mirror of
https://github.com/bab2min/Kiwi.git
synced 2026-06-17 01:54:27 +00:00
336 lines
11 KiB
C++
336 lines
11 KiB
C++
#include "gtest/gtest.h"
|
|
#include <kiwi/Kiwi.h>
|
|
#include "common.h"
|
|
#include "../src/Combiner.h"
|
|
|
|
using namespace kiwi;
|
|
|
|
TEST(KiwiCpp, Combiner)
|
|
{
|
|
cmb::CompiledRule rule;
|
|
{
|
|
cmb::RuleSet crs;
|
|
std::ifstream ifs{ MODEL_PATH "/combiningRule.txt" };
|
|
crs.loadRules(ifs);
|
|
rule = crs.compile();
|
|
}
|
|
|
|
EXPECT_EQ(rule.combine(u"이", POSTag::vcp, u"다", POSTag::ec, CondVowel::vowel)[0], u"다");
|
|
EXPECT_EQ(rule.combine(u"이", POSTag::vcp, u"었", POSTag::ec, CondVowel::vowel)[0], u"였");
|
|
EXPECT_EQ(rule.combine(u"이", POSTag::vcp, u"ᆫ지도", POSTag::ec, CondVowel::vowel)[0], u"ᆫ지도");
|
|
EXPECT_EQ(rule.combine(u"이", POSTag::vcp, u"ᆫ가", POSTag::ec, CondVowel::vowel)[0], u"ᆫ가");
|
|
|
|
EXPECT_EQ(rule.combine(u"ᆯ", POSTag::p, u"ᆯ", POSTag::etm, CondVowel::vowel)[0], u"ᆯ");
|
|
|
|
EXPECT_EQ(rule.combine(u"이르", POSTag::vv, u"어", POSTag::ec)[0], u"이르러");
|
|
EXPECT_EQ(rule.combine(u"푸", POSTag::vv, u"어", POSTag::ec)[0], u"퍼");
|
|
EXPECT_EQ(rule.combine(u"따르", POSTag::vv, u"어", POSTag::ec)[0], u"따라");
|
|
EXPECT_EQ(rule.combine(u"돕", POSTag::vv, u"어", POSTag::ec)[0], u"도와");
|
|
EXPECT_EQ(rule.combine(u"하", POSTag::vv, u"도록", POSTag::ec)[0], u"토록");
|
|
EXPECT_EQ(rule.combine(u"하", POSTag::vv, u"어", POSTag::ec)[0], u"해");
|
|
|
|
EXPECT_EQ(rule.combine(u"타이르", POSTag::p, u"어", POSTag::ec)[0], u"타일러");
|
|
EXPECT_EQ(rule.combine(u"가르", POSTag::p, u"어", POSTag::ec)[0], u"갈라");
|
|
|
|
EXPECT_EQ(rule.combine(u"나", POSTag::np, u"가", POSTag::jks)[0], u"내가");
|
|
|
|
EXPECT_EQ(rule.combine(u"시", POSTag::ep, u"어용", POSTag::ef)[0], u"세용");
|
|
}
|
|
|
|
Kiwi& reuseKiwiInstance()
|
|
{
|
|
static Kiwi kiwi = KiwiBuilder{ MODEL_PATH }.build();
|
|
return kiwi;
|
|
}
|
|
|
|
TEST(KiwiCpp, InitClose)
|
|
{
|
|
Kiwi& kiwi = reuseKiwiInstance();
|
|
}
|
|
|
|
TEST(KiwiCpp, BuilderAddWords)
|
|
{
|
|
KiwiBuilder builder{ MODEL_PATH };
|
|
EXPECT_TRUE(builder.addWord(KWORD, POSTag::nnp, 0.0));
|
|
Kiwi kiwi = builder.build();
|
|
|
|
auto res = kiwi.analyze(KWORD, Match::all);
|
|
EXPECT_EQ(res.first[0].str, KWORD);
|
|
}
|
|
|
|
#define TEST_SENT u"이 예쁜 꽃은 독을 품었지만 진짜 아름다움을 가지고 있어요."
|
|
|
|
TEST(KiwiCpp, AnalyzeWithNone)
|
|
{
|
|
Kiwi kiwi = KiwiBuilder{ MODEL_PATH, 0, BuildOption::none }.build();
|
|
kiwi.analyze(TEST_SENT, Match::all);
|
|
}
|
|
|
|
TEST(KiwiCpp, AnalyzeWithIntegrateAllomorph)
|
|
{
|
|
Kiwi kiwi = KiwiBuilder{ MODEL_PATH, 0, BuildOption::integrateAllomorph }.build();
|
|
kiwi.analyze(TEST_SENT, Match::all);
|
|
}
|
|
|
|
TEST(KiwiCpp, AnalyzeWithLoadDefaultDict)
|
|
{
|
|
Kiwi kiwi = KiwiBuilder{ MODEL_PATH, 0, BuildOption::loadDefaultDict }.build();
|
|
kiwi.analyze(TEST_SENT, Match::all);
|
|
}
|
|
|
|
TEST(KiwiCpp, AnalyzeMultithread)
|
|
{
|
|
auto data = loadTestCorpus();
|
|
std::vector<TokenResult> results;
|
|
Kiwi kiwi = KiwiBuilder{ MODEL_PATH, 2 }.build();
|
|
size_t idx = 0;
|
|
kiwi.analyze(1, [&]() -> std::u16string
|
|
{
|
|
if (idx >= data.size()) return {};
|
|
return utf8To16(data[idx++]);
|
|
}, [&](std::vector<TokenResult>&& res)
|
|
{
|
|
results.emplace_back(std::move(res[0]));
|
|
}, Match::all);
|
|
EXPECT_EQ(data.size(), results.size());
|
|
}
|
|
|
|
TEST(KiwiCpp, AnalyzeError01)
|
|
{
|
|
Kiwi& kiwi = reuseKiwiInstance();
|
|
TokenResult res = kiwi.analyze(u"갔는데", Match::all);
|
|
EXPECT_EQ(res.first[0].str, std::u16string{ u"가" });
|
|
res = kiwi.analyze(u"잤는데", Match::all);
|
|
EXPECT_EQ(res.first[0].str, std::u16string{ u"자" });
|
|
}
|
|
|
|
TEST(KiwiCpp, NormalizeCoda)
|
|
{
|
|
Kiwi& kiwi = reuseKiwiInstance();
|
|
TokenResult res = kiwi.analyze(u"키윜ㅋㅋ", Match::allWithNormalizing);
|
|
EXPECT_EQ(res.first[1].str, std::u16string{ u"ㅋㅋㅋ" });
|
|
res = kiwi.analyze(u"키윟ㅎ", Match::allWithNormalizing);
|
|
EXPECT_EQ(res.first[1].str, std::u16string{ u"ㅎㅎ" });
|
|
res = kiwi.analyze(u"키윅ㄱ", Match::allWithNormalizing);
|
|
EXPECT_EQ(res.first[1].str, std::u16string{ u"ㄱㄱ" });
|
|
res = kiwi.analyze(u"키윈ㄴㄴ", Match::allWithNormalizing);
|
|
EXPECT_EQ(res.first[1].str, std::u16string{ u"ㄴㄴㄴ" });
|
|
res = kiwi.analyze(u"키윊ㅎㅎ", Match::allWithNormalizing);
|
|
EXPECT_EQ(res.first[2].str, std::u16string{ u"ㅎㅎ" });
|
|
res = kiwi.analyze(u"키윍ㄱㄱ", Match::allWithNormalizing);
|
|
EXPECT_EQ(res.first[2].str, std::u16string{ u"ㄱㄱ" });
|
|
}
|
|
|
|
TEST(KiwiCpp, AnalyzeWithWordPosition)
|
|
{
|
|
std::u16string testSentence = u"나 정말 배불렄ㅋㅋ";
|
|
Kiwi kiwi = KiwiBuilder{ MODEL_PATH, 0, BuildOption::none }.build();
|
|
TokenResult tokenResult = kiwi.analyze(testSentence, Match::all);
|
|
std::vector<TokenInfo> tokenInfoList = tokenResult.first;
|
|
|
|
EXPECT_EQ(tokenInfoList[0].wordPosition, 0);
|
|
EXPECT_EQ(tokenInfoList[1].wordPosition, 1);
|
|
EXPECT_EQ(tokenInfoList[2].wordPosition, 2);
|
|
EXPECT_EQ(tokenInfoList[3].wordPosition, 2);
|
|
}
|
|
|
|
TEST(KiwiCpp, Issue57_BuilderAddWord)
|
|
{
|
|
{
|
|
KiwiBuilder builder{ MODEL_PATH };
|
|
builder.addWord(u"울트라리스크", POSTag::nnp, 3.0);
|
|
builder.addWord(u"파일즈", POSTag::nnp, 0.0);
|
|
Kiwi kiwi = builder.build();
|
|
TokenResult res = kiwi.analyze(u"울트라리스크가 뭐야?", Match::all);
|
|
EXPECT_EQ(res.first[0].str, std::u16string{ u"울트라리스크" });
|
|
}
|
|
|
|
{
|
|
KiwiBuilder builder{ MODEL_PATH };
|
|
builder.addWord(u"파일즈", POSTag::nnp, 0.0);
|
|
builder.addWord(u"울트라리스크", POSTag::nnp, 3.0);
|
|
Kiwi kiwi = builder.build();
|
|
TokenResult res = kiwi.analyze(u"울트라리스크가 뭐야?", Match::all);
|
|
EXPECT_EQ(res.first[0].str, std::u16string{ u"울트라리스크" });
|
|
}
|
|
}
|
|
|
|
TEST(KiwiCpp, PositionAndLength)
|
|
{
|
|
Kiwi& kiwi = reuseKiwiInstance();
|
|
|
|
{
|
|
auto tokens = kiwi.analyze(u"자랑했던", Match::all).first;
|
|
EXPECT_EQ(tokens[0].position, 0);
|
|
EXPECT_EQ(tokens[0].length, 2);
|
|
EXPECT_EQ(tokens[1].position, 2);
|
|
EXPECT_EQ(tokens[1].length, 1);
|
|
EXPECT_EQ(tokens[2].position, 2);
|
|
EXPECT_EQ(tokens[2].length, 1);
|
|
EXPECT_EQ(tokens[3].position, 3);
|
|
EXPECT_EQ(tokens[3].length, 1);
|
|
}
|
|
{
|
|
auto tokens = kiwi.analyze(u"이르렀다", Match::all).first;
|
|
EXPECT_EQ(tokens[0].position, 0);
|
|
EXPECT_EQ(tokens[0].length, 2);
|
|
EXPECT_EQ(tokens[1].position, 2);
|
|
EXPECT_EQ(tokens[1].length, 1);
|
|
EXPECT_EQ(tokens[2].position, 3);
|
|
EXPECT_EQ(tokens[2].length, 1);
|
|
}
|
|
{
|
|
auto tokens = kiwi.analyze(u"다다랐다", Match::all).first;
|
|
EXPECT_EQ(tokens[0].position, 0);
|
|
EXPECT_EQ(tokens[0].length, 3);
|
|
EXPECT_EQ(tokens[1].position, 2);
|
|
EXPECT_EQ(tokens[1].length, 1);
|
|
EXPECT_EQ(tokens[2].position, 3);
|
|
EXPECT_EQ(tokens[2].length, 1);
|
|
}
|
|
{
|
|
auto tokens = kiwi.analyze(u"바다다!", Match::all).first;
|
|
EXPECT_EQ(tokens[0].position, 0);
|
|
EXPECT_EQ(tokens[0].length, 2);
|
|
EXPECT_EQ(tokens[1].position, 2);
|
|
EXPECT_EQ(tokens[1].length, 0);
|
|
EXPECT_EQ(tokens[2].position, 2);
|
|
EXPECT_EQ(tokens[2].length, 1);
|
|
}
|
|
}
|
|
|
|
TEST(KiwiCpp, TokenProbs)
|
|
{
|
|
Kiwi& kiwi = reuseKiwiInstance();
|
|
auto tokens = kiwi.analyze(u"넘어갈 뻔 했답니다 강남역 맛집 토끼정", Match::all).first;
|
|
}
|
|
|
|
TEST(KiwiCpp, Issue71_SentenceSplit_u16)
|
|
{
|
|
Kiwi& kiwi = reuseKiwiInstance();
|
|
|
|
std::u16string str = u"다녀온 후기\n\n<강남 토끼정에 다녀왔습니다.> 음식도 맛있었어요 다만 역시 토끼정 본점 답죠?ㅎㅅㅎ 그 맛이 크으.. 아주 맛있었음...! ^^";
|
|
std::vector<std::pair<size_t, size_t>> sentRanges = kiwi.splitIntoSents(str);
|
|
std::vector<std::u16string> sents;
|
|
for (auto& p : sentRanges)
|
|
{
|
|
sents.emplace_back(str.substr(p.first, p.second - p.first));
|
|
}
|
|
|
|
EXPECT_EQ(sents[0], u"다녀온 후기");
|
|
EXPECT_EQ(sents[1], u"<강남 토끼정에 다녀왔습니다.>");
|
|
EXPECT_EQ(sents[2], u"음식도 맛있었어요");
|
|
EXPECT_EQ(sents[3], u"다만 역시 토끼정 본점 답죠?ㅎㅅㅎ");
|
|
EXPECT_EQ(sents[4], u"그 맛이 크으..");
|
|
EXPECT_EQ(sents[5], u"아주 맛있었음...! ^^");
|
|
}
|
|
|
|
TEST(KiwiCpp, Issue71_SentenceSplit_u8)
|
|
{
|
|
Kiwi& kiwi = reuseKiwiInstance();
|
|
|
|
std::string str = u8"다녀온 후기\n\n<강남 토끼정에 다녀왔습니다.> 음식도 맛있었어요 다만 역시 토끼정 본점 답죠?ㅎㅅㅎ 그 맛이 크으.. 아주 맛있었음...! ^^";
|
|
std::vector<std::pair<size_t, size_t>> sentRanges = kiwi.splitIntoSents(str);
|
|
std::vector<std::string> sents;
|
|
for (auto& p : sentRanges)
|
|
{
|
|
sents.emplace_back(str.substr(p.first, p.second - p.first));
|
|
}
|
|
|
|
EXPECT_EQ(sents[0], u8"다녀온 후기");
|
|
EXPECT_EQ(sents[1], u8"<강남 토끼정에 다녀왔습니다.>");
|
|
EXPECT_EQ(sents[2], u8"음식도 맛있었어요");
|
|
EXPECT_EQ(sents[3], u8"다만 역시 토끼정 본점 답죠?ㅎㅅㅎ");
|
|
EXPECT_EQ(sents[4], u8"그 맛이 크으..");
|
|
EXPECT_EQ(sents[5], u8"아주 맛있었음...! ^^");
|
|
}
|
|
|
|
TEST(KiwiCpp, AddRule)
|
|
{
|
|
Kiwi& okiwi = reuseKiwiInstance();
|
|
auto ores = okiwi.analyze(u"했어요! 하잖아요! 할까요?", Match::allWithNormalizing);
|
|
|
|
{
|
|
KiwiBuilder builder{ MODEL_PATH };
|
|
auto inserted = builder.addRule(POSTag::ef, [](std::u16string input)
|
|
{
|
|
if (input.back() == u'요')
|
|
{
|
|
input.back() = u'용';
|
|
}
|
|
return input;
|
|
}, 0);
|
|
|
|
Kiwi kiwi = builder.build();
|
|
auto res = kiwi.analyze(u"했어용! 하잖아용! 할까용?", Match::allWithNormalizing);
|
|
|
|
EXPECT_EQ(ores.second, res.second);
|
|
}
|
|
|
|
{
|
|
KiwiBuilder builder{ MODEL_PATH };
|
|
auto inserted = builder.addRule(POSTag::ef, [](std::u16string input)
|
|
{
|
|
if (input.back() == u'요')
|
|
{
|
|
input.back() = u'용';
|
|
}
|
|
return input;
|
|
}, -1);
|
|
|
|
Kiwi kiwi = builder.build();
|
|
auto res = kiwi.analyze(u"했어용! 하잖아용! 할까용?", Match::allWithNormalizing);
|
|
|
|
EXPECT_FLOAT_EQ(ores.second -3, res.second);
|
|
}
|
|
}
|
|
|
|
TEST(KiwiCpp, AddPreAnalyzedWord)
|
|
{
|
|
Kiwi& okiwi = reuseKiwiInstance();
|
|
auto ores = okiwi.analyze("팅겼어...", Match::allWithNormalizing);
|
|
|
|
KiwiBuilder builder{ MODEL_PATH };
|
|
std::vector<std::pair<std::u16string, POSTag>> analyzed;
|
|
analyzed.emplace_back(u"팅기", POSTag::vv);
|
|
analyzed.emplace_back(u"었", POSTag::ep);
|
|
analyzed.emplace_back(u"어", POSTag::ef);
|
|
|
|
EXPECT_THROW(builder.addPreAnalyzedWord(u"팅겼어", analyzed), UnknownMorphemeException);
|
|
|
|
builder.addWord(u"팅기", POSTag::vv);
|
|
builder.addPreAnalyzedWord(u"팅겼어", analyzed);
|
|
|
|
Kiwi kiwi = builder.build();
|
|
auto res = kiwi.analyze("팅겼어...", Match::allWithNormalizing);
|
|
|
|
EXPECT_EQ(res.first[0].str, u"팅기");
|
|
EXPECT_EQ(res.first[0].tag, POSTag::vv);
|
|
EXPECT_EQ(res.first[1].str, u"었");
|
|
EXPECT_EQ(res.first[1].tag, POSTag::ep);
|
|
EXPECT_EQ(res.first[2].str, u"어");
|
|
EXPECT_EQ(res.first[2].tag, POSTag::ef);
|
|
EXPECT_EQ(res.first[3].str, u"...");
|
|
EXPECT_EQ(res.first[3].tag, POSTag::sf);
|
|
}
|
|
|
|
TEST(KiwiCpp, JoinAffix)
|
|
{
|
|
Kiwi& kiwi = reuseKiwiInstance();
|
|
auto sample = u"사랑스러운 풋사과들아! 배송됐니";
|
|
auto ores = kiwi.analyze(sample, Match::none);
|
|
auto res0 = kiwi.analyze(sample, Match::joinNounPrefix);
|
|
EXPECT_EQ(res0.first[3].str, u"풋사과");
|
|
auto res1 = kiwi.analyze(sample, Match::joinNounSuffix);
|
|
EXPECT_EQ(res1.first[4].str, u"사과들");
|
|
auto res2 = kiwi.analyze(sample, Match::joinNounPrefix | Match::joinNounSuffix);
|
|
EXPECT_EQ(res2.first[3].str, u"풋사과들");
|
|
auto res3 = kiwi.analyze(sample, Match::joinAdjSuffix);
|
|
EXPECT_EQ(res3.first[0].str, u"사랑스럽");
|
|
auto res4 = kiwi.analyze(sample, Match::joinVerbSuffix);
|
|
EXPECT_EQ(res4.first[8].str, u"배송되");
|
|
auto res5 = kiwi.analyze(sample, Match::joinAffix);
|
|
EXPECT_EQ(res5.first[0].str, u"사랑스럽");
|
|
EXPECT_EQ(res5.first[2].str, u"풋사과들");
|
|
EXPECT_EQ(res5.first[5].str, u"배송되");
|
|
}
|