Moved the input stream, together with line/column info, into its own class, which allowed some other stuff just to pass the stream, and not have to be a member of Scanner.

exp.cpp

@@ -34,13 +34,12 @@ namespace YAML
 		// Escape
 		// . Translates the next 'codeLength' characters into a hex number and returns the result.
 		// . Throws if it's not actually hex.
-		std::string Escape(std::istream& in, int& length, int codeLength)
+		std::string Escape(Stream& in, int codeLength)
 		{
 			// grab string
-			length += codeLength;
 			std::string str;
 			for(int i=0;i<codeLength;i++)
-				str += in.get();
+				str += in.GetChar();
 
 			// get the value
 			unsigned value = ParseHex(str);
@@ -64,18 +63,14 @@ namespace YAML
 		// Escape
 		// . Escapes the sequence starting 'in' (it must begin with a '\' or single quote)
 		//   and returns the result.
-		// . Fills 'length' with how many characters we ate.
 		// . Throws if it's an unknown escape character.
-		std::string Escape(std::istream& in, int& length)
+		std::string Escape(Stream& in)
 		{
-			// slash + character
-			length = 2;
-
 			// eat slash
-			char escape = in.get();
+			char escape = in.GetChar();
 
 			// switch on escape character
-			char ch = in.get();
+			char ch = in.GetChar();
 
 			// first do single quote, since it's easier
 			if(escape == '\'' && ch == '\'')
@@ -101,9 +96,9 @@ namespace YAML
 				case '_': return "\xC2\xA0";  // #xA0
 				case 'L': return "\xE2\x80\xA8";  // LS (#x2028)
 				case 'P': return "\xE2\x80\xA9";  // PS (#x2029)
-				case 'x': return Escape(in, length, 2);
-				case 'u': return Escape(in, length, 4);
-				case 'U': return Escape(in, length, 8);
+				case 'x': return Escape(in, 2);
+				case 'u': return Escape(in, 4);
+				case 'U': return Escape(in, 8);
 			}
 
 			throw UnknownEscapeSequence(ch);

exp.h

@@ -3,6 +3,7 @@
 #include "regex.h"
 #include <string>
 #include <ios>
+#include "stream.h"
 
 namespace YAML
 {
@@ -50,7 +51,7 @@ namespace YAML
 		const RegEx Chomp = (ChompIndicator + Digit) || (Digit + ChompIndicator) || ChompIndicator || Digit;
 
 		// and some functions
-		std::string Escape(std::istream& in, int& length);
+		std::string Escape(Stream& in);
 	}
 
 	namespace Keys

main.cpp

@@ -3,22 +3,6 @@
 
 int main()
 {
-	YAML::RegEx alpha = YAML::RegEx('a', 'z') || YAML::RegEx('A', 'Z');
-	alpha.Matches("a");
-	alpha.Matches("d");
-	alpha.Matches("F");
-	alpha.Matches("0");
-	alpha.Matches("5");
-	alpha.Matches(" ");
-
-	YAML::RegEx blank = YAML::RegEx(' ') || YAML::RegEx('\t');
-	YAML::RegEx docstart = YAML::RegEx("---") + (blank || YAML::RegEx(EOF) || YAML::RegEx());
-	docstart.Matches("--- ");
-	docstart.Matches("... ");
-	docstart.Matches("----");
-	docstart.Matches("---\t");
-	docstart.Matches("---");
-
 	YAML::Document doc("test.yaml");
 
 	return 0;

scanner.cpp

@@ -7,8 +7,7 @@
 namespace YAML
 {
 	Scanner::Scanner(std::istream& in)
-		: INPUT(in), m_startedStream(false), m_endedStream(false), m_simpleKeyAllowed(false), m_flowLevel(0),
-		m_line(0), m_column(0)
+		: INPUT(in), m_startedStream(false), m_endedStream(false), m_simpleKeyAllowed(false), m_flowLevel(0)
 	{
 	}
 
@@ -27,45 +26,6 @@ namespace YAML
 	///////////////////////////////////////////////////////////////////////
 	// Misc. helpers
 
-	// GetChar
-	// . Extracts a character from the stream and updates our position
-	char Scanner::GetChar()
-	{
-		char ch = INPUT.get();
-		m_column++;
-		if(ch == '\n') {
-			m_column = 0;
-			m_line++;
-		}
-		return ch;
-	}
-
-	// GetChar
-	// . Extracts 'n' characters from the stream and updates our position
-	std::string Scanner::GetChar(int n)
-	{
-		std::string ret;
-		for(int i=0;i<n;i++)
-			ret += GetChar();
-		return ret;
-	}
-
-	// Eat
-	// . Eats 'n' characters and updates our position.
-	void Scanner::Eat(int n)
-	{
-		for(int i=0;i<n;i++)
-			GetChar();
-	}
-
-	// GetLineBreak
-	// . Eats with no checking
-	void Scanner::EatLineBreak()
-	{
-		Eat(1);
-		m_column = 0;
-	}
-
 	// IsWhitespaceToBeEaten
 	// . We can eat whitespace if:
 	//   1. It's a space
@@ -88,7 +48,7 @@ namespace YAML
 	bool Scanner::IsDocumentStart()
 	{
 		// needs to be at the start of a new line
-		if(m_column != 0)
+		if(INPUT.column != 0)
 			return false;
 
 		return Exp::DocStart.Matches(INPUT);
@@ -98,7 +58,7 @@ namespace YAML
 	bool Scanner::IsDocumentEnd()
 	{
 		// needs to be at the start of a new line
-		if(m_column != 0)
+		if(INPUT.column != 0)
 			return false;
 
 		return Exp::DocEnd.Matches(INPUT);
@@ -166,7 +126,7 @@ namespace YAML
 		ValidateSimpleKey();
 
 		// maybe need to end some blocks
-		PopIndentTo(m_column);
+		PopIndentTo(INPUT.column);
 
 		// *****
 		// And now branch based on the next few characters!
@@ -236,13 +196,13 @@ namespace YAML
 		while(1) {
 			// first eat whitespace
 			while(IsWhitespaceToBeEaten(INPUT.peek()))
-				Eat(1);
+				INPUT.Eat(1);
 
 			// then eat a comment
 			if(Exp::Comment.Matches(INPUT)) {
 				// eat until line break
 				while(INPUT && !Exp::Break.Matches(INPUT))
-					Eat(1);
+					INPUT.Eat(1);
 			}
 
 			// if it's NOT a line break, then we're done!
@@ -250,7 +210,7 @@ namespace YAML
 				break;
 
 			// otherwise, let's eat the line break and keep going
-			EatLineBreak();
+			INPUT.EatLineBreak();
 
 			// oh yeah, and let's get rid of that simple key
 			ValidateSimpleKey();

scanner.h

@@ -6,6 +6,7 @@
 #include <stack>
 #include <set>
 #include "regex.h"
+#include "stream.h"
 
 namespace YAML
 {
@@ -31,11 +32,6 @@ namespace YAML
 		void Scan();
 
 	private:
-		char GetChar();
-		std::string GetChar(int n);
-		void Eat(int n = 1);
-		void EatLineBreak();
-
 		bool IsWhitespaceToBeEaten(char ch);
 		bool IsDocumentStart();
 		bool IsDocumentEnd();
@@ -44,9 +40,6 @@ namespace YAML
 		bool IsValue();
 		bool IsPlainScalar();
 
-		void GetBlockIndentation(int& indent, std::string& breaks);
-		std::string ScanScalar(RegEx end, bool eatEnd, int indent, char escape, bool fold, bool eatLeadingWhitespace, bool trimTrailingSpaces, int chomp);
-
 		struct SimpleKey {
 			SimpleKey(int pos_, int line_, int column_, int flowLevel_);
 
@@ -63,8 +56,7 @@ namespace YAML
 
 	private:
 		// the stream
-		std::istream& INPUT;
-		int m_line, m_column;
+		Stream INPUT;
 
 		// the output (tokens)
 		std::queue <Token *> m_tokens;

scanscalar.cpp

@@ -142,7 +142,7 @@ namespace YAML
 		if(m_simpleKeyAllowed)
 			InsertSimpleKey();
 
-		pToken->value = ScanScalar(end, false, indent, 0, true, true, true, 0);
+		pToken->value = ScanScalar(INPUT, end, false, indent, 0, true, true, true, 0);
 
 		m_simpleKeyAllowed = false;
 		if(true/*info.leadingBlanks*/)
@@ -221,7 +221,7 @@ namespace YAML
 		//}
 
 		// eat single or double quote
-		char quote = GetChar();
+		char quote = INPUT.GetChar();
 		pToken->single = (quote == '\'');
 
 		RegEx end = (pToken->single ? RegEx(quote) && !Exp::EscSingleQuote : RegEx(quote));
@@ -231,7 +231,7 @@ namespace YAML
 		if(m_simpleKeyAllowed)
 			InsertSimpleKey();
 
-		pToken->value = ScanScalar(end, true, 0, escape, true, true, false, 0);
+		pToken->value = ScanScalar(INPUT, end, true, 0, escape, true, true, false, 0);
 		m_simpleKeyAllowed = false;
 
 		return pToken;
@@ -243,39 +243,39 @@ namespace YAML
 		WhitespaceInfo info;
 
 		// eat block indicator ('|' or '>')
-		char indicator = GetChar();
+		char indicator = INPUT.GetChar();
 		info.fold = (indicator == Keys::FoldedScalar);
 
 		// eat chomping/indentation indicators
 		int n = Exp::Chomp.Match(INPUT);
 		for(int i=0;i<n;i++)
-			info.SetChompers(GetChar());
+			info.SetChompers(INPUT.GetChar());
 
 		// first eat whitespace
 		while(Exp::Blank.Matches(INPUT))
-			Eat(1);
+			INPUT.Eat(1);
 
 		// and comments to the end of the line
 		if(Exp::Comment.Matches(INPUT))
 			while(INPUT && !Exp::Break.Matches(INPUT))
-				Eat(1);
+				INPUT.Eat(1);
 
 		// if it's not a line break, then we ran into a bad character inline
 		if(INPUT && !Exp::Break.Matches(INPUT))
 			throw UnexpectedCharacterInBlockScalar();
 
 		// and eat that baby
-		EatLineBreak();
+		INPUT.EatLineBreak();
 
 		// set the initial indentation
 		int indent = info.increment;
 		if(info.increment && m_indents.top() >= 0)
 			indent += m_indents.top();
 
-		GetBlockIndentation(indent, info.trailingBreaks);
+		GetBlockIndentation(INPUT, indent, info.trailingBreaks, m_indents.top());
 
 		bool eatLeadingWhitespace = false;
-		pToken->value = ScanScalar(RegEx(), false, indent, 0, info.fold, eatLeadingWhitespace, false, info.chomp);
+		pToken->value = ScanScalar(INPUT, RegEx(), false, indent, 0, info.fold, eatLeadingWhitespace, false, info.chomp);
 
 		// simple keys always ok after block scalars (since we're gonna start a new line anyways)
 		m_simpleKeyAllowed = true;
@@ -286,20 +286,20 @@ namespace YAML
 	// . Helper to scanning a block scalar.
 	// . Eats leading *indentation* zeros (i.e., those that come before 'indent'),
 	//   and updates 'indent' (if it hasn't been set yet).
-	void Scanner::GetBlockIndentation(int& indent, std::string& breaks)
+	void GetBlockIndentation(Stream& INPUT, int& indent, std::string& breaks, int topIndent)
 	{
 		int maxIndent = 0;
 
 		while(1) {
 			// eat as many indentation spaces as we can
-			while((indent == 0 || m_column < indent) && INPUT.peek() == ' ')
-				Eat(1);
+			while((indent == 0 || INPUT.column < indent) && INPUT.peek() == ' ')
+				INPUT.Eat(1);
 
-			if(m_column > maxIndent)
-				maxIndent = m_column;
+			if(INPUT.column > maxIndent)
+				maxIndent = INPUT.column;
 
 			// do we need more indentation, but we've got a tab?
-			if((indent == 0 || m_column < indent) && INPUT.peek() == '\t')
+			if((indent == 0 || INPUT.column < indent) && INPUT.peek() == '\t')
 				throw IllegalTabInScalar();   // TODO: are literal scalar lines allowed to have tabs here?
 
 			// is this a non-empty line?
@@ -308,21 +308,21 @@ namespace YAML
 
 			// otherwise, eat the line break and move on
 			int n = Exp::Break.Match(INPUT);
-			breaks += GetChar(n);
+			breaks += INPUT.GetChar(n);
 		}
 
 		// finally, set the indentation
 		if(indent == 0) {
 			indent = maxIndent;
-			if(indent < m_indents.top() + 1)
-				indent = m_indents.top() + 1;
+			if(indent < topIndent + 1)
+				indent = topIndent + 1;
 			if(indent < 1)
 				indent = 1;
 		}
 	}
 
 	// ScanScalar
-	std::string Scanner::ScanScalar(RegEx end, bool eatEnd, int indent, char escape, bool fold, bool eatLeadingWhitespace, bool trimTrailingSpaces, int chomp)
+	std::string ScanScalar(Stream& INPUT, RegEx end, bool eatEnd, int indent, char escape, bool fold, bool eatLeadingWhitespace, bool trimTrailingSpaces, int chomp)
 	{
 		bool emptyLine = false, moreIndented = false;
 		std::string scalar;
@@ -337,20 +337,18 @@ namespace YAML
 				// escaped newline? (only if we're escaping on slash)
 				if(escape == '\\' && Exp::EscBreak.Matches(INPUT)) {
 					int n = Exp::EscBreak.Match(INPUT);
-					Eat(n);
+					INPUT.Eat(n);
 					continue;
 				}
 
 				// escape this?
 				if(INPUT.peek() == escape) {
-					int length = 0;
-					scalar += Exp::Escape(INPUT, length);
-					m_column += length;
+					scalar += Exp::Escape(INPUT);
 					continue;
 				}
 
 				// otherwise, just add the damn character
-				scalar += GetChar();
+				scalar += INPUT.GetChar();
 			}
 
 			// eof? if we're looking to eat something, then we throw
@@ -364,26 +362,26 @@ namespace YAML
 			int n = end.Match(INPUT);
 			if(n >= 0) {
 				if(eatEnd)
-					Eat(n);
+					INPUT.Eat(n);
 				break;
 			}
 
 			// ********************************
 			// Phase #2: eat line ending
 			n = Exp::Break.Match(INPUT);
-			Eat(n);
+			INPUT.Eat(n);
 
 			// ********************************
 			// Phase #3: scan initial spaces
 
 			// first the required indentation
-			while(INPUT.peek() == ' ' && m_column < indent)
-				Eat(1);
+			while(INPUT.peek() == ' ' && INPUT.column < indent)
+				INPUT.Eat(1);
 
 			// and then the rest of the whitespace
 			if(eatLeadingWhitespace) {
 				while(Exp::Blank.Matches(INPUT))
-					Eat(1);
+					INPUT.Eat(1);
 			}
 
 			// was this an empty line?
@@ -399,7 +397,7 @@ namespace YAML
 			moreIndented = nextMoreIndented;
 
 			// are we done via indentation?
-			if(!emptyLine && m_column < indent)
+			if(!emptyLine && INPUT.column < indent)
 				break;
 		}
 

scanscalar.h

@@ -1,9 +1,14 @@
 #pragma once
 
 #include <string>
+#include "regex.h"
+#include "stream.h"
 
 namespace YAML
 {
+	void GetBlockIndentation(Stream& INPUT, int& indent, std::string& breaks, int topIndent);
+	std::string ScanScalar(Stream& INPUT, RegEx end, bool eatEnd, int indent, char escape, bool fold, bool eatLeadingWhitespace, bool trimTrailingSpaces, int chomp);
+
 	struct WhitespaceInfo {
 		WhitespaceInfo();
 

scantoken.cpp

@@ -22,8 +22,8 @@ namespace YAML
 	template <> StreamEndToken *Scanner::ScanToken(StreamEndToken *pToken)
 	{
 		// force newline
-		if(m_column > 0)
-			m_column = 0;
+		if(INPUT.column > 0)
+			INPUT.column = 0;
 
 		PopIndentTo(-1);
 		ValidateAllSimpleKeys();
@@ -37,12 +37,12 @@ namespace YAML
 	// DocumentStartToken
 	template <> DocumentStartToken *Scanner::ScanToken(DocumentStartToken *pToken)
 	{
-		PopIndentTo(m_column);
+		PopIndentTo(INPUT.column);
 		ValidateAllSimpleKeys();
 		m_simpleKeyAllowed = false;
 
 		// eat
-		Eat(3);
+		INPUT.Eat(3);
 		return pToken;
 	}
 
@@ -54,7 +54,7 @@ namespace YAML
 		m_simpleKeyAllowed = false;
 
 		// eat
-		Eat(3);
+		INPUT.Eat(3);
 		return pToken;
 	}
 
@@ -67,7 +67,7 @@ namespace YAML
 		m_simpleKeyAllowed = true;
 
 		// eat
-		Eat(1);
+		INPUT.Eat(1);
 		return pToken;
 	}
 
@@ -80,7 +80,7 @@ namespace YAML
 		m_simpleKeyAllowed = true;
 
 		// eat
-		Eat(1);
+		INPUT.Eat(1);
 		return pToken;
 	}
 
@@ -94,7 +94,7 @@ namespace YAML
 		m_simpleKeyAllowed = false;
 
 		// eat
-		Eat(1);
+		INPUT.Eat(1);
 		return pToken;
 	}
 
@@ -108,7 +108,7 @@ namespace YAML
 		m_simpleKeyAllowed = false;
 
 		// eat
-		Eat(1);
+		INPUT.Eat(1);
 		return pToken;
 	}
 
@@ -118,7 +118,7 @@ namespace YAML
 		m_simpleKeyAllowed = true;
 
 		// eat
-		Eat(1);
+		INPUT.Eat(1);
 		return pToken;
 	}
 
@@ -133,11 +133,11 @@ namespace YAML
 		if(!m_simpleKeyAllowed)
 			throw IllegalBlockEntry();
 
-		PushIndentTo(m_column, true);
+		PushIndentTo(INPUT.column, true);
 		m_simpleKeyAllowed = true;
 
 		// eat
-		Eat(1);
+		INPUT.Eat(1);
 		return pToken;
 	}
 
@@ -149,7 +149,7 @@ namespace YAML
 			if(!m_simpleKeyAllowed)
 				throw IllegalMapKey();
 
-			PushIndentTo(m_column, false);
+			PushIndentTo(INPUT.column, false);
 		}
 
 		// can only put a simple key here if we're in block context
@@ -159,7 +159,7 @@ namespace YAML
 			m_simpleKeyAllowed = false;
 
 		// eat
-		Eat(1);
+		INPUT.Eat(1);
 		return pToken;
 	}
 
@@ -176,7 +176,7 @@ namespace YAML
 				if(!m_simpleKeyAllowed)
 					throw IllegalMapValue();
 
-				PushIndentTo(m_column, false);
+				PushIndentTo(INPUT.column, false);
 			}
 
 			// can only put a simple key here if we're in block context
@@ -187,7 +187,7 @@ namespace YAML
 		}
 
 		// eat
-		Eat(1);
+		INPUT.Eat(1);
 		return pToken;
 	}
 
@@ -200,13 +200,13 @@ namespace YAML
 		m_simpleKeyAllowed = false;
 
 		// eat the indicator
-		char indicator = GetChar();
+		char indicator = INPUT.GetChar();
 		pToken->alias = (indicator == Keys::Alias);
 
 		// now eat the content
 		std::string tag;
 		while(Exp::AlphaNumeric.Matches(INPUT))
-			tag += GetChar();
+			tag += INPUT.GetChar();
 
 		// we need to have read SOMETHING!
 		if(tag.empty())

simplekey.cpp

@@ -34,10 +34,10 @@ namespace YAML
 	//   and saves it on a stack.
 	void Scanner::InsertSimpleKey()
 	{
-		SimpleKey key(INPUT.tellg(), m_line, m_column, m_flowLevel);
+		SimpleKey key(INPUT.pos(), INPUT.line, INPUT.column, m_flowLevel);
 
 		// first add a map start, if necessary
-		key.pMapStart = PushIndentTo(m_column, false);
+		key.pMapStart = PushIndentTo(INPUT.column, false);
 		if(key.pMapStart)
 			key.pMapStart->isValid = false;
 //		else
@@ -79,7 +79,7 @@ namespace YAML
 			isValid = false;
 
 		// also needs to be less than 1024 characters and inline
-		if(m_line != key.line || (int) INPUT.tellg() - key.pos > 1024)
+		if(INPUT.line != key.line || INPUT.pos() - key.pos > 1024)
 			isValid = false;
 
 		// invalidate key

stream.cpp

@@ -0,0 +1,43 @@
+#include "stream.h"
+
+namespace YAML
+{
+	// GetChar
+	// . Extracts a character from the stream and updates our position
+	char Stream::GetChar()
+	{
+		char ch = input.get();
+		column++;
+		if(ch == '\n') {
+			column = 0;
+			line++;
+		}
+		return ch;
+	}
+
+	// GetChar
+	// . Extracts 'n' characters from the stream and updates our position
+	std::string Stream::GetChar(int n)
+	{
+		std::string ret;
+		for(int i=0;i<n;i++)
+			ret += GetChar();
+		return ret;
+	}
+
+	// Eat
+	// . Eats 'n' characters and updates our position.
+	void Stream::Eat(int n)
+	{
+		for(int i=0;i<n;i++)
+			GetChar();
+	}
+
+	// GetLineBreak
+	// . Eats with no checking
+	void Stream::EatLineBreak()
+	{
+		Eat(1);
+		column = 0;
+	}
+}

stream.h

@@ -0,0 +1,26 @@
+#pragma once
+
+#include <ios>
+#include <string>
+
+namespace YAML
+{
+	struct Stream
+	{
+		Stream(std::istream& input_): input(input_), line(0), column(0) {}
+
+		char peek() { return input.peek(); }
+		int pos() const { return input.tellg(); }
+		operator std::istream& () { return input; }
+		operator bool() { return input.good(); }
+		bool operator !() { return !input; }
+
+		char GetChar();
+		std::string GetChar(int n);
+		void Eat(int n = 1);
+		void EatLineBreak();
+
+		std::istream& input;
+		int line, column;
+	};
+}

test.yaml

@@ -1,4 +1,12 @@
 ---
-- "quoted scalar that contains
----
-    the document start!"
+- "quoted scalar\twith a tab\nand a newline"
+- 'This is Jesse''s single quote!'
+- |
+  here's a literal:
+  #include <iostream>
+  
+  int main()
+  {
+     std::cout << "Hello World!\n";
+     return 0;
+  }

yaml-reader.vcproj

@@ -217,6 +217,10 @@
 				RelativePath=".\simplekey.cpp"
 				>
 			</File>
+			<File
+				RelativePath=".\stream.cpp"
+				>
+			</File>
 		</Filter>
 		<Filter
 			Name="Header Files"
@@ -271,6 +275,10 @@
 				RelativePath=".\sequence.h"
 				>
 			</File>
+			<File
+				RelativePath=".\stream.h"
+				>
+			</File>
 			<File
 				RelativePath=".\token.h"
 				>