return [len(uce[1].split()) for uce in res]
def getconcorde(self, uces) :
- return self.cuces.execute('select * from uces where id IN (%s);' % ', '.join([`i` for i in uces]))
+ return self.cuces.execute('select * from uces where id IN (%s) ORDER BY id;' % ', '.join([`i` for i in uces]))
def getuciconcorde(self, ucis) :
uces = [[val,[uce.ident for uce in self.ucis[val].uces]] for val in ucis]
self.corpus.pathout = PathOut(filename = parametres_corpus['originalpath'], dirout = parametres_corpus['pathout'])
self.corpus.pathout.createdir(parametres_corpus['pathout'])
self.corpus.parametres['uuid'] = str(uuid4())
- self.corpus.parametres['corpus_name'] = os.path.split(self.corpus.parametres['pathout'])[1]
+ self.corpus.parametres['corpus_name'] = parametres_corpus['corpus_name']#os.path.split(self.corpus.parametres['pathout'])[1]
self.corpus.parametres['type'] = 'corpus'
if self.corpus.parametres['keep_ponct'] :
self.ponctuation_espace = [' ', '']
self.dlg = dlg
self.ori = corpus
self.infile = None
- self.corpus = Corpus(self, corpus.parametres)
+ self.corpus = Corpus(self, {'type' : 'corpus', 'originalpath' : corpus.parametres['originalpath'], 'encoding' : corpus.parametres['encoding']})
self.last = 0
self.encoding = corpus.parametres['encoding']
self.corpus.parametres['corpus_name'] = parametres['corpus_name']
idpara += 1
elif parametres.get('fromcluster', False) :
pass
+ elif parametres.get('fromuceids', False) :
+ print 'fromuceids'
+ dictucekeep = dict(zip(parametres['uceids'], parametres['uceids']))
+ idpara = 0
+ for uci in self.ori.ucis :
+ if uci.paras == [] :
+ keepuces = [uce for uce in uci.uces if uce.ident in dictucekeep]
+ if keepuces != [] :
+ uci.uces = keepuces
+ self.corpus.ucis.append(uci)
+ idpara += 1
+ else :
+ newuces = []
+ newpara = []
+ for et in uci.paras :
+ keepuces = [uce for uce in uci.uces if uce.para == idpara]
+ idpara += 1
+ if keepuces != [] :
+ newuces += keepuces
+ newpara.append(et)
+ if newuces != [] :
+ uci.uces = newuces
+ uci.paras = newpara
+ self.corpus.ucis.append(uci)
+
#create database
self.connect()
self.dobuild()
self.dlg = dlg
parametres = DoConf(os.path.join(self.parent.UserConfigPath,'corpus.cfg')).getoptions('corpus')
parametres['pathout'] = PathOut(parent.filename, 'corpus').mkdirout()
+ parametres['corpus_name'] = os.path.split(parametres['pathout'])[1]
dial = CorpusPref(parent, parametres)
dial.CenterOnParent()
dial.txtpath.SetLabel(parent.filename)
parametres['meta'] = corpus.make_etoiles()
elif parametres.get('fromtheme', False) :
parametres['meta'] = corpus.make_themes()
+ else :
+ parametres['meta'] = []
parametres['meta'].sort()
dial = SubTextFromMetaDial(parent, parametres)
self.res = dial.ShowModal()