Merge pull request #192 from bab2min/dev/issue191

앞쪽의 특수 문자가 형태소와 잘못 결합하여 분석되는 버그 수정
bab2min · Oct 12, 2024 · c1da90c · c1da90c
2 parents c75659c + 5933afc
commit c1da90c
Show file tree

Hide file tree

Showing 2 changed files with 14 additions and 1 deletion.
diff --git a/src/KTrie.cpp b/src/KTrie.cpp
@@ -653,7 +653,8 @@ size_t kiwi::splitByTrie(
 				const auto scanStart = max(endPosMap[nBeginWithMultiplier].first, (uint32_t)1), scanEnd = endPosMap[nBeginWithMultiplier].second;
 				const bool longestMatched = scanStart < scanEnd && any_of(out.begin() + scanStart, out.begin() + scanEnd, [&](const KGraphNode& g)
 				{
-					return nBeginWithMultiplier == g.endPos && lastSpecialEndPos == g.endPos - (g.uform.empty() ? g.form->sizeWithoutSpace() : g.uform.size()) * posMultiplier;
+					const auto start = g.endPos - (g.uform.empty() ? g.form->sizeWithoutSpace() : g.uform.size()) * posMultiplier;
+					return nBeginWithMultiplier == g.endPos && (lastSpecialEndPos == start || specialStartPos == start);
 				});
 
 				// insert unknown form 

diff --git a/test/test_cpp.cpp b/test/test_cpp.cpp
@@ -361,6 +361,18 @@ TEST(KiwiCpp, UserTag)
 	EXPECT_EQ(tokens[7].tag, POSTag::user2);
 }
 
+TEST(KiwiCpp, STagPrefix)
+{
+	Kiwi& kiwi = reuseKiwiInstance();
+	auto res = kiwi.analyze(u"자신있는 지역은 `후분양`으로 나올듯 싶습니다.", Match::allWithNormalizing).first;
+	EXPECT_EQ(res[0].str, u"자신");
+	EXPECT_EQ(res[1].str, u"있");
+	EXPECT_EQ(res[2].str, u"는");
+	EXPECT_EQ(res[3].str, u"지역");
+	EXPECT_EQ(res[4].str, u"은");
+	EXPECT_EQ(res[5].str, u"`");
+}
+
 TEST(KiwiCpp, HSDataset)
 {
 	KiwiBuilder kw{ MODEL_PATH, 0, BuildOption::default_, };