// integrate word segmentor into LTP
int LTP::wordseg(XML4NLP & xml) {
if (xml.QueryNote(NOTE_WORD)) {
return 0;
}
//
int ret = splitSentence_dummy(xml);
if (0 != ret) {
ERROR_LOG("in LTP::wordseg, failed to perform split sentence preprocess.");
return ret;
}
// get the segmentor pointer
void * segmentor = _resource.GetSegmentor();
if (0 == segmentor) {
ERROR_LOG("in LTP::wordseg, failed to init a segmentor");
return kWordsegError;
}
int stnsNum = xml.CountSentenceInDocument();
if (0 == stnsNum) {
ERROR_LOG("in LTP::wordseg, number of sentence equals 0");
return kEmptyStringError;
}
for (int i = 0; i < stnsNum; ++ i) {
std::string strStn = xml.GetSentence(i);
std::vector<std::string> vctWords;
if (ltp::strutils::codecs::length(strStn) > MAX_SENTENCE_LEN) {
ERROR_LOG("in LTP::wordseg, input sentence is too long");
return kSentenceTooLongError;
}
if (0 == segmentor_segment(segmentor, strStn, vctWords)) {
ERROR_LOG("in LTP::wordseg, failed to perform word segment on \"%s\"",
strStn.c_str());
return kWordsegError;
}
if (0 != xml.SetWordsToSentence(vctWords, i)) {
ERROR_LOG("in LTP::wordseg, failed to write segment result to xml");
return kWriteXmlError;
}
}
xml.SetNote(NOTE_WORD);
return 0;
}
