From de29068110e3aee790661ca070073d9b60354c93 Mon Sep 17 00:00:00 2001
From: Jesse Beder <jbeder+github@gmail.com>
Date: Fri, 27 Jun 2008 19:07:30 +0000
Subject: [PATCH] Added stream input to the regular expressions, greatly
 simplifying the usage (in particular, we no longer have to specify the number
 of characters to be checked).

---
 regex.cpp   | 214 +++++++++++++++++++++++++++++++++++++++++-----------
 regex.h     |  45 ++++++++++-
 scanner.cpp |  56 +++++---------
 scanner.h   |   2 -
 4 files changed, 232 insertions(+), 85 deletions(-)

diff --git a/regex.cpp b/regex.cpp
index a2907f5..f7830b7 100644
--- a/regex.cpp
+++ b/regex.cpp
@@ -2,30 +2,60 @@
 
 namespace YAML
 {
-	RegEx::RegEx(REGEX_OP op): m_op(op)
+	RegEx::RegEx(REGEX_OP op): m_op(op), m_pOp(0)
 	{
+		SetOp();
 	}
 
-	RegEx::RegEx(): m_op(REGEX_EMPTY)
+	RegEx::RegEx(const RegEx& rhs): m_pOp(0)
 	{
+		m_op = rhs.m_op;
+		m_a = rhs.m_a;
+		m_z = rhs.m_z;
+		m_params = rhs.m_params;
+
+		SetOp();
 	}
 
-	RegEx::RegEx(char ch): m_op(REGEX_MATCH), m_a(ch)
+	RegEx::RegEx(): m_op(REGEX_EMPTY), m_pOp(0)
 	{
+		SetOp();
 	}
 
-	RegEx::RegEx(char a, char z): m_op(REGEX_RANGE), m_a(a), m_z(z)
+	RegEx::RegEx(char ch): m_op(REGEX_MATCH), m_pOp(0), m_a(ch)
 	{
+		SetOp();
 	}
 
-	RegEx::RegEx(const std::string& str, REGEX_OP op): m_op(op)
+	RegEx::RegEx(char a, char z): m_op(REGEX_RANGE), m_pOp(0), m_a(a), m_z(z)
+	{
+		SetOp();
+	}
+
+	RegEx::RegEx(const std::string& str, REGEX_OP op): m_op(op), m_pOp(0)
 	{
 		for(unsigned i=0;i<str.size();i++)
 			m_params.push_back(RegEx(str[0]));
+
+		SetOp();
 	}
 
 	RegEx::~RegEx()
 	{
+		delete m_pOp;
+	}
+
+	void RegEx::SetOp()
+	{
+		delete m_pOp;
+		m_pOp = 0;
+		switch(m_op) {
+			case REGEX_MATCH: m_pOp = new MatchOperator; break;
+			case REGEX_RANGE: m_pOp = new RangeOperator; break;
+			case REGEX_OR: m_pOp = new OrOperator; break;
+			case REGEX_NOT: m_pOp = new NotOperator; break;
+			case REGEX_SEQ: m_pOp = new SeqOperator; break;
+		}
 	}
 
 	bool RegEx::Matches(char ch) const
@@ -40,6 +70,11 @@ namespace YAML
 		return Match(str) >= 0;
 	}
 
+	bool RegEx::Matches(std::istream& in) const
+	{
+		return Match(in) >= 0;
+	}
+
 	// Match
 	// . Matches the given string against this regular expression.
 	// . Returns the number of characters matched.
@@ -49,44 +84,36 @@ namespace YAML
 	//   but that of course matches zero characters).
 	int RegEx::Match(const std::string& str) const
 	{
-		switch(m_op) {
-			case REGEX_EMPTY:
-				if(str.empty())
-					return 0;
-				return -1;
-			case REGEX_MATCH:
-				if(str.empty() || str[0] != m_a)
-					return -1;
-				return 1;
-			case REGEX_RANGE:
-				if(str.empty() || m_a > str[0] || m_z < str[0])
-					return -1;
-				return 1;
-			case REGEX_NOT:
-				if(m_params.empty())
-					return false;
-				if(m_params[0].Match(str) >= 0)
-					return -1;
-				return 1;
-			case REGEX_OR:
-				for(unsigned i=0;i<m_params.size();i++) {
-					int n = m_params[i].Match(str);
-					if(n >= 0)
-						return n;
-				}
-				return -1;
-			case REGEX_SEQ:
-				int offset = 0;
-				for(unsigned i=0;i<m_params.size();i++) {
-					int n = m_params[i].Match(str.substr(offset));
-					if(n == -1)
-						return -1;
-					offset += n;
-				}
-				return offset;
-		}
+		if(!m_pOp)
+			return -1;
 
-		return -1;
+		return m_pOp->Match(str, *this);
+
+			//case REGEX_EMPTY:
+			//	if(str.empty())
+			//		return 0;
+			//	return -1;
+	}
+
+	// Match
+	// . The stream version does the same thing as the string version;
+	//   REMEMBER that we only match from the start of the stream!
+	// . Note: the istream is not a const reference, but we guarantee
+	//   that the pointer will be in the same spot, and we'll clear its
+	//   flags before we end.
+	int RegEx::Match(std::istream& in) const
+	{
+		if(!m_pOp)
+			return -1;
+
+		int pos = in.tellg();
+		int ret = m_pOp->Match(in, *this);
+
+		// reset input stream!
+		in.clear();
+		in.seekg(pos);
+
+		return ret;
 	}
 
 	RegEx operator ! (const RegEx& ex)
@@ -111,4 +138,107 @@ namespace YAML
 		ret.m_params.push_back(ex2);
 		return ret;
 	}
+
+	//////////////////////////////////////////////////////////////////////////////
+	// Operators
+
+	// MatchOperator
+	int RegEx::MatchOperator::Match(const std::string& str, const RegEx& regex) const
+	{
+		if(str.empty() || str[0] != regex.m_a)
+			return -1;
+		return 1;
+	}
+
+	
+	int RegEx::MatchOperator::Match(std::istream& in, const RegEx& regex) const
+	{
+		if(!in || in.peek() != regex.m_a)
+			return -1;
+		return 1;
+	}
+
+	// RangeOperator
+	int RegEx::RangeOperator::Match(const std::string& str, const RegEx& regex) const
+	{
+		if(str.empty() || regex.m_a > str[0] || regex.m_z < str[0])
+			return -1;
+		return 1;
+	}
+
+	int RegEx::RangeOperator::Match(std::istream& in, const RegEx& regex) const
+	{
+		if(!in || regex.m_a > in.peek() || regex.m_z < in.peek())
+			return -1;
+		return 1;
+	}
+
+	// OrOperator
+	int RegEx::OrOperator::Match(const std::string& str, const RegEx& regex) const
+	{
+		for(unsigned i=0;i<regex.m_params.size();i++) {
+			int n = regex.m_params[i].Match(str);
+			if(n >= 0)
+				return n;
+		}
+		return -1;
+	}
+
+	int RegEx::OrOperator::Match(std::istream& in, const RegEx& regex) const
+	{
+		for(unsigned i=0;i<regex.m_params.size();i++) {
+			int n = regex.m_params[i].Match(in);
+			if(n >= 0)
+				return n;
+		}
+		return -1;
+	}
+
+	// NotOperator
+	int RegEx::NotOperator::Match(const std::string& str, const RegEx& regex) const
+	{
+		if(regex.m_params.empty())
+			return -1;
+		if(regex.m_params[0].Match(str) >= 0)
+			return -1;
+		return 1;
+	}
+
+	int RegEx::NotOperator::Match(std::istream& in, const RegEx& regex) const
+	{
+		if(regex.m_params.empty())
+			return -1;
+		if(regex.m_params[0].Match(in) >= 0)
+			return -1;
+		return 1;
+	}
+
+	// SeqOperator
+	int RegEx::SeqOperator::Match(const std::string& str, const RegEx& regex) const
+	{
+		int offset = 0;
+		for(unsigned i=0;i<regex.m_params.size();i++) {
+			int n = regex.m_params[i].Match(str.substr(offset));
+			if(n == -1)
+				return -1;
+			offset += n;
+		}
+		return offset;
+	}
+	
+	int RegEx::SeqOperator::Match(std::istream& in, const RegEx& regex) const
+	{
+		int offset = 0;
+		for(unsigned i=0;i<regex.m_params.size();i++) {
+			int n = regex.m_params[i].Match(in);
+			if(n == -1)
+				return -1;
+
+			offset += n;
+			in.seekg(n, std::ios_base::cur);
+		}
+
+		return offset;
+	}
 }
+
diff --git a/regex.h b/regex.h
index 08d7acc..b063d1c 100644
--- a/regex.h
+++ b/regex.h
@@ -2,6 +2,7 @@
 
 #include <vector>
 #include <string>
+#include <ios>
 
 namespace YAML
 {
@@ -10,17 +11,55 @@ namespace YAML
 	// simplified regular expressions
 	// . Only straightforward matches (no repeated characters)
 	// . Only matches from start of string
-	class RegEx {
+	class RegEx
+	{
+	private:
+		struct Operator {
+			virtual ~Operator() {}
+			virtual int Match(const std::string& str, const RegEx& regex) const = 0;
+			virtual int Match(std::istream& in, const RegEx& regex) const = 0;
+		};
+
+		struct MatchOperator: public Operator {
+			virtual int Match(const std::string& str, const RegEx& regex) const;
+			virtual int Match(std::istream& in, const RegEx& regex) const;
+		};
+
+		struct RangeOperator: public Operator {
+			virtual int Match(const std::string& str, const RegEx& regex) const;
+			virtual int Match(std::istream& in, const RegEx& regex) const;
+		};
+
+		struct OrOperator: public Operator {
+			virtual int Match(const std::string& str, const RegEx& regex) const;
+			virtual int Match(std::istream& in, const RegEx& regex) const;
+		};
+
+		struct NotOperator: public Operator {
+			virtual int Match(const std::string& str, const RegEx& regex) const;
+			virtual int Match(std::istream& in, const RegEx& regex) const;
+		};
+
+		struct SeqOperator: public Operator {
+			virtual int Match(const std::string& str, const RegEx& regex) const;
+			virtual int Match(std::istream& in, const RegEx& regex) const;
+		};
+
 	public:
+		friend struct Operator;
+
 		RegEx();
 		RegEx(char ch);
 		RegEx(char a, char z);
-		RegEx(const std::string& str, REGEX_OP op = REGEX_SEQ); 
+		RegEx(const std::string& str, REGEX_OP op = REGEX_SEQ);
+		RegEx(const RegEx& rhs);
 		~RegEx();
 
 		bool Matches(char ch) const;
 		bool Matches(const std::string& str) const;
+		bool Matches(std::istream& in) const;
 		int Match(const std::string& str) const;
+		int Match(std::istream& in) const;
 
 		friend RegEx operator ! (const RegEx& ex);
 		friend RegEx operator || (const RegEx& ex1, const RegEx& ex2);
@@ -28,9 +67,11 @@ namespace YAML
 
 	private:
 		RegEx(REGEX_OP op);
+		void SetOp();
 
 	private:
 		REGEX_OP m_op;
+		Operator *m_pOp;
 		char m_a, m_z;
 		std::vector <RegEx> m_params;
 	};
diff --git a/scanner.cpp b/scanner.cpp
index e1495f5..f656d32 100644
--- a/scanner.cpp
+++ b/scanner.cpp
@@ -48,22 +48,6 @@ namespace YAML
 		}
 	}
 
-	// Peek
-	// . Peeks at the next 'n' characters and returns them in a string.
-	std::string Scanner::Peek(int n)
-	{
-		std::string ret;
-
-		int pos = INPUT.tellg();
-		for(int i=0;i<n;i++)
-			ret += INPUT.get();
-
-		INPUT.clear();
-		INPUT.seekg(pos);
-
-		return ret;
-	}
-
 	// GetLineBreak
 	// . Eats with no checking
 	void Scanner::EatLineBreak()
@@ -97,7 +81,7 @@ namespace YAML
 		if(m_column != 0)
 			return false;
 
-		return Exp::DocStart.Matches(Peek(4));
+		return Exp::DocStart.Matches(INPUT);
 	}
 
 	// IsDocumentEnd
@@ -107,41 +91,37 @@ namespace YAML
 		if(m_column != 0)
 			return false;
 
-		return Exp::DocEnd.Matches(Peek(4));
+		return Exp::DocEnd.Matches(INPUT);
 	}
 
 	// IsBlockEntry
 	bool Scanner::IsBlockEntry()
 	{
-		return Exp::BlockEntry.Matches(Peek(2));
+		return Exp::BlockEntry.Matches(INPUT);
 	}
 
 	// IsKey
 	bool Scanner::IsKey()
 	{
-		std::string next = Peek(2);
 		if(m_flowLevel > 0)
-			return Exp::KeyInFlow.Matches(next);
-		return Exp::Key.Matches(next);
+			return Exp::KeyInFlow.Matches(INPUT);
+		return Exp::Key.Matches(INPUT);
 	}
 
 	// IsValue
 	bool Scanner::IsValue()
 	{
-		std::string next = Peek(2);
 		if(m_flowLevel > 0)
-			return Exp::ValueInFlow.Matches(next);
-		return Exp::Value.Matches(next);
+			return Exp::ValueInFlow.Matches(INPUT);
+		return Exp::Value.Matches(INPUT);
 	}
 
 	// IsPlainScalar
-	// . Rules:
 	bool Scanner::IsPlainScalar()
 	{
-		std::string next = Peek(2);
 		if(m_flowLevel > 0)
-			return Exp::PlainScalarInFlow.Matches(next);
-		return Exp::PlainScalar.Matches(next);
+			return Exp::PlainScalarInFlow.Matches(INPUT);
+		return Exp::PlainScalar.Matches(INPUT);
 	}
 
 	///////////////////////////////////////////////////////////////////////
@@ -368,21 +348,19 @@ namespace YAML
 				break;
 
 			// comment
-			if(Exp::Comment.Matches(INPUT.peek()))
+			if(Exp::Comment.Matches(INPUT))
 				break;
 
 			// first eat non-blanks
-			while(INPUT && !Exp::BlankOrBreak.Matches(INPUT.peek())) {
-				std::string next = Peek(2);
-
+			while(INPUT && !Exp::BlankOrBreak.Matches(INPUT)) {
 				// illegal colon in flow context
-				if(m_flowLevel > 0 && Exp::IllegalColonInScalar.Matches(next))
+				if(m_flowLevel > 0 && Exp::IllegalColonInScalar.Matches(INPUT))
 					throw IllegalScalar();
 
 				// characters that might end the scalar
-				if(m_flowLevel > 0 && Exp::EndScalarInFlow.Matches(next))
+				if(m_flowLevel > 0 && Exp::EndScalarInFlow.Matches(INPUT))
 					break;
-				if(m_flowLevel == 0 && Exp::EndScalar.Matches(next))
+				if(m_flowLevel == 0 && Exp::EndScalar.Matches(INPUT))
 					break;
 
 				if(leadingBlanks) {
@@ -409,12 +387,12 @@ namespace YAML
 			}
 
 			// did we hit a non-blank character that ended us?
-			if(!Exp::BlankOrBreak.Matches(INPUT.peek()))
+			if(!Exp::BlankOrBreak.Matches(INPUT))
 				break;
 
 			// now eat blanks
-			while(INPUT && Exp::BlankOrBreak.Matches(INPUT.peek())) {
-				if(Exp::Blank.Matches(INPUT.peek())) {
+			while(INPUT && Exp::BlankOrBreak.Matches(INPUT)) {
+				if(Exp::Blank.Matches(INPUT)) {
 					if(leadingBlanks && m_column <= m_indents.top())
 						throw IllegalTabInScalar();
 
diff --git a/scanner.h b/scanner.h
index 94cd88a..4a91e50 100644
--- a/scanner.h
+++ b/scanner.h
@@ -71,8 +71,6 @@ namespace YAML
 	private:
 		char GetChar();
 		void Eat(int n = 1);
-		std::string Peek(int n);
-
 		void EatLineBreak();
 
 		bool IsWhitespaceToBeEaten(char ch);