# -*- coding: utf-8 -*-
#Author: Pierre Ratinaud
#Copyright (c) 2010 Pierre Ratinaud
-#Lisense: GNU/GPL
+#License: GNU/GPL
import os
from optparse import OptionParser
# log.warning('ATTENTION gethapaxuces')
# MakeUciStat(corpus)
#corpus.gethapaxuces()
- ucisize = corpus.getucisize()
- ucisize = [`val` for val in ucisize]
- uciet = [[val.split('_')[1] for val in uci.etoiles[1:]] for uci in corpus.ucis]
- #for line in uciet :
- # print '\t'.join(line)
+ # ucisize = corpus.getucisize()
+ # ucisize = [`val` for val in ucisize]
+ #uciet = [uci.etoiles[1:] for uci in corpus.ucis]
+ uceet = [corpus.ucis[uce.uci].etoiles[1:] for uci in corpus.ucis for uce in uci.uces]
+ print uceet[0:10]
+ for line in uceet :
+ print '\t'.join(line)
#res = zip(uciet, ucisize)
- res = [uciet[i] + [ucisize[i]] for i, val in enumerate(uciet)]
- print res[0:10]
- with open('ucisize.csv', 'w') as f :
- f.write('\n'.join(['\t'.join(val) for val in res]))
+ # res = [uciet[i] + [ucisize[i]] for i, val in enumerate(uciet)]
+ # print res[0:10]
+ #ucesize = corpus.getucesize()
+ #print ucesize[0:40]
+ #with open('sentences_size.csv', 'w') as f :
+ # f.write('\n'.join([`val` for val in ucesize]))
# self.content = f.read()
#self.content = self.content.replace('\r','')
if options.type_analyse == 'alceste' :