clu-ling / py-processors

                # our old sequence continues

                current = re.sub('(B-|I-)','', str(current))

                tok = re.sub('(B-|I-)','', str(tok))

                if tok == current:

                    end = i

                # our old sequence has ended

                else:

                    # do we have a previous NE?

                    if current != Sentence.O:

                        end = i

                        named_entity = ' '.join(self.words[start:end])

                        entity_dict[current].append(named_entity)

                    # update our book-keeping vars

                    current = tok

                    start = i

                    end = None

        # this might be empty

        return entity_dict

            if tok == Sentence.O:

                # did we have an entity with the last token?

                current = re.sub('(B-|I-)','', str(current))

                if current == Sentence.O:

                    continue

                else:

                    # the last sequence has ended

                    end = i

                    # store the entity

                    named_entity = ' '.join(self.words[start:end])

                    entity_dict[current].append(named_entity)

                    # reset our book-keeping vars

                    current = Sentence.O

                    start = None

                    end = None

            # we have a tag!

            else:

                # our old sequence continues