added kiwi::IOException & kiwi::FormatException

bab2min · bab2min · commit a72a3735a65b · 2024-02-15T02:50:42.000+09:00
diff --git a/include/kiwi/Types.h b/include/kiwi/Types.h
@@ -82,6 +82,18 @@ namespace kiwi
 		using std::runtime_error::runtime_error;
 	};
 
+	class IOException : public Exception
+	{
+	public:
+		using Exception::Exception;
+	};
+
+	class FormatException : public Exception
+	{
+	public:
+		using Exception::Exception;
+	};
+
 	class UnicodeException : public Exception
 	{
 	public:
diff --git a/src/FileUtils.cpp b/src/FileUtils.cpp
@@ -23,7 +23,7 @@ namespace kiwi
 		}
 		catch (const ios_base::failure&)
 		{
-			throw Exception{ "Cannot open file : " + filePath };
+			throw IOException{ "Cannot open file : " + filePath };
 		}
 		f.exceptions(exc);
 		return f;
@@ -43,7 +43,7 @@ namespace kiwi
 		}
 		catch (const ios_base::failure&)
 		{
-			throw Exception{ "Cannot open file : " + filePath };
+			throw IOException{ "Cannot open file : " + filePath };
 		}
 		f.exceptions(exc);
 		return f;
diff --git a/src/Form.cpp b/src/Form.cpp
@@ -1,4 +1,4 @@
-#include <cassert>
+﻿#include <cassert>
 #include <algorithm>
 #include <kiwi/Utils.h>
 #include <kiwi/Form.h>
diff --git a/src/KiwiBuilder.cpp b/src/KiwiBuilder.cpp
@@ -142,7 +142,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem
 		auto fields = split(wstr, u'\t');
 		if (fields.size() < 3)
 		{
-			throw Exception{ "wrong line: " + line };
+			throw FormatException{ "wrong line: " + line };
 		}
 
 		auto form = normalizeHangul(fields[0]);
@@ -180,7 +180,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem
 				}
 				else if (f == u"non_vowel")
 				{
-					if (cvowel != CondVowel::none) throw Exception{ "wrong line: " + line };
+					if (cvowel != CondVowel::none) throw FormatException{ "wrong line: " + line };
 					cvowel = CondVowel::non_vowel;
 					if (i + 1 < fields.size())
 					{
@@ -189,7 +189,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem
 				}
 				else if (f == u"vocalic")
 				{
-					if (cvowel != CondVowel::none) throw Exception{ "wrong line: " + line };
+					if (cvowel != CondVowel::none) throw FormatException{ "wrong line: " + line };
 					cvowel = CondVowel::vocalic;
 					if (i + 1 < fields.size())
 					{
@@ -198,18 +198,18 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem
 				}
 				else if (f == u"non_adj")
 				{
-					if (cpolar != CondPolarity::none) throw Exception{ "wrong line: " + line };
+					if (cpolar != CondPolarity::none) throw FormatException{ "wrong line: " + line };
 					cpolar = CondPolarity::non_adj;
 				}
 				else if (f.starts_with(u"complex "))
 				{
-					if (complex) throw Exception{ "wrong line: " + line };
+					if (complex) throw FormatException{ "wrong line: " + line };
 					complex = true;
 					complexChunks.emplace(make_pair(form, tag), f.substr(8).to_string());
 				}
 				else if (f[0] == u'=')
 				{
-					if (!origMorphemeOfAlias.empty()) throw Exception{ "wrong line: " + line };
+					if (!origMorphemeOfAlias.empty()) throw FormatException{ "wrong line: " + line };
 					if (f[1] == u'=')
 					{
 						origMorphemeOfAlias = normalizeHangul(f.substr(2));
@@ -233,7 +233,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem
 				}
 				else if (f[0] == u'~')
 				{
-					if (!origMorphemeOfAlias.empty()) throw Exception{ "wrong line: " + line };
+					if (!origMorphemeOfAlias.empty()) throw FormatException{ "wrong line: " + line };
 					if (f.find('/') == f.npos)
 					{
 						origMorphemeOfAlias = normalizeHangul(f.substr(1));
@@ -248,7 +248,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem
 				}
 				else if (f[0] == u'-')
 				{
-					if (altWeight < 0) throw Exception{ "wrong line: " + line };
+					if (altWeight < 0) throw FormatException{ "wrong line: " + line };
 					altWeight = stof(f.begin(), f.end());
 				}
 				else if (f[0] == '>')
@@ -270,7 +270,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem
 				}
 				else
 				{
-					throw Exception{ "wrong line: " + line };
+					throw FormatException{ "wrong line: " + line };
 				}
 			}
 		}
@@ -306,12 +306,12 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem
 		
 		if (it != morphMap.end() && it2 != morphMap.end())
 		{
-			throw Exception{ "ambiguous base morpheme: " + utf16To8(p.origForm) + "/" + tagToString(clearIrregular(p.origTag)) };
+			throw FormatException{ "ambiguous base morpheme: " + utf16To8(p.origForm) + "/" + tagToString(clearIrregular(p.origTag)) };
 		}
 		it = (it == morphMap.end()) ? it2 : it;
 		if (it == morphMap.end())
 		{
-			throw Exception{ "cannot find base morpheme: " + utf16To8(p.origForm) + "/" + tagToString(p.origTag) };
+				throw FormatException{ "cannot find base morpheme: " + utf16To8(p.origForm) + "/" + tagToString(p.origTag) };
 		}
 		p.origMorphId = it->second.first;
 		if (!p.addAlias) continue;
@@ -385,27 +385,27 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem
 
 		if (fd.back().size() != (fd.size() - 1) * 2)
 		{
-			throw Exception{ "wrong position information : " + utf16To8(fd[0]) + " " + utf16To8(fd.back())};
+			throw FormatException{ "wrong position information : " + utf16To8(fd[0]) + " " + utf16To8(fd.back())};
 		}
 		auto posMap = normalizeHangulWithPosition(joinHangul(it->first.first)).second;
 		for (size_t i = 0; i < fd.size() - 1; ++i)
 		{
 			auto f = split(fd[i], u'/');
-			if (f.size() != 2) throw Exception{ "wrong format of morpheme : " + utf16To8(fd[i]) };
+			if (f.size() != 2) throw FormatException{ "wrong format of morpheme : " + utf16To8(fd[i]) };
 			auto norm = normalizeHangul(f[0]);
 			auto tag = toPOSTag(f[1]);
 			auto it = morphMap.find(make_pair(norm, tag));
 			if (it == morphMap.end())
 			{
-				throw Exception{ "cannot find morpheme : " + utf16To8(fd[i]) };
+				throw FormatException{ "cannot find morpheme : " + utf16To8(fd[i]) };
 			}
 			size_t lmId = it->second.first;
 			if (!morphemes[lmId].kform)
 			{
 				auto it = longTailMap.find(make_pair(norm, tag));
 				if (it == longTailMap.end())
 				{
-					throw Exception{ "cannot find morpheme : " + utf16To8(fd[i]) };
+					throw FormatException{ "cannot find morpheme : " + utf16To8(fd[i]) };
 				}
 				lmId = it->second;
 			}
@@ -1514,7 +1514,7 @@ size_t KiwiBuilder::loadDictionary(const string& dictPath)
 		size_t fieldSize = split(wstr, u'\t', fields.begin(), 2) - fields.begin();
 		if (fieldSize < 2)
 		{
-			throw Exception("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);
+			throw FormatException("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);
 		}
 
 		while (!fields[0].empty() && fields[0][0] == ' ') fields[0] = fields[0].substr(1);
@@ -1536,26 +1536,26 @@ size_t KiwiBuilder::loadDictionary(const string& dictPath)
 				size_t p = m.rfind(u'/');
 				if (p == m.npos)
 				{
-					throw Exception("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);
+					throw FormatException("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);
 				}
 				auto pos = toPOSTag(m.substr(p + 1));
 				if (pos == POSTag::max)
 				{
-					throw Exception("[loadUserDictionary] Unknown Tag '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));
+					throw FormatException("[loadUserDictionary] Unknown Tag '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));
 				}
 				morphemes.emplace_back(m.substr(0, p), pos);
 			}
 
 			if (fields[0].empty())
 			{
-				throw Exception("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);
+				throw FormatException("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);
 			}
 
 			if (fields[0].back() == '$')
 			{
 				if (morphemes.size() > 1)
 				{
-					throw Exception("[loadUserDictionary] Replace rule cannot have 2 or more forms '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));
+					throw FormatException("[loadUserDictionary] Replace rule cannot have 2 or more forms '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));
 				}
 
 				auto suffix = fields[0].substr(0, fields[0].size() - 1);
@@ -1583,7 +1583,7 @@ size_t KiwiBuilder::loadDictionary(const string& dictPath)
 			auto pos = toPOSTag(fields[1]);
 			if (pos == POSTag::max)
 			{
-				throw Exception("[loadUserDictionary] Unknown Tag '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));
+				throw FormatException("[loadUserDictionary] Unknown Tag '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));
 			}
 			addedCnt += addWord(fields[0], pos, score).second ? 1 : 0;
 		}

Original file line number	Diff line number	Diff line change
`@@ -23,7 +23,7 @@ namespace kiwi`
`23`	`23`	`}`
`24`	`24`	`catch (const ios_base::failure&)`
`25`	`25`	`{`
`26`		`- throw Exception{ "Cannot open file : " + filePath };`
	`26`	`+ throw IOException{ "Cannot open file : " + filePath };`
`27`	`27`	`}`
`28`	`28`	`f.exceptions(exc);`
`29`	`29`	`return f;`
`@@ -43,7 +43,7 @@ namespace kiwi`
`43`	`43`	`}`
`44`	`44`	`catch (const ios_base::failure&)`
`45`	`45`	`{`
`46`		`- throw Exception{ "Cannot open file : " + filePath };`
	`46`	`+ throw IOException{ "Cannot open file : " + filePath };`
`47`	`47`	`}`
`48`	`48`	`f.exceptions(exc);`
`49`	`49`	`return f;`
Original file line number	Diff line number	Diff line change
`@@ -1,4 +1,4 @@`
`1`		`-#include <cassert>`
	`1`	`+#include <cassert>`
`2`	`2`	`#include <algorithm>`
`3`	`3`	`#include <kiwi/Utils.h>`
`4`	`4`	`#include <kiwi/Form.h>`
Original file line number	Diff line number	Diff line change
`@@ -142,7 +142,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem`
`142`	`142`	`auto fields = split(wstr, u'\t');`
`143`	`143`	`if (fields.size() < 3)`
`144`	`144`	`{`
`145`		`- throw Exception{ "wrong line: " + line };`
	`145`	`+ throw FormatException{ "wrong line: " + line };`
`146`	`146`	`}`
`147`	`147`
`148`	`148`	`auto form = normalizeHangul(fields[0]);`
`@@ -180,7 +180,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem`
`180`	`180`	`}`
`181`	`181`	`else if (f == u"non_vowel")`
`182`	`182`	`{`
`183`		`- if (cvowel != CondVowel::none) throw Exception{ "wrong line: " + line };`
	`183`	`+ if (cvowel != CondVowel::none) throw FormatException{ "wrong line: " + line };`
`184`	`184`	`cvowel = CondVowel::non_vowel;`
`185`	`185`	`if (i + 1 < fields.size())`
`186`	`186`	`{`
`@@ -189,7 +189,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem`
`189`	`189`	`}`
`190`	`190`	`else if (f == u"vocalic")`
`191`	`191`	`{`
`192`		`- if (cvowel != CondVowel::none) throw Exception{ "wrong line: " + line };`
	`192`	`+ if (cvowel != CondVowel::none) throw FormatException{ "wrong line: " + line };`
`193`	`193`	`cvowel = CondVowel::vocalic;`
`194`	`194`	`if (i + 1 < fields.size())`
`195`	`195`	`{`
`@@ -198,18 +198,18 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem`
`198`	`198`	`}`
`199`	`199`	`else if (f == u"non_adj")`
`200`	`200`	`{`
`201`		`- if (cpolar != CondPolarity::none) throw Exception{ "wrong line: " + line };`
	`201`	`+ if (cpolar != CondPolarity::none) throw FormatException{ "wrong line: " + line };`
`202`	`202`	`cpolar = CondPolarity::non_adj;`
`203`	`203`	`}`
`204`	`204`	`else if (f.starts_with(u"complex "))`
`205`	`205`	`{`
`206`		`- if (complex) throw Exception{ "wrong line: " + line };`
	`206`	`+ if (complex) throw FormatException{ "wrong line: " + line };`
`207`	`207`	`complex = true;`
`208`	`208`	`complexChunks.emplace(make_pair(form, tag), f.substr(8).to_string());`
`209`	`209`	`}`
`210`	`210`	`else if (f[0] == u'=')`
`211`	`211`	`{`
`212`		`- if (!origMorphemeOfAlias.empty()) throw Exception{ "wrong line: " + line };`
	`212`	`+ if (!origMorphemeOfAlias.empty()) throw FormatException{ "wrong line: " + line };`
`213`	`213`	`if (f[1] == u'=')`
`214`	`214`	`{`
`215`	`215`	`origMorphemeOfAlias = normalizeHangul(f.substr(2));`
`@@ -233,7 +233,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem`
`233`	`233`	`}`
`234`	`234`	`else if (f[0] == u'~')`
`235`	`235`	`{`
`236`		`- if (!origMorphemeOfAlias.empty()) throw Exception{ "wrong line: " + line };`
	`236`	`+ if (!origMorphemeOfAlias.empty()) throw FormatException{ "wrong line: " + line };`
`237`	`237`	`if (f.find('/') == f.npos)`
`238`	`238`	`{`
`239`	`239`	`origMorphemeOfAlias = normalizeHangul(f.substr(1));`
`@@ -248,7 +248,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem`
`248`	`248`	`}`
`249`	`249`	`else if (f[0] == u'-')`
`250`	`250`	`{`
`251`		`- if (altWeight < 0) throw Exception{ "wrong line: " + line };`
	`251`	`+ if (altWeight < 0) throw FormatException{ "wrong line: " + line };`
`252`	`252`	`altWeight = stof(f.begin(), f.end());`
`253`	`253`	`}`
`254`	`254`	`else if (f[0] == '>')`
`@@ -270,7 +270,7 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem`
`270`	`270`	`}`
`271`	`271`	`else`
`272`	`272`	`{`
`273`		`- throw Exception{ "wrong line: " + line };`
	`273`	`+ throw FormatException{ "wrong line: " + line };`
`274`	`274`	`}`
`275`	`275`	`}`
`276`	`276`	`}`
`@@ -306,12 +306,12 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem`
`306`	`306`
`307`	`307`	`if (it != morphMap.end() && it2 != morphMap.end())`
`308`	`308`	`{`
`309`		`- throw Exception{ "ambiguous base morpheme: " + utf16To8(p.origForm) + "/" + tagToString(clearIrregular(p.origTag)) };`
	`309`	`+ throw FormatException{ "ambiguous base morpheme: " + utf16To8(p.origForm) + "/" + tagToString(clearIrregular(p.origTag)) };`
`310`	`310`	`}`
`311`	`311`	`it = (it == morphMap.end()) ? it2 : it;`
`312`	`312`	`if (it == morphMap.end())`
`313`	`313`	`{`
`314`		`- throw Exception{ "cannot find base morpheme: " + utf16To8(p.origForm) + "/" + tagToString(p.origTag) };`
	`314`	`+ throw FormatException{ "cannot find base morpheme: " + utf16To8(p.origForm) + "/" + tagToString(p.origTag) };`
`315`	`315`	`}`
`316`	`316`	`p.origMorphId = it->second.first;`
`317`	`317`	`if (!p.addAlias) continue;`
`@@ -385,27 +385,27 @@ auto KiwiBuilder::loadMorphemesFromTxt(std::istream& is, Fn&& filter) -> Morphem`
`385`	`385`
`386`	`386`	`if (fd.back().size() != (fd.size() - 1) * 2)`
`387`	`387`	`{`
`388`		`- throw Exception{ "wrong position information : " + utf16To8(fd[0]) + " " + utf16To8(fd.back())};`
	`388`	`+ throw FormatException{ "wrong position information : " + utf16To8(fd[0]) + " " + utf16To8(fd.back())};`
`389`	`389`	`}`
`390`	`390`	`auto posMap = normalizeHangulWithPosition(joinHangul(it->first.first)).second;`
`391`	`391`	`for (size_t i = 0; i < fd.size() - 1; ++i)`
`392`	`392`	`{`
`393`	`393`	`auto f = split(fd[i], u'/');`
`394`		`- if (f.size() != 2) throw Exception{ "wrong format of morpheme : " + utf16To8(fd[i]) };`
	`394`	`+ if (f.size() != 2) throw FormatException{ "wrong format of morpheme : " + utf16To8(fd[i]) };`
`395`	`395`	`auto norm = normalizeHangul(f[0]);`
`396`	`396`	`auto tag = toPOSTag(f[1]);`
`397`	`397`	`auto it = morphMap.find(make_pair(norm, tag));`
`398`	`398`	`if (it == morphMap.end())`
`399`	`399`	`{`
`400`		`- throw Exception{ "cannot find morpheme : " + utf16To8(fd[i]) };`
	`400`	`+ throw FormatException{ "cannot find morpheme : " + utf16To8(fd[i]) };`
`401`	`401`	`}`
`402`	`402`	`size_t lmId = it->second.first;`
`403`	`403`	`if (!morphemes[lmId].kform)`
`404`	`404`	`{`
`405`	`405`	`auto it = longTailMap.find(make_pair(norm, tag));`
`406`	`406`	`if (it == longTailMap.end())`
`407`	`407`	`{`
`408`		`- throw Exception{ "cannot find morpheme : " + utf16To8(fd[i]) };`
	`408`	`+ throw FormatException{ "cannot find morpheme : " + utf16To8(fd[i]) };`
`409`	`409`	`}`
`410`	`410`	`lmId = it->second;`
`411`	`411`	`}`
`@@ -1514,7 +1514,7 @@ size_t KiwiBuilder::loadDictionary(const string& dictPath)`
`1514`	`1514`	`size_t fieldSize = split(wstr, u'\t', fields.begin(), 2) - fields.begin();`
`1515`	`1515`	`if (fieldSize < 2)`
`1516`	`1516`	`{`
`1517`		`- throw Exception("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);`
	`1517`	`+ throw FormatException("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);`
`1518`	`1518`	`}`
`1519`	`1519`
`1520`	`1520`	`while (!fields[0].empty() && fields[0][0] == ' ') fields[0] = fields[0].substr(1);`
`@@ -1536,26 +1536,26 @@ size_t KiwiBuilder::loadDictionary(const string& dictPath)`
`1536`	`1536`	`size_t p = m.rfind(u'/');`
`1537`	`1537`	`if (p == m.npos)`
`1538`	`1538`	`{`
`1539`		`- throw Exception("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);`
	`1539`	`+ throw FormatException("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);`
`1540`	`1540`	`}`
`1541`	`1541`	`auto pos = toPOSTag(m.substr(p + 1));`
`1542`	`1542`	`if (pos == POSTag::max)`
`1543`	`1543`	`{`
`1544`		`- throw Exception("[loadUserDictionary] Unknown Tag '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));`
	`1544`	`+ throw FormatException("[loadUserDictionary] Unknown Tag '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));`
`1545`	`1545`	`}`
`1546`	`1546`	`morphemes.emplace_back(m.substr(0, p), pos);`
`1547`	`1547`	`}`
`1548`	`1548`
`1549`	`1549`	`if (fields[0].empty())`
`1550`	`1550`	`{`
`1551`		`- throw Exception("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);`
	`1551`	`+ throw FormatException("[loadUserDictionary] Wrong dictionary format at line " + to_string(lineNo) + " : " + line);`
`1552`	`1552`	`}`
`1553`	`1553`
`1554`	`1554`	`if (fields[0].back() == '$')`
`1555`	`1555`	`{`
`1556`	`1556`	`if (morphemes.size() > 1)`
`1557`	`1557`	`{`
`1558`		`- throw Exception("[loadUserDictionary] Replace rule cannot have 2 or more forms '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));`
	`1558`	`+ throw FormatException("[loadUserDictionary] Replace rule cannot have 2 or more forms '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));`
`1559`	`1559`	`}`
`1560`	`1560`
`1561`	`1561`	`auto suffix = fields[0].substr(0, fields[0].size() - 1);`
`@@ -1583,7 +1583,7 @@ size_t KiwiBuilder::loadDictionary(const string& dictPath)`
`1583`	`1583`	`auto pos = toPOSTag(fields[1]);`
`1584`	`1584`	`if (pos == POSTag::max)`
`1585`	`1585`	`{`
`1586`		`- throw Exception("[loadUserDictionary] Unknown Tag '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));`
	`1586`	`+ throw FormatException("[loadUserDictionary] Unknown Tag '" + utf16To8(fields[1]) + "' at line " + to_string(lineNo));`
`1587`	`1587`	`}`
`1588`	`1588`	`addedCnt += addWord(fields[0], pos, score).second ? 1 : 0;`
`1589`	`1589`	`}`