX', u'2003\tJUN', u'1995\tMEM', u'2009\tREG', u'2002\tTRV', 
         u'2008\tX', u'1992\tX', u'1992\tY', u'2009\tHOU', u'1996\tPSY', u'1996\tART', u'2000\tCHR', u'2000\tART']
freqs = [12, 1, 35, 6, 1, 2, 4, 4, 8, 2, 1, 1, 1, 1, 1]

def normalize(words, freqs):
    """
    normalizes two vectors (one for independent variables and one for dependent ones) 

    arguments:
    words -- ['x1\ty1', 'x2\ty2',..., 'xN\tyN']
    freqs -- [v1,...,vN]
    """
    def custom_cmp(a, b, idx=0):
        ans = cmp(a[0][idx], b[0][idx])
        if ans == 0 and idx + 1 < len(a):
            ans = custom_cmp(a, b, idx + 1)
        return ans
    return sorted(zip([tuple(re.split(r'\s+', v)) for v in words], freqs), cmp=custom_cmp)


if __name__ == '__main__':
    data = normalize(words, freqs)
    print(data)