lemuceeff[uce] = lemuceeff.get(uce, 0) + eff[i]
return lemuceeff
+ def getlemclustereff(self, lem, cluster) :
+ return len(list(set(self.lc[cluster]).intersection(self.getlemuces(lem))))
+
def getlemeff(self, lem) :
return self.lems[lem].freq
res = self.getalluces()
return [len(uce[1].split()) for uce in res]
-# def getlemseff(self) :
-# if self.idformes is None :
-# self.make_idformes()
-# return dict([[lem, sum([self.idformes[forme].freq for forme in self.lems[lem]])] for lem in self.lems])
-
-# def getlemsefftype(self) :
-# if self.idformes is None :
-# self.make_idformes()
-# if self.lems is None :
-# self.make_lems()
-# return dict([[lem, [sum([self.idformes[forme].freq for forme in self.lems[lem]]), '', self.idformes[self.lems[lem].keys()[0]].gram]] for lem in self.lems])
-
def getconcorde(self, uces) :
return self.cuces.execute('select * from uces where id IN (%s);' % ', '.join([`i` for i in uces]))
lim -= 1
else :
stop = nbmax - 1
- log.info('nb actives = %i - eff min = %i ' % (stop, lim))
+ lim = effs[stop]
+ log.info('nb actives = %i - eff min = %i ' % (stop + 1, lim))
return [val[1] for val in allactives[0:stop + 1]], lim
def make_and_write_profile(self, actives, ucecl, fileout) :
for word in uce :
self.last += 1
self.corpus.add_word(word)
- #if self.dlg is not None :
- # if self.limitshow > self.count :
- # self.dlg.Pulse('textes : %i - segments : %i' % (iduci + 1, iduce +1))
- # self.count += 1
- # self.limitshow = 0
- # else :
- # self.limitshow = self.last / 100000
log.debug(' '.join([`iduci`,`idpara`,`iduce`]))
if self.last > self.lim :
self.backup_uce()
if douce :
out = []
reste, texte_uce, suite = self.decouper(self.prep_txt(txt), self.ucesize + 15, self.ucesize)
-# print 'reste'
-# print reste
-# print 'texte_uce'
-# print texte_uce
-# print 'suite'
-# print suite
while reste :
uce = ' '.join([val for val in texte_uce if val not in self.ponctuation_espace])
if uce != '' :
out.append(uce)
reste, texte_uce, suite = self.decouper(suite, self.ucesize + 15, self.ucesize)
-# print 'reste'
-# print reste
-# print 'texte_uce'
-# print texte_uce
-# print 'suite'
-# print suite
-
uce = ' '.join([val for val in texte_uce if val not in self.ponctuation_espace])
if uce != '' :
- #print 'RESTEE UUCEEEEEEEEEEEEE', uce
out.append(uce)
return out
else :
ReadLexique(self.parent, lang = parametres['lang'])
self.parent.expressions = ReadDicoAsDico(self.parent.DictPath.get(parametres['lang']+'_exp', 'french_exp'))
self.parametres = parametres
+ else :
+ if self.dlg is not None :
+ self.dlg.Destroy()
dial.Destroy()
def doanalyse(self) :