Fix #205

Author: bab2min

include/kiwi/Form.h

@@ -228,17 +228,18 @@ namespace kiwi
 		uint32_t formId = 0;
 		float scoreHash = 0;
 		uint32_t typoId = 0;
+		uint16_t numSpaces = 0;
 		CondVowel leftCond = CondVowel::none;
 
 		TypoForm() = default;
 
-		TypoForm(const std::tuple<uint32_t, float, CondVowel>& p)
-			: formId{ std::get<0>(p) }, scoreHash{ std::get<1>(p) }, leftCond{ std::get<2>(p) }
+		TypoForm(const std::tuple<uint32_t, float, uint16_t, CondVowel>& p)
+			: formId{ std::get<0>(p) }, scoreHash{ std::get<1>(p) }, numSpaces{ std::get<2>(p)}, leftCond{std::get<3>(p)}
 		{
 		}
 
-		TypoForm(uint32_t _formId, float _score = 0, bool _hash = 0, uint32_t _typoId = 0, CondVowel _leftCond = CondVowel::none)
-			: formId{ _formId }, scoreHash{ _hash ? -_score : _score }, typoId{ _typoId }, leftCond{ _leftCond }
+		TypoForm(uint32_t _formId, float _score = 0, bool _hash = 0, uint32_t _typoId = 0, uint16_t _numSpaces = 0, CondVowel _leftCond = CondVowel::none)
+			: formId{ _formId }, scoreHash{ _hash ? -_score : _score }, typoId{ _typoId }, numSpaces{ _numSpaces }, leftCond{ _leftCond }
 		{
 		}
 

src/KTrie.cpp

@@ -50,18 +50,21 @@ namespace kiwi
 		uint32_t start = 0;
 		uint32_t typoId = 0;
 		uint32_t end = 0; // only used in continual typo tolerant mode
+		uint32_t numSpaces = 0;
 
 		FormCandidate(const Form* _form = nullptr, 
 			float _cost = 0, 
 			uint32_t _start = 0, 
 			uint32_t _typoId = 0, 
 			uint32_t _end = 0, 
+			uint32_t _numSpaces = 0,
 			uint32_t = 0)
 			: form{ _form }, 
 			cost{ _cost }, 
 			start{ _start }, 
 			typoId{ _typoId }, 
-			end{ _end }
+			end{ _end },
+			numSpaces{ _numSpaces }
 		{}
 
 		size_t getStartPos(size_t ) const
@@ -86,7 +89,7 @@ namespace kiwi
 
 		size_t getFormSizeWithTypos(const size_t* typoPtrs) const
 		{
-			return typoPtrs[typoId + 1] - typoPtrs[typoId];
+			return typoPtrs[typoId + 1] - typoPtrs[typoId] + numSpaces;
 		}
 
 		bool operator==(const Form* f) const
@@ -100,7 +103,7 @@ namespace kiwi
 	{
 		const Form* form = nullptr;
 
-		FormCandidate(const Form* _form = nullptr, float = 0, uint32_t = 0, uint32_t = 0, uint32_t = 0, uint32_t = 0)
+		FormCandidate(const Form* _form = nullptr, float = 0, uint32_t = 0, uint32_t = 0, uint32_t = 0, uint32_t = 0, uint32_t = 0)
 			: form{ _form }
 		{}
 
@@ -146,8 +149,9 @@ namespace kiwi
 			uint32_t _start = 0,
 			uint32_t _typoId = 0,
 			uint32_t _end = 0,
+			uint32_t _numSpaces = 0,
 			uint32_t _lengthenedSize = 0)
-			: FormCandidate<typoTolerant, continualTypoTolerant, false>{ _form, _cost, _start, _typoId, _end, _lengthenedSize },
+			: FormCandidate<typoTolerant, continualTypoTolerant, false>{ _form, _cost, _start, _typoId, _end, _numSpaces, _lengthenedSize },
 			lengthenedSize{ _lengthenedSize }
 		{}
 
@@ -203,6 +207,7 @@ namespace kiwi
 						startPosition ? startPosition : ((nonSpaces.size() - typoFormSize) * posMultiplier), 
 						tCand->typoId, 
 						endPosition, 
+						tCand->numSpaces,
 						lengthenedSize);
 				}
 				if (tCand[0].hash() != tCand[1].hash()) break;

src/KiwiBuilder.cpp

@@ -2028,7 +2028,7 @@ Kiwi KiwiBuilder::build(const TypoTransformer& typos, float typoCostThreshold) c
 	// 오타 교정이 있는 경우 가능한 모든 오타에 대해 Trie 생성
 	else
 	{
-		using TypoInfo = tuple<uint32_t, float, CondVowel>;
+		using TypoInfo = tuple<uint32_t, float, uint16_t, CondVowel>;
 		UnorderedMap<KString, Vector<TypoInfo>> typoGroup;
 		auto ptypos = typos.prepare();
 		ret.continualTypoCost = ptypos.getContinualTypoCost();
@@ -2043,12 +2043,12 @@ Kiwi KiwiBuilder::build(const TypoTransformer& typos, float typoCostThreshold) c
 				for (auto t : ptypos._generate(f->form, typoCostThreshold))
 				{
 					if (t.leftCond != CondVowel::none && f->vowel != CondVowel::none && t.leftCond != f->vowel) continue;
-					typoGroup[removeSpace(t.str)].emplace_back(f - ret.forms.data(), t.cost, t.leftCond);
+					typoGroup[removeSpace(t.str)].emplace_back(f - ret.forms.data(), t.cost, f->numSpaces, t.leftCond);
 				}
 			}
 			else
 			{
-				typoGroup[removeSpace(f->form)].emplace_back(f - ret.forms.data(), 0, CondVowel::none);
+				typoGroup[removeSpace(f->form)].emplace_back(f - ret.forms.data(), 0, f->numSpaces, CondVowel::none);
 			}
 		}
 
@@ -2107,7 +2107,7 @@ Kiwi KiwiBuilder::build(const TypoTransformer& typos, float typoCostThreshold) c
 			estimatedNodeSize += f->first.size() - commonPrefix;
 			prevForm = &f->first;
 		}
-		ret.typoForms.emplace_back(0, 0, hash);
+		ret.typoForms.emplace_back(0, 0, 0, hash);
 		ret.typoPtrs.emplace_back(ret.typoPool.size());
 		formTrie.reserveMore(estimatedNodeSize);