Add more test cases for Z_SIOT

bab2min · Nov 20, 2024 · 8a938f1 · 8a938f1
1 parent 5f2e935
commit 8a938f1
Showing 1 changed file with 35 additions and 0 deletions.
diff --git a/test/test_cpp.cpp b/test/test_cpp.cpp
@@ -988,6 +988,12 @@ TEST(KiwiCpp, ZCoda)
 TEST(KiwiCpp, ZSiot)
 {
 	Kiwi& kiwi = reuseKiwiInstance();
+
+	auto resSplit = kiwi.analyze(u"찰랑찰랑한 머릿결과 볼륨감", Match::allWithNormalizing | Match::splitSaisiot);
+	EXPECT_EQ(resSplit.first.size(), 8);
+	EXPECT_EQ(resSplit.first[3].str, u"머리");
+	EXPECT_EQ(resSplit.first[4].tag, POSTag::z_siot);
+	EXPECT_EQ(resSplit.first[5].str, u"결");
 
 	for (auto s : {u"하굣길", u"만둣국", u"나뭇잎", u"세숫물", u"고춧가루", u"시곗바늘", u"사글셋방"})
 	{
@@ -1014,6 +1020,35 @@ TEST(KiwiCpp, ZSiot)
 	}
 }
 
+TEST(KiwiCpp, ZSiotWithTypo)
+{
+	Kiwi kiwi = KiwiBuilder{ MODEL_PATH, 0, BuildOption::default_, }.build(getDefaultTypoSet(DefaultTypoSet::basicTypoSetWithContinual));
+
+	for (auto s : { u"하굣길", u"만둣국", u"나뭇잎", u"세숫물", u"고춧가루", u"시곗바늘", u"사글셋방" })
+	{
+		auto resNone = kiwi.analyze(s, Match::allWithNormalizing);
+		auto resSplit = kiwi.analyze(s, Match::allWithNormalizing | Match::splitSaisiot);
+		auto resMerge = kiwi.analyze(s, Match::allWithNormalizing | Match::mergeSaisiot);
+		EXPECT_FALSE(std::any_of(resNone.first.begin(), resNone.first.end(), [](const TokenInfo& token) { return token.tag == POSTag::z_siot; }));
+		EXPECT_EQ(resSplit.first.size(), 3);
+		EXPECT_EQ(resSplit.first[0].tag, POSTag::nng);
+		EXPECT_EQ(resSplit.first[1].tag, POSTag::z_siot);
+		EXPECT_EQ(resSplit.first[2].tag, POSTag::nng);
+		EXPECT_EQ(resMerge.first.size(), 1);
+		EXPECT_EQ(resMerge.first[0].tag, POSTag::nng);
+	}
+
+	for (auto s : { u"발렛 파킹", u"미닛" })
+	{
+		auto resNone = kiwi.analyze(s, Match::allWithNormalizing);
+		auto resSplit = kiwi.analyze(s, Match::allWithNormalizing | Match::splitSaisiot);
+		auto resMerge = kiwi.analyze(s, Match::allWithNormalizing | Match::mergeSaisiot);
+		EXPECT_EQ(resNone.second, resSplit.second);
+		EXPECT_EQ(resNone.second, resMerge.second);
+		EXPECT_FALSE(std::any_of(resSplit.first.begin(), resSplit.first.end(), [](const TokenInfo& token) { return token.tag == POSTag::z_siot; }));
+	}
+}
+
 TEST(KiwiCpp, AnalyzeWithWordPosition)
 {
 	std::u16string testSentence = u"나 정말 배불렄ㅋㅋ";