// integrate postagger into LTP
int LTP::postag(XML4NLP & xml) {
if ( xml.QueryNote(NOTE_POS) ) {
return 0;
}
// dependency
int ret = wordseg(xml);
if (0 != ret) {
ERROR_LOG("in LTP::postag, failed to perform word segment preprocess");
return ret;
}
void * postagger = _resource.GetPostagger();
if (0 == postagger) {
ERROR_LOG("in LTP::postag, failed to init a postagger");
return kPostagError;
}
int stnsNum = xml.CountSentenceInDocument();
if (0 == stnsNum) {
ERROR_LOG("in LTP::postag, number of sentence equals 0");
return kEmptyStringError;
}
for (int i = 0; i < stnsNum; ++i) {
vector<string> vecWord;
vector<string> vecPOS;
xml.GetWordsFromSentence(vecWord, i);
if (0 == vecWord.size()) {
ERROR_LOG("Input sentence is empty.");
return kEmptyStringError;
}
if (vecWord.size() > MAX_WORDS_NUM) {
ERROR_LOG("Input sentence is too long.");
return kSentenceTooLongError;
}
if (0 == postagger_postag(postagger, vecWord, vecPOS)) {
ERROR_LOG("in LTP::postag, failed to perform postag on sent. #%d", i+1);
return kPostagError;
}
if (xml.SetPOSsToSentence(vecPOS, i) != 0) {
ERROR_LOG("in LTP::postag, failed to write postag result to xml");
return kWriteXmlError;
}
}
xml.SetNote(NOTE_POS);
return 0;
}
int LTP::srl(XML4NLP & xml) {
if ( xml.QueryNote(NOTE_SRL) ) return 0;
// dependency
int ret = ner(xml);
if (0 != ret) {
ERROR_LOG("in LTP::srl, failed to perform ner preprocess");
return ret;
}
ret = parser(xml);
if (0 != ret) {
ERROR_LOG("in LTP::srl, failed to perform parsing preprocess");
return ret;
}
int stnsNum = xml.CountSentenceInDocument();
if (stnsNum == 0) {
ERROR_LOG("in LTP::srl, number of sentence equals 0");
return kEmptyStringError;
}
for (int i = 0; i < stnsNum; ++i) {
vector<string> vecWord;
vector<string> vecPOS;
vector<string> vecNE;
vector< pair<int, string> > vecParse;
vector< pair<int, vector< pair<string, pair< int, int > > > > > vecSRLResult;
if (xml.GetWordsFromSentence(vecWord, i) != 0) {
ERROR_LOG("in LTP::ner, failed to get words from xml");
return kReadXmlError;
}
if (xml.GetPOSsFromSentence(vecPOS, i) != 0) {
ERROR_LOG("in LTP::ner, failed to get postags from xml");
return kReadXmlError;
}
if (xml.GetNEsFromSentence(vecNE, i) != 0) {
ERROR_LOG("in LTP::ner, failed to get ner result from xml");
return kReadXmlError;
}
if (xml.GetParsesFromSentence(vecParse, i) != 0) {
ERROR_LOG("in LTP::ner, failed to get parsing result from xml");
return kReadXmlError;
}
if (0 != srl_dosrl(vecWord, vecPOS, vecParse, vecSRLResult)) {
ERROR_LOG("in LTP::srl, failed to perform srl on sent. #%d", i+1);
return kSRLError;
}
int j = 0;
for (; j < vecSRLResult.size(); ++j) {
vector<string> vecType;
vector< pair<int, int> > vecBegEnd;
int k = 0;
for (; k < vecSRLResult[j].second.size(); ++k) {
vecType.push_back(vecSRLResult[j].second[k].first);
vecBegEnd.push_back(vecSRLResult[j].second[k].second);
}
if (0 != xml.SetPredArgToWord(i, vecSRLResult[j].first, vecType, vecBegEnd)) {
return kWriteXmlError;
}
}
}
xml.SetNote(NOTE_SRL);
return 0;
}
int LTP::parser(XML4NLP & xml) {
if ( xml.QueryNote(NOTE_PARSER) ) return 0;
int ret = postag(xml);
if (0 != ret) {
ERROR_LOG("in LTP::parser, failed to perform postag preprocessing");
return ret;
}
void * parser = _resource.GetParser();
if (parser == NULL) {
ERROR_LOG("in LTP::parser, failed to init a parser");
return kParserError;
}
int stnsNum = xml.CountSentenceInDocument();
if (stnsNum == 0) {
ERROR_LOG("in LTP::parser, number of sentences equals 0");
return kEmptyStringError;
}
for (int i = 0; i < stnsNum; ++i) {
std::vector<std::string> vecWord;
std::vector<std::string> vecPOS;
std::vector<int> vecHead;
std::vector<std::string> vecRel;
if (xml.GetWordsFromSentence(vecWord, i) != 0) {
ERROR_LOG("in LTP::parser, failed to get words from xml");
return kReadXmlError;
}
if (xml.GetPOSsFromSentence(vecPOS, i) != 0) {
ERROR_LOG("in LTP::parser, failed to get postags from xml");
return kReadXmlError;
}
if (0 == vecWord.size()) {
ERROR_LOG("Input sentence is empty.");
return kEmptyStringError;
}
if (vecWord.size() > MAX_WORDS_NUM) {
ERROR_LOG("Input sentence is too long.");
return kSentenceTooLongError;
}
if (-1 == parser_parse(parser, vecWord, vecPOS, vecHead, vecRel)) {
ERROR_LOG("in LTP::parser, failed to perform parse on sent. #%d", i+1);
return kParserError;
}
if (0 != xml.SetParsesToSentence(vecHead, vecRel, i)) {
ERROR_LOG("in LTP::parser, failed to write parse result to xml");
return kWriteXmlError;
}
}
xml.SetNote(NOTE_PARSER);
return 0;
}
// perform ner over xml
int LTP::ner(XML4NLP & xml) {
if ( xml.QueryNote(NOTE_NE) ) {
return 0;
}
// dependency
int ret = postag(xml);
if (0 != ret) {
ERROR_LOG("in LTP::ner, failed to perform postag preprocess");
return ret;
}
void * ner = _resource.GetNER();
if (NULL == ner) {
ERROR_LOG("in LTP::ner, failed to init a ner.");
return kNERError;
}
int stnsNum = xml.CountSentenceInDocument();
if (stnsNum == 0) {
ERROR_LOG("in LTP::ner, number of sentence equals 0");
return kEmptyStringError;
}
for (int i = 0; i < stnsNum; ++ i) {
vector<string> vecWord;
vector<string> vecPOS;
vector<string> vecNETag;
if (xml.GetWordsFromSentence(vecWord, i) != 0) {
ERROR_LOG("in LTP::ner, failed to get words from xml");
return kReadXmlError;
}
if (xml.GetPOSsFromSentence(vecPOS, i) != 0) {
ERROR_LOG("in LTP::ner, failed to get postags from xml");
return kNERError;
}
if (0 == vecWord.size()) {
ERROR_LOG("Input sentence is empty.");
return kEmptyStringError;
}
if (vecWord.size() > MAX_WORDS_NUM) {
ERROR_LOG("Input sentence is too long.");
return kSentenceTooLongError;
}
if (0 == ner_recognize(ner, vecWord, vecPOS, vecNETag)) {
ERROR_LOG("in LTP::ner, failed to perform ner on sent. #%d", i+1);
return kNERError;
}
xml.SetNEsToSentence(vecNETag, i);
}
xml.SetNote(NOTE_NE);
return 0;
}
static std::string xml2jsonstr(const XML4NLP & xml, std::string str_type) {
Json::Value root;
int paragraphNum = xml.CountParagraphInDocument();
for (int pid = 0; pid < paragraphNum; ++ pid) {
Json::Value paragraph;
int stnsNum = xml.CountSentenceInParagraph(pid);
for (int sid = 0; sid < stnsNum; ++sid) {
Json::Value sentence;
std::vector<std::string> vecWord;
std::vector<std::string> vecPOS;
std::vector<std::string> vecNETag;
std::vector<std::pair<int, std::string>> vecParse;
//std::vector<std::vector<std::string>> vecSemResult;
std::vector<std::vector<std::pair<int, std::string>>> vecSemResult;
std::vector<std::pair<int, std::vector<std::pair<const char *, std::pair< int, int > > > > > vecSRLResult;
// seg
xml.GetWordsFromSentence(vecWord, pid, sid);
// postag
if (str_type == LTP_SERVICE_NAME_POSTAG
|| str_type == LTP_SERVICE_NAME_NER
|| str_type == LTP_SERVICE_NAME_DEPPARSE
|| str_type == LTP_SERVICE_NAME_SRL
|| str_type == LTP_SERVICE_NAME_ALL) {
xml.GetPOSsFromSentence(vecPOS, pid, sid);
}
// ner
if (str_type == LTP_SERVICE_NAME_NER
|| str_type == LTP_SERVICE_NAME_SRL
|| str_type == LTP_SERVICE_NAME_ALL) {
xml.GetNEsFromSentence(vecNETag, pid, sid);
}
// dp
if (str_type == LTP_SERVICE_NAME_DEPPARSE
|| str_type == LTP_SERVICE_NAME_SRL
|| str_type == LTP_SERVICE_NAME_ALL) {
xml.GetParsesFromSentence(vecParse, pid, sid);
}
// srl
if (str_type == LTP_SERVICE_NAME_SRL
|| str_type == LTP_SERVICE_NAME_ALL) {
// get by word
}
for (int wid = 0; wid < vecWord.size(); ++wid) {
Json::Value word;
word["id"] = wid;
word["cont"] = vecWord[wid];
// postag
if (str_type == LTP_SERVICE_NAME_POSTAG
|| str_type == LTP_SERVICE_NAME_NER
|| str_type == LTP_SERVICE_NAME_DEPPARSE
|| str_type == LTP_SERVICE_NAME_SRL
|| str_type == LTP_SERVICE_NAME_ALL) {
word["pos"] = vecPOS[wid];
}
// ner
if (str_type == LTP_SERVICE_NAME_NER
|| str_type == LTP_SERVICE_NAME_SRL
|| str_type == LTP_SERVICE_NAME_ALL) {
word["ne"] = vecNETag[wid];
}
// dp
if (str_type == LTP_SERVICE_NAME_DEPPARSE
|| str_type == LTP_SERVICE_NAME_SRL
|| str_type == LTP_SERVICE_NAME_ALL) {
word["parent"] = vecParse[wid].first;
word["relate"] = vecParse[wid].second;
}
// srl
if (str_type == LTP_SERVICE_NAME_SRL
|| str_type == LTP_SERVICE_NAME_ALL) {
Json::Value args;
std::vector<std::string> vecType;
std::vector<std::pair<int, int>> vecBegEnd;
xml.GetPredArgToWord(pid, sid, wid, vecType, vecBegEnd);
if (vecType.size() != 0) {
for (int arg_id = 0; arg_id < vecType.size(); ++arg_id) {
Json::Value arg;
arg["id"] = arg_id;
arg["type"] = vecType[arg_id];
arg["beg"] = vecBegEnd[arg_id].first;
arg["end"] = vecBegEnd[arg_id].second;
args.append(arg);
}
} else {
args.resize(0);
}
word["arg"] = args;
}
sentence.append(word);
}
paragraph.append(sentence);
} // sentence
root.append(paragraph);
} // paragraph
return root.toStyledString();
}