X-Git-Url: https://git.njae.me.uk/?a=blobdiff_plain;f=segment.py;h=08bc0aa84244bf222655defba7b4a0773f87ebcf;hb=38f3707a2337652adf616b47b2f35f3f20d433bd;hp=e4b019f4c8248d8647f938fea8295c132308de0b;hpb=792bef4fa890a8c834ddd83ab9a573d0e2a75dc9;p=cipher-tools.git diff --git a/segment.py b/segment.py index e4b019f..08bc0aa 100644 --- a/segment.py +++ b/segment.py @@ -3,6 +3,8 @@ import string import collections from math import log10 import itertools +import sys +sys.setrecursionlimit(1000000) def memo(f): "Memoize function f." @@ -58,7 +60,5 @@ def avoid_long_words(key, N): """ return -log10((N * 10**(len(key) - 2))) -# N = 1024908267229 ## Number of tokens - Pw = Pdist(datafile('count_1w.txt'), avoid_long_words)