bow_utils.py 文件源码

def add(self, desc):
        if self.built_: 
            return

        if self.vocab_len_ < self.N_:
            Nd = len(desc)
            st, end = self.vocab_len_, min(self.vocab_len_ + Nd, self.N_)
            self.vocab_data_[st:end] = desc[:end-st]
            self.vocab_len_ += len(desc)
            print('Vocabulary building: {:}/{:}'.format(self.vocab_len_, self.N_))
        else: 
            print('Vocabulary built')
            self.built_ = True

        # else: 
        #     # Build vocab if not built already
        #     self.voc_.build(self.vocab_data_, self.K_)
        #     self.vocab_ = self.voc_.getCentroids()

        #     sz = self.vocab_.shape[:2]
        #     if sz[0] != self.K_ or sz[1] != self.D_: 
        #         raise RuntimeError('Voc error! KxD={:}x{:}, expected'.format(sz[0],sz[1],self.K_,self.D_))

        #     self.save('vocab.yaml.gz')