'''
Created on Dec 5, 2016

@author: David
'''
import codecs

file = "f:/parll data/data/syllable_len_dist_nz.txt"
f = codecs.open(file, "r", "utf-8")

for line in f:
    rl = line.rstrip()[1:-1]
    tfreq = 0
    tsum = 0
    for upair in rl.split(","):
        feat,freq = (int(x.strip()) for x in upair.split(":"))
        tfreq += freq
        tsum += feat * freq
    avg_len = tsum/tfreq
    print(avg_len)