X-Git-Url: http://iramuteq.org/git?p=iramuteq;a=blobdiff_plain;f=corpus.py;h=d26a8a376dd96e3fa4d7be12a9c47bce6584a9ef;hp=db258131fdb070723d06ae797e44942e48500163;hb=bd8d0a889d1d393e64a6d768dc14e9c639a0df8c;hpb=eb798ba97d92c40656cedc5414974a9edf7a332c;ds=sidebyside diff --git a/corpus.py b/corpus.py index db25813..d26a8a3 100644 --- a/corpus.py +++ b/corpus.py @@ -482,6 +482,38 @@ class Corpus : if lem : guce = ' '.join([self.formes[forme].lem for forme in guce.split()]) f.write(guce.encode(self.parametres['syscoding']) + '\n\n') + + def export_owledge(self, rep, classe, lem = False, uci = False) : + sts = self.lc[classe - 1] + if not uci : + res = self.getconcorde(sts) + self.make_iduces() + else : + res = self.getuciconcorde(sts) + for uce in res : + ident = uce[0] + guce = uce[1] + outf = '.'.join([`ident`, 'txt']) + outf = os.path.join(rep, outf) + if lem : + guce = ' '.join([self.formes[forme].lem for forme in guce.split()]) + with open(outf, 'w') as f : + f.write(guce.encode('cp1252', errors = 'replace')) + + def export_tropes(self, fileout, classe, lem = False, uci = False) : + sts = self.lc[classe - 1] + if not uci : + res = self.getconcorde(sts) + self.make_iduces() + else : + res = self.getuciconcorde(sts) + with open(fileout, 'w') as f : + for uce in res : + guce = uce[1] + if lem : + guce = ' '.join([self.formes[forme].lem for forme in guce.split()]) + f.write(guce.encode('cp1252', errors = 'replace')) + f.write('\n') def make_and_write_sparse_matrix_from_uces(self, actives, outfile, listuce = False) : log.info('make_and_write_sparse_matrix_from_uces %s' % outfile)