Switched the Exp:: regexes to functions that lazily evaluate their regexes

2025-09-09 12:41:17 +00:00 · 2009-11-04 22:56:59 +00:00
parent 3779e4255d
commit 9a21a3ec8d
6 changed files with 182 additions and 83 deletions
--- a/src/emitterutils.cpp
+++ b/src/emitterutils.cpp
@@ -129,7 +129,7 @@ namespace YAML
 			
 			bool IsValidPlainScalar(const std::string& str, bool inFlow, bool allowOnlyAscii) {
 				// first check the start
-				const RegEx& start = (inFlow ? Exp::PlainScalarInFlow : Exp::PlainScalar);
+				const RegEx& start = (inFlow ? Exp::PlainScalarInFlow() : Exp::PlainScalar());
 				if(!start.Matches(str))
 					return false;
 				
@@ -138,12 +138,12 @@ namespace YAML
 					return false;

 				// then check until something is disallowed
-				const RegEx& disallowed = (inFlow ? Exp::EndScalarInFlow : Exp::EndScalar)
-				                          || (Exp::BlankOrBreak + Exp::Comment)
-				                          || Exp::NotPrintable
-				                          || Exp::Utf8_ByteOrderMark
-				                          || Exp::Break
-				                          || Exp::Tab;
+				const RegEx& disallowed = (inFlow ? Exp::EndScalarInFlow() : Exp::EndScalar())
+				                          || (Exp::BlankOrBreak() + Exp::Comment())
+				                          || Exp::NotPrintable()
+				                          || Exp::Utf8_ByteOrderMark()
+				                          || Exp::Break()
+				                          || Exp::Tab();
 				StringCharSource buffer(str.c_str(), str.size());
 				while(buffer) {
 					if(disallowed.Matches(buffer))
@@ -299,7 +299,7 @@ namespace YAML
 			out << "!<";
 			StringCharSource buffer(str.c_str(), str.size());
 			while(buffer) {
-				int n = Exp::URI.Match(buffer);
+				int n = Exp::URI().Match(buffer);
 				if(n <= 0)
 					return false;

--- a/src/exp.h
+++ b/src/exp.h
@@ -17,54 +17,153 @@ namespace YAML
 	namespace Exp
 	{
 		// misc
-		const RegEx Space = RegEx(' ');
-		const RegEx Tab = RegEx('\t');
-		const RegEx Blank = Space || Tab;
-		const RegEx Break = RegEx('\n') || RegEx("\r\n");
-		const RegEx BlankOrBreak = Blank || Break;
-		const RegEx Digit = RegEx('0', '9');
-		const RegEx Alpha = RegEx('a', 'z') || RegEx('A', 'Z');
-		const RegEx AlphaNumeric = Alpha || Digit;
-		const RegEx Word = AlphaNumeric || RegEx('-');
-		const RegEx Hex = Digit || RegEx('A', 'F') || RegEx('a', 'f');
+		inline const RegEx& Space() {
+			static const RegEx e = RegEx(' ');
+			return e;
+		}
+		inline const RegEx& Tab() {
+			static const RegEx e = RegEx('\t');
+			return e;
+		}
+		inline const RegEx& Blank() {
+			static const RegEx e = Space() || Tab();
+			return e;
+		}
+		inline const RegEx& Break() {
+			static const RegEx e = RegEx('\n') || RegEx("\r\n");
+			return e;
+		}
+		inline const RegEx& BlankOrBreak() {
+			static const RegEx e = Blank() || Break();
+			return e;
+		}
+		inline const RegEx& Digit() {
+			static const RegEx e = RegEx('0', '9');
+			return e;
+		}
+		inline const RegEx& Alpha() {
+			static const RegEx e = RegEx('a', 'z') || RegEx('A', 'Z');
+			return e;
+		}
+		inline const RegEx& AlphaNumeric() {
+			static const RegEx e = Alpha() || Digit();
+			return e;
+		}
+		inline const RegEx& Word() {
+			static const RegEx e = AlphaNumeric() || RegEx('-');
+			return e;
+		}
+		inline const RegEx& Hex() {
+			static const RegEx e = Digit() || RegEx('A', 'F') || RegEx('a', 'f');
+			return e;
+		}
 		// Valid Unicode code points that are not part of c-printable (YAML 1.2, sec. 5.1)
-		const RegEx NotPrintable = RegEx(0) || 
-			RegEx("\x01\x02\x03\x04\x05\x06\x07\x08\x0B\x0C\x7F", REGEX_OR) || 
-			RegEx(0x0E, 0x1F) ||
-			(RegEx('\xC2') + (RegEx('\x80', '\x84') || RegEx('\x86', '\x9F')));
-		const RegEx Utf8_ByteOrderMark = RegEx("\xEF\xBB\xBF");
+		inline const RegEx& NotPrintable() {
+			static const RegEx e = RegEx(0) || 
+				RegEx("\x01\x02\x03\x04\x05\x06\x07\x08\x0B\x0C\x7F", REGEX_OR) || 
+				RegEx(0x0E, 0x1F) ||
+				(RegEx('\xC2') + (RegEx('\x80', '\x84') || RegEx('\x86', '\x9F')));
+			return e;
+		}
+		inline const RegEx& Utf8_ByteOrderMark() {
+			static const RegEx e = RegEx("\xEF\xBB\xBF");
+			return e;
+		}

 		// actual tags

-		const RegEx DocStart = RegEx("---") + (BlankOrBreak || RegEx());
-		const RegEx DocEnd = RegEx("...") + (BlankOrBreak || RegEx());
-		const RegEx DocIndicator = DocStart || DocEnd;
-		const RegEx BlockEntry = RegEx('-') + (BlankOrBreak || RegEx());
-		const RegEx Key = RegEx('?'),
-		            KeyInFlow = RegEx('?') + BlankOrBreak;
-		const RegEx Value = RegEx(':') + (BlankOrBreak || RegEx()),
-		            ValueInFlow = RegEx(':') + (BlankOrBreak || RegEx(",}", REGEX_OR)),
-		            ValueInJSONFlow = RegEx(':');
-		const RegEx Comment = RegEx('#');
-		const RegEx AnchorEnd = RegEx("?:,]}%@`", REGEX_OR) || BlankOrBreak;
-		const RegEx URI = Word || RegEx("#;/?:@&=+$,_.!~*'()[]", REGEX_OR) || (RegEx('%') + Hex + Hex);
-		const RegEx Tag = Word || RegEx("#;/?:@&=+$_.~*'", REGEX_OR) || (RegEx('%') + Hex + Hex);
+		inline const RegEx& DocStart() {
+			static const RegEx e = RegEx("---") + (BlankOrBreak() || RegEx());
+			return e;
+		}
+		inline const RegEx& DocEnd() {
+			static const RegEx e = RegEx("...") + (BlankOrBreak() || RegEx());
+			return e;
+		}
+		inline const RegEx& DocIndicator() {
+			static const RegEx e = DocStart || DocEnd;
+			return e;
+		}
+		inline const RegEx& BlockEntry() {
+			static const RegEx e = RegEx('-') + (BlankOrBreak() || RegEx());
+			return e;
+		}
+		inline const RegEx& Key() {
+			static const RegEx e = RegEx('?');
+			return e;
+		}
+		inline const RegEx& KeyInFlow() {
+			static const RegEx e = RegEx('?') + BlankOrBreak();
+			return e;
+		}
+		inline const RegEx& Value() {
+			static const RegEx e = RegEx(':') + (BlankOrBreak() || RegEx());
+			return e;
+		}
+		inline const RegEx& ValueInFlow() {
+			static const RegEx e = RegEx(':') + (BlankOrBreak() || RegEx(",}", REGEX_OR));
+			return e;
+		}
+		inline const RegEx& ValueInJSONFlow() {
+			static const RegEx e = RegEx(':');
+			return e;
+		}
+		inline const RegEx Comment() {
+			static const RegEx e = RegEx('#');
+			return e;
+		}
+		inline const RegEx& AnchorEnd() {
+			static const RegEx e = RegEx("?:,]}%@`", REGEX_OR) || BlankOrBreak();
+			return e;
+		}
+		inline const RegEx& URI() {
+			static const RegEx e = Word() || RegEx("#;/?:@&=+$,_.!~*'()[]", REGEX_OR) || (RegEx('%') + Hex() + Hex());
+			return e;
+		}
+		inline const RegEx& Tag() {
+			static const RegEx e = Word() || RegEx("#;/?:@&=+$_.~*'", REGEX_OR) || (RegEx('%') + Hex() + Hex());
+			return e;
+		}

 		// Plain scalar rules:
 		// . Cannot start with a blank.
 		// . Can never start with any of , [ ] { } # & * ! | > \' \" % @ `
 		// . In the block context - ? : must be not be followed with a space.
 		// . In the flow context ? is illegal and : and - must not be followed with a space.
-		const RegEx PlainScalar = !(BlankOrBreak || RegEx(",[]{}#&*!|>\'\"%@`", REGEX_OR) || (RegEx("-?:", REGEX_OR) + Blank)),
-		            PlainScalarInFlow = !(BlankOrBreak || RegEx("?,[]{}#&*!|>\'\"%@`", REGEX_OR) || (RegEx("-:", REGEX_OR) + Blank));
-		const RegEx EndScalar = RegEx(':') + (BlankOrBreak || RegEx()),
-		            EndScalarInFlow = (RegEx(':') + (BlankOrBreak || RegEx(",]}", REGEX_OR))) || RegEx(",?[]{}", REGEX_OR);
+		inline const RegEx& PlainScalar() {
+			static const RegEx e = !(BlankOrBreak() || RegEx(",[]{}#&*!|>\'\"%@`", REGEX_OR) || (RegEx("-?:", REGEX_OR) + Blank()));
+			return e;
+		}
+		inline const RegEx& PlainScalarInFlow() {
+			static const RegEx e = !(BlankOrBreak() || RegEx("?,[]{}#&*!|>\'\"%@`", REGEX_OR) || (RegEx("-:", REGEX_OR) + Blank()));
+			return e;
+		}
+		inline const RegEx& EndScalar() {
+			static const RegEx e = RegEx(':') + (BlankOrBreak() || RegEx());
+			return e;
+		}
+		inline const RegEx& EndScalarInFlow() {
+			static const RegEx e = (RegEx(':') + (BlankOrBreak() || RegEx(",]}", REGEX_OR))) || RegEx(",?[]{}", REGEX_OR);
+			return e;
+		}

-		const RegEx EscSingleQuote = RegEx("\'\'");
-		const RegEx EscBreak = RegEx('\\') + Break;
+		inline const RegEx& EscSingleQuote() {
+			static const RegEx e = RegEx("\'\'");
+			return e;
+		}
+		inline const RegEx& EscBreak() {
+			static const RegEx e = RegEx('\\') + Break();
+			return e;
+		}

-		const RegEx ChompIndicator = RegEx("+-", REGEX_OR);
-		const RegEx Chomp = (ChompIndicator + Digit) || (Digit + ChompIndicator) || ChompIndicator || Digit;
+		inline const RegEx& ChompIndicator() {
+			static const RegEx e = RegEx("+-", REGEX_OR);
+			return e;
+		}
+		inline const RegEx& Chomp() {
+			static const RegEx e = (ChompIndicator() + Digit()) || (Digit() + ChompIndicator()) || ChompIndicator() || Digit();
+			return e;
+		}

 		// and some functions
 		std::string Escape(Stream& in);
--- a/src/scanner.cpp
+++ b/src/scanner.cpp
@@ -119,10 +119,10 @@ namespace YAML
 			return ScanDirective();

 		// document token
-		if(INPUT.column() == 0 && Exp::DocStart.Matches(INPUT))
+		if(INPUT.column() == 0 && Exp::DocStart().Matches(INPUT))
 			return ScanDocStart();

-		if(INPUT.column() == 0 && Exp::DocEnd.Matches(INPUT))
+		if(INPUT.column() == 0 && Exp::DocEnd().Matches(INPUT))
 			return ScanDocEnd();

 		// flow start/end/entry
@@ -136,10 +136,10 @@ namespace YAML
 			return ScanFlowEntry();

 		// block/map stuff
-		if(Exp::BlockEntry.Matches(INPUT))
+		if(Exp::BlockEntry().Matches(INPUT))
 			return ScanBlockEntry();

-		if((InBlockContext() ? Exp::Key : Exp::KeyInFlow).Matches(INPUT))
+		if((InBlockContext() ? Exp::Key() : Exp::KeyInFlow()).Matches(INPUT))
 			return ScanKey();

 		if(GetValueRegex().Matches(INPUT))
@@ -161,7 +161,7 @@ namespace YAML
 			return ScanQuotedScalar();

 		// plain scalars
-		if((InBlockContext() ? Exp::PlainScalar : Exp::PlainScalarInFlow).Matches(INPUT))
+		if((InBlockContext() ? Exp::PlainScalar() : Exp::PlainScalarInFlow()).Matches(INPUT))
 			return ScanPlainScalar();

 		// don't know what it is!
@@ -175,24 +175,24 @@ namespace YAML
 		while(1) {
 			// first eat whitespace
 			while(INPUT && IsWhitespaceToBeEaten(INPUT.peek())) {
-				if(InBlockContext() && Exp::Tab.Matches(INPUT))
+				if(InBlockContext() && Exp::Tab().Matches(INPUT))
 					m_simpleKeyAllowed = false;
 				INPUT.eat(1);
 			}

 			// then eat a comment
-			if(Exp::Comment.Matches(INPUT)) {
+			if(Exp::Comment().Matches(INPUT)) {
 				// eat until line break
-				while(INPUT && !Exp::Break.Matches(INPUT))
+				while(INPUT && !Exp::Break().Matches(INPUT))
 					INPUT.eat(1);
 			}

 			// if it's NOT a line break, then we're done!
-			if(!Exp::Break.Matches(INPUT))
+			if(!Exp::Break().Matches(INPUT))
 				break;

 			// otherwise, let's eat the line break and keep going
-			int n = Exp::Break.Match(INPUT);
+			int n = Exp::Break().Match(INPUT);
 			INPUT.eat(n);

 			// oh yeah, and let's get rid of that simple key
@@ -231,9 +231,9 @@ namespace YAML
 	const RegEx& Scanner::GetValueRegex() const
 	{
 		if(InBlockContext())
-			return Exp::Value;
+			return Exp::Value();
 		
-		return m_canBeJSONFlow ? Exp::ValueInJSONFlow : Exp::ValueInFlow;
+		return m_canBeJSONFlow ? Exp::ValueInJSONFlow() : Exp::ValueInFlow();
 	}

 	// StartStream
@@ -323,7 +323,7 @@ namespace YAML
 			const IndentMarker& indent = *m_indents.top();
 			if(indent.column < INPUT.column())
 				break;
-			if(indent.column == INPUT.column() && !(indent.type == IndentMarker::SEQ && !Exp::BlockEntry.Matches(INPUT)))
+			if(indent.column == INPUT.column() && !(indent.type == IndentMarker::SEQ && !Exp::BlockEntry().Matches(INPUT)))
 				break;
 				
 			PopIndent();
--- a/src/scanscalar.cpp
+++ b/src/scanscalar.cpp
@@ -32,12 +32,12 @@ namespace YAML
 			
 			std::size_t lastNonWhitespaceChar = scalar.size();
 			bool escapedNewline = false;
-			while(!params.end.Matches(INPUT) && !Exp::Break.Matches(INPUT)) {
+			while(!params.end.Matches(INPUT) && !Exp::Break().Matches(INPUT)) {
 				if(!INPUT)
 					break;

 				// document indicator?
-				if(INPUT.column() == 0 && Exp::DocIndicator.Matches(INPUT)) {
+				if(INPUT.column() == 0 && Exp::DocIndicator().Matches(INPUT)) {
 					if(params.onDocIndicator == BREAK)
 						break;
 					else if(params.onDocIndicator == THROW)
@@ -48,7 +48,7 @@ namespace YAML
 				pastOpeningBreak = true;

 				// escaped newline? (only if we're escaping on slash)
-				if(params.escape == '\\' && Exp::EscBreak.Matches(INPUT)) {
+				if(params.escape == '\\' && Exp::EscBreak().Matches(INPUT)) {
 					// eat escape character and get out (but preserve trailing whitespace!)
 					INPUT.get();
 					lastNonWhitespaceChar = scalar.size();
@@ -78,7 +78,7 @@ namespace YAML
 			}

 			// doc indicator?
-			if(params.onDocIndicator == BREAK && INPUT.column() == 0 && Exp::DocIndicator.Matches(INPUT))
+			if(params.onDocIndicator == BREAK && INPUT.column() == 0 && Exp::DocIndicator().Matches(INPUT))
 				break;

 			// are we done via character match?
@@ -95,7 +95,7 @@ namespace YAML
 			
 			// ********************************
 			// Phase #2: eat line ending
-			n = Exp::Break.Match(INPUT);
+			n = Exp::Break().Match(INPUT);
 			INPUT.eat(n);

 			// ********************************
@@ -110,7 +110,7 @@ namespace YAML
 				params.indent = std::max(params.indent, INPUT.column());

 			// and then the rest of the whitespace
-			while(Exp::Blank.Matches(INPUT)) {
+			while(Exp::Blank().Matches(INPUT)) {
 				// we check for tabs that masquerade as indentation
 				if(INPUT.peek() == '\t'&& INPUT.column() < params.indent && params.onTabInIndentation == THROW)
 					throw ParserException(INPUT.mark(), ErrorMsg::TAB_IN_INDENTATION);
@@ -122,8 +122,8 @@ namespace YAML
 			}

 			// was this an empty line?
-			bool nextEmptyLine = Exp::Break.Matches(INPUT);
-			bool nextMoreIndented = Exp::Blank.Matches(INPUT);
+			bool nextEmptyLine = Exp::Break().Matches(INPUT);
+			bool nextMoreIndented = Exp::Blank().Matches(INPUT);
 			if(params.fold == FOLD_BLOCK && foldedNewlineCount == 0 && nextEmptyLine)
 				foldedNewlineStartedMoreIndented = moreIndented;

--- a/src/scantag.cpp
+++ b/src/scantag.cpp
@@ -19,7 +19,7 @@ namespace YAML
 				return tag;
 			}
 		
-			int n = Exp::URI.Match(INPUT);
+			int n = Exp::URI().Match(INPUT);
 			if(n <= 0)
 				break;
 			
@@ -44,7 +44,7 @@ namespace YAML

 			int n = 0;
 			if(canBeHandle) {
-				n = Exp::Word.Match(INPUT);
+				n = Exp::Word().Match(INPUT);
 				if(n <= 0) {
 					canBeHandle = false;
 					firstNonWordChar = INPUT.mark();
@@ -52,7 +52,7 @@ namespace YAML
 			}
 			
 			if(!canBeHandle)
-				n = Exp::Tag.Match(INPUT);
+				n = Exp::Tag().Match(INPUT);

 			if(n <= 0)
 				break;
@@ -68,7 +68,7 @@ namespace YAML
 		std::string tag;
 		
 		while(INPUT) {
-			int n = Exp::Tag.Match(INPUT);
+			int n = Exp::Tag().Match(INPUT);
 			if(n <= 0)
 				break;
 			
--- a/src/scantoken.cpp
+++ b/src/scantoken.cpp
@@ -31,22 +31,22 @@ namespace YAML
 		INPUT.eat(1);

 		// read name
-		while(INPUT && !Exp::BlankOrBreak.Matches(INPUT))
+		while(INPUT && !Exp::BlankOrBreak().Matches(INPUT))
 			token.value += INPUT.get();

 		// read parameters
 		while(1) {
 			// first get rid of whitespace
-			while(Exp::Blank.Matches(INPUT))
+			while(Exp::Blank().Matches(INPUT))
 				INPUT.eat(1);

 			// break on newline or comment
-			if(!INPUT || Exp::Break.Matches(INPUT) || Exp::Comment.Matches(INPUT))
+			if(!INPUT || Exp::Break().Matches(INPUT) || Exp::Comment().Matches(INPUT))
 				break;

 			// now read parameter
 			std::string param;
-			while(INPUT && !Exp::BlankOrBreak.Matches(INPUT))
+			while(INPUT && !Exp::BlankOrBreak().Matches(INPUT))
 				param += INPUT.get();

 			token.params.push_back(param);
@@ -238,7 +238,7 @@ namespace YAML
 		alias = (indicator == Keys::Alias);

 		// now eat the content
-		while(Exp::AlphaNumeric.Matches(INPUT))
+		while(Exp::AlphaNumeric().Matches(INPUT))
 			name += INPUT.get();

 		// we need to have read SOMETHING!
@@ -246,7 +246,7 @@ namespace YAML
 			throw ParserException(INPUT.mark(), alias ? ErrorMsg::ALIAS_NOT_FOUND : ErrorMsg::ANCHOR_NOT_FOUND);

 		// and needs to end correctly
-		if(INPUT && !Exp::AnchorEnd.Matches(INPUT))
+		if(INPUT && !Exp::AnchorEnd().Matches(INPUT))
 			throw ParserException(INPUT.mark(), alias ? ErrorMsg::CHAR_IN_ALIAS : ErrorMsg::CHAR_IN_ANCHOR);

 		// and we're done
@@ -297,7 +297,7 @@ namespace YAML

 		// set up the scanning parameters
 		ScanScalarParams params;
-		params.end = (InFlowContext() ? Exp::EndScalarInFlow : Exp::EndScalar) || (Exp::BlankOrBreak + Exp::Comment);
+		params.end = (InFlowContext() ? Exp::EndScalarInFlow() : Exp::EndScalar()) || (Exp::BlankOrBreak() + Exp::Comment());
 		params.eatEnd = false;
 		params.indent = (InFlowContext() ? 0 : GetTopIndent() + 1);
 		params.fold = FOLD_FLOW;
@@ -337,7 +337,7 @@ namespace YAML

 		// setup the scanning parameters
 		ScanScalarParams params;
-		params.end = (single ? RegEx(quote) && !Exp::EscSingleQuote : RegEx(quote));
+		params.end = (single ? RegEx(quote) && !Exp::EscSingleQuote() : RegEx(quote));
 		params.eatEnd = true;
 		params.escape = (single ? '\'' : '\\');
 		params.indent = 0;
@@ -384,14 +384,14 @@ namespace YAML

 		// eat chomping/indentation indicators
 		params.chomp = CLIP;
-		int n = Exp::Chomp.Match(INPUT);
+		int n = Exp::Chomp().Match(INPUT);
 		for(int i=0;i<n;i++) {
 			char ch = INPUT.get();
 			if(ch == '+')
 				params.chomp = KEEP;
 			else if(ch == '-')
 				params.chomp = STRIP;
-			else if(Exp::Digit.Matches(ch)) {
+			else if(Exp::Digit().Matches(ch)) {
 				if(ch == '0')
 					throw ParserException(INPUT.mark(), ErrorMsg::ZERO_INDENT_IN_BLOCK);

@@ -401,16 +401,16 @@ namespace YAML
 		}

 		// now eat whitespace
-		while(Exp::Blank.Matches(INPUT))
+		while(Exp::Blank().Matches(INPUT))
 			INPUT.eat(1);

 		// and comments to the end of the line
-		if(Exp::Comment.Matches(INPUT))
-			while(INPUT && !Exp::Break.Matches(INPUT))
+		if(Exp::Comment().Matches(INPUT))
+			while(INPUT && !Exp::Break().Matches(INPUT))
 				INPUT.eat(1);

 		// if it's not a line break, then we ran into a bad character inline
-		if(INPUT && !Exp::Break.Matches(INPUT))
+		if(INPUT && !Exp::Break().Matches(INPUT))
 			throw ParserException(INPUT.mark(), ErrorMsg::CHAR_IN_BLOCK);

 		// set the initial indentation