Started the scanner.

document.cpp

@@ -1,6 +1,7 @@
 #include "document.h"
 #include "node.h"
 #include "parser.h"
+#include "scanner.h"
 #include <fstream>
 
 namespace YAML
@@ -30,10 +31,11 @@ namespace YAML
 		Clear();
 
 		std::ifstream fin(fileName.c_str());
-		Parser parser(fin);
+		Scanner scanner(fin);
-		if(!parser)
+		scanner.Scan();
-			return;
+//		if(!scanner)
+//			return;
 
-		m_pRoot = parser.ReadNextNode();
+//		m_pRoot = parser.ReadNextNode();
 	}
 }

scanner.cpp

@@ -0,0 +1,275 @@
+#include "scanner.h"
+#include "token.h"
+
+namespace YAML
+{
+	Scanner::Scanner(std::istream& in)
+		: INPUT(in), m_startedStream(false), m_simpleKeyAllowed(false), m_flowLevel(0), m_column(0)
+	{
+	}
+
+	Scanner::~Scanner()
+	{
+	}
+
+	///////////////////////////////////////////////////////////////////////
+	// Misc. helpers
+
+	// GetChar
+	// . Extracts a character from the stream and updates our position
+	char Scanner::GetChar()
+	{
+		m_column++;
+		return INPUT.get();
+	}
+
+	// GetLineBreak
+	// . Eats with no checking
+	void Scanner::EatLineBreak()
+	{
+		m_column = 0;
+		INPUT.get();
+	}
+
+	// EatDocumentStart
+	// . Eats with no checking
+	void Scanner::EatDocumentStart()
+	{
+		INPUT.get();
+		INPUT.get();
+		INPUT.get();
+	}
+
+	// EatDocumentEnd
+	// . Eats with no checking
+	void Scanner::EatDocumentEnd()
+	{
+		INPUT.get();
+		INPUT.get();
+		INPUT.get();
+	}
+
+	// IsWhitespaceToBeEaten
+	// . We can eat whitespace if:
+	//   1. It's a space
+	//   2. It's a tab, and we're either:
+	//      a. In the flow context
+	//      b. In the block context but not where a simple key could be allowed
+	//         (i.e., not at the beginning of a line, or following '-', '?', or ':')
+	bool Scanner::IsWhitespaceToBeEaten()
+	{
+		char ch = INPUT.peek();
+
+		if(ch == ' ')
+			return true;
+
+		if(ch == '\t' && (m_flowLevel >= 0 || !m_simpleKeyAllowed))
+			return true;
+
+		return false;
+	}
+
+	// IsLineBreak
+	bool Scanner::IsLineBreak()
+	{
+		char ch = INPUT.peek();
+		return ch == '\n'; // TODO: More types of line breaks
+	}
+
+	// IsBlank
+	bool Scanner::IsBlank()
+	{
+		char ch = INPUT.peek();
+		return IsLineBreak() || ch == ' ' || ch == '\t' || ch == EOF;
+	}
+
+	// IsDocumentStart
+	bool Scanner::IsDocumentStart()
+	{
+		// needs to be at the start of a new line
+		if(m_column != 0)
+			return false;
+
+		// then needs '---'
+		for(int i=0;i<3;i++) {
+			if(INPUT.peek() != '-') {
+				// first put 'em back
+				for(int j=0;j<i;j++)
+					INPUT.putback('-');
+
+				// and return
+				return false;
+			}
+			INPUT.get();
+		}
+
+		// then needs a blank character (or eof)
+		if(!IsBlank()) {
+			// put 'em back
+			for(int i=0;i<3;i++)
+				INPUT.putback('-');
+
+			// and return
+			return false;
+		}
+
+		// finally, put 'em back and go
+		for(int i=0;i<3;i++)
+			INPUT.putback('-');
+
+		return true;
+	}
+
+	// IsDocumentEnd
+	bool Scanner::IsDocumentEnd()
+	{
+		// needs to be at the start of a new line
+		if(m_column != 0)
+			return false;
+
+		// then needs '...'
+		for(int i=0;i<3;i++) {
+			if(INPUT.peek() != '.') {
+				// first put 'em back
+				for(int j=0;j<i;j++)
+					INPUT.putback('.');
+
+				// and return
+				return false;
+			}
+			INPUT.get();
+		}
+
+		// then needs a blank character (or eof)
+		if(!IsBlank()) {
+			// put 'em back
+			for(int i=0;i<3;i++)
+				INPUT.putback('.');
+
+			// and return
+			return false;
+		}
+
+		// finally, put 'em back and go
+		for(int i=0;i<3;i++)
+			INPUT.putback('-');
+
+		return true;
+	}
+
+	///////////////////////////////////////////////////////////////////////
+	// Specialization for scanning specific tokens
+
+	// StreamStartToken
+	template <> StreamStartToken *Scanner::ScanToken(StreamStartToken *pToken)
+	{
+		m_startedStream = true;
+		m_simpleKeyAllowed = true;
+		m_indents.push(-1);
+
+		return pToken;
+	}
+
+	// StreamEndToken
+	template <> StreamEndToken *Scanner::ScanToken(StreamEndToken *pToken)
+	{
+		// force newline
+		if(m_column > 0)
+			m_column = 0;
+
+		// TODO: unroll indentation
+		// TODO: "reset simple keys"
+
+		m_simpleKeyAllowed = false;
+
+		return pToken;
+	}
+
+	// DocumentStartToken
+	template <> DocumentStartToken *Scanner::ScanToken(DocumentStartToken *pToken)
+	{
+		// TODO: unroll indentation
+		// TODO: reset simple keys
+
+		m_simpleKeyAllowed = false;
+
+		// eat it
+		EatDocumentStart();
+
+		return pToken;
+	}
+
+	// DocumentEndToken
+	template <> DocumentEndToken *Scanner::ScanToken(DocumentEndToken *pToken)
+	{
+		// TODO: unroll indentation
+		// TODO: reset simple keys
+
+		m_simpleKeyAllowed = false;
+
+		// eat it
+		EatDocumentEnd();
+
+		return pToken;
+	}
+
+	///////////////////////////////////////////////////////////////////////
+	// The main scanning function
+
+	Token *Scanner::ScanNextToken()
+	{
+		if(!m_startedStream)
+			return ScanToken(new StreamStartToken);
+
+		ScanToNextToken();
+		// TODO: remove "obsolete potential simple keys"
+		// TODO: unroll indent
+
+		if(INPUT.peek() == EOF)
+			return ScanToken(new StreamEndToken);
+
+		if(IsDocumentStart())
+			return ScanToken(new DocumentStartToken);
+
+		if(IsDocumentEnd())
+			return ScanToken(new DocumentEndToken);
+
+		return 0;
+	}
+
+	// ScanToNextToken
+	// . Eats input until we reach the next token-like thing.
+	void Scanner::ScanToNextToken()
+	{
+		while(1) {
+			// first eat whitespace
+			while(IsWhitespaceToBeEaten())
+				INPUT.get();
+
+			// then eat a comment
+			if(INPUT.peek() == Keys::Comment) {
+				// eat until line break
+				while(INPUT && !IsLineBreak())
+					INPUT.get();
+			}
+
+			// if it's NOT a line break, then we're done!
+			if(!IsLineBreak())
+				break;
+
+			// otherwise, let's eat the line break and keep going
+			EatLineBreak();
+
+			// new line - we may be able to accept a simple key now
+			if(m_flowLevel == 0)
+				m_simpleKeyAllowed = true;
+        }
+	}
+
+	// temporary function for testing
+	void Scanner::Scan()
+	{
+		while(Token *pToken = ScanNextToken())
+			delete pToken;
+	}
+}

scanner.h

@@ -0,0 +1,55 @@
+#pragma once
+
+#include <ios>
+#include <string>
+#include <queue>
+#include <stack>
+
+namespace YAML
+{
+	class Token;
+
+	namespace Keys
+	{
+		const char Comment = '#';
+	}
+
+	class Scanner
+	{
+	public:
+		Scanner(std::istream& in);
+		~Scanner();
+
+		Token *ScanNextToken();
+		void ScanToNextToken();
+
+		void Scan();
+
+	private:
+		char GetChar();
+		void EatLineBreak();
+		void EatDocumentStart();
+		void EatDocumentEnd();
+
+		bool IsWhitespaceToBeEaten();
+		bool IsLineBreak();
+		bool IsBlank();
+		bool IsDocumentStart();
+		bool IsDocumentEnd();
+		template <typename T> T *ScanToken(T *pToken);
+
+	private:
+		// the stream
+		std::istream& INPUT;
+		int m_column;
+
+		// the output (tokens)
+		std::queue <Token *> m_tokens;
+
+		// state info
+		bool m_startedStream;
+		bool m_simpleKeyAllowed;
+		int m_flowLevel;                // number of unclosed '[' and '{' indicators
+		std::stack <int> m_indents;
+	};
+}

token.h

@@ -0,0 +1,10 @@
+#pragma once
+
+namespace YAML
+{
+	class Token {};
+	class StreamStartToken: public Token {};
+	class StreamEndToken: public Token {};
+	class DocumentStartToken: public Token {};
+	class DocumentEndToken: public Token {};
+}

yaml-reader.vcproj

@@ -189,6 +189,10 @@
 				RelativePath=".\scalar.cpp"
 				>
 			</File>
+			<File
+				RelativePath=".\scanner.cpp"
+				>
+			</File>
 			<File
 				RelativePath=".\sequence.cpp"
 				>
@@ -223,10 +227,18 @@
 				RelativePath=".\scalar.h"
 				>
 			</File>
+			<File
+				RelativePath=".\scanner.h"
+				>
+			</File>
 			<File
 				RelativePath=".\sequence.h"
 				>
 			</File>
+			<File
+				RelativePath=".\token.h"
+				>
+			</File>
 		</Filter>
 		<Filter
 			Name="Resource Files"