TEST(postag_interface_unittest, test_speed) {
void * engine = postagger_create_postagger("./ltp_data/pos.model");
EXPECT_TRUE(NULL != engine);
std::ifstream ifs("./test_data/unittest/test_data.segmented");
std::string line;
std::string word;
std::vector<std::string> words;
std::vector<std::string> tags;
int nr_tokens = 0;
long start_time = clock();
while (std::getline(ifs, line, '\n')) {
std::stringstream S(line);
words.clear();
tags.clear();
while (S >> word) {
words.push_back(word);
}
postagger_postag(engine, words, tags);
nr_tokens += words.size();
}
double throughput_per_millisecond = (nr_tokens / ((clock() -start_time) / 1000.));
std::cerr << throughput_per_millisecond << std::endl;
postagger_release_postagger(engine);
}
std::vector<std::string> postag(const std::vector<std::string>& words) {
std::vector<std::string> ret;
if (model == NULL) {
std::cerr << "Postagger: Model not loaded!" << std::endl;
} else {
postagger_postag(model, words, ret);
}
return ret;
}
TEST(postag_interface_unittest, test_empty_list) {
void * engine = postagger_create_postagger("./ltp_data/pos.model");
EXPECT_TRUE(NULL != engine);
std::vector<std::string> words;
std::vector<std::string> tags;
int nr_words = postagger_postag(engine, words, tags);
EXPECT_EQ(0, nr_words);
postagger_release_postagger(engine);
}
// 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;
}
TEST(postag_interface_unittest, test_normal) {
void * engine = postagger_create_postagger("./ltp_data/pos.model");
EXPECT_TRUE(NULL != engine);
std::vector<std::string> words;
std::vector<std::string> tags;
for (int i = 0; i < kNumNormalWords; ++ i) {
words.push_back(kNormalWords[i]);
}
int nr_words = postagger_postag(engine, words, tags);
// tagged words should be greater than 4
EXPECT_GT(nr_words, 0);
postagger_release_postagger(engine);
}
TEST(postag_interface_unittest, test_empty_word) {
void * engine = postagger_create_postagger("./ltp_data/pos.model");
EXPECT_TRUE(NULL != engine);
std::vector<std::string> words;
std::vector<std::string> tags;
for (int i = 0; i < kNumNormalWords; ++ i) {
if (i == 2) {
words.push_back("");
} else {
words.push_back(kNormalWords[i]);
}
}
int nr_words = postagger_postag(engine, words, tags);
EXPECT_EQ(0, nr_words);
postagger_release_postagger(engine);
}
