Added api sketch

src/aliasmanager.cpp

@@ -0,0 +1,29 @@
+#include "yaml-cpp/aliasmanager.h"
+#include "yaml-cpp/node.h"
+#include <cassert>
+#include <sstream>
+
+namespace YAML
+{
+	AliasManager::AliasManager(): m_curAnchor(0)
+	{
+	}
+
+	void AliasManager::RegisterReference(const Node& node)
+	{
+		m_anchorByIdentity.insert(std::make_pair(&node, _CreateNewAnchor()));
+	}
+
+	anchor_t AliasManager::LookupAnchor(const Node& node) const
+	{
+		AnchorByIdentity::const_iterator it = m_anchorByIdentity.find(&node);
+		if(it == m_anchorByIdentity.end())
+			return 0;
+		return it->second;
+	}
+
+	anchor_t AliasManager::_CreateNewAnchor()
+	{
+		return ++m_curAnchor;
+	}
+}

src/collectionstack.h

@@ -0,0 +1,35 @@
+#ifndef COLLECTIONSTACK_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define COLLECTIONSTACK_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include <stack>
+#include <cassert>
+
+namespace YAML
+{
+	struct CollectionType {
+		enum value { None, BlockMap, BlockSeq, FlowMap, FlowSeq, CompactMap };
+	};
+
+	class CollectionStack
+	{
+	public:
+		CollectionType::value GetCurCollectionType() const {
+			if(collectionStack.empty())
+				return CollectionType::None;
+			return collectionStack.top();
+		}
+		
+		void PushCollectionType(CollectionType::value type) { collectionStack.push(type); }
+		void PopCollectionType(CollectionType::value type) { assert(type == GetCurCollectionType()); collectionStack.pop(); }
+		
+	private:
+		std::stack<CollectionType::value> collectionStack;
+	};
+}
+
+#endif // COLLECTIONSTACK_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/contrib/graphbuilder.cpp

@@ -0,0 +1,16 @@
+#include "yaml-cpp/parser.h"
+#include "yaml-cpp/contrib/graphbuilder.h"
+#include "graphbuilderadapter.h"
+
+namespace YAML
+{
+  void *BuildGraphOfNextDocument(Parser& parser, GraphBuilderInterface& graphBuilder)
+  {
+    GraphBuilderAdapter eventHandler(graphBuilder);
+    if (parser.HandleNextDocument(eventHandler)) {
+      return eventHandler.RootNode();
+    } else {
+      return NULL;
+    }
+  }
+}

src/contrib/graphbuilderadapter.cpp

@@ -0,0 +1,96 @@
+#include "graphbuilderadapter.h"
+
+namespace YAML
+{
+  int GraphBuilderAdapter::ContainerFrame::sequenceMarker;
+  
+  void GraphBuilderAdapter::OnNull(const Mark& mark, anchor_t anchor)
+  {
+    void *pParent = GetCurrentParent();
+    void *pNode = m_builder.NewNull(mark, pParent);
+    RegisterAnchor(anchor, pNode);
+    
+    DispositionNode(pNode);
+  }
+  
+  void GraphBuilderAdapter::OnAlias(const Mark& mark, anchor_t anchor)
+  {
+    void *pReffedNode = m_anchors.Get(anchor);
+    DispositionNode(m_builder.AnchorReference(mark, pReffedNode));
+  }
+  
+  void GraphBuilderAdapter::OnScalar(const Mark& mark, const std::string& tag, anchor_t anchor, const std::string& value)
+  {
+    void *pParent = GetCurrentParent();
+    void *pNode = m_builder.NewScalar(mark, tag, pParent, value);
+    RegisterAnchor(anchor, pNode);
+    
+    DispositionNode(pNode);
+  }
+  
+  void GraphBuilderAdapter::OnSequenceStart(const Mark& mark, const std::string& tag, anchor_t anchor)
+  {
+    void *pNode = m_builder.NewSequence(mark, tag, GetCurrentParent());
+    m_containers.push(ContainerFrame(pNode));
+    RegisterAnchor(anchor, pNode);
+  }
+  
+  void GraphBuilderAdapter::OnSequenceEnd()
+  {
+    void *pSequence = m_containers.top().pContainer;
+    m_containers.pop();
+    
+    DispositionNode(pSequence);
+  }
+  
+  void GraphBuilderAdapter::OnMapStart(const Mark& mark, const std::string& tag, anchor_t anchor)
+  {
+    void *pNode = m_builder.NewMap(mark, tag, GetCurrentParent());
+    m_containers.push(ContainerFrame(pNode, m_pKeyNode));
+    m_pKeyNode = NULL;
+    RegisterAnchor(anchor, pNode);
+  }
+  
+  void GraphBuilderAdapter::OnMapEnd()
+  {
+    void *pMap = m_containers.top().pContainer;
+    m_pKeyNode = m_containers.top().pPrevKeyNode;
+    m_containers.pop();
+    DispositionNode(pMap);
+  }
+  
+  void *GraphBuilderAdapter::GetCurrentParent() const
+  {
+    if (m_containers.empty()) {
+      return NULL;
+    }
+    return m_containers.top().pContainer;
+  }
+  
+  void GraphBuilderAdapter::RegisterAnchor(anchor_t anchor, void *pNode)
+  {
+    if (anchor) {
+      m_anchors.Register(anchor, pNode);
+    }
+  }
+  
+  void GraphBuilderAdapter::DispositionNode(void *pNode)
+  {
+    if (m_containers.empty()) {
+      m_pRootNode = pNode;
+      return;
+    }
+    
+    void *pContainer = m_containers.top().pContainer;
+    if (m_containers.top().isMap()) {
+      if (m_pKeyNode) {
+        m_builder.AssignInMap(pContainer, m_pKeyNode, pNode);
+        m_pKeyNode = NULL;
+      } else {
+        m_pKeyNode = pNode;
+      }
+    } else {
+      m_builder.AppendToSequence(pContainer, pNode);
+    }
+  }
+}

src/contrib/graphbuilderadapter.h

@@ -0,0 +1,73 @@
+#ifndef GRAPHBUILDERADAPTER_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define GRAPHBUILDERADAPTER_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+#include <cstdlib>
+#include <map>
+#include <stack>
+#include "yaml-cpp/eventhandler.h"
+#include "yaml-cpp/contrib/anchordict.h"
+#include "yaml-cpp/contrib/graphbuilder.h"
+
+namespace YAML
+{
+  class GraphBuilderAdapter : public EventHandler
+  {
+  public:
+    GraphBuilderAdapter(GraphBuilderInterface& builder)
+    : m_builder(builder), m_pRootNode(NULL), m_pKeyNode(NULL)
+    {
+    }
+    
+    virtual void OnDocumentStart(const Mark& mark) {(void)mark;}
+    virtual void OnDocumentEnd() {}
+    
+    virtual void OnNull(const Mark& mark, anchor_t anchor);
+    virtual void OnAlias(const Mark& mark, anchor_t anchor);
+    virtual void OnScalar(const Mark& mark, const std::string& tag, anchor_t anchor, const std::string& value);
+    
+    virtual void OnSequenceStart(const Mark& mark, const std::string& tag, anchor_t anchor);
+    virtual void OnSequenceEnd();
+    
+    virtual void OnMapStart(const Mark& mark, const std::string& tag, anchor_t anchor);
+    virtual void OnMapEnd();
+    
+    void *RootNode() const {return m_pRootNode;}
+  
+  private:
+    struct ContainerFrame
+    {
+      ContainerFrame(void *pSequence)
+      : pContainer(pSequence), pPrevKeyNode(&sequenceMarker)
+      {}
+      ContainerFrame(void *pMap, void* pPrevKeyNode)
+      : pContainer(pMap), pPrevKeyNode(pPrevKeyNode)
+      {}
+      
+      void *pContainer;
+      void *pPrevKeyNode;
+      
+      bool isMap() const {return pPrevKeyNode != &sequenceMarker;}
+    
+    private:
+      static int sequenceMarker;
+    };
+    typedef std::stack<ContainerFrame> ContainerStack;
+    typedef AnchorDict<void*> AnchorMap;
+    
+    GraphBuilderInterface& m_builder;
+    ContainerStack m_containers;
+    AnchorMap m_anchors;
+    void *m_pRootNode;
+    void *m_pKeyNode;
+    
+    void *GetCurrentParent() const;
+    void RegisterAnchor(anchor_t anchor, void *pNode);
+    void DispositionNode(void *pNode);
+  };
+}
+
+#endif // GRAPHBUILDERADAPTER_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/conversion.cpp

@@ -0,0 +1,89 @@
+#include "yaml-cpp/conversion.h"
+#include <algorithm>
+
+////////////////////////////////////////////////////////////////
+// Specializations for converting a string to specific types
+
+namespace
+{
+	// we're not gonna mess with the mess that is all the isupper/etc. functions
+	bool IsLower(char ch) { return 'a' <= ch && ch <= 'z'; }
+	bool IsUpper(char ch) { return 'A' <= ch && ch <= 'Z'; }
+	char ToLower(char ch) { return IsUpper(ch) ? ch + 'a' - 'A' : ch; }
+
+	std::string tolower(const std::string& str)
+	{
+		std::string s(str);
+		std::transform(s.begin(), s.end(), s.begin(), ToLower);
+		return s;
+	}
+
+	template <typename T>
+	bool IsEntirely(const std::string& str, T func)
+	{
+		for(std::size_t i=0;i<str.size();i++)
+			if(!func(str[i]))
+				return false;
+
+		return true;
+	}
+
+	// IsFlexibleCase
+	// . Returns true if 'str' is:
+	//   . UPPERCASE
+	//   . lowercase
+	//   . Capitalized
+	bool IsFlexibleCase(const std::string& str)
+	{
+		if(str.empty())
+			return true;
+
+		if(IsEntirely(str, IsLower))
+			return true;
+
+		bool firstcaps = IsUpper(str[0]);
+		std::string rest = str.substr(1);
+		return firstcaps && (IsEntirely(rest, IsLower) || IsEntirely(rest, IsUpper));
+	}
+}
+
+namespace YAML
+{
+	bool Convert(const std::string& input, bool& b)
+	{
+		// we can't use iostream bool extraction operators as they don't
+		// recognize all possible values in the table below (taken from
+		// http://yaml.org/type/bool.html)
+		static const struct {
+			std::string truename, falsename;
+		} names[] = {
+			{ "y", "n" },
+			{ "yes", "no" },
+			{ "true", "false" },
+			{ "on", "off" },
+		};
+
+		if(!IsFlexibleCase(input))
+			return false;
+
+		for(unsigned i=0;i<sizeof(names)/sizeof(names[0]);i++) {
+			if(names[i].truename == tolower(input)) {
+				b = true;
+				return true;
+			}
+
+			if(names[i].falsename == tolower(input)) {
+				b = false;
+				return true;
+			}
+		}
+
+		return false;
+	}
+	
+	bool Convert(const std::string& input, _Null& /*output*/)
+	{
+		return input.empty() || input == "~" || input == "null" || input == "Null" || input == "NULL";
+	}
+}
+

src/directives.cpp

@@ -0,0 +1,24 @@
+#include "directives.h"
+
+namespace YAML
+{
+	Directives::Directives()
+	{
+		// version
+		version.isDefault = true;
+		version.major = 1;
+		version.minor = 2;
+	}
+	
+	const std::string Directives::TranslateTagHandle(const std::string& handle) const
+	{
+		std::map <std::string, std::string>::const_iterator it = tags.find(handle);
+		if(it == tags.end()) {
+			if(handle == "!!")
+				return "tag:yaml.org,2002:";
+			return handle;
+		}
+		
+		return it->second;
+	}
+}

src/directives.h

@@ -0,0 +1,29 @@
+#ifndef DIRECTIVES_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define DIRECTIVES_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include <string>
+#include <map>
+
+namespace YAML
+{
+	struct Version {
+		bool isDefault;
+		int major, minor;
+	};
+	
+	struct Directives {
+		Directives();
+		
+		const std::string TranslateTagHandle(const std::string& handle) const;
+
+		Version version;
+		std::map<std::string, std::string> tags;
+	};
+}
+
+#endif // DIRECTIVES_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/emitfromevents.cpp

@@ -0,0 +1,105 @@
+#include "yaml-cpp/emitfromevents.h"
+#include "yaml-cpp/emitter.h"
+#include "yaml-cpp/null.h"
+#include <cassert>
+#include <sstream>
+
+namespace {
+	std::string ToString(YAML::anchor_t anchor) {
+		std::stringstream stream;
+		stream << anchor;
+		return stream.str();
+	}
+}
+
+namespace YAML
+{
+	EmitFromEvents::EmitFromEvents(Emitter& emitter): m_emitter(emitter)
+	{
+	}
+	
+	void EmitFromEvents::OnDocumentStart(const Mark&)
+	{
+	}
+	
+	void EmitFromEvents::OnDocumentEnd()
+	{
+	}
+	
+	void EmitFromEvents::OnNull(const Mark&, anchor_t anchor)
+	{
+		BeginNode();
+		EmitProps("", anchor);
+		m_emitter << Null;
+	}
+	
+	void EmitFromEvents::OnAlias(const Mark&, anchor_t anchor)
+	{
+		BeginNode();
+		m_emitter << Alias(ToString(anchor));
+	}
+	
+	void EmitFromEvents::OnScalar(const Mark&, const std::string& tag, anchor_t anchor, const std::string& value)
+	{
+		BeginNode();
+		EmitProps(tag, anchor);
+		m_emitter << value;
+	}
+	
+	void EmitFromEvents::OnSequenceStart(const Mark&, const std::string& tag, anchor_t anchor)
+	{
+		BeginNode();
+		EmitProps(tag, anchor);
+		m_emitter << BeginSeq;
+		m_stateStack.push(State::WaitingForSequenceEntry);
+	}
+	
+	void EmitFromEvents::OnSequenceEnd()
+	{
+		m_emitter << EndSeq;
+		assert(m_stateStack.top() == State::WaitingForSequenceEntry);
+		m_stateStack.pop();
+	}
+	
+	void EmitFromEvents::OnMapStart(const Mark&, const std::string& tag, anchor_t anchor)
+	{
+		BeginNode();
+		EmitProps(tag, anchor);
+		m_emitter << BeginMap;
+		m_stateStack.push(State::WaitingForKey);
+	}
+
+	void EmitFromEvents::OnMapEnd()
+	{
+		m_emitter << EndMap;
+		assert(m_stateStack.top() == State::WaitingForKey);
+		m_stateStack.pop();
+	}
+
+	void EmitFromEvents::BeginNode()
+	{
+		if(m_stateStack.empty())
+			return;
+		
+		switch(m_stateStack.top()) {
+			case State::WaitingForKey:
+				m_emitter << Key;
+				m_stateStack.top() = State::WaitingForValue;
+				break;
+			case State::WaitingForValue:
+				m_emitter << Value;
+				m_stateStack.top() = State::WaitingForKey;
+				break;
+			default:
+				break;
+		}
+	}
+	
+	void EmitFromEvents::EmitProps(const std::string& tag, anchor_t anchor)
+	{
+		if(!tag.empty() && tag != "?")
+			m_emitter << VerbatimTag(tag);
+		if(anchor)
+			m_emitter << Anchor(ToString(anchor));
+	}
+}

src/emitter.cpp

@@ -0,0 +1,838 @@
+#include "yaml-cpp/emitter.h"
+#include "emitterstate.h"
+#include "emitterutils.h"
+#include "indentation.h"
+#include "yaml-cpp/exceptions.h"
+#include <sstream>
+
+namespace YAML
+{	
+	Emitter::Emitter(): m_pState(new EmitterState)
+	{
+	}
+	
+	Emitter::~Emitter()
+	{
+	}
+	
+	const char *Emitter::c_str() const
+	{
+		return m_stream.str();
+	}
+	
+	unsigned Emitter::size() const
+	{
+		return m_stream.pos();
+	}
+	
+	// state checking
+	bool Emitter::good() const
+	{
+		return m_pState->good();
+	}
+	
+	const std::string Emitter::GetLastError() const
+	{
+		return m_pState->GetLastError();
+	}
+
+	// global setters
+	bool Emitter::SetOutputCharset(EMITTER_MANIP value)
+	{
+		return m_pState->SetOutputCharset(value, GLOBAL);
+	}
+
+	bool Emitter::SetStringFormat(EMITTER_MANIP value)
+	{
+		return m_pState->SetStringFormat(value, GLOBAL);
+	}
+	
+	bool Emitter::SetBoolFormat(EMITTER_MANIP value)
+	{
+		bool ok = false;
+		if(m_pState->SetBoolFormat(value, GLOBAL))
+			ok = true;
+		if(m_pState->SetBoolCaseFormat(value, GLOBAL))
+			ok = true;
+		if(m_pState->SetBoolLengthFormat(value, GLOBAL))
+			ok = true;
+		return ok;
+	}
+	
+	bool Emitter::SetIntBase(EMITTER_MANIP value)
+	{
+		return m_pState->SetIntFormat(value, GLOBAL);
+	}
+	
+	bool Emitter::SetSeqFormat(EMITTER_MANIP value)
+	{
+		return m_pState->SetFlowType(GT_SEQ, value, GLOBAL);
+	}
+	
+	bool Emitter::SetMapFormat(EMITTER_MANIP value)
+	{
+		bool ok = false;
+		if(m_pState->SetFlowType(GT_MAP, value, GLOBAL))
+			ok = true;
+		if(m_pState->SetMapKeyFormat(value, GLOBAL))
+			ok = true;
+		return ok;
+	}
+	
+	bool Emitter::SetIndent(unsigned n)
+	{
+		return m_pState->SetIndent(n, GLOBAL);
+	}
+	
+	bool Emitter::SetPreCommentIndent(unsigned n)
+	{
+		return m_pState->SetPreCommentIndent(n, GLOBAL);
+	}
+	
+	bool Emitter::SetPostCommentIndent(unsigned n)
+	{
+		return m_pState->SetPostCommentIndent(n, GLOBAL);
+	}
+
+	// SetLocalValue
+	// . Either start/end a group, or set a modifier locally
+	Emitter& Emitter::SetLocalValue(EMITTER_MANIP value)
+	{
+		if(!good())
+			return *this;
+		
+		switch(value) {
+			case BeginDoc:
+				EmitBeginDoc();
+				break;
+			case EndDoc:
+				EmitEndDoc();
+				break;
+			case BeginSeq:
+				EmitBeginSeq();
+				break;
+			case EndSeq:
+				EmitEndSeq();
+				break;
+			case BeginMap:
+				EmitBeginMap();
+				break;
+			case EndMap:
+				EmitEndMap();
+				break;
+			case Key:
+				EmitKey();
+				break;
+			case Value:
+				EmitValue();
+				break;
+			case TagByKind:
+				EmitKindTag();
+				break;
+			case Newline:
+				EmitNewline();
+				break;
+			default:
+				m_pState->SetLocalValue(value);
+				break;
+		}
+		return *this;
+	}
+	
+	Emitter& Emitter::SetLocalIndent(const _Indent& indent)
+	{
+		m_pState->SetIndent(indent.value, LOCAL);
+		return *this;
+	}
+
+	// GotoNextPreAtomicState
+	// . Runs the state machine, emitting if necessary, and returns 'true' if done (i.e., ready to emit an atom)
+	bool Emitter::GotoNextPreAtomicState()
+	{
+		if(!good())
+			return true;
+		
+		unsigned curIndent = m_pState->GetCurIndent();
+		
+		EMITTER_STATE curState = m_pState->GetCurState();
+		switch(curState) {
+				// document-level
+			case ES_WAITING_FOR_DOC:
+				m_pState->SwitchState(ES_WRITING_DOC);
+				return true;
+			case ES_WRITING_DOC:
+				return true;
+			case ES_DONE_WITH_DOC:
+				EmitBeginDoc();
+				return false;
+				
+				// block sequence
+			case ES_WAITING_FOR_BLOCK_SEQ_ENTRY:
+				m_stream << IndentTo(curIndent) << "-";
+				m_pState->RequireSoftSeparation();
+				m_pState->SwitchState(ES_WRITING_BLOCK_SEQ_ENTRY);
+				return true;
+			case ES_WRITING_BLOCK_SEQ_ENTRY:
+				return true;
+			case ES_DONE_WITH_BLOCK_SEQ_ENTRY:
+				m_stream << '\n';
+				m_pState->SwitchState(ES_WAITING_FOR_BLOCK_SEQ_ENTRY);
+				return false;
+				
+				// flow sequence
+			case ES_WAITING_FOR_FLOW_SEQ_ENTRY:
+				m_pState->SwitchState(ES_WRITING_FLOW_SEQ_ENTRY);
+				return true;
+			case ES_WRITING_FLOW_SEQ_ENTRY:
+				return true;
+			case ES_DONE_WITH_FLOW_SEQ_ENTRY:
+				EmitSeparationIfNecessary();
+				m_stream << ',';
+				m_pState->RequireSoftSeparation();
+				m_pState->SwitchState(ES_WAITING_FOR_FLOW_SEQ_ENTRY);
+				return false;
+				
+				// block map
+			case ES_WAITING_FOR_BLOCK_MAP_ENTRY:
+				m_pState->SetError(ErrorMsg::EXPECTED_KEY_TOKEN);
+				return true;
+			case ES_WAITING_FOR_BLOCK_MAP_KEY:
+				if(m_pState->CurrentlyInLongKey()) {
+					m_stream << IndentTo(curIndent) << '?';
+					m_pState->RequireSoftSeparation();
+				}
+				m_pState->SwitchState(ES_WRITING_BLOCK_MAP_KEY);
+				return true;
+			case ES_WRITING_BLOCK_MAP_KEY:
+				return true;
+			case ES_DONE_WITH_BLOCK_MAP_KEY:
+				m_pState->SetError(ErrorMsg::EXPECTED_VALUE_TOKEN);
+				return true;
+			case ES_WAITING_FOR_BLOCK_MAP_VALUE:
+				m_pState->SwitchState(ES_WRITING_BLOCK_MAP_VALUE);
+				return true;
+			case ES_WRITING_BLOCK_MAP_VALUE:
+				return true;
+			case ES_DONE_WITH_BLOCK_MAP_VALUE:
+				m_pState->SetError(ErrorMsg::EXPECTED_KEY_TOKEN);
+				return true;
+				
+				// flow map
+			case ES_WAITING_FOR_FLOW_MAP_ENTRY:
+				m_pState->SetError(ErrorMsg::EXPECTED_KEY_TOKEN);
+				return true;
+			case ES_WAITING_FOR_FLOW_MAP_KEY:
+				EmitSeparationIfNecessary();
+				m_pState->SwitchState(ES_WRITING_FLOW_MAP_KEY);
+				if(m_pState->CurrentlyInLongKey()) {
+					m_stream << '?';
+					m_pState->RequireSoftSeparation();
+				}
+				return true;
+			case ES_WRITING_FLOW_MAP_KEY:
+				return true;
+			case ES_DONE_WITH_FLOW_MAP_KEY:
+				m_pState->SetError(ErrorMsg::EXPECTED_VALUE_TOKEN);
+				return true;
+			case ES_WAITING_FOR_FLOW_MAP_VALUE:
+				EmitSeparationIfNecessary();
+				m_stream << ':';
+				m_pState->RequireSoftSeparation();
+				m_pState->SwitchState(ES_WRITING_FLOW_MAP_VALUE);
+				return true;
+			case ES_WRITING_FLOW_MAP_VALUE:
+				return true;
+			case ES_DONE_WITH_FLOW_MAP_VALUE:
+				m_pState->SetError(ErrorMsg::EXPECTED_KEY_TOKEN);
+				return true;
+			default:
+				assert(false);
+		}
+
+		assert(false);
+		return true;
+	}
+	
+	// PreAtomicWrite
+	// . Depending on the emitter state, write to the stream to get it
+	//   in position to do an atomic write (e.g., scalar, sequence, or map)
+	void Emitter::PreAtomicWrite()
+	{
+		if(!good())
+			return;
+		
+		while(!GotoNextPreAtomicState())
+			;
+	}
+	
+	// PostAtomicWrite
+	// . Clean up
+	void Emitter::PostAtomicWrite()
+	{
+		if(!good())
+			return;
+		
+		EMITTER_STATE curState = m_pState->GetCurState();
+		switch(curState) {
+				// document-level
+			case ES_WRITING_DOC:
+				m_pState->SwitchState(ES_DONE_WITH_DOC);
+				break;
+
+				// block seq
+			case ES_WRITING_BLOCK_SEQ_ENTRY:
+				m_pState->SwitchState(ES_DONE_WITH_BLOCK_SEQ_ENTRY);
+				break;
+				
+				// flow seq
+			case ES_WRITING_FLOW_SEQ_ENTRY:
+				m_pState->SwitchState(ES_DONE_WITH_FLOW_SEQ_ENTRY);
+				break;
+				
+				// block map
+			case ES_WRITING_BLOCK_MAP_KEY:
+				if(!m_pState->CurrentlyInLongKey()) {
+					m_stream << ':';
+					m_pState->RequireSoftSeparation();
+				}
+				m_pState->SwitchState(ES_DONE_WITH_BLOCK_MAP_KEY);
+				break;
+			case ES_WRITING_BLOCK_MAP_VALUE:
+				m_pState->SwitchState(ES_DONE_WITH_BLOCK_MAP_VALUE);
+				break;
+				
+				// flow map
+			case ES_WRITING_FLOW_MAP_KEY:
+				m_pState->SwitchState(ES_DONE_WITH_FLOW_MAP_KEY);
+				break;
+			case ES_WRITING_FLOW_MAP_VALUE:
+				m_pState->SwitchState(ES_DONE_WITH_FLOW_MAP_VALUE);
+				break;
+			default:
+				assert(false);
+		};
+				
+		m_pState->ClearModifiedSettings();
+	}
+	
+	// EmitSeparationIfNecessary
+	void Emitter::EmitSeparationIfNecessary()
+	{
+		if(!good())
+			return;
+		
+		if(m_pState->RequiresSoftSeparation())
+			m_stream << ' ';
+		else if(m_pState->RequiresHardSeparation())
+			m_stream << '\n';
+		m_pState->UnsetSeparation();
+	}
+	
+	// EmitBeginDoc
+	void Emitter::EmitBeginDoc()
+	{
+		if(!good())
+			return;
+		
+		EMITTER_STATE curState = m_pState->GetCurState();
+		if(curState != ES_WAITING_FOR_DOC && curState != ES_WRITING_DOC && curState != ES_DONE_WITH_DOC) {
+			m_pState->SetError("Unexpected begin document");
+			return;
+		}
+		
+		if(curState == ES_WRITING_DOC || curState == ES_DONE_WITH_DOC)
+			m_stream << '\n';		
+		m_stream << "---\n";
+
+		m_pState->UnsetSeparation();
+		m_pState->SwitchState(ES_WAITING_FOR_DOC);
+	}
+	
+	// EmitEndDoc
+	void Emitter::EmitEndDoc()
+	{
+		if(!good())
+			return;
+
+		
+		EMITTER_STATE curState = m_pState->GetCurState();
+		if(curState != ES_WAITING_FOR_DOC && curState != ES_WRITING_DOC && curState != ES_DONE_WITH_DOC) {
+			m_pState->SetError("Unexpected end document");
+			return;
+		}
+		
+		if(curState == ES_WRITING_DOC || curState == ES_DONE_WITH_DOC)
+			m_stream << '\n';		
+		m_stream << "...\n";
+		
+		m_pState->UnsetSeparation();
+		m_pState->SwitchState(ES_WAITING_FOR_DOC);
+	}
+
+	// EmitBeginSeq
+	void Emitter::EmitBeginSeq()
+	{
+		if(!good())
+			return;
+		
+		// must have a long key if we're emitting a sequence
+		m_pState->StartLongKey();
+		
+		PreAtomicWrite();
+		
+		EMITTER_STATE curState = m_pState->GetCurState();
+		EMITTER_MANIP flowType = m_pState->GetFlowType(GT_SEQ);
+		if(flowType == Block) {
+			if(curState == ES_WRITING_BLOCK_SEQ_ENTRY ||
+			   curState == ES_WRITING_BLOCK_MAP_KEY || curState == ES_WRITING_BLOCK_MAP_VALUE ||
+			   curState == ES_WRITING_DOC
+			) {
+				if(m_pState->RequiresHardSeparation() || curState != ES_WRITING_DOC) {
+					m_stream << "\n";
+					m_pState->UnsetSeparation();
+				}
+			}
+			m_pState->PushState(ES_WAITING_FOR_BLOCK_SEQ_ENTRY);
+		} else if(flowType == Flow) {
+			EmitSeparationIfNecessary();
+			m_stream << "[";
+			m_pState->PushState(ES_WAITING_FOR_FLOW_SEQ_ENTRY);
+		} else
+			assert(false);
+
+		m_pState->BeginGroup(GT_SEQ);
+	}
+	
+	// EmitEndSeq
+	void Emitter::EmitEndSeq()
+	{
+		if(!good())
+			return;
+		
+		if(m_pState->GetCurGroupType() != GT_SEQ)
+			return m_pState->SetError(ErrorMsg::UNEXPECTED_END_SEQ);
+		
+		EMITTER_STATE curState = m_pState->GetCurState();
+		FLOW_TYPE flowType = m_pState->GetCurGroupFlowType();
+		if(flowType == FT_BLOCK) {
+			// Note: block sequences are *not* allowed to be empty, but we convert it
+			//       to a flow sequence if it is
+			assert(curState == ES_DONE_WITH_BLOCK_SEQ_ENTRY || curState == ES_WAITING_FOR_BLOCK_SEQ_ENTRY);
+			if(curState == ES_WAITING_FOR_BLOCK_SEQ_ENTRY) {
+				// Note: only one of these will actually output anything for a given situation
+				EmitSeparationIfNecessary();
+				unsigned curIndent = m_pState->GetCurIndent();
+				m_stream << IndentTo(curIndent);
+
+				m_stream << "[]";
+			}
+		} else if(flowType == FT_FLOW) {
+			// Note: flow sequences are allowed to be empty
+			assert(curState == ES_DONE_WITH_FLOW_SEQ_ENTRY || curState == ES_WAITING_FOR_FLOW_SEQ_ENTRY);
+			m_stream << "]";
+		} else
+			assert(false);
+		
+		m_pState->PopState();
+		m_pState->EndGroup(GT_SEQ);
+
+		PostAtomicWrite();
+	}
+	
+	// EmitBeginMap
+	void Emitter::EmitBeginMap()
+	{
+		if(!good())
+			return;
+		
+		// must have a long key if we're emitting a map
+		m_pState->StartLongKey();
+
+		PreAtomicWrite();
+
+		EMITTER_STATE curState = m_pState->GetCurState();
+		EMITTER_MANIP flowType = m_pState->GetFlowType(GT_MAP);
+		if(flowType == Block) {
+			if(curState == ES_WRITING_BLOCK_SEQ_ENTRY ||
+			   curState == ES_WRITING_BLOCK_MAP_KEY || curState == ES_WRITING_BLOCK_MAP_VALUE ||
+			   curState == ES_WRITING_DOC
+			) {
+				if(m_pState->RequiresHardSeparation() || (curState != ES_WRITING_DOC && curState != ES_WRITING_BLOCK_SEQ_ENTRY)) {
+					m_stream << "\n";
+					m_pState->UnsetSeparation();
+				}
+			}
+			m_pState->PushState(ES_WAITING_FOR_BLOCK_MAP_ENTRY);
+		} else if(flowType == Flow) {
+			EmitSeparationIfNecessary();
+			m_stream << "{";
+			m_pState->PushState(ES_WAITING_FOR_FLOW_MAP_ENTRY);
+		} else
+			assert(false);
+		
+		m_pState->BeginGroup(GT_MAP);
+	}
+	
+	// EmitEndMap
+	void Emitter::EmitEndMap()
+	{
+		if(!good())
+			return;
+		
+		if(m_pState->GetCurGroupType() != GT_MAP)
+			return m_pState->SetError(ErrorMsg::UNEXPECTED_END_MAP);
+
+		EMITTER_STATE curState = m_pState->GetCurState();
+		FLOW_TYPE flowType = m_pState->GetCurGroupFlowType();
+		if(flowType == FT_BLOCK) {
+			// Note: block sequences are *not* allowed to be empty, but we convert it
+			//       to a flow sequence if it is
+			assert(curState == ES_DONE_WITH_BLOCK_MAP_VALUE || curState == ES_WAITING_FOR_BLOCK_MAP_ENTRY);
+			if(curState == ES_WAITING_FOR_BLOCK_MAP_ENTRY) {
+				// Note: only one of these will actually output anything for a given situation
+				EmitSeparationIfNecessary();
+				unsigned curIndent = m_pState->GetCurIndent();
+				m_stream << IndentTo(curIndent);
+				m_stream << "{}";
+			}
+		} else if(flowType == FT_FLOW) {
+			// Note: flow maps are allowed to be empty
+			assert(curState == ES_DONE_WITH_FLOW_MAP_VALUE || curState == ES_WAITING_FOR_FLOW_MAP_ENTRY);
+			EmitSeparationIfNecessary();
+			m_stream << "}";
+		} else
+			assert(false);
+		
+		m_pState->PopState();
+		m_pState->EndGroup(GT_MAP);
+		
+		PostAtomicWrite();
+	}
+	
+	// EmitKey
+	void Emitter::EmitKey()
+	{
+		if(!good())
+			return;
+		
+		EMITTER_STATE curState = m_pState->GetCurState();
+		FLOW_TYPE flowType = m_pState->GetCurGroupFlowType();
+		if(curState != ES_WAITING_FOR_BLOCK_MAP_ENTRY && curState != ES_DONE_WITH_BLOCK_MAP_VALUE
+		   && curState != ES_WAITING_FOR_FLOW_MAP_ENTRY && curState != ES_DONE_WITH_FLOW_MAP_VALUE)
+			return m_pState->SetError(ErrorMsg::UNEXPECTED_KEY_TOKEN);
+
+		if(flowType == FT_BLOCK) {
+			if(curState == ES_DONE_WITH_BLOCK_MAP_VALUE)
+				m_stream << '\n';
+			unsigned curIndent = m_pState->GetCurIndent();
+			m_stream << IndentTo(curIndent);
+			m_pState->UnsetSeparation();
+			m_pState->SwitchState(ES_WAITING_FOR_BLOCK_MAP_KEY);
+		} else if(flowType == FT_FLOW) {
+			EmitSeparationIfNecessary();
+			if(curState == ES_DONE_WITH_FLOW_MAP_VALUE) {
+				m_stream << ',';
+				m_pState->RequireSoftSeparation();
+			}
+			m_pState->SwitchState(ES_WAITING_FOR_FLOW_MAP_KEY);
+		} else
+			assert(false);
+		
+		if(m_pState->GetMapKeyFormat() == LongKey)
+			m_pState->StartLongKey();
+		else if(m_pState->GetMapKeyFormat() == Auto)
+			m_pState->StartSimpleKey();
+		else
+			assert(false);
+	}
+	
+	// EmitValue
+	void Emitter::EmitValue()
+	{
+		if(!good())
+			return;
+
+		EMITTER_STATE curState = m_pState->GetCurState();
+		FLOW_TYPE flowType = m_pState->GetCurGroupFlowType();
+		if(curState != ES_DONE_WITH_BLOCK_MAP_KEY && curState != ES_DONE_WITH_FLOW_MAP_KEY)
+			return m_pState->SetError(ErrorMsg::UNEXPECTED_VALUE_TOKEN);
+
+		if(flowType == FT_BLOCK) {
+			if(m_pState->CurrentlyInLongKey()) {
+				m_stream << '\n';
+				m_stream << IndentTo(m_pState->GetCurIndent());
+				m_stream << ':';
+				m_pState->RequireSoftSeparation();
+			}
+			m_pState->SwitchState(ES_WAITING_FOR_BLOCK_MAP_VALUE);
+		} else if(flowType == FT_FLOW) {
+			m_pState->SwitchState(ES_WAITING_FOR_FLOW_MAP_VALUE);
+		} else
+			assert(false);
+	}
+
+	// EmitNewline
+	void Emitter::EmitNewline()
+	{
+		if(!good())
+			return;
+
+		if(CanEmitNewline()) {
+			m_stream << '\n';
+			m_pState->UnsetSeparation();
+		}
+	}
+
+	bool Emitter::CanEmitNewline() const
+	{
+		FLOW_TYPE flowType = m_pState->GetCurGroupFlowType();
+		if(flowType == FT_BLOCK && m_pState->CurrentlyInLongKey())
+			return true;
+
+		EMITTER_STATE curState = m_pState->GetCurState();
+		return curState != ES_DONE_WITH_BLOCK_MAP_KEY && curState != ES_WAITING_FOR_BLOCK_MAP_VALUE && curState != ES_WRITING_BLOCK_MAP_VALUE;
+	}
+
+	// *******************************************************************************************
+	// overloads of Write
+	
+	Emitter& Emitter::Write(const std::string& str)
+	{
+		if(!good())
+			return *this;
+		
+		// literal scalars must use long keys
+		if(m_pState->GetStringFormat() == Literal && m_pState->GetCurGroupFlowType() != FT_FLOW)
+			m_pState->StartLongKey();
+		
+		PreAtomicWrite();
+		EmitSeparationIfNecessary();
+		
+		bool escapeNonAscii = m_pState->GetOutputCharset() == EscapeNonAscii;
+		EMITTER_MANIP strFmt = m_pState->GetStringFormat();
+		FLOW_TYPE flowType = m_pState->GetCurGroupFlowType();
+		unsigned curIndent = m_pState->GetCurIndent();
+
+		switch(strFmt) {
+			case Auto:
+				Utils::WriteString(m_stream, str, flowType == FT_FLOW, escapeNonAscii);
+				break;
+			case SingleQuoted:
+				if(!Utils::WriteSingleQuotedString(m_stream, str)) {
+					m_pState->SetError(ErrorMsg::SINGLE_QUOTED_CHAR);
+					return *this;
+				}
+				break;
+			case DoubleQuoted:
+				Utils::WriteDoubleQuotedString(m_stream, str, escapeNonAscii);
+				break;
+			case Literal:
+				if(flowType == FT_FLOW)
+					Utils::WriteString(m_stream, str, flowType == FT_FLOW, escapeNonAscii);
+				else
+					Utils::WriteLiteralString(m_stream, str, curIndent + m_pState->GetIndent());
+				break;
+			default:
+				assert(false);
+		}
+		
+		PostAtomicWrite();
+		return *this;
+	}
+
+	void Emitter::PreWriteIntegralType(std::stringstream& str)
+	{
+		PreAtomicWrite();
+		EmitSeparationIfNecessary();
+		
+		EMITTER_MANIP intFmt = m_pState->GetIntFormat();
+		switch(intFmt) {
+			case Dec:
+				str << std::dec;
+				break;
+			case Hex:
+				str << std::hex;
+				break;
+				case Oct:
+				str << std::oct;
+				break;
+			default:
+				assert(false);
+		}
+	}
+
+	void Emitter::PreWriteStreamable(std::stringstream& str)
+	{
+		PreAtomicWrite();
+		EmitSeparationIfNecessary();
+		str.precision(15);
+	}
+
+	void Emitter::PostWriteIntegralType(const std::stringstream& str)
+	{
+		m_stream << str.str();
+		PostAtomicWrite();
+	}
+
+	void Emitter::PostWriteStreamable(const std::stringstream& str)
+	{
+		m_stream << str.str();
+		PostAtomicWrite();
+	}
+
+	const char *Emitter::ComputeFullBoolName(bool b) const
+	{
+		const EMITTER_MANIP mainFmt = (m_pState->GetBoolLengthFormat() == ShortBool ? YesNoBool : m_pState->GetBoolFormat());
+		const EMITTER_MANIP caseFmt = m_pState->GetBoolCaseFormat();
+		switch(mainFmt) {
+			case YesNoBool:
+				switch(caseFmt) {
+					case UpperCase: return b ? "YES" : "NO";
+					case CamelCase: return b ? "Yes" : "No";
+					case LowerCase: return b ? "yes" : "no";
+					default: break;
+				}
+				break;
+			case OnOffBool:
+				switch(caseFmt) {
+					case UpperCase: return b ? "ON" : "OFF";
+					case CamelCase: return b ? "On" : "Off";
+					case LowerCase: return b ? "on" : "off";
+					default: break;
+				}
+				break;
+			case TrueFalseBool:
+				switch(caseFmt) {
+					case UpperCase: return b ? "TRUE" : "FALSE";
+					case CamelCase: return b ? "True" : "False";
+					case LowerCase: return b ? "true" : "false";
+					default: break;
+				}
+				break;
+			default:
+				break;
+		}
+		return b ? "y" : "n"; // should never get here, but it can't hurt to give these answers
+	}
+
+	Emitter& Emitter::Write(bool b)
+	{
+		if(!good())
+			return *this;
+		
+		PreAtomicWrite();
+		EmitSeparationIfNecessary();
+	
+		const char *name = ComputeFullBoolName(b);
+		if(m_pState->GetBoolLengthFormat() == ShortBool)
+			m_stream << name[0];
+		else
+			m_stream << name;
+
+		PostAtomicWrite();
+		return *this;
+	}
+
+	Emitter& Emitter::Write(const _Alias& alias)
+	{
+		if(!good())
+			return *this;
+		
+		PreAtomicWrite();
+		EmitSeparationIfNecessary();
+		if(!Utils::WriteAlias(m_stream, alias.content)) {
+			m_pState->SetError(ErrorMsg::INVALID_ALIAS);
+			return *this;
+		}
+		PostAtomicWrite();
+		return *this;
+	}
+	
+	Emitter& Emitter::Write(const _Anchor& anchor)
+	{
+		if(!good())
+			return *this;
+		
+		PreAtomicWrite();
+		EmitSeparationIfNecessary();
+		if(!Utils::WriteAnchor(m_stream, anchor.content)) {
+			m_pState->SetError(ErrorMsg::INVALID_ANCHOR);
+			return *this;
+		}
+		m_pState->RequireHardSeparation();
+		// Note: no PostAtomicWrite() because we need another value for this node
+		return *this;
+	}
+	
+	Emitter& Emitter::Write(const _Tag& tag)
+	{
+		if(!good())
+			return *this;
+
+		PreAtomicWrite();
+		EmitSeparationIfNecessary();
+		
+		bool success = false;
+		if(tag.type == _Tag::Type::Verbatim)
+			success = Utils::WriteTag(m_stream, tag.content, true);
+		else if(tag.type == _Tag::Type::PrimaryHandle)
+			success = Utils::WriteTag(m_stream, tag.content, false);
+		else
+			success = Utils::WriteTagWithPrefix(m_stream, tag.prefix, tag.content);
+		
+		if(!success) {
+			m_pState->SetError(ErrorMsg::INVALID_TAG);
+			return *this;
+		}
+		
+		m_pState->RequireHardSeparation();
+		// Note: no PostAtomicWrite() because we need another value for this node
+		return *this;
+	}
+
+	void Emitter::EmitKindTag()
+	{
+		Write(LocalTag(""));
+	}
+
+	Emitter& Emitter::Write(const _Comment& comment)
+	{
+		if(!good())
+			return *this;
+		
+		if(m_stream.col() > 0)
+			m_stream << Indentation(m_pState->GetPreCommentIndent());
+		Utils::WriteComment(m_stream, comment.content, m_pState->GetPostCommentIndent());
+		m_pState->RequireHardSeparation();
+		m_pState->ForceHardSeparation();
+		
+		return *this;
+	}
+
+	Emitter& Emitter::Write(const _Null& /*null*/)
+	{
+		if(!good())
+			return *this;
+		
+		PreAtomicWrite();
+		EmitSeparationIfNecessary();
+		m_stream << "~";
+		PostAtomicWrite();
+		return *this;
+	}
+
+	Emitter& Emitter::Write(const _Binary& binary)
+	{
+		Write(SecondaryTag("binary"));
+
+		if(!good())
+			return *this;
+		
+		PreAtomicWrite();
+		EmitSeparationIfNecessary();
+		Utils::WriteBinary(m_stream, binary.data, binary.size);
+		PostAtomicWrite();
+		return *this;
+	}
+}
+

src/emitterstate.cpp

@@ -0,0 +1,265 @@
+#include "emitterstate.h"
+#include "yaml-cpp/exceptions.h"
+
+namespace YAML
+{
+	EmitterState::EmitterState(): m_isGood(true), m_curIndent(0), m_requiresSoftSeparation(false), m_requiresHardSeparation(false)
+	{
+		// start up
+		m_stateStack.push(ES_WAITING_FOR_DOC);
+		
+		// set default global manipulators
+		m_charset.set(EmitNonAscii);
+		m_strFmt.set(Auto);
+		m_boolFmt.set(TrueFalseBool);
+		m_boolLengthFmt.set(LongBool);
+		m_boolCaseFmt.set(LowerCase);
+		m_intFmt.set(Dec);
+		m_indent.set(2);
+		m_preCommentIndent.set(2);
+		m_postCommentIndent.set(1);
+		m_seqFmt.set(Block);
+		m_mapFmt.set(Block);
+		m_mapKeyFmt.set(Auto);
+	}
+	
+	EmitterState::~EmitterState()
+	{
+	}
+
+	// SetLocalValue
+	// . We blindly tries to set all possible formatters to this value
+	// . Only the ones that make sense will be accepted
+	void EmitterState::SetLocalValue(EMITTER_MANIP value)
+	{
+		SetOutputCharset(value, LOCAL);
+		SetStringFormat(value, LOCAL);
+		SetBoolFormat(value, LOCAL);
+		SetBoolCaseFormat(value, LOCAL);
+		SetBoolLengthFormat(value, LOCAL);
+		SetIntFormat(value, LOCAL);
+		SetFlowType(GT_SEQ, value, LOCAL);
+		SetFlowType(GT_MAP, value, LOCAL);
+		SetMapKeyFormat(value, LOCAL);
+	}
+	
+	void EmitterState::BeginGroup(GROUP_TYPE type)
+	{
+		unsigned lastIndent = (m_groups.empty() ? 0 : m_groups.top().indent);
+		m_curIndent += lastIndent;
+		
+		std::auto_ptr<Group> pGroup(new Group(type));
+		
+		// transfer settings (which last until this group is done)
+		pGroup->modifiedSettings = m_modifiedSettings;
+
+		// set up group
+		pGroup->flow = GetFlowType(type);
+		pGroup->indent = GetIndent();
+		pGroup->usingLongKey = (GetMapKeyFormat() == LongKey ? true : false);
+
+		m_groups.push(pGroup);
+	}
+	
+	void EmitterState::EndGroup(GROUP_TYPE type)
+	{
+		if(m_groups.empty())
+			return SetError(ErrorMsg::UNMATCHED_GROUP_TAG);
+		
+		// get rid of the current group
+		{
+			std::auto_ptr<Group> pFinishedGroup = m_groups.pop();
+			if(pFinishedGroup->type != type)
+				return SetError(ErrorMsg::UNMATCHED_GROUP_TAG);
+		}
+
+		// reset old settings
+		unsigned lastIndent = (m_groups.empty() ? 0 : m_groups.top().indent);
+		assert(m_curIndent >= lastIndent);
+		m_curIndent -= lastIndent;
+		
+		// some global settings that we changed may have been overridden
+		// by a local setting we just popped, so we need to restore them
+		m_globalModifiedSettings.restore();
+	}
+		
+	GROUP_TYPE EmitterState::GetCurGroupType() const
+	{
+		if(m_groups.empty())
+			return GT_NONE;
+		
+		return m_groups.top().type;
+	}
+	
+	FLOW_TYPE EmitterState::GetCurGroupFlowType() const
+	{
+		if(m_groups.empty())
+			return FT_NONE;
+		
+		return (m_groups.top().flow == Flow ? FT_FLOW : FT_BLOCK);
+	}
+	
+	bool EmitterState::CurrentlyInLongKey()
+	{
+		if(m_groups.empty())
+			return false;
+		return m_groups.top().usingLongKey;
+	}
+	
+	void EmitterState::StartLongKey()
+	{
+		if(!m_groups.empty())
+			m_groups.top().usingLongKey = true;
+	}
+	
+	void EmitterState::StartSimpleKey()
+	{
+		if(!m_groups.empty())
+			m_groups.top().usingLongKey = false;
+	}
+
+	void EmitterState::ClearModifiedSettings()
+	{
+		m_modifiedSettings.clear();
+	}
+
+	bool EmitterState::SetOutputCharset(EMITTER_MANIP value, FMT_SCOPE scope)
+	{
+		switch(value) {
+			case EmitNonAscii:
+			case EscapeNonAscii:
+				_Set(m_charset, value, scope);
+				return true;
+			default:
+				return false;
+		}
+	}
+	
+	bool EmitterState::SetStringFormat(EMITTER_MANIP value, FMT_SCOPE scope)
+	{
+		switch(value) {
+			case Auto:
+			case SingleQuoted:
+			case DoubleQuoted:
+			case Literal:
+				_Set(m_strFmt, value, scope);
+				return true;
+			default:
+				return false;
+		}
+	}
+	
+	bool EmitterState::SetBoolFormat(EMITTER_MANIP value, FMT_SCOPE scope)
+	{
+		switch(value) {
+			case OnOffBool:
+			case TrueFalseBool:
+			case YesNoBool:
+				_Set(m_boolFmt, value, scope);
+				return true;
+			default:
+				return false;
+		}
+	}
+
+	bool EmitterState::SetBoolLengthFormat(EMITTER_MANIP value, FMT_SCOPE scope)
+	{
+		switch(value) {
+			case LongBool:
+			case ShortBool:
+				_Set(m_boolLengthFmt, value, scope);
+				return true;
+			default:
+				return false;
+		}
+	}
+
+	bool EmitterState::SetBoolCaseFormat(EMITTER_MANIP value, FMT_SCOPE scope)
+	{
+		switch(value) {
+			case UpperCase:
+			case LowerCase:
+			case CamelCase:
+				_Set(m_boolCaseFmt, value, scope);
+				return true;
+			default:
+				return false;
+		}
+	}
+
+	bool EmitterState::SetIntFormat(EMITTER_MANIP value, FMT_SCOPE scope)
+	{
+		switch(value) {
+			case Dec:
+			case Hex:
+			case Oct:
+				_Set(m_intFmt, value, scope);
+				return true;
+			default:
+				return false;
+		}
+	}
+
+	bool EmitterState::SetIndent(unsigned value, FMT_SCOPE scope)
+	{
+		if(value == 0)
+			return false;
+		
+		_Set(m_indent, value, scope);
+		return true;
+	}
+
+	bool EmitterState::SetPreCommentIndent(unsigned value, FMT_SCOPE scope)
+	{
+		if(value == 0)
+			return false;
+		
+		_Set(m_preCommentIndent, value, scope);
+		return true;
+	}
+	
+	bool EmitterState::SetPostCommentIndent(unsigned value, FMT_SCOPE scope)
+	{
+		if(value == 0)
+			return false;
+		
+		_Set(m_postCommentIndent, value, scope);
+		return true;
+	}
+
+	bool EmitterState::SetFlowType(GROUP_TYPE groupType, EMITTER_MANIP value, FMT_SCOPE scope)
+	{
+		switch(value) {
+			case Block:
+			case Flow:
+				_Set(groupType == GT_SEQ ? m_seqFmt : m_mapFmt, value, scope);
+				return true;
+			default:
+				return false;
+		}
+	}
+
+	EMITTER_MANIP EmitterState::GetFlowType(GROUP_TYPE groupType) const
+	{
+		// force flow style if we're currently in a flow
+		FLOW_TYPE flowType = GetCurGroupFlowType();
+		if(flowType == FT_FLOW)
+			return Flow;
+		
+		// otherwise, go with what's asked of use
+		return (groupType == GT_SEQ ? m_seqFmt.get() : m_mapFmt.get());
+	}
+	
+	bool EmitterState::SetMapKeyFormat(EMITTER_MANIP value, FMT_SCOPE scope)
+	{
+		switch(value) {
+			case Auto:
+			case LongKey:
+				_Set(m_mapKeyFmt, value, scope);
+				return true;
+			default:
+				return false;
+		}
+	}
+}
+

src/emitterstate.h

@@ -0,0 +1,210 @@
+#ifndef EMITTERSTATE_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define EMITTERSTATE_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "ptr_stack.h"
+#include "setting.h"
+#include "yaml-cpp/emittermanip.h"
+#include <cassert>
+#include <vector>
+#include <stack>
+#include <memory>
+
+namespace YAML
+{
+	enum FMT_SCOPE {
+		LOCAL,
+		GLOBAL
+	};
+	
+	enum GROUP_TYPE {
+		GT_NONE,
+		GT_SEQ,
+		GT_MAP
+	};
+	
+	enum FLOW_TYPE {
+		FT_NONE,
+		FT_FLOW,
+		FT_BLOCK
+	};
+	
+	enum NODE_STATE {
+		NS_START,
+		NS_READY_FOR_ATOM,
+		NS_END
+	};
+	
+	enum EMITTER_STATE {
+		ES_WAITING_FOR_DOC,
+		ES_WRITING_DOC,
+		ES_DONE_WITH_DOC,
+		
+		// block seq
+		ES_WAITING_FOR_BLOCK_SEQ_ENTRY,
+		ES_WRITING_BLOCK_SEQ_ENTRY,
+		ES_DONE_WITH_BLOCK_SEQ_ENTRY,
+		
+		// flow seq
+		ES_WAITING_FOR_FLOW_SEQ_ENTRY,
+		ES_WRITING_FLOW_SEQ_ENTRY,
+		ES_DONE_WITH_FLOW_SEQ_ENTRY,
+		
+		// block map
+		ES_WAITING_FOR_BLOCK_MAP_ENTRY,
+		ES_WAITING_FOR_BLOCK_MAP_KEY,
+		ES_WRITING_BLOCK_MAP_KEY,
+		ES_DONE_WITH_BLOCK_MAP_KEY,
+		ES_WAITING_FOR_BLOCK_MAP_VALUE,
+		ES_WRITING_BLOCK_MAP_VALUE,
+		ES_DONE_WITH_BLOCK_MAP_VALUE,
+		
+		// flow map
+		ES_WAITING_FOR_FLOW_MAP_ENTRY,
+		ES_WAITING_FOR_FLOW_MAP_KEY,
+		ES_WRITING_FLOW_MAP_KEY,
+		ES_DONE_WITH_FLOW_MAP_KEY,
+		ES_WAITING_FOR_FLOW_MAP_VALUE,
+		ES_WRITING_FLOW_MAP_VALUE,
+		ES_DONE_WITH_FLOW_MAP_VALUE
+	};
+	
+	class EmitterState
+	{
+	public:
+		EmitterState();
+		~EmitterState();
+		
+		// basic state checking
+		bool good() const { return m_isGood; }
+		const std::string GetLastError() const { return m_lastError; }
+		void SetError(const std::string& error) { m_isGood = false; m_lastError = error; }
+		
+		// main state of the machine
+		EMITTER_STATE GetCurState() const { return m_stateStack.top(); }
+		void SwitchState(EMITTER_STATE state) { PopState(); PushState(state); }
+		void PushState(EMITTER_STATE state) { m_stateStack.push(state); }
+		void PopState() { m_stateStack.pop(); }
+		
+		void SetLocalValue(EMITTER_MANIP value);
+		
+		// group handling
+		void BeginGroup(GROUP_TYPE type);
+		void EndGroup(GROUP_TYPE type);
+		
+		GROUP_TYPE GetCurGroupType() const;
+		FLOW_TYPE GetCurGroupFlowType() const;
+		int GetCurIndent() const { return m_curIndent; }
+		
+		bool CurrentlyInLongKey();
+		void StartLongKey();
+		void StartSimpleKey();
+
+		bool RequiresSoftSeparation() const { return m_requiresSoftSeparation; }
+		bool RequiresHardSeparation() const { return m_requiresHardSeparation; }
+		void RequireSoftSeparation() { m_requiresSoftSeparation = true; }
+		void RequireHardSeparation() { m_requiresSoftSeparation = true; m_requiresHardSeparation = true; }
+		void ForceHardSeparation() { m_requiresSoftSeparation = false; }
+		void UnsetSeparation() { m_requiresSoftSeparation = false; m_requiresHardSeparation = false; }
+
+		void ClearModifiedSettings();
+
+		// formatters
+		bool SetOutputCharset(EMITTER_MANIP value, FMT_SCOPE scope);
+		EMITTER_MANIP GetOutputCharset() const { return m_charset.get(); }
+
+		bool SetStringFormat(EMITTER_MANIP value, FMT_SCOPE scope);
+		EMITTER_MANIP GetStringFormat() const { return m_strFmt.get(); }
+		
+		bool SetBoolFormat(EMITTER_MANIP value, FMT_SCOPE scope);
+		EMITTER_MANIP GetBoolFormat() const { return m_boolFmt.get(); }
+
+		bool SetBoolLengthFormat(EMITTER_MANIP value, FMT_SCOPE scope);
+		EMITTER_MANIP GetBoolLengthFormat() const { return m_boolLengthFmt.get(); }
+
+		bool SetBoolCaseFormat(EMITTER_MANIP value, FMT_SCOPE scope);
+		EMITTER_MANIP GetBoolCaseFormat() const { return m_boolCaseFmt.get(); }
+
+		bool SetIntFormat(EMITTER_MANIP value, FMT_SCOPE scope);
+		EMITTER_MANIP GetIntFormat() const { return m_intFmt.get(); }
+
+		bool SetIndent(unsigned value, FMT_SCOPE scope);
+		int GetIndent() const { return m_indent.get(); }
+		
+		bool SetPreCommentIndent(unsigned value, FMT_SCOPE scope);
+		int GetPreCommentIndent() const { return m_preCommentIndent.get(); }
+		bool SetPostCommentIndent(unsigned value, FMT_SCOPE scope);
+		int GetPostCommentIndent() const { return m_postCommentIndent.get(); }
+		
+		bool SetFlowType(GROUP_TYPE groupType, EMITTER_MANIP value, FMT_SCOPE scope);
+		EMITTER_MANIP GetFlowType(GROUP_TYPE groupType) const;
+		
+		bool SetMapKeyFormat(EMITTER_MANIP value, FMT_SCOPE scope);
+		EMITTER_MANIP GetMapKeyFormat() const { return m_mapKeyFmt.get(); }
+		
+	private:
+		template <typename T>
+		void _Set(Setting<T>& fmt, T value, FMT_SCOPE scope);
+		
+	private:
+		// basic state ok?
+		bool m_isGood;
+		std::string m_lastError;
+		
+		// other state
+		std::stack<EMITTER_STATE> m_stateStack;
+		
+		Setting<EMITTER_MANIP> m_charset;
+		Setting<EMITTER_MANIP> m_strFmt;
+		Setting<EMITTER_MANIP> m_boolFmt;
+		Setting<EMITTER_MANIP> m_boolLengthFmt;
+		Setting<EMITTER_MANIP> m_boolCaseFmt;
+		Setting<EMITTER_MANIP> m_intFmt;
+		Setting<unsigned> m_indent;
+		Setting<unsigned> m_preCommentIndent, m_postCommentIndent;
+		Setting<EMITTER_MANIP> m_seqFmt;
+		Setting<EMITTER_MANIP> m_mapFmt;
+		Setting<EMITTER_MANIP> m_mapKeyFmt;
+		
+		SettingChanges m_modifiedSettings;
+		SettingChanges m_globalModifiedSettings;
+		
+		struct Group {
+			Group(GROUP_TYPE type_): type(type_), usingLongKey(false), indent(0) {}
+			
+			GROUP_TYPE type;
+			EMITTER_MANIP flow;
+			bool usingLongKey;
+			int indent;
+			
+			SettingChanges modifiedSettings;
+		};
+
+		ptr_stack<Group> m_groups;
+		unsigned m_curIndent;
+		bool m_requiresSoftSeparation;
+		bool m_requiresHardSeparation;
+	};
+
+	template <typename T>
+	void EmitterState::_Set(Setting<T>& fmt, T value, FMT_SCOPE scope) {
+		switch(scope) {
+			case LOCAL:
+				m_modifiedSettings.push(fmt.set(value));
+				break;
+			case GLOBAL:
+				fmt.set(value);
+				m_globalModifiedSettings.push(fmt.set(value));  // this pushes an identity set, so when we restore,
+				                                                // it restores to the value here, and not the previous one
+				break;
+			default:
+				assert(false);
+		}
+	}
+}
+
+#endif // EMITTERSTATE_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/emitterutils.cpp

@@ -0,0 +1,385 @@
+#include "emitterutils.h"
+#include "exp.h"
+#include "indentation.h"
+#include "yaml-cpp/exceptions.h"
+#include "stringsource.h"
+#include <sstream>
+#include <iomanip>
+
+namespace YAML
+{
+	namespace Utils
+	{
+		namespace {
+			enum {REPLACEMENT_CHARACTER = 0xFFFD};
+
+			bool IsAnchorChar(int ch) { // test for ns-anchor-char
+				switch (ch) {
+					case ',': case '[': case ']': case '{': case '}': // c-flow-indicator
+					case ' ': case '\t': // s-white
+					case 0xFEFF: // c-byte-order-mark
+					case 0xA: case 0xD: // b-char
+						return false;
+					case 0x85:
+						return true;
+				}
+
+				if (ch < 0x20)
+					return false;
+
+				if (ch < 0x7E)
+					return true;
+
+				if (ch < 0xA0)
+					return false;
+				if (ch >= 0xD800 && ch <= 0xDFFF)
+					return false;
+				if ((ch & 0xFFFE) == 0xFFFE)
+					return false;
+				if ((ch >= 0xFDD0) && (ch <= 0xFDEF))
+					return false;
+				if (ch > 0x10FFFF)
+					return false;
+
+				return true;
+			}
+			
+			int Utf8BytesIndicated(char ch) {
+				int byteVal = static_cast<unsigned char>(ch);
+				switch (byteVal >> 4) {
+					case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:
+						return 1;
+					case 12: case 13:
+						return 2;
+					case 14:
+						return 3;
+					case 15:
+						return 4;
+					default:
+					  return -1;
+				}
+			}
+
+			bool IsTrailingByte(char ch) {
+				return (ch & 0xC0) == 0x80;
+			}
+			
+			bool GetNextCodePointAndAdvance(int& codePoint, std::string::const_iterator& first, std::string::const_iterator last) {
+				if (first == last)
+					return false;
+				
+				int nBytes = Utf8BytesIndicated(*first);
+				if (nBytes < 1) {
+					// Bad lead byte
+					++first;
+					codePoint = REPLACEMENT_CHARACTER;
+					return true;
+				}
+				
+				if (nBytes == 1) {
+					codePoint = *first++;
+					return true;
+				}
+				
+				// Gather bits from trailing bytes
+				codePoint = static_cast<unsigned char>(*first) & ~(0xFF << (7 - nBytes));
+				++first;
+				--nBytes;
+				for (; nBytes > 0; ++first, --nBytes) {
+					if ((first == last) || !IsTrailingByte(*first)) {
+						codePoint = REPLACEMENT_CHARACTER;
+						break;
+					}
+					codePoint <<= 6;
+					codePoint |= *first & 0x3F;
+				}
+
+				// Check for illegal code points
+				if (codePoint > 0x10FFFF)
+					codePoint = REPLACEMENT_CHARACTER;
+				else if (codePoint >= 0xD800 && codePoint <= 0xDFFF)
+					codePoint = REPLACEMENT_CHARACTER;
+				else if ((codePoint & 0xFFFE) == 0xFFFE)
+					codePoint = REPLACEMENT_CHARACTER;
+				else if (codePoint >= 0xFDD0 && codePoint <= 0xFDEF)
+					codePoint = REPLACEMENT_CHARACTER;
+				return true;
+			}
+			
+			void WriteCodePoint(ostream& out, int codePoint) {
+				if (codePoint < 0 || codePoint > 0x10FFFF) {
+					codePoint = REPLACEMENT_CHARACTER;
+				}
+				if (codePoint < 0x7F) {
+					out << static_cast<char>(codePoint);
+				} else if (codePoint < 0x7FF) {
+					out << static_cast<char>(0xC0 | (codePoint >> 6))
+					    << static_cast<char>(0x80 | (codePoint & 0x3F));
+				} else if (codePoint < 0xFFFF) {
+					out << static_cast<char>(0xE0 | (codePoint >> 12))
+					    << static_cast<char>(0x80 | ((codePoint >> 6) & 0x3F))
+					    << static_cast<char>(0x80 | (codePoint & 0x3F));
+				} else {
+					out << static_cast<char>(0xF0 | (codePoint >> 18))
+					    << static_cast<char>(0x80 | ((codePoint >> 12) & 0x3F))
+					    << static_cast<char>(0x80 | ((codePoint >> 6) & 0x3F))
+					    << static_cast<char>(0x80 | (codePoint & 0x3F));
+				}
+			}
+			
+			bool IsValidPlainScalar(const std::string& str, bool inFlow, bool allowOnlyAscii) {
+				// first check the start
+				const RegEx& start = (inFlow ? Exp::PlainScalarInFlow() : Exp::PlainScalar());
+				if(!start.Matches(str))
+					return false;
+				
+				// and check the end for plain whitespace (which can't be faithfully kept in a plain scalar)
+				if(!str.empty() && *str.rbegin() == ' ')
+					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();
+				StringCharSource buffer(str.c_str(), str.size());
+				while(buffer) {
+					if(disallowed.Matches(buffer))
+						return false;
+					if(allowOnlyAscii && (0x7F < static_cast<unsigned char>(buffer[0]))) 
+						return false;
+					++buffer;
+				}
+				
+				return true;
+			}
+
+			void WriteDoubleQuoteEscapeSequence(ostream& out, int codePoint) {
+				static const char hexDigits[] = "0123456789abcdef";
+
+				char escSeq[] = "\\U00000000";
+				int digits = 8;
+				if (codePoint < 0xFF) {
+					escSeq[1] = 'x';
+					digits = 2;
+				} else if (codePoint < 0xFFFF) {
+					escSeq[1] = 'u';
+					digits = 4;
+				}
+
+				// Write digits into the escape sequence
+				int i = 2;
+				for (; digits > 0; --digits, ++i) {
+					escSeq[i] = hexDigits[(codePoint >> (4 * (digits - 1))) & 0xF];
+				}
+
+				escSeq[i] = 0; // terminate with NUL character
+				out << escSeq;
+			}
+
+			bool WriteAliasName(ostream& out, const std::string& str) {
+				int codePoint;
+				for(std::string::const_iterator i = str.begin();
+					GetNextCodePointAndAdvance(codePoint, i, str.end());
+					)
+				{
+					if (!IsAnchorChar(codePoint))
+						return false;
+
+					WriteCodePoint(out, codePoint);
+				}
+				return true;
+			}
+		}
+		
+		bool WriteString(ostream& out, const std::string& str, bool inFlow, bool escapeNonAscii)
+		{
+			if(IsValidPlainScalar(str, inFlow, escapeNonAscii)) {
+				out << str;
+				return true;
+			} else
+				return WriteDoubleQuotedString(out, str, escapeNonAscii);
+		}
+		
+		bool WriteSingleQuotedString(ostream& out, const std::string& str)
+		{
+			out << "'";
+			int codePoint;
+			for(std::string::const_iterator i = str.begin();
+				GetNextCodePointAndAdvance(codePoint, i, str.end());
+				) 
+			{
+				if (codePoint == '\n')
+					return false;  // We can't handle a new line and the attendant indentation yet
+
+				if (codePoint == '\'')
+					out << "''";
+				else
+					WriteCodePoint(out, codePoint);
+			}
+			out << "'";
+			return true;
+		}
+		
+		bool WriteDoubleQuotedString(ostream& out, const std::string& str, bool escapeNonAscii)
+		{
+			out << "\"";
+			int codePoint;
+			for(std::string::const_iterator i = str.begin();
+				GetNextCodePointAndAdvance(codePoint, i, str.end());
+				) 
+			{
+				if (codePoint == '\"')
+					out << "\\\"";
+				else if (codePoint == '\\')
+					out << "\\\\";
+				else if (codePoint < 0x20 || (codePoint >= 0x80 && codePoint <= 0xA0)) // Control characters and non-breaking space
+					WriteDoubleQuoteEscapeSequence(out, codePoint);
+				else if (codePoint == 0xFEFF) // Byte order marks (ZWNS) should be escaped (YAML 1.2, sec. 5.2)	
+					WriteDoubleQuoteEscapeSequence(out, codePoint);
+				else if (escapeNonAscii && codePoint > 0x7E)
+					WriteDoubleQuoteEscapeSequence(out, codePoint);
+				else
+					WriteCodePoint(out, codePoint);
+			}
+			out << "\"";
+			return true;
+		}
+
+		bool WriteLiteralString(ostream& out, const std::string& str, int indent)
+		{
+			out << "|\n";
+			out << IndentTo(indent);
+			int codePoint;
+			for(std::string::const_iterator i = str.begin();
+				GetNextCodePointAndAdvance(codePoint, i, str.end());
+				)
+			{
+				if (codePoint == '\n')
+				  out << "\n" << IndentTo(indent);
+				else
+				  WriteCodePoint(out, codePoint);
+			}
+			return true;
+		}
+		
+		bool WriteComment(ostream& out, const std::string& str, int postCommentIndent)
+		{
+			const unsigned curIndent = out.col();
+			out << "#" << Indentation(postCommentIndent);
+			int codePoint;
+			for(std::string::const_iterator i = str.begin();
+				GetNextCodePointAndAdvance(codePoint, i, str.end());
+				)
+			{
+				if(codePoint == '\n')
+					out << "\n" << IndentTo(curIndent) << "#" << Indentation(postCommentIndent);
+				else
+					WriteCodePoint(out, codePoint);
+			}
+			return true;
+		}
+
+		bool WriteAlias(ostream& out, const std::string& str)
+		{
+			out << "*";
+			return WriteAliasName(out, str);
+		}
+		
+		bool WriteAnchor(ostream& out, const std::string& str)
+		{
+			out << "&";
+			return WriteAliasName(out, str);
+		}
+
+		bool WriteTag(ostream& out, const std::string& str, bool verbatim)
+		{
+			out << (verbatim ? "!<" : "!");
+			StringCharSource buffer(str.c_str(), str.size());
+			const RegEx& reValid = verbatim ? Exp::URI() : Exp::Tag();
+			while(buffer) {
+				int n = reValid.Match(buffer);
+				if(n <= 0)
+					return false;
+
+				while(--n >= 0) {
+					out << buffer[0];
+					++buffer;
+				}
+			}
+			if (verbatim)
+				out << ">";
+			return true;
+		}
+
+		bool WriteTagWithPrefix(ostream& out, const std::string& prefix, const std::string& tag)
+		{
+			out << "!";
+			StringCharSource prefixBuffer(prefix.c_str(), prefix.size());
+			while(prefixBuffer) {
+				int n = Exp::URI().Match(prefixBuffer);
+				if(n <= 0)
+					return false;
+				
+				while(--n >= 0) {
+					out << prefixBuffer[0];
+					++prefixBuffer;
+				}
+			}
+
+			out << "!";
+			StringCharSource tagBuffer(tag.c_str(), tag.size());
+			while(tagBuffer) {
+				int n = Exp::Tag().Match(tagBuffer);
+				if(n <= 0)
+					return false;
+				
+				while(--n >= 0) {
+					out << tagBuffer[0];
+					++tagBuffer;
+				}
+			}
+			return true;
+		}
+
+		bool WriteBinary(ostream& out, const char *data, std::size_t size)
+		{
+			static const char encoding[] = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789+/";
+			const char PAD = '=';
+			
+			out << "\"";
+			std::size_t chunks = size / 3;
+			std::size_t remainder = size % 3;
+
+			for(std::size_t i=0;i<chunks;i++, data += 3) {
+				out << encoding[data[0] >> 2];
+				out << encoding[((data[0] & 0x3) << 4) | (data[1] >> 4)];
+				out << encoding[((data[1] & 0xf) << 2) | (data[2] >> 6)];
+				out << encoding[data[2] & 0x3f];
+			}
+			
+			switch(remainder) {
+				case 0:
+					break;
+				case 1:
+					out << encoding[data[0] >> 2];
+					out << encoding[((data[0] & 0x3) << 4)];
+					out << PAD;
+					out << PAD;
+					break;
+				case 2:
+					out << encoding[data[0] >> 2];
+					out << encoding[((data[0] & 0x3) << 4) | (data[1] >> 4)];
+					out << encoding[((data[1] & 0xf) << 2)];
+					out << PAD;
+					break;
+			}
+			
+			out << "\"";
+			return true;
+		}
+	}
+}
+

src/emitterutils.h

@@ -0,0 +1,29 @@
+#ifndef EMITTERUTILS_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define EMITTERUTILS_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "yaml-cpp/ostream.h"
+#include <string>
+
+namespace YAML
+{
+	namespace Utils
+	{
+		bool WriteString(ostream& out, const std::string& str, bool inFlow, bool escapeNonAscii);
+		bool WriteSingleQuotedString(ostream& out, const std::string& str);
+		bool WriteDoubleQuotedString(ostream& out, const std::string& str, bool escapeNonAscii);
+		bool WriteLiteralString(ostream& out, const std::string& str, int indent);
+		bool WriteComment(ostream& out, const std::string& str, int postCommentIndent);
+		bool WriteAlias(ostream& out, const std::string& str);
+		bool WriteAnchor(ostream& out, const std::string& str);
+		bool WriteTag(ostream& out, const std::string& str, bool verbatim);
+		bool WriteTagWithPrefix(ostream& out, const std::string& prefix, const std::string& tag);
+		bool WriteBinary(ostream& out, const char *data, std::size_t size);
+	}
+}
+
+#endif // EMITTERUTILS_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/exp.cpp

@@ -0,0 +1,113 @@
+#include "exp.h"
+#include "yaml-cpp/exceptions.h"
+#include <sstream>
+
+namespace YAML
+{
+	namespace Exp
+	{
+		unsigned ParseHex(const std::string& str, const Mark& mark)
+		{
+			unsigned value = 0;
+			for(std::size_t i=0;i<str.size();i++) {
+				char ch = str[i];
+				int digit = 0;
+				if('a' <= ch && ch <= 'f')
+					digit = ch - 'a' + 10;
+				else if('A' <= ch && ch <= 'F')
+					digit = ch - 'A' + 10;
+				else if('0' <= ch && ch <= '9')
+					digit = ch - '0';
+				else
+					throw ParserException(mark, ErrorMsg::INVALID_HEX);
+
+				value = (value << 4) + digit;
+			}
+
+			return value;
+		}
+
+		std::string Str(unsigned ch)
+		{
+			return std::string(1, static_cast<char>(ch));
+		}
+
+		// Escape
+		// . Translates the next 'codeLength' characters into a hex number and returns the result.
+		// . Throws if it's not actually hex.
+		std::string Escape(Stream& in, int codeLength)
+		{
+			// grab string
+			std::string str;
+			for(int i=0;i<codeLength;i++)
+				str += in.get();
+
+			// get the value
+			unsigned value = ParseHex(str, in.mark());
+
+			// legal unicode?
+			if((value >= 0xD800 && value <= 0xDFFF) || value > 0x10FFFF) {
+				std::stringstream msg;
+				msg << ErrorMsg::INVALID_UNICODE << value;
+				throw ParserException(in.mark(), msg.str());
+			}
+
+			// now break it up into chars
+			if(value <= 0x7F)
+				return Str(value);
+			else if(value <= 0x7FF)
+				return Str(0xC0 + (value >> 6)) + Str(0x80 + (value & 0x3F));
+			else if(value <= 0xFFFF)
+				return Str(0xE0 + (value >> 12)) + Str(0x80 + ((value >> 6) & 0x3F)) + Str(0x80 + (value & 0x3F));
+			else
+				return Str(0xF0 + (value >> 18)) + Str(0x80 + ((value >> 12) & 0x3F)) +
+					Str(0x80 + ((value >> 6) & 0x3F)) + Str(0x80 + (value & 0x3F));
+		}
+
+		// Escape
+		// . Escapes the sequence starting 'in' (it must begin with a '\' or single quote)
+		//   and returns the result.
+		// . Throws if it's an unknown escape character.
+		std::string Escape(Stream& in)
+		{
+			// eat slash
+			char escape = in.get();
+
+			// switch on escape character
+			char ch = in.get();
+
+			// first do single quote, since it's easier
+			if(escape == '\'' && ch == '\'')
+				return "\'";
+
+			// now do the slash (we're not gonna check if it's a slash - you better pass one!)
+			switch(ch) {
+				case '0': return std::string(1, '\x00');
+				case 'a': return "\x07";
+				case 'b': return "\x08";
+				case 't':
+                case '\t': return "\x09";
+				case 'n': return "\x0A";
+				case 'v': return "\x0B";
+				case 'f': return "\x0C";
+				case 'r': return "\x0D";
+				case 'e': return "\x1B";
+				case ' ': return "\x20";
+				case '\"': return "\"";
+				case '\'': return "\'";
+				case '\\': return "\\";
+				case '/': return "/";
+				case 'N': return "\x85";
+				case '_': return "\xA0";
+				case 'L': return "\xE2\x80\xA8";  // LS (#x2028)
+				case 'P': return "\xE2\x80\xA9";  // PS (#x2029)
+				case 'x': return Escape(in, 2);
+				case 'u': return Escape(in, 4);
+				case 'U': return Escape(in, 8);
+			}
+
+			std::stringstream msg;
+			throw ParserException(in.mark(), std::string(ErrorMsg::INVALID_ESCAPE) + ch);
+		}
+	}
+}

src/exp.h

@@ -0,0 +1,196 @@
+#ifndef EXP_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define EXP_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "regex.h"
+#include <string>
+#include <ios>
+#include "stream.h"
+
+namespace YAML
+{
+	////////////////////////////////////////////////////////////////////////////////
+	// Here we store a bunch of expressions for matching different parts of the file.
+
+	namespace Exp
+	{
+		// misc
+		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)
+		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
+
+		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& Anchor() {
+			static const RegEx e = !(RegEx("[]{},", REGEX_OR) || BlankOrBreak());
+			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.
+		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;
+		}
+
+		inline const RegEx& EscSingleQuote() {
+			static const RegEx e = RegEx("\'\'");
+			return e;
+		}
+		inline const RegEx& EscBreak() {
+			static const RegEx e = RegEx('\\') + Break();
+			return e;
+		}
+
+		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);
+	}
+
+	namespace Keys
+	{
+		const char Directive = '%';
+		const char FlowSeqStart = '[';
+		const char FlowSeqEnd = ']';
+		const char FlowMapStart = '{';
+		const char FlowMapEnd = '}';
+		const char FlowEntry = ',';
+		const char Alias = '*';
+		const char Anchor = '&';
+		const char Tag = '!';
+		const char LiteralScalar = '|';
+		const char FoldedScalar = '>';
+		const char VerbatimTagStart = '<';
+		const char VerbatimTagEnd = '>';
+	}
+}
+
+#endif // EXP_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/indentation.h

@@ -0,0 +1,38 @@
+#ifndef INDENTATION_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define INDENTATION_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "yaml-cpp/ostream.h"
+#include <iostream>
+
+namespace YAML
+{
+	struct Indentation {
+		Indentation(unsigned n_): n(n_) {}
+		unsigned n;
+	};
+	
+	inline ostream& operator << (ostream& out, const Indentation& indent) {
+		for(unsigned i=0;i<indent.n;i++)
+			out << ' ';
+		return out;
+	}
+
+	struct IndentTo {
+		IndentTo(unsigned n_): n(n_) {}
+		unsigned n;
+	};
+	
+	inline ostream& operator << (ostream& out, const IndentTo& indent) {
+		while(out.col() < indent.n)
+			out << ' ';
+		return out;
+	}
+}
+
+
+#endif // INDENTATION_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/iterator.cpp

@@ -0,0 +1,103 @@
+#include "yaml-cpp/node.h"
+#include "yaml-cpp/exceptions.h"
+#include "iterpriv.h"
+
+namespace YAML
+{
+	Iterator::Iterator(): m_pData(new IterPriv)
+	{
+	}
+
+	Iterator::Iterator(std::auto_ptr<IterPriv> pData): m_pData(pData)
+	{
+	}
+
+	Iterator::Iterator(const Iterator& rhs): m_pData(new IterPriv(*rhs.m_pData))
+	{
+	}
+
+	Iterator& Iterator::operator = (const Iterator& rhs)
+	{
+		if(this == &rhs)
+			return *this;
+
+		m_pData.reset(new IterPriv(*rhs.m_pData));
+		return *this;
+	}
+
+	Iterator::~Iterator()
+	{
+	}
+
+	Iterator& Iterator::operator ++ ()
+	{
+		if(m_pData->type == IterPriv::IT_SEQ)
+			++m_pData->seqIter;
+		else if(m_pData->type == IterPriv::IT_MAP)
+			++m_pData->mapIter;
+
+		return *this;
+	}
+
+	Iterator Iterator::operator ++ (int)
+	{
+		Iterator temp = *this;
+
+		if(m_pData->type == IterPriv::IT_SEQ)
+			++m_pData->seqIter;
+		else if(m_pData->type == IterPriv::IT_MAP)
+			++m_pData->mapIter;
+
+		return temp;
+	}
+
+	const Node& Iterator::operator * () const
+	{
+		if(m_pData->type == IterPriv::IT_SEQ)
+			return **m_pData->seqIter;
+
+		throw BadDereference();
+	}
+
+	const Node *Iterator::operator -> () const
+	{
+		if(m_pData->type == IterPriv::IT_SEQ)
+			return *m_pData->seqIter;
+
+		throw BadDereference();
+	}
+
+	const Node& Iterator::first() const
+	{
+		if(m_pData->type == IterPriv::IT_MAP)
+			return *m_pData->mapIter->first;
+
+		throw BadDereference();
+	}
+
+	const Node& Iterator::second() const
+	{
+		if(m_pData->type == IterPriv::IT_MAP)
+			return *m_pData->mapIter->second;
+
+		throw BadDereference();
+	}
+
+	bool operator == (const Iterator& it, const Iterator& jt)
+	{
+		if(it.m_pData->type != jt.m_pData->type)
+			return false;
+
+		if(it.m_pData->type == IterPriv::IT_SEQ)
+			return it.m_pData->seqIter == jt.m_pData->seqIter;
+		else if(it.m_pData->type == IterPriv::IT_MAP)
+			return it.m_pData->mapIter == jt.m_pData->mapIter;
+
+		return true;
+	}
+
+	bool operator != (const Iterator& it, const Iterator& jt)
+	{
+		return !(it == jt);
+	}
+}

src/iterpriv.h

@@ -0,0 +1,33 @@
+#ifndef ITERPRIV_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define ITERPRIV_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "yaml-cpp/ltnode.h"
+#include <vector>
+#include <map>
+
+namespace YAML
+{
+	class Node;
+
+	// IterPriv
+	// . The implementation for iterators - essentially a union of sequence and map iterators.
+	struct IterPriv
+	{
+		IterPriv(): type(IT_NONE) {}
+		IterPriv(std::vector <Node *>::const_iterator it): type(IT_SEQ), seqIter(it) {}
+		IterPriv(std::map <Node *, Node *, ltnode>::const_iterator it): type(IT_MAP), mapIter(it) {}
+
+		enum ITER_TYPE { IT_NONE, IT_SEQ, IT_MAP };
+		ITER_TYPE type;
+
+		std::vector <Node *>::const_iterator seqIter;
+		std::map <Node *, Node *, ltnode>::const_iterator mapIter;
+	};
+}
+
+#endif // ITERPRIV_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/node.cpp

@@ -0,0 +1,269 @@
+#include "yaml-cpp/node.h"
+#include "iterpriv.h"
+#include "nodebuilder.h"
+#include "nodeownership.h"
+#include "scanner.h"
+#include "tag.h"
+#include "token.h"
+#include "yaml-cpp/aliasmanager.h"
+#include "yaml-cpp/emitfromevents.h"
+#include "yaml-cpp/emitter.h"
+#include "yaml-cpp/eventhandler.h"
+#include <cassert>
+#include <stdexcept>
+
+namespace YAML
+{
+	bool ltnode::operator()(const Node *pNode1, const Node *pNode2) const {
+		return *pNode1 < *pNode2;
+	}
+
+	Node::Node(): m_pOwnership(new NodeOwnership), m_type(NodeType::Null)
+	{
+	}
+
+	Node::Node(NodeOwnership& owner): m_pOwnership(new NodeOwnership(&owner)), m_type(NodeType::Null)
+	{
+	}
+
+	Node::~Node()
+	{
+		Clear();
+	}
+
+	void Node::Clear()
+	{
+		m_pOwnership.reset(new NodeOwnership);
+		m_type = NodeType::Null;
+		m_tag.clear();
+		m_scalarData.clear();
+		m_seqData.clear();
+		m_mapData.clear();
+	}
+	
+	bool Node::IsAliased() const
+	{
+		return m_pOwnership->IsAliased(*this);
+	}
+
+	Node& Node::CreateNode()
+	{
+		return m_pOwnership->Create();
+	}
+
+	std::auto_ptr<Node> Node::Clone() const
+	{
+		std::auto_ptr<Node> pNode(new Node);
+		NodeBuilder nodeBuilder(*pNode);
+		EmitEvents(nodeBuilder);
+		return pNode;
+	}
+
+	void Node::EmitEvents(EventHandler& eventHandler) const
+	{
+		eventHandler.OnDocumentStart(m_mark);
+		AliasManager am;
+		EmitEvents(am, eventHandler);
+		eventHandler.OnDocumentEnd();
+	}
+
+	void Node::EmitEvents(AliasManager& am, EventHandler& eventHandler) const
+	{
+		anchor_t anchor = NullAnchor;
+		if(IsAliased()) {
+			anchor = am.LookupAnchor(*this);
+			if(anchor) {
+				eventHandler.OnAlias(m_mark, anchor);
+				return;
+			}
+			
+			am.RegisterReference(*this);
+			anchor = am.LookupAnchor(*this);
+		}
+		
+		switch(m_type) {
+			case NodeType::Null:
+				eventHandler.OnNull(m_mark, anchor);
+				break;
+			case NodeType::Scalar:
+				eventHandler.OnScalar(m_mark, m_tag, anchor, m_scalarData);
+				break;
+			case NodeType::Sequence:
+				eventHandler.OnSequenceStart(m_mark, m_tag, anchor);
+				for(std::size_t i=0;i<m_seqData.size();i++)
+					m_seqData[i]->EmitEvents(am, eventHandler);
+				eventHandler.OnSequenceEnd();
+				break;
+			case NodeType::Map:
+				eventHandler.OnMapStart(m_mark, m_tag, anchor);
+				for(node_map::const_iterator it=m_mapData.begin();it!=m_mapData.end();++it) {
+					it->first->EmitEvents(am, eventHandler);
+					it->second->EmitEvents(am, eventHandler);
+				}
+				eventHandler.OnMapEnd();
+				break;
+		}
+	}
+
+	void Node::Init(NodeType::value type, const Mark& mark, const std::string& tag)
+	{
+		Clear();
+		m_mark = mark;
+		m_type = type;
+		m_tag = tag;
+	}
+
+	void Node::MarkAsAliased()
+	{
+		m_pOwnership->MarkAsAliased(*this);
+	}
+	
+	void Node::SetScalarData(const std::string& data)
+	{
+		assert(m_type == NodeType::Scalar); // TODO: throw?
+		m_scalarData = data;
+	}
+
+	void Node::Append(Node& node)
+	{
+		assert(m_type == NodeType::Sequence); // TODO: throw?
+		m_seqData.push_back(&node);
+	}
+	
+	void Node::Insert(Node& key, Node& value)
+	{
+		assert(m_type == NodeType::Map); // TODO: throw?
+		m_mapData[&key] = &value;
+	}
+
+	// begin
+	// Returns an iterator to the beginning of this (sequence or map).
+	Iterator Node::begin() const
+	{
+		switch(m_type) {
+			case NodeType::Null:
+			case NodeType::Scalar:
+				return Iterator();
+			case NodeType::Sequence:
+				return Iterator(std::auto_ptr<IterPriv>(new IterPriv(m_seqData.begin())));
+			case NodeType::Map:
+				return Iterator(std::auto_ptr<IterPriv>(new IterPriv(m_mapData.begin())));
+		}
+		
+		assert(false);
+		return Iterator();
+	}
+
+	// end
+	// . Returns an iterator to the end of this (sequence or map).
+	Iterator Node::end() const
+	{
+		switch(m_type) {
+			case NodeType::Null:
+			case NodeType::Scalar:
+				return Iterator();
+			case NodeType::Sequence:
+				return Iterator(std::auto_ptr<IterPriv>(new IterPriv(m_seqData.end())));
+			case NodeType::Map:
+				return Iterator(std::auto_ptr<IterPriv>(new IterPriv(m_mapData.end())));
+		}
+		
+		assert(false);
+		return Iterator();
+	}
+
+	// size
+	// . Returns the size of a sequence or map node
+	// . Otherwise, returns zero.
+	std::size_t Node::size() const
+	{
+		switch(m_type) {
+			case NodeType::Null:
+			case NodeType::Scalar:
+				return 0;
+			case NodeType::Sequence:
+				return m_seqData.size();
+			case NodeType::Map:
+				return m_mapData.size();
+		}
+		
+		assert(false);
+		return 0;
+	}
+
+	const Node *Node::FindAtIndex(std::size_t i) const
+	{
+		if(m_type == NodeType::Sequence)
+			return m_seqData[i];
+		return 0;
+	}
+
+	bool Node::GetScalar(std::string& s) const
+	{
+		switch(m_type) {
+			case NodeType::Null:
+				s = "~";
+				return true;
+			case NodeType::Scalar:
+				s = m_scalarData;
+				return true;
+			case NodeType::Sequence:
+			case NodeType::Map:
+				return false;
+		}
+		
+		assert(false);
+		return false;
+	}
+
+	Emitter& operator << (Emitter& out, const Node& node)
+	{
+		EmitFromEvents emitFromEvents(out);
+		node.EmitEvents(emitFromEvents);
+		return out;
+	}
+
+	int Node::Compare(const Node& rhs) const
+	{
+		if(m_type != rhs.m_type)
+			return rhs.m_type - m_type;
+		
+		switch(m_type) {
+			case NodeType::Null:
+				return 0;
+			case NodeType::Scalar:
+				return m_scalarData.compare(rhs.m_scalarData);
+			case NodeType::Sequence:
+				if(m_seqData.size() < rhs.m_seqData.size())
+					return 1;
+				else if(m_seqData.size() > rhs.m_seqData.size())
+					return -1;
+				for(std::size_t i=0;i<m_seqData.size();i++)
+					if(int cmp = m_seqData[i]->Compare(*rhs.m_seqData[i]))
+						return cmp;
+				return 0;
+			case NodeType::Map:
+				if(m_mapData.size() < rhs.m_mapData.size())
+					return 1;
+				else if(m_mapData.size() > rhs.m_mapData.size())
+					return -1;
+				node_map::const_iterator it = m_mapData.begin();
+				node_map::const_iterator jt = rhs.m_mapData.begin();
+				for(;it!=m_mapData.end() && jt!=rhs.m_mapData.end();it++, jt++) {
+					if(int cmp = it->first->Compare(*jt->first))
+						return cmp;
+					if(int cmp = it->second->Compare(*jt->second))
+						return cmp;
+				}
+				return 0;
+		}
+		
+		assert(false);
+		return 0;
+	}
+
+	bool operator < (const Node& n1, const Node& n2)
+	{
+		return n1.Compare(n2) < 0;
+	}
+}

src/nodebuilder.cpp

@@ -0,0 +1,145 @@
+#include "nodebuilder.h"
+#include "yaml-cpp/mark.h"
+#include "yaml-cpp/node.h"
+#include <cassert>
+
+namespace YAML
+{
+	NodeBuilder::NodeBuilder(Node& root): m_root(root), m_initializedRoot(false), m_finished(false)
+	{
+		m_root.Clear();
+		m_anchors.push_back(0); // since the anchors start at 1
+	}
+	
+	NodeBuilder::~NodeBuilder()
+	{
+	}
+
+	void NodeBuilder::OnDocumentStart(const Mark&)
+	{
+	}
+
+	void NodeBuilder::OnDocumentEnd()
+	{
+		assert(m_finished);
+	}
+
+	void NodeBuilder::OnNull(const Mark& mark, anchor_t anchor)
+	{
+		Node& node = Push(anchor);
+		node.Init(NodeType::Null, mark, "");
+		Pop();
+	}
+
+	void NodeBuilder::OnAlias(const Mark& /*mark*/, anchor_t anchor)
+	{
+		Node& node = *m_anchors[anchor];
+		Insert(node);
+		node.MarkAsAliased();
+	}
+
+	void NodeBuilder::OnScalar(const Mark& mark, const std::string& tag, anchor_t anchor, const std::string& value)
+	{
+		Node& node = Push(anchor);
+		node.Init(NodeType::Scalar, mark, tag);
+		node.SetScalarData(value);
+		Pop();
+	}
+
+	void NodeBuilder::OnSequenceStart(const Mark& mark, const std::string& tag, anchor_t anchor)
+	{
+		Node& node = Push(anchor);
+		node.Init(NodeType::Sequence, mark, tag);
+	}
+
+	void NodeBuilder::OnSequenceEnd()
+	{
+		Pop();
+	}
+
+	void NodeBuilder::OnMapStart(const Mark& mark, const std::string& tag, anchor_t anchor)
+	{
+		Node& node = Push(anchor);
+		node.Init(NodeType::Map, mark, tag);
+		m_didPushKey.push(false);
+	}
+
+	void NodeBuilder::OnMapEnd()
+	{
+		m_didPushKey.pop();
+		Pop();
+	}
+	
+	Node& NodeBuilder::Push(anchor_t anchor)
+	{
+		Node& node = Push();
+		RegisterAnchor(anchor, node);
+		return node;
+	}
+	
+	Node& NodeBuilder::Push()
+	{
+		if(!m_initializedRoot) {
+			m_initializedRoot = true;
+			return m_root;
+		}
+		
+		Node& node = m_root.CreateNode();
+		m_stack.push(&node);
+		return node;
+	}
+	
+	Node& NodeBuilder::Top()
+	{
+		return m_stack.empty() ? m_root : *m_stack.top();
+	}
+	
+	void NodeBuilder::Pop()
+	{
+		assert(!m_finished);
+		if(m_stack.empty()) {
+			m_finished = true;
+			return;
+		}
+		
+		Node& node = *m_stack.top();
+		m_stack.pop();
+		Insert(node);
+	}
+	
+	void NodeBuilder::Insert(Node& node)
+	{
+		Node& curTop = Top();
+		switch(curTop.Type()) {
+			case NodeType::Null:
+			case NodeType::Scalar:
+				assert(false);
+				break;
+			case NodeType::Sequence:
+				curTop.Append(node);
+				break;
+			case NodeType::Map:
+				assert(!m_didPushKey.empty());
+				if(m_didPushKey.top()) {
+					assert(!m_pendingKeys.empty());
+
+					Node& key = *m_pendingKeys.top();
+					m_pendingKeys.pop();
+					curTop.Insert(key, node);
+					m_didPushKey.top() = false;
+				} else {
+					m_pendingKeys.push(&node);
+					m_didPushKey.top() = true;
+				}
+				break;
+		}
+	}
+
+	void NodeBuilder::RegisterAnchor(anchor_t anchor, Node& node)
+	{
+		if(anchor) {
+			assert(anchor == m_anchors.size());
+			m_anchors.push_back(&node);
+		}
+	}
+}

src/nodebuilder.h

@@ -0,0 +1,61 @@
+#ifndef NODEBUILDER_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define NODEBUILDER_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+#include "yaml-cpp/eventhandler.h"
+#include <map>
+#include <memory>
+#include <stack>
+#include <vector>
+
+namespace YAML
+{
+	class Node;
+	
+	class NodeBuilder: public EventHandler
+	{
+	public:
+		explicit NodeBuilder(Node& root);
+		virtual ~NodeBuilder();
+
+		virtual void OnDocumentStart(const Mark& mark);
+		virtual void OnDocumentEnd();
+		
+		virtual void OnNull(const Mark& mark, anchor_t anchor);
+		virtual void OnAlias(const Mark& mark, anchor_t anchor);
+		virtual void OnScalar(const Mark& mark, const std::string& tag, anchor_t anchor, const std::string& value);
+		
+		virtual void OnSequenceStart(const Mark& mark, const std::string& tag, anchor_t anchor);
+		virtual void OnSequenceEnd();
+		
+		virtual void OnMapStart(const Mark& mark, const std::string& tag, anchor_t anchor);
+		virtual void OnMapEnd();
+		
+	private:
+		Node& Push(anchor_t anchor);
+		Node& Push();
+		Node& Top();
+		void Pop();
+
+		void Insert(Node& node);
+		void RegisterAnchor(anchor_t anchor, Node& node);
+		
+	private:
+		Node& m_root;
+		bool m_initializedRoot;
+		bool m_finished;
+		
+		std::stack<Node *> m_stack;
+		std::stack<Node *> m_pendingKeys;
+		std::stack<bool> m_didPushKey;
+
+		typedef std::vector<Node *> Anchors;
+		Anchors m_anchors;
+	};
+}
+
+#endif // NODEBUILDER_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+

src/nodeownership.cpp

@@ -0,0 +1,31 @@
+#include "nodeownership.h"
+#include "yaml-cpp/node.h"
+
+namespace YAML
+{
+	NodeOwnership::NodeOwnership(NodeOwnership *pOwner): m_pOwner(pOwner)
+	{
+		if(!m_pOwner)
+			m_pOwner = this;
+	}
+	
+	NodeOwnership::~NodeOwnership()
+	{
+	}
+
+	Node& NodeOwnership::_Create()
+	{
+		m_nodes.push_back(std::auto_ptr<Node>(new Node));
+		return m_nodes.back();
+	}
+
+	void NodeOwnership::_MarkAsAliased(const Node& node)
+	{
+		m_aliasedNodes.insert(&node);
+	}
+
+	bool NodeOwnership::_IsAliased(const Node& node) const
+	{
+		return m_aliasedNodes.count(&node) > 0;
+	}
+}

src/nodeownership.h

@@ -0,0 +1,39 @@
+#ifndef NODE_OWNERSHIP_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define NODE_OWNERSHIP_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "yaml-cpp/noncopyable.h"
+#include "ptr_vector.h"
+#include <set>
+
+namespace YAML
+{
+	class Node;
+	
+	class NodeOwnership: private noncopyable
+	{
+	public:
+		explicit NodeOwnership(NodeOwnership *pOwner = 0);
+		~NodeOwnership();
+		
+		Node& Create() { return m_pOwner->_Create(); }
+		void MarkAsAliased(const Node& node) { m_pOwner->_MarkAsAliased(node); }
+		bool IsAliased(const Node& node) const { return m_pOwner->_IsAliased(node); }
+		
+	private:
+		Node& _Create();
+		void _MarkAsAliased(const Node& node);
+		bool _IsAliased(const Node& node) const;
+		
+	private:
+		ptr_vector<Node> m_nodes;
+		std::set<const Node *> m_aliasedNodes;
+		NodeOwnership *m_pOwner;
+	};
+}
+
+#endif // NODE_OWNERSHIP_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/null.cpp

@@ -0,0 +1,12 @@
+#include "yaml-cpp/null.h"
+#include "yaml-cpp/node.h"
+
+namespace YAML
+{
+	_Null Null;
+
+	bool IsNull(const Node& node)
+	{
+		return node.Read(Null);
+	}
+}

src/ostream.cpp

@@ -0,0 +1,63 @@
+#include "yaml-cpp/ostream.h"
+#include <cstring>
+
+namespace YAML
+{
+	ostream::ostream(): m_buffer(0), m_pos(0), m_size(0), m_row(0), m_col(0)
+	{
+		reserve(1024);
+	}
+	
+	ostream::~ostream()
+	{
+		delete [] m_buffer;
+	}
+	
+	void ostream::reserve(unsigned size)
+	{
+		if(size <= m_size)
+			return;
+		
+		char *newBuffer = new char[size];
+		std::memset(newBuffer, 0, size * sizeof(char));
+		std::memcpy(newBuffer, m_buffer, m_size * sizeof(char));
+		delete [] m_buffer;
+		m_buffer = newBuffer;
+		m_size = size;
+	}
+	
+	void ostream::put(char ch)
+	{
+		if(m_pos >= m_size - 1)   // an extra space for the NULL terminator
+			reserve(m_size * 2);
+		
+		m_buffer[m_pos] = ch;
+		m_pos++;
+		
+		if(ch == '\n') {
+			m_row++;
+			m_col = 0;
+		} else
+			m_col++;
+	}
+
+	ostream& operator << (ostream& out, const char *str)
+	{
+		std::size_t length = std::strlen(str);
+		for(std::size_t i=0;i<length;i++)
+			out.put(str[i]);
+		return out;
+	}
+	
+	ostream& operator << (ostream& out, const std::string& str)
+	{
+		out << str.c_str();
+		return out;
+	}
+	
+	ostream& operator << (ostream& out, char ch)
+	{
+		out.put(ch);
+		return out;
+	}
+}

src/parser.cpp

@@ -0,0 +1,152 @@
+#include "yaml-cpp/parser.h"
+#include "directives.h"
+#include "yaml-cpp/eventhandler.h"
+#include "yaml-cpp/exceptions.h"
+#include "yaml-cpp/node.h"
+#include "nodebuilder.h"
+#include "scanner.h"
+#include "singledocparser.h"
+#include "tag.h"
+#include "token.h"
+#include <sstream>
+#include <cstdio>
+
+namespace YAML
+{
+	Parser::Parser()
+	{
+	}
+	
+	Parser::Parser(std::istream& in)
+	{
+		Load(in);
+	}
+
+	Parser::~Parser()
+	{
+	}
+
+	Parser::operator bool() const
+	{
+		return m_pScanner.get() && !m_pScanner->empty();
+	}
+
+	void Parser::Load(std::istream& in)
+	{
+		m_pScanner.reset(new Scanner(in));
+		m_pDirectives.reset(new Directives);
+	}
+
+	// HandleNextDocument
+	// . Handles the next document
+	// . Throws a ParserException on error.
+	// . Returns false if there are no more documents
+	bool Parser::HandleNextDocument(EventHandler& eventHandler)
+	{
+		if(!m_pScanner.get())
+			return false;
+
+		ParseDirectives();
+		if(m_pScanner->empty())
+			return false;
+		
+		SingleDocParser sdp(*m_pScanner, *m_pDirectives);
+		sdp.HandleDocument(eventHandler);
+		return true;
+	}
+
+	// GetNextDocument
+	// . Reads the next document in the queue (of tokens).
+	// . Throws a ParserException on error.
+	bool Parser::GetNextDocument(Node& document)
+	{
+		NodeBuilder builder(document);
+		return HandleNextDocument(builder);
+	}
+
+	// ParseDirectives
+	// . Reads any directives that are next in the queue.
+	void Parser::ParseDirectives()
+	{
+		bool readDirective = false;
+
+		while(1) {
+			if(m_pScanner->empty())
+				break;
+
+			Token& token = m_pScanner->peek();
+			if(token.type != Token::DIRECTIVE)
+				break;
+
+			// we keep the directives from the last document if none are specified;
+			// but if any directives are specific, then we reset them
+			if(!readDirective)
+				m_pDirectives.reset(new Directives);
+
+			readDirective = true;
+			HandleDirective(token);
+			m_pScanner->pop();
+		}
+	}
+
+	void Parser::HandleDirective(const Token& token)
+	{
+		if(token.value == "YAML")
+			HandleYamlDirective(token);
+		else if(token.value == "TAG")
+			HandleTagDirective(token);
+	}
+
+	// HandleYamlDirective
+	// . Should be of the form 'major.minor' (like a version number)
+	void Parser::HandleYamlDirective(const Token& token)
+	{
+		if(token.params.size() != 1)
+			throw ParserException(token.mark, ErrorMsg::YAML_DIRECTIVE_ARGS);
+		
+		if(!m_pDirectives->version.isDefault)
+			throw ParserException(token.mark, ErrorMsg::REPEATED_YAML_DIRECTIVE);
+
+		std::stringstream str(token.params[0]);
+		str >> m_pDirectives->version.major;
+		str.get();
+		str >> m_pDirectives->version.minor;
+		if(!str || str.peek() != EOF)
+			throw ParserException(token.mark, std::string(ErrorMsg::YAML_VERSION) + token.params[0]);
+
+		if(m_pDirectives->version.major > 1)
+			throw ParserException(token.mark, ErrorMsg::YAML_MAJOR_VERSION);
+
+		m_pDirectives->version.isDefault = false;
+		// TODO: warning on major == 1, minor > 2?
+	}
+
+	// HandleTagDirective
+	// . Should be of the form 'handle prefix', where 'handle' is converted to 'prefix' in the file.
+	void Parser::HandleTagDirective(const Token& token)
+	{
+		if(token.params.size() != 2)
+			throw ParserException(token.mark, ErrorMsg::TAG_DIRECTIVE_ARGS);
+
+		const std::string& handle = token.params[0];
+		const std::string& prefix = token.params[1];
+		if(m_pDirectives->tags.find(handle) != m_pDirectives->tags.end())
+			throw ParserException(token.mark, ErrorMsg::REPEATED_TAG_DIRECTIVE);
+		
+		m_pDirectives->tags[handle] = prefix;
+	}
+
+	void Parser::PrintTokens(std::ostream& out)
+	{
+		if(!m_pScanner.get())
+			return;
+		
+		while(1) {
+			if(m_pScanner->empty())
+				break;
+
+			out << m_pScanner->peek() << "\n";
+			m_pScanner->pop();
+		}
+	}
+}

src/ptr_stack.h

@@ -0,0 +1,46 @@
+#ifndef PTR_STACK_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define PTR_STACK_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+#include "yaml-cpp/noncopyable.h"
+#include <cstddef>
+#include <cstdlib>
+#include <memory>
+#include <vector>
+
+template <typename T>
+class ptr_stack: private YAML::noncopyable
+{
+public:
+	ptr_stack() {}
+	~ptr_stack() { clear(); }
+	
+	void clear() {
+		for(unsigned i=0;i<m_data.size();i++)
+			delete m_data[i];
+		m_data.clear();
+	}
+	
+	std::size_t size() const { return m_data.size(); }
+	bool empty() const { return m_data.empty(); }
+	
+	void push(std::auto_ptr<T> t) {
+		m_data.push_back(NULL);
+		m_data.back() = t.release();
+	}
+	std::auto_ptr<T> pop() {
+		std::auto_ptr<T> t(m_data.back());
+		m_data.pop_back();
+		return t;
+	}
+	T& top() { return *m_data.back(); }
+	const T& top() const { return *m_data.back(); }
+	
+private:
+	std::vector<T*> m_data;
+};
+
+#endif // PTR_STACK_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/ptr_vector.h

@@ -0,0 +1,47 @@
+#ifndef PTR_VECTOR_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define PTR_VECTOR_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+#include "yaml-cpp/noncopyable.h"
+#include <cstddef>
+#include <cstdlib>
+#include <memory>
+#include <vector>
+
+namespace YAML {
+	
+	template <typename T>
+	class ptr_vector: private YAML::noncopyable
+	{
+	public:
+		ptr_vector() {}
+		~ptr_vector() { clear(); }
+		
+		void clear() {
+			for(unsigned i=0;i<m_data.size();i++)
+				delete m_data[i];
+			m_data.clear();
+		}
+		
+		std::size_t size() const { return m_data.size(); }
+		bool empty() const { return m_data.empty(); }
+		
+		void push_back(std::auto_ptr<T> t) {
+			m_data.push_back(NULL);
+			m_data.back() = t.release();
+		}
+		T& operator[](std::size_t i) { return *m_data[i]; }
+		const T& operator[](std::size_t i) const { return *m_data[i]; }
+		
+		T& back() { return *m_data.back(); }
+		const T& back() const { return *m_data.back(); }
+
+	private:
+		std::vector<T*> m_data;
+	};
+}
+
+#endif // PTR_VECTOR_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/regex.cpp

@@ -0,0 +1,60 @@
+#include "regex.h"
+
+namespace YAML
+{
+	// constructors
+	RegEx::RegEx(): m_op(REGEX_EMPTY)
+	{
+	}
+	
+	RegEx::RegEx(REGEX_OP op): m_op(op)
+	{
+	}
+	
+	RegEx::RegEx(char ch): m_op(REGEX_MATCH), m_a(ch)
+	{
+	}
+	
+	RegEx::RegEx(char a, char z): m_op(REGEX_RANGE), m_a(a), m_z(z)
+	{
+	}
+	
+	RegEx::RegEx(const std::string& str, REGEX_OP op): m_op(op)
+	{
+		for(std::size_t i=0;i<str.size();i++)
+			m_params.push_back(RegEx(str[i]));
+	}
+	
+	// combination constructors
+	RegEx operator ! (const RegEx& ex)
+	{
+		RegEx ret(REGEX_NOT);
+		ret.m_params.push_back(ex);
+		return ret;
+	}
+	
+	RegEx operator || (const RegEx& ex1, const RegEx& ex2)
+	{
+		RegEx ret(REGEX_OR);
+		ret.m_params.push_back(ex1);
+		ret.m_params.push_back(ex2);
+		return ret;
+	}
+	
+	RegEx operator && (const RegEx& ex1, const RegEx& ex2)
+	{
+		RegEx ret(REGEX_AND);
+		ret.m_params.push_back(ex1);
+		ret.m_params.push_back(ex2);
+		return ret;
+	}
+	
+	RegEx operator + (const RegEx& ex1, const RegEx& ex2)
+	{
+		RegEx ret(REGEX_SEQ);
+		ret.m_params.push_back(ex1);
+		ret.m_params.push_back(ex2);
+		return ret;
+	}	
+}
+

src/regex.h

@@ -0,0 +1,67 @@
+#ifndef REGEX_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define REGEX_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include <vector>
+#include <string>
+
+namespace YAML
+{
+	class Stream;
+
+	enum REGEX_OP { REGEX_EMPTY, REGEX_MATCH, REGEX_RANGE, REGEX_OR, REGEX_AND, REGEX_NOT, REGEX_SEQ };
+
+	// simplified regular expressions
+	// . Only straightforward matches (no repeated characters)
+	// . Only matches from start of string
+	class RegEx
+	{
+	public:
+		RegEx();
+		RegEx(char ch);
+		RegEx(char a, char z);
+		RegEx(const std::string& str, REGEX_OP op = REGEX_SEQ);
+		~RegEx() {}
+
+		friend RegEx operator ! (const RegEx& ex);
+		friend RegEx operator || (const RegEx& ex1, const RegEx& ex2);
+		friend RegEx operator && (const RegEx& ex1, const RegEx& ex2);
+		friend RegEx operator + (const RegEx& ex1, const RegEx& ex2);
+		
+		bool Matches(char ch) const;
+		bool Matches(const std::string& str) const;
+		bool Matches(const Stream& in) const;
+		template <typename Source> bool Matches(const Source& source) const;
+
+		int Match(const std::string& str) const;
+		int Match(const Stream& in) const;
+		template <typename Source> int Match(const Source& source) const;
+
+	private:
+		RegEx(REGEX_OP op);
+		
+		template <typename Source> bool IsValidSource(const Source& source) const;
+		template <typename Source> int MatchUnchecked(const Source& source) const;
+
+		template <typename Source> int MatchOpEmpty(const Source& source) const;
+		template <typename Source> int MatchOpMatch(const Source& source) const;
+		template <typename Source> int MatchOpRange(const Source& source) const;
+		template <typename Source> int MatchOpOr(const Source& source) const;
+		template <typename Source> int MatchOpAnd(const Source& source) const;
+		template <typename Source> int MatchOpNot(const Source& source) const;
+		template <typename Source> int MatchOpSeq(const Source& source) const;
+
+	private:
+		REGEX_OP m_op;
+		char m_a, m_z;
+		std::vector <RegEx> m_params;
+	};
+}
+
+#include "regeximpl.h"
+
+#endif // REGEX_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/regeximpl.h

@@ -0,0 +1,186 @@
+#ifndef REGEXIMPL_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define REGEXIMPL_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "stream.h"
+#include "stringsource.h"
+#include "streamcharsource.h"
+
+namespace YAML
+{
+	// query matches
+	inline bool RegEx::Matches(char ch) const {
+		std::string str;
+		str += ch;
+		return Matches(str);
+	}
+	
+	inline bool RegEx::Matches(const std::string& str) const {
+		return Match(str) >= 0;
+	}
+	
+	inline bool RegEx::Matches(const Stream& in) const {
+		return Match(in) >= 0;
+	}
+
+	template <typename Source>
+	inline bool RegEx::Matches(const Source& source) const {
+		return Match(source) >= 0;
+	}
+
+	// Match
+	// . Matches the given string against this regular expression.
+	// . Returns the number of characters matched.
+	// . Returns -1 if no characters were matched (the reason for
+	//   not returning zero is that we may have an empty regex
+	//   which is ALWAYS successful at matching zero characters).
+	// . REMEMBER that we only match from the start of the buffer!	
+	inline int RegEx::Match(const std::string& str) const
+	{
+		StringCharSource source(str.c_str(), str.size());
+		return Match(source);
+	}
+	
+	inline int RegEx::Match(const Stream& in) const
+	{
+		StreamCharSource source(in);
+		return Match(source);
+	}
+
+	template <typename Source>
+	inline bool RegEx::IsValidSource(const Source& source) const
+	{
+		return source;
+	}
+	
+	template<>
+	inline bool RegEx::IsValidSource<StringCharSource>(const StringCharSource&source) const
+	{
+		switch(m_op) {
+			case REGEX_MATCH:
+			case REGEX_RANGE:
+				return source;
+			default:
+				return true;
+		}
+	}
+
+	template <typename Source>
+	inline int RegEx::Match(const Source& source) const
+	{
+		return IsValidSource(source) ? MatchUnchecked(source) : -1;
+	}
+	
+	template <typename Source>
+	inline int RegEx::MatchUnchecked(const Source& source) const
+	{
+		switch(m_op) {
+			case REGEX_EMPTY:
+				return MatchOpEmpty(source);
+			case REGEX_MATCH:
+				return MatchOpMatch(source);
+			case REGEX_RANGE:
+				return MatchOpRange(source);
+			case REGEX_OR:
+				return MatchOpOr(source);
+			case REGEX_AND:
+				return MatchOpAnd(source);
+			case REGEX_NOT:
+				return MatchOpNot(source);
+			case REGEX_SEQ:
+				return MatchOpSeq(source);
+		}
+		
+		return -1;
+	}
+
+	//////////////////////////////////////////////////////////////////////////////
+	// Operators
+	// Note: the convention MatchOp*<Source> is that we can assume IsSourceValid(source).
+	//       So we do all our checks *before* we call these functions
+	
+	// EmptyOperator
+	template <typename Source>
+	inline int RegEx::MatchOpEmpty(const Source& source) const {
+		return source[0] == Stream::eof() ? 0 : -1;
+	}
+	
+	template <>
+	inline int RegEx::MatchOpEmpty<StringCharSource>(const StringCharSource& source) const {
+		return !source ? 0 : -1;  // the empty regex only is successful on the empty string
+	}
+
+	// MatchOperator
+	template <typename Source>
+	inline int RegEx::MatchOpMatch(const Source& source) const {
+		if(source[0] != m_a)
+			return -1;
+		return 1;
+	}
+	
+	// RangeOperator
+	template <typename Source>
+	inline int RegEx::MatchOpRange(const Source& source) const {
+		if(m_a > source[0] || m_z < source[0])
+			return -1;
+		return 1;
+	}
+	
+	// OrOperator
+	template <typename Source>
+	inline int RegEx::MatchOpOr(const Source& source) const {
+		for(std::size_t i=0;i<m_params.size();i++) {
+			int n = m_params[i].MatchUnchecked(source);
+			if(n >= 0)
+				return n;
+		}
+		return -1;
+	}
+	
+	// AndOperator
+	// Note: 'AND' is a little funny, since we may be required to match things
+	//       of different lengths. If we find a match, we return the length of
+	//       the FIRST entry on the list.
+	template <typename Source>
+	inline int RegEx::MatchOpAnd(const Source& source) const {
+		int first = -1;
+		for(std::size_t i=0;i<m_params.size();i++) {
+			int n = m_params[i].MatchUnchecked(source);
+			if(n == -1)
+				return -1;
+			if(i == 0)
+				first = n;
+		}
+		return first;
+	}
+	
+	// NotOperator
+	template <typename Source>
+	inline int RegEx::MatchOpNot(const Source& source) const {
+		if(m_params.empty())
+			return -1;
+		if(m_params[0].MatchUnchecked(source) >= 0)
+			return -1;
+		return 1;
+	}
+	
+	// SeqOperator
+	template <typename Source>
+	inline int RegEx::MatchOpSeq(const Source& source) const {
+		int offset = 0;
+		for(std::size_t i=0;i<m_params.size();i++) {
+			int n = m_params[i].Match(source + offset); // note Match, not MatchUnchecked because we need to check validity after the offset
+			if(n == -1)
+				return -1;
+			offset += n;
+		}
+		
+		return offset;
+	}
+}
+
+#endif // REGEXIMPL_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/scanner.cpp

@@ -0,0 +1,387 @@
+#include "scanner.h"
+#include "token.h"
+#include "yaml-cpp/exceptions.h"
+#include "exp.h"
+#include <cassert>
+#include <memory>
+
+namespace YAML
+{
+	Scanner::Scanner(std::istream& in)
+		: INPUT(in), m_startedStream(false), m_endedStream(false), m_simpleKeyAllowed(false), m_canBeJSONFlow(false)
+	{
+	}
+
+	Scanner::~Scanner()
+	{
+	}
+
+	// empty
+	// . Returns true if there are no more tokens to be read
+	bool Scanner::empty()
+	{
+		EnsureTokensInQueue();
+		return m_tokens.empty();
+	}
+
+	// pop
+	// . Simply removes the next token on the queue.
+	void Scanner::pop()
+	{
+		EnsureTokensInQueue();
+		if(!m_tokens.empty())
+			m_tokens.pop();
+	}
+
+	// peek
+	// . Returns (but does not remove) the next token on the queue.
+	Token& Scanner::peek()
+	{
+		EnsureTokensInQueue();
+		assert(!m_tokens.empty());  // should we be asserting here? I mean, we really just be checking
+		                            // if it's empty before peeking.
+
+#if 0
+		static Token *pLast = 0;
+		if(pLast != &m_tokens.front())
+			std::cerr << "peek: " << m_tokens.front() << "\n";
+		pLast = &m_tokens.front();
+#endif
+
+		return m_tokens.front();
+	}
+
+	// EnsureTokensInQueue
+	// . Scan until there's a valid token at the front of the queue,
+	//   or we're sure the queue is empty.
+	void Scanner::EnsureTokensInQueue()
+	{
+		while(1) {
+			if(!m_tokens.empty()) {
+				Token& token = m_tokens.front();
+
+				// if this guy's valid, then we're done
+				if(token.status == Token::VALID)
+					return;
+
+				// here's where we clean up the impossible tokens
+				if(token.status == Token::INVALID) {
+					m_tokens.pop();
+					continue;
+				}
+
+				// note: what's left are the unverified tokens
+			}
+
+			// no token? maybe we've actually finished
+			if(m_endedStream)
+				return;
+
+			// no? then scan...
+			ScanNextToken();
+		}
+	}
+
+	// ScanNextToken
+	// . The main scanning function; here we branch out and
+	//   scan whatever the next token should be.
+	void Scanner::ScanNextToken()
+	{
+		if(m_endedStream)
+			return;
+
+		if(!m_startedStream)
+			return StartStream();
+
+		// get rid of whitespace, etc. (in between tokens it should be irrelevent)
+		ScanToNextToken();
+
+		// maybe need to end some blocks
+		PopIndentToHere();
+
+		// *****
+		// And now branch based on the next few characters!
+		// *****
+		
+		// end of stream
+		if(!INPUT)
+			return EndStream();
+
+		if(INPUT.column() == 0 && INPUT.peek() == Keys::Directive)
+			return ScanDirective();
+
+		// document token
+		if(INPUT.column() == 0 && Exp::DocStart().Matches(INPUT))
+			return ScanDocStart();
+
+		if(INPUT.column() == 0 && Exp::DocEnd().Matches(INPUT))
+			return ScanDocEnd();
+
+		// flow start/end/entry
+		if(INPUT.peek() == Keys::FlowSeqStart || INPUT.peek() == Keys::FlowMapStart)
+			return ScanFlowStart();
+
+		if(INPUT.peek() == Keys::FlowSeqEnd || INPUT.peek() == Keys::FlowMapEnd)
+			return ScanFlowEnd();
+	
+		if(INPUT.peek() == Keys::FlowEntry)
+			return ScanFlowEntry();
+
+		// block/map stuff
+		if(Exp::BlockEntry().Matches(INPUT))
+			return ScanBlockEntry();
+
+		if((InBlockContext() ? Exp::Key() : Exp::KeyInFlow()).Matches(INPUT))
+			return ScanKey();
+
+		if(GetValueRegex().Matches(INPUT))
+			return ScanValue();
+
+		// alias/anchor
+		if(INPUT.peek() == Keys::Alias || INPUT.peek() == Keys::Anchor)
+			return ScanAnchorOrAlias();
+
+		// tag
+		if(INPUT.peek() == Keys::Tag)
+			return ScanTag();
+
+		// special scalars
+		if(InBlockContext() && (INPUT.peek() == Keys::LiteralScalar || INPUT.peek() == Keys::FoldedScalar))
+			return ScanBlockScalar();
+
+		if(INPUT.peek() == '\'' || INPUT.peek() == '\"')
+			return ScanQuotedScalar();
+
+		// plain scalars
+		if((InBlockContext() ? Exp::PlainScalar() : Exp::PlainScalarInFlow()).Matches(INPUT))
+			return ScanPlainScalar();
+
+		// don't know what it is!
+		throw ParserException(INPUT.mark(), ErrorMsg::UNKNOWN_TOKEN);
+	}
+
+	// ScanToNextToken
+	// . Eats input until we reach the next token-like thing.
+	void Scanner::ScanToNextToken()
+	{
+		while(1) {
+			// first eat whitespace
+			while(INPUT && IsWhitespaceToBeEaten(INPUT.peek())) {
+				if(InBlockContext() && Exp::Tab().Matches(INPUT))
+					m_simpleKeyAllowed = false;
+				INPUT.eat(1);
+			}
+
+			// then eat a comment
+			if(Exp::Comment().Matches(INPUT)) {
+				// eat until line break
+				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))
+				break;
+
+			// otherwise, let's eat the line break and keep going
+			int n = Exp::Break().Match(INPUT);
+			INPUT.eat(n);
+
+			// oh yeah, and let's get rid of that simple key
+			InvalidateSimpleKey();
+
+			// new line - we may be able to accept a simple key now
+			if(InBlockContext())
+				m_simpleKeyAllowed = true;
+        }
+	}
+
+	///////////////////////////////////////////////////////////////////////
+	// Misc. helpers
+
+	// IsWhitespaceToBeEaten
+	// . We can eat whitespace if it's a space or tab
+	// . Note: originally tabs in block context couldn't be eaten
+	//         "where a simple key could be allowed
+	//         (i.e., not at the beginning of a line, or following '-', '?', or ':')"
+	//   I think this is wrong, since tabs can be non-content whitespace; it's just
+	//   that they can't contribute to indentation, so once you've seen a tab in a
+	//   line, you can't start a simple key
+	bool Scanner::IsWhitespaceToBeEaten(char ch)
+	{
+		if(ch == ' ')
+			return true;
+
+		if(ch == '\t')
+			return true;
+
+		return false;
+	}
+
+	// GetValueRegex
+	// . Get the appropriate regex to check if it's a value token
+	const RegEx& Scanner::GetValueRegex() const
+	{
+		if(InBlockContext())
+			return Exp::Value();
+		
+		return m_canBeJSONFlow ? Exp::ValueInJSONFlow() : Exp::ValueInFlow();
+	}
+
+	// StartStream
+	// . Set the initial conditions for starting a stream.
+	void Scanner::StartStream()
+	{
+		m_startedStream = true;
+		m_simpleKeyAllowed = true;
+		std::auto_ptr<IndentMarker> pIndent(new IndentMarker(-1, IndentMarker::NONE));
+		m_indentRefs.push_back(pIndent);
+		m_indents.push(&m_indentRefs.back());
+	}
+
+	// EndStream
+	// . Close out the stream, finish up, etc.
+	void Scanner::EndStream()
+	{
+		// force newline
+		if(INPUT.column() > 0)
+			INPUT.ResetColumn();
+
+		PopAllIndents();
+		PopAllSimpleKeys();
+
+		m_simpleKeyAllowed = false;
+		m_endedStream = true;
+	}
+
+	Token *Scanner::PushToken(Token::TYPE type)
+	{
+		m_tokens.push(Token(type, INPUT.mark()));
+		return &m_tokens.back();
+	}
+
+	Token::TYPE Scanner::GetStartTokenFor(IndentMarker::INDENT_TYPE type) const
+	{
+		switch(type) {
+			case IndentMarker::SEQ: return Token::BLOCK_SEQ_START;
+			case IndentMarker::MAP: return Token::BLOCK_MAP_START;
+			case IndentMarker::NONE: assert(false); break;
+		}
+		assert(false);
+		throw std::runtime_error("yaml-cpp: internal error, invalid indent type");
+	}
+
+	// PushIndentTo
+	// . Pushes an indentation onto the stack, and enqueues the
+	//   proper token (sequence start or mapping start).
+	// . Returns the indent marker it generates (if any).
+	Scanner::IndentMarker *Scanner::PushIndentTo(int column, IndentMarker::INDENT_TYPE type)
+	{
+		// are we in flow?
+		if(InFlowContext())
+			return 0;
+		
+		std::auto_ptr<IndentMarker> pIndent(new IndentMarker(column, type));
+		IndentMarker& indent = *pIndent;
+		const IndentMarker& lastIndent = *m_indents.top();
+
+		// is this actually an indentation?
+		if(indent.column < lastIndent.column)
+			return 0;
+		if(indent.column == lastIndent.column && !(indent.type == IndentMarker::SEQ && lastIndent.type == IndentMarker::MAP))
+			return 0;
+
+		// push a start token
+		indent.pStartToken = PushToken(GetStartTokenFor(type));
+
+		// and then the indent
+		m_indents.push(&indent);
+		m_indentRefs.push_back(pIndent);
+		return &m_indentRefs.back();
+	}
+
+	// PopIndentToHere
+	// . Pops indentations off the stack until we reach the current indentation level,
+	//   and enqueues the proper token each time.
+	// . Then pops all invalid indentations off.
+	void Scanner::PopIndentToHere()
+	{
+		// are we in flow?
+		if(InFlowContext())
+			return;
+
+		// now pop away
+		while(!m_indents.empty()) {
+			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)))
+				break;
+				
+			PopIndent();
+		}
+		
+		while(!m_indents.empty() && m_indents.top()->status == IndentMarker::INVALID)
+			PopIndent();
+	}
+	
+	// PopAllIndents
+	// . Pops all indentations (except for the base empty one) off the stack,
+	//   and enqueues the proper token each time.
+	void Scanner::PopAllIndents()
+	{
+		// are we in flow?
+		if(InFlowContext())
+			return;
+
+		// now pop away
+		while(!m_indents.empty()) {
+			const IndentMarker& indent = *m_indents.top();
+			if(indent.type == IndentMarker::NONE)
+				break;
+			
+			PopIndent();
+		}
+	}
+	
+	// PopIndent
+	// . Pops a single indent, pushing the proper token
+	void Scanner::PopIndent()
+	{
+		const IndentMarker& indent = *m_indents.top();
+		m_indents.pop();
+
+		if(indent.status != IndentMarker::VALID) {
+			InvalidateSimpleKey();
+			return;
+		}
+		
+		if(indent.type == IndentMarker::SEQ)
+			m_tokens.push(Token(Token::BLOCK_SEQ_END, INPUT.mark()));
+		else if(indent.type == IndentMarker::MAP)
+			m_tokens.push(Token(Token::BLOCK_MAP_END, INPUT.mark()));
+	}
+
+	// GetTopIndent
+	int Scanner::GetTopIndent() const
+	{
+		if(m_indents.empty())
+			return 0;
+		return m_indents.top()->column;
+	}
+
+	// ThrowParserException
+	// . Throws a ParserException with the current token location
+	//   (if available).
+	// . Does not parse any more tokens.
+	void Scanner::ThrowParserException(const std::string& msg) const
+	{
+		Mark mark = Mark::null();
+		if(!m_tokens.empty()) {
+			const Token& token = m_tokens.front();
+			mark = token.mark;
+		}
+		throw ParserException(mark, msg);
+	}
+}
+

src/scanner.h

@@ -0,0 +1,132 @@
+#ifndef SCANNER_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define SCANNER_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include <ios>
+#include <string>
+#include <queue>
+#include <stack>
+#include <set>
+#include <map>
+#include "ptr_vector.h"
+#include "stream.h"
+#include "token.h"
+
+namespace YAML
+{
+	class Node;
+	class RegEx;
+
+	class Scanner
+	{
+	public:
+		Scanner(std::istream& in);
+		~Scanner();
+
+		// token queue management (hopefully this looks kinda stl-ish)
+		bool empty();
+		void pop();
+		Token& peek();
+
+	private:
+		struct IndentMarker {
+			enum INDENT_TYPE { MAP, SEQ, NONE };
+			enum STATUS { VALID, INVALID, UNKNOWN };
+			IndentMarker(int column_, INDENT_TYPE type_): column(column_), type(type_), status(VALID), pStartToken(0) {}
+		
+			int column;
+			INDENT_TYPE type;
+			STATUS status;
+			Token *pStartToken;
+		};
+		
+		enum FLOW_MARKER { FLOW_MAP, FLOW_SEQ };
+	
+	private:	
+		// scanning
+		void EnsureTokensInQueue();
+		void ScanNextToken();
+		void ScanToNextToken();
+		void StartStream();
+		void EndStream();
+		Token *PushToken(Token::TYPE type);
+		
+		bool InFlowContext() const { return !m_flows.empty(); }
+		bool InBlockContext() const { return m_flows.empty(); }
+		int GetFlowLevel() const { return m_flows.size(); }
+		
+		Token::TYPE GetStartTokenFor(IndentMarker::INDENT_TYPE type) const;
+		IndentMarker *PushIndentTo(int column, IndentMarker::INDENT_TYPE type);
+		void PopIndentToHere();
+		void PopAllIndents();
+		void PopIndent();
+		int GetTopIndent() const;
+
+		// checking input
+		bool CanInsertPotentialSimpleKey() const;
+		bool ExistsActiveSimpleKey() const;
+		void InsertPotentialSimpleKey();
+		void InvalidateSimpleKey();
+		bool VerifySimpleKey();
+		void PopAllSimpleKeys();
+		
+		void ThrowParserException(const std::string& msg) const;
+
+		bool IsWhitespaceToBeEaten(char ch);
+		const RegEx& GetValueRegex() const;
+
+		struct SimpleKey {
+			SimpleKey(const Mark& mark_, int flowLevel_);
+
+			void Validate();
+			void Invalidate();
+			
+			Mark mark;
+			int flowLevel;
+			IndentMarker *pIndent;
+			Token *pMapStart, *pKey;
+		};
+
+		// and the tokens
+		void ScanDirective();
+		void ScanDocStart();
+		void ScanDocEnd();
+		void ScanBlockSeqStart();
+		void ScanBlockMapSTart();
+		void ScanBlockEnd();
+		void ScanBlockEntry();
+		void ScanFlowStart();
+		void ScanFlowEnd();
+		void ScanFlowEntry();
+		void ScanKey();
+		void ScanValue();
+		void ScanAnchorOrAlias();
+		void ScanTag();
+		void ScanPlainScalar();
+		void ScanQuotedScalar();
+		void ScanBlockScalar();
+
+	private:
+		// the stream
+		Stream INPUT;
+
+		// the output (tokens)
+		std::queue<Token> m_tokens;
+
+		// state info
+		bool m_startedStream, m_endedStream;
+		bool m_simpleKeyAllowed;
+		bool m_canBeJSONFlow;
+		std::stack<SimpleKey> m_simpleKeys;
+		std::stack<IndentMarker *> m_indents;
+		ptr_vector<IndentMarker> m_indentRefs; // for "garbage collection"
+		std::stack<FLOW_MARKER> m_flows;
+	};
+}
+
+#endif // SCANNER_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+

src/scanscalar.cpp

@@ -0,0 +1,199 @@
+#include "scanscalar.h"
+#include "scanner.h"
+#include "exp.h"
+#include "yaml-cpp/exceptions.h"
+#include "token.h"
+
+namespace YAML
+{
+	// ScanScalar
+	// . This is where the scalar magic happens.
+	//
+	// . We do the scanning in three phases:
+	//   1. Scan until newline
+	//   2. Eat newline
+	//   3. Scan leading blanks.
+	//
+	// . Depending on the parameters given, we store or stop
+	//   and different places in the above flow.
+	std::string ScanScalar(Stream& INPUT, ScanScalarParams& params)
+	{
+		bool foundNonEmptyLine = false;
+		bool pastOpeningBreak = (params.fold == FOLD_FLOW);
+		bool emptyLine = false, moreIndented = false;
+		int foldedNewlineCount = 0;
+		bool foldedNewlineStartedMoreIndented = false;
+		std::string scalar;
+		params.leadingSpaces = false;
+
+		while(INPUT) {
+			// ********************************
+			// Phase #1: scan until line ending
+			
+			std::size_t lastNonWhitespaceChar = scalar.size();
+			bool escapedNewline = false;
+			while(!params.end.Matches(INPUT) && !Exp::Break().Matches(INPUT)) {
+				if(!INPUT)
+					break;
+
+				// document indicator?
+				if(INPUT.column() == 0 && Exp::DocIndicator().Matches(INPUT)) {
+					if(params.onDocIndicator == BREAK)
+						break;
+					else if(params.onDocIndicator == THROW)
+						throw ParserException(INPUT.mark(), ErrorMsg::DOC_IN_SCALAR);
+				}
+
+				foundNonEmptyLine = true;
+				pastOpeningBreak = true;
+
+				// escaped newline? (only if we're escaping on slash)
+				if(params.escape == '\\' && Exp::EscBreak().Matches(INPUT)) {
+					// eat escape character and get out (but preserve trailing whitespace!)
+					INPUT.get();
+					lastNonWhitespaceChar = scalar.size();
+					escapedNewline = true;
+					break;
+				}
+
+				// escape this?
+				if(INPUT.peek() == params.escape) {
+					scalar += Exp::Escape(INPUT);
+					lastNonWhitespaceChar = scalar.size();
+					continue;
+				}
+
+				// otherwise, just add the damn character
+				char ch = INPUT.get();
+				scalar += ch;
+				if(ch != ' ' && ch != '\t')
+					lastNonWhitespaceChar = scalar.size();
+			}
+
+			// eof? if we're looking to eat something, then we throw
+			if(!INPUT) {
+				if(params.eatEnd)
+					throw ParserException(INPUT.mark(), ErrorMsg::EOF_IN_SCALAR);
+				break;
+			}
+
+			// doc indicator?
+			if(params.onDocIndicator == BREAK && INPUT.column() == 0 && Exp::DocIndicator().Matches(INPUT))
+				break;
+
+			// are we done via character match?
+			int n = params.end.Match(INPUT);
+			if(n >= 0) {
+				if(params.eatEnd)
+					INPUT.eat(n);
+				break;
+			}
+			
+			// do we remove trailing whitespace?
+			if(params.fold == FOLD_FLOW)
+				scalar.erase(lastNonWhitespaceChar);
+			
+			// ********************************
+			// Phase #2: eat line ending
+			n = Exp::Break().Match(INPUT);
+			INPUT.eat(n);
+
+			// ********************************
+			// Phase #3: scan initial spaces
+
+			// first the required indentation
+			while(INPUT.peek() == ' ' && (INPUT.column() < params.indent || (params.detectIndent && !foundNonEmptyLine)))
+				INPUT.eat(1);
+
+			// update indent if we're auto-detecting
+			if(params.detectIndent && !foundNonEmptyLine)
+				params.indent = std::max(params.indent, INPUT.column());
+
+			// and then the rest of the whitespace
+			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);
+
+				if(!params.eatLeadingWhitespace)
+					break;
+
+				INPUT.eat(1);
+			}
+
+			// was this an empty line?
+			bool nextEmptyLine = Exp::Break().Matches(INPUT);
+			bool nextMoreIndented = Exp::Blank().Matches(INPUT);
+			if(params.fold == FOLD_BLOCK && foldedNewlineCount == 0 && nextEmptyLine)
+				foldedNewlineStartedMoreIndented = moreIndented;
+
+			// for block scalars, we always start with a newline, so we should ignore it (not fold or keep)
+			if(pastOpeningBreak) {
+				switch(params.fold) {
+					case DONT_FOLD:
+						scalar += "\n";
+						break;
+					case FOLD_BLOCK:
+						if(!emptyLine && !nextEmptyLine && !moreIndented && !nextMoreIndented && INPUT.column() >= params.indent)
+							scalar += " ";
+						else if(nextEmptyLine)
+							foldedNewlineCount++;
+						else
+							scalar += "\n";
+						
+						if(!nextEmptyLine && foldedNewlineCount > 0) {
+							scalar += std::string(foldedNewlineCount - 1, '\n');
+							if(foldedNewlineStartedMoreIndented || nextMoreIndented | !foundNonEmptyLine)
+								scalar += "\n";
+							foldedNewlineCount = 0;
+						}
+						break;
+					case FOLD_FLOW:
+						if(nextEmptyLine)
+							scalar += "\n";
+						else if(!emptyLine && !nextEmptyLine && !escapedNewline)
+							scalar += " ";
+						break;
+				}
+			}
+
+			emptyLine = nextEmptyLine;
+			moreIndented = nextMoreIndented;
+			pastOpeningBreak = true;
+
+			// are we done via indentation?
+			if(!emptyLine && INPUT.column() < params.indent) {
+				params.leadingSpaces = true;
+				break;
+			}
+		}
+
+		// post-processing
+		if(params.trimTrailingSpaces) {
+			std::size_t pos = scalar.find_last_not_of(' ');
+			if(pos < scalar.size())
+				scalar.erase(pos + 1);
+		}
+
+		switch(params.chomp) {
+			case CLIP: {
+				const std::size_t pos = scalar.find_last_not_of('\n');
+				if(pos == std::string::npos)
+					scalar.erase();
+				else if(pos + 1 < scalar.size())
+					scalar.erase(pos + 2);
+			} break;
+			case STRIP: {
+				const std::size_t pos = scalar.find_last_not_of('\n');
+				if(pos == std::string::npos)
+					scalar.erase();
+				else if(pos < scalar.size())
+					scalar.erase(pos + 1);
+			} break;
+			default:
+				break;
+		}
+
+		return scalar;
+	}
+}

src/scanscalar.h

@@ -0,0 +1,45 @@
+#ifndef SCANSCALAR_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define SCANSCALAR_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include <string>
+#include "regex.h"
+#include "stream.h"
+
+namespace YAML
+{
+	enum CHOMP { STRIP = -1, CLIP, KEEP };
+	enum ACTION { NONE, BREAK, THROW };
+	enum FOLD { DONT_FOLD, FOLD_BLOCK, FOLD_FLOW };
+
+	struct ScanScalarParams {
+		ScanScalarParams(): eatEnd(false), indent(0), detectIndent(false), eatLeadingWhitespace(0), escape(0), fold(DONT_FOLD),
+			trimTrailingSpaces(0), chomp(CLIP), onDocIndicator(NONE), onTabInIndentation(NONE), leadingSpaces(false) {}
+
+		// input:
+		RegEx end;                      // what condition ends this scalar?
+		bool eatEnd;                    // should we eat that condition when we see it?
+		int indent;                     // what level of indentation should be eaten and ignored?
+		bool detectIndent;              // should we try to autodetect the indent?
+		bool eatLeadingWhitespace;      // should we continue eating this delicious indentation after 'indent' spaces?
+		char escape;                    // what character do we escape on (i.e., slash or single quote) (0 for none)
+		FOLD fold;                      // how do we fold line ends?
+		bool trimTrailingSpaces;        // do we remove all trailing spaces (at the very end)
+		CHOMP chomp;                    // do we strip, clip, or keep trailing newlines (at the very end)
+		                                //   Note: strip means kill all, clip means keep at most one, keep means keep all
+		ACTION onDocIndicator;          // what do we do if we see a document indicator?
+		ACTION onTabInIndentation;      // what do we do if we see a tab where we should be seeing indentation spaces
+
+		// output:
+		bool leadingSpaces;
+	};
+
+	std::string ScanScalar(Stream& INPUT, ScanScalarParams& info);
+}
+
+#endif // SCANSCALAR_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+

src/scantag.cpp

@@ -0,0 +1,84 @@
+#include "scanner.h"
+#include "regex.h"
+#include "exp.h"
+#include "yaml-cpp/exceptions.h"
+
+namespace YAML
+{
+	const std::string ScanVerbatimTag(Stream& INPUT)
+	{
+		std::string tag;
+		
+		// eat the start character
+		INPUT.get();
+		
+		while(INPUT) {
+			if(INPUT.peek() == Keys::VerbatimTagEnd) {
+				// eat the end character
+				INPUT.get();
+				return tag;
+			}
+		
+			int n = Exp::URI().Match(INPUT);
+			if(n <= 0)
+				break;
+			
+			tag += INPUT.get(n);
+		}
+
+		throw ParserException(INPUT.mark(), ErrorMsg::END_OF_VERBATIM_TAG);
+	}
+	
+	const std::string ScanTagHandle(Stream& INPUT, bool& canBeHandle)
+	{
+		std::string tag;
+		canBeHandle = true;
+		Mark firstNonWordChar;
+		
+		while(INPUT) {
+			if(INPUT.peek() == Keys::Tag) {
+				if(!canBeHandle)
+					throw ParserException(firstNonWordChar, ErrorMsg::CHAR_IN_TAG_HANDLE);
+				break;
+			}
+
+			int n = 0;
+			if(canBeHandle) {
+				n = Exp::Word().Match(INPUT);
+				if(n <= 0) {
+					canBeHandle = false;
+					firstNonWordChar = INPUT.mark();
+				}
+			}
+			
+			if(!canBeHandle)
+				n = Exp::Tag().Match(INPUT);
+
+			if(n <= 0)
+				break;
+			
+			tag += INPUT.get(n);
+		}
+
+		return tag;
+	}
+	
+	const std::string ScanTagSuffix(Stream& INPUT)
+	{
+		std::string tag;
+		
+		while(INPUT) {
+			int n = Exp::Tag().Match(INPUT);
+			if(n <= 0)
+				break;
+			
+			tag += INPUT.get(n);
+		}
+		
+		if(tag.empty())
+			throw ParserException(INPUT.mark(), ErrorMsg::TAG_WITH_NO_SUFFIX);
+		
+		return tag;
+	}
+}
+

src/scantag.h

@@ -0,0 +1,20 @@
+#ifndef SCANTAG_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define SCANTAG_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include <string>
+#include "stream.h"
+
+namespace YAML
+{
+	const std::string ScanVerbatimTag(Stream& INPUT);
+	const std::string ScanTagHandle(Stream& INPUT, bool& canBeHandle);
+	const std::string ScanTagSuffix(Stream& INPUT);
+}
+
+#endif // SCANTAG_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+

src/scantoken.cpp

@@ -0,0 +1,439 @@
+#include "scanner.h"
+#include "token.h"
+#include "yaml-cpp/exceptions.h"
+#include "exp.h"
+#include "scanscalar.h"
+#include "scantag.h"
+#include "tag.h"
+#include <sstream>
+
+namespace YAML
+{
+	///////////////////////////////////////////////////////////////////////
+	// Specialization for scanning specific tokens
+
+	// Directive
+	// . Note: no semantic checking is done here (that's for the parser to do)
+	void Scanner::ScanDirective()
+	{
+		std::string name;
+		std::vector <std::string> params;
+
+		// pop indents and simple keys
+		PopAllIndents();
+		PopAllSimpleKeys();
+
+		m_simpleKeyAllowed = false;
+		m_canBeJSONFlow = false;
+
+		// store pos and eat indicator
+		Token token(Token::DIRECTIVE, INPUT.mark());
+		INPUT.eat(1);
+
+		// read name
+		while(INPUT && !Exp::BlankOrBreak().Matches(INPUT))
+			token.value += INPUT.get();
+
+		// read parameters
+		while(1) {
+			// first get rid of whitespace
+			while(Exp::Blank().Matches(INPUT))
+				INPUT.eat(1);
+
+			// break on newline or comment
+			if(!INPUT || Exp::Break().Matches(INPUT) || Exp::Comment().Matches(INPUT))
+				break;
+
+			// now read parameter
+			std::string param;
+			while(INPUT && !Exp::BlankOrBreak().Matches(INPUT))
+				param += INPUT.get();
+
+			token.params.push_back(param);
+		}
+		
+		m_tokens.push(token);
+	}
+
+	// DocStart
+	void Scanner::ScanDocStart()
+	{
+		PopAllIndents();
+		PopAllSimpleKeys();
+		m_simpleKeyAllowed = false;
+		m_canBeJSONFlow = false;
+
+		// eat
+		Mark mark = INPUT.mark();
+		INPUT.eat(3);
+		m_tokens.push(Token(Token::DOC_START, mark));
+	}
+
+	// DocEnd
+	void Scanner::ScanDocEnd()
+	{
+		PopAllIndents();
+		PopAllSimpleKeys();
+		m_simpleKeyAllowed = false;
+		m_canBeJSONFlow = false;
+
+		// eat
+		Mark mark = INPUT.mark();
+		INPUT.eat(3);
+		m_tokens.push(Token(Token::DOC_END, mark));
+	}
+
+	// FlowStart
+	void Scanner::ScanFlowStart()
+	{
+		// flows can be simple keys
+		InsertPotentialSimpleKey();
+		m_simpleKeyAllowed = true;
+		m_canBeJSONFlow = false;
+
+		// eat
+		Mark mark = INPUT.mark();
+		char ch = INPUT.get();
+		FLOW_MARKER flowType = (ch == Keys::FlowSeqStart ? FLOW_SEQ : FLOW_MAP);
+		m_flows.push(flowType);
+		Token::TYPE type = (flowType == FLOW_SEQ ? Token::FLOW_SEQ_START : Token::FLOW_MAP_START);
+		m_tokens.push(Token(type, mark));
+	}
+
+	// FlowEnd
+	void Scanner::ScanFlowEnd()
+	{
+		if(InBlockContext())
+			throw ParserException(INPUT.mark(), ErrorMsg::FLOW_END);
+
+		// we might have a solo entry in the flow context
+		if(InFlowContext()) {
+			if(m_flows.top() == FLOW_MAP && VerifySimpleKey())
+				m_tokens.push(Token(Token::VALUE, INPUT.mark()));
+			else if(m_flows.top() == FLOW_SEQ)
+				InvalidateSimpleKey();
+		}
+
+		m_simpleKeyAllowed = false;
+		m_canBeJSONFlow = true;
+
+		// eat
+		Mark mark = INPUT.mark();
+		char ch = INPUT.get();
+
+		// check that it matches the start
+		FLOW_MARKER flowType = (ch == Keys::FlowSeqEnd ? FLOW_SEQ : FLOW_MAP);
+		if(m_flows.top() != flowType)
+			throw ParserException(mark, ErrorMsg::FLOW_END);
+		m_flows.pop();
+		
+		Token::TYPE type = (flowType ? Token::FLOW_SEQ_END : Token::FLOW_MAP_END);
+		m_tokens.push(Token(type, mark));
+	}
+
+	// FlowEntry
+	void Scanner::ScanFlowEntry()
+	{
+		// we might have a solo entry in the flow context
+		if(InFlowContext()) {
+			if(m_flows.top() == FLOW_MAP && VerifySimpleKey())
+				m_tokens.push(Token(Token::VALUE, INPUT.mark()));
+			else if(m_flows.top() == FLOW_SEQ)
+				InvalidateSimpleKey();
+		}
+		
+		m_simpleKeyAllowed = true;
+		m_canBeJSONFlow = false;
+
+		// eat
+		Mark mark = INPUT.mark();
+		INPUT.eat(1);
+		m_tokens.push(Token(Token::FLOW_ENTRY, mark));
+	}
+
+	// BlockEntry
+	void Scanner::ScanBlockEntry()
+	{
+		// we better be in the block context!
+		if(InFlowContext())
+			throw ParserException(INPUT.mark(), ErrorMsg::BLOCK_ENTRY);
+
+		// can we put it here?
+		if(!m_simpleKeyAllowed)
+			throw ParserException(INPUT.mark(), ErrorMsg::BLOCK_ENTRY);
+
+		PushIndentTo(INPUT.column(), IndentMarker::SEQ);
+		m_simpleKeyAllowed = true;
+		m_canBeJSONFlow = false;
+
+		// eat
+		Mark mark = INPUT.mark();
+		INPUT.eat(1);
+		m_tokens.push(Token(Token::BLOCK_ENTRY, mark));
+	}
+
+	// Key
+	void Scanner::ScanKey()
+	{
+		// handle keys diffently in the block context (and manage indents)
+		if(InBlockContext()) {
+			if(!m_simpleKeyAllowed)
+				throw ParserException(INPUT.mark(), ErrorMsg::MAP_KEY);
+
+			PushIndentTo(INPUT.column(), IndentMarker::MAP);
+		}
+
+		// can only put a simple key here if we're in block context
+		m_simpleKeyAllowed = InBlockContext();
+
+		// eat
+		Mark mark = INPUT.mark();
+		INPUT.eat(1);
+		m_tokens.push(Token(Token::KEY, mark));
+	}
+
+	// Value
+	void Scanner::ScanValue()
+	{
+		// and check that simple key
+		bool isSimpleKey = VerifySimpleKey();
+		m_canBeJSONFlow = false;
+		
+		if(isSimpleKey) {
+			// can't follow a simple key with another simple key (dunno why, though - it seems fine)
+			m_simpleKeyAllowed = false;
+		} else {
+			// handle values diffently in the block context (and manage indents)
+			if(InBlockContext()) {
+				if(!m_simpleKeyAllowed)
+					throw ParserException(INPUT.mark(), ErrorMsg::MAP_VALUE);
+
+				PushIndentTo(INPUT.column(), IndentMarker::MAP);
+			}
+
+			// can only put a simple key here if we're in block context
+			m_simpleKeyAllowed = InBlockContext();
+		}
+
+		// eat
+		Mark mark = INPUT.mark();
+		INPUT.eat(1);
+		m_tokens.push(Token(Token::VALUE, mark));
+	}
+
+	// AnchorOrAlias
+	void Scanner::ScanAnchorOrAlias()
+	{
+		bool alias;
+		std::string name;
+
+		// insert a potential simple key
+		InsertPotentialSimpleKey();
+		m_simpleKeyAllowed = false;
+		m_canBeJSONFlow = false;
+
+		// eat the indicator
+		Mark mark = INPUT.mark();
+		char indicator = INPUT.get();
+		alias = (indicator == Keys::Alias);
+
+		// now eat the content
+		while(INPUT && Exp::Anchor().Matches(INPUT))
+			name += INPUT.get();
+
+		// we need to have read SOMETHING!
+		if(name.empty())
+			throw ParserException(INPUT.mark(), alias ? ErrorMsg::ALIAS_NOT_FOUND : ErrorMsg::ANCHOR_NOT_FOUND);
+
+		// and needs to end correctly
+		if(INPUT && !Exp::AnchorEnd().Matches(INPUT))
+			throw ParserException(INPUT.mark(), alias ? ErrorMsg::CHAR_IN_ALIAS : ErrorMsg::CHAR_IN_ANCHOR);
+
+		// and we're done
+		Token token(alias ? Token::ALIAS : Token::ANCHOR, mark);
+		token.value = name;
+		m_tokens.push(token);
+	}
+
+	// Tag
+	void Scanner::ScanTag()
+	{
+		// insert a potential simple key
+		InsertPotentialSimpleKey();
+		m_simpleKeyAllowed = false;
+		m_canBeJSONFlow = false;
+
+		Token token(Token::TAG, INPUT.mark());
+
+		// eat the indicator
+		INPUT.get();
+		
+		if(INPUT && INPUT.peek() == Keys::VerbatimTagStart){
+			std::string tag = ScanVerbatimTag(INPUT);
+
+			token.value = tag;
+			token.data = Tag::VERBATIM;
+		} else {
+			bool canBeHandle;
+			token.value = ScanTagHandle(INPUT, canBeHandle);
+			if(!canBeHandle && token.value.empty())
+				token.data = Tag::NON_SPECIFIC;
+			else if(token.value.empty())
+				token.data = Tag::SECONDARY_HANDLE;
+			else
+				token.data = Tag::PRIMARY_HANDLE;
+			
+			// is there a suffix?
+			if(canBeHandle && INPUT.peek() == Keys::Tag) {
+				// eat the indicator
+				INPUT.get();
+				token.params.push_back(ScanTagSuffix(INPUT));
+				token.data = Tag::NAMED_HANDLE;
+			}
+		}
+
+		m_tokens.push(token);
+	}
+
+	// PlainScalar
+	void Scanner::ScanPlainScalar()
+	{
+		std::string scalar;
+
+		// set up the scanning parameters
+		ScanScalarParams params;
+		params.end = (InFlowContext() ? Exp::EndScalarInFlow() : Exp::EndScalar()) || (Exp::BlankOrBreak() + Exp::Comment());
+		params.eatEnd = false;
+		params.indent = (InFlowContext() ? 0 : GetTopIndent() + 1);
+		params.fold = FOLD_FLOW;
+		params.eatLeadingWhitespace = true;
+		params.trimTrailingSpaces = true;
+		params.chomp = STRIP;
+		params.onDocIndicator = BREAK;
+		params.onTabInIndentation = THROW;
+
+		// insert a potential simple key
+		InsertPotentialSimpleKey();
+
+		Mark mark = INPUT.mark();
+		scalar = ScanScalar(INPUT, params);
+
+		// can have a simple key only if we ended the scalar by starting a new line
+		m_simpleKeyAllowed = params.leadingSpaces;
+		m_canBeJSONFlow = false;
+
+		// finally, check and see if we ended on an illegal character
+		//if(Exp::IllegalCharInScalar.Matches(INPUT))
+		//	throw ParserException(INPUT.mark(), ErrorMsg::CHAR_IN_SCALAR);
+
+		Token token(Token::PLAIN_SCALAR, mark);
+		token.value = scalar;
+		m_tokens.push(token);
+	}
+
+	// QuotedScalar
+	void Scanner::ScanQuotedScalar()
+	{
+		std::string scalar;
+
+		// peek at single or double quote (don't eat because we need to preserve (for the time being) the input position)
+		char quote = INPUT.peek();
+		bool single = (quote == '\'');
+
+		// setup the scanning parameters
+		ScanScalarParams params;
+		params.end = (single ? RegEx(quote) && !Exp::EscSingleQuote() : RegEx(quote));
+		params.eatEnd = true;
+		params.escape = (single ? '\'' : '\\');
+		params.indent = 0;
+		params.fold = FOLD_FLOW;
+		params.eatLeadingWhitespace = true;
+		params.trimTrailingSpaces = false;
+		params.chomp = CLIP;
+		params.onDocIndicator = THROW;
+
+		// insert a potential simple key
+		InsertPotentialSimpleKey();
+
+		Mark mark = INPUT.mark();
+
+		// now eat that opening quote
+		INPUT.get();
+		
+		// and scan
+		scalar = ScanScalar(INPUT, params);
+		m_simpleKeyAllowed = false;
+		m_canBeJSONFlow = true;
+
+		Token token(Token::NON_PLAIN_SCALAR, mark);
+		token.value = scalar;
+		m_tokens.push(token);
+	}
+
+	// BlockScalarToken
+	// . These need a little extra processing beforehand.
+	// . We need to scan the line where the indicator is (this doesn't count as part of the scalar),
+	//   and then we need to figure out what level of indentation we'll be using.
+	void Scanner::ScanBlockScalar()
+	{
+		std::string scalar;
+
+		ScanScalarParams params;
+		params.indent = 1;
+		params.detectIndent = true;
+
+		// eat block indicator ('|' or '>')
+		Mark mark = INPUT.mark();
+		char indicator = INPUT.get();
+		params.fold = (indicator == Keys::FoldedScalar ? FOLD_BLOCK : DONT_FOLD);
+
+		// eat chomping/indentation indicators
+		params.chomp = CLIP;
+		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)) {
+				if(ch == '0')
+					throw ParserException(INPUT.mark(), ErrorMsg::ZERO_INDENT_IN_BLOCK);
+
+				params.indent = ch - '0';
+				params.detectIndent = false;
+			}
+		}
+
+		// now eat whitespace
+		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))
+				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 ParserException(INPUT.mark(), ErrorMsg::CHAR_IN_BLOCK);
+
+		// set the initial indentation
+		if(GetTopIndent() >= 0)
+			params.indent += GetTopIndent();
+
+		params.eatLeadingWhitespace = false;
+		params.trimTrailingSpaces = false;
+		params.onTabInIndentation = THROW;
+
+		scalar = ScanScalar(INPUT, params);
+
+		// simple keys always ok after block scalars (since we're gonna start a new line anyways)
+		m_simpleKeyAllowed = true;
+		m_canBeJSONFlow = false;
+
+		Token token(Token::NON_PLAIN_SCALAR, mark);
+		token.value = scalar;
+		m_tokens.push(token);
+	}
+}

src/setting.h

@@ -0,0 +1,105 @@
+#ifndef SETTING_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define SETTING_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include <memory>
+#include <vector>
+#include "yaml-cpp/noncopyable.h"
+
+namespace YAML
+{
+	class SettingChangeBase;
+	
+	template <typename T>
+	class Setting
+	{
+	public:
+		Setting(): m_value() {}
+		
+		const T get() const { return m_value; }
+		std::auto_ptr <SettingChangeBase> set(const T& value);
+		void restore(const Setting<T>& oldSetting) {
+			m_value = oldSetting.get();
+		}
+		
+	private:
+		T m_value;
+	};
+
+	class SettingChangeBase
+	{
+	public:
+		virtual ~SettingChangeBase() {}
+		virtual void pop() = 0;
+	};
+	
+	template <typename T>
+	class SettingChange: public SettingChangeBase
+	{
+	public:
+		SettingChange(Setting<T> *pSetting): m_pCurSetting(pSetting) {
+			// copy old setting to save its state
+			m_oldSetting = *pSetting;
+		}
+		
+		virtual void pop() {
+			m_pCurSetting->restore(m_oldSetting);
+		}
+
+	private:
+		Setting<T> *m_pCurSetting;
+		Setting<T> m_oldSetting;
+	};
+
+	template <typename T>
+	inline std::auto_ptr <SettingChangeBase> Setting<T>::set(const T& value) {
+		std::auto_ptr <SettingChangeBase> pChange(new SettingChange<T> (this));
+		m_value = value;
+		return pChange;
+	}
+	
+	class SettingChanges: private noncopyable
+	{
+	public:
+		SettingChanges() {}
+		~SettingChanges() { clear(); }
+		
+		void clear() {
+			restore();
+			
+			for(setting_changes::const_iterator it=m_settingChanges.begin();it!=m_settingChanges.end();++it)
+				delete *it;
+			m_settingChanges.clear();
+		}
+		
+		void restore() {
+			for(setting_changes::const_iterator it=m_settingChanges.begin();it!=m_settingChanges.end();++it)
+				(*it)->pop();
+		}
+		
+		void push(std::auto_ptr <SettingChangeBase> pSettingChange) {
+			m_settingChanges.push_back(pSettingChange.release());
+		}
+		
+		// like std::auto_ptr - assignment is transfer of ownership
+		SettingChanges& operator = (SettingChanges& rhs) {
+			if(this == &rhs)
+				return *this;
+			
+			clear();
+			m_settingChanges = rhs.m_settingChanges;
+			rhs.m_settingChanges.clear();
+			return *this;
+		}
+		
+	private:
+		typedef std::vector <SettingChangeBase *> setting_changes;
+		setting_changes m_settingChanges;
+	};
+}
+
+#endif // SETTING_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/simplekey.cpp

@@ -0,0 +1,139 @@
+#include "scanner.h"
+#include "token.h"
+#include "yaml-cpp/exceptions.h"
+#include "exp.h"
+
+namespace YAML
+{
+	Scanner::SimpleKey::SimpleKey(const Mark& mark_, int flowLevel_)
+		: mark(mark_), flowLevel(flowLevel_), pIndent(0), pMapStart(0), pKey(0)
+	{
+	}
+
+	void Scanner::SimpleKey::Validate()
+	{
+		// Note: pIndent will *not* be garbage here;
+		//       we "garbage collect" them so we can
+		//       always refer to them
+		if(pIndent)
+			pIndent->status = IndentMarker::VALID;
+		if(pMapStart)
+			pMapStart->status = Token::VALID;
+		if(pKey)
+			pKey->status = Token::VALID;
+	}
+
+	void Scanner::SimpleKey::Invalidate()
+	{
+		if(pIndent)
+			pIndent->status = IndentMarker::INVALID;
+		if(pMapStart)
+			pMapStart->status = Token::INVALID;
+		if(pKey)
+			pKey->status = Token::INVALID;
+	}
+	
+	// CanInsertPotentialSimpleKey
+	bool Scanner::CanInsertPotentialSimpleKey() const
+	{
+		if(!m_simpleKeyAllowed)
+			return false;
+
+		return !ExistsActiveSimpleKey();
+	}
+
+	// ExistsActiveSimpleKey
+	// . Returns true if there's a potential simple key at our flow level
+	//   (there's allowed at most one per flow level, i.e., at the start of the flow start token)
+	bool Scanner::ExistsActiveSimpleKey() const
+	{
+		if(m_simpleKeys.empty())
+			return false;
+		
+		const SimpleKey& key = m_simpleKeys.top();
+		return key.flowLevel == GetFlowLevel();
+	}
+
+	// InsertPotentialSimpleKey
+	// . If we can, add a potential simple key to the queue,
+	//   and save it on a stack.
+	void Scanner::InsertPotentialSimpleKey()
+	{
+		if(!CanInsertPotentialSimpleKey())
+			return;
+		
+		SimpleKey key(INPUT.mark(), GetFlowLevel());
+
+		// first add a map start, if necessary
+		if(InBlockContext()) {
+			key.pIndent = PushIndentTo(INPUT.column(), IndentMarker::MAP);
+			if(key.pIndent) {
+				key.pIndent->status = IndentMarker::UNKNOWN;
+				key.pMapStart = key.pIndent->pStartToken;
+				key.pMapStart->status = Token::UNVERIFIED;
+			}
+		}
+
+		// then add the (now unverified) key
+		m_tokens.push(Token(Token::KEY, INPUT.mark()));
+		key.pKey = &m_tokens.back();
+		key.pKey->status = Token::UNVERIFIED;
+
+		m_simpleKeys.push(key);
+	}
+
+	// InvalidateSimpleKey
+	// . Automatically invalidate the simple key in our flow level
+	void Scanner::InvalidateSimpleKey()
+	{
+		if(m_simpleKeys.empty())
+			return;
+		
+		// grab top key
+		SimpleKey& key = m_simpleKeys.top();
+		if(key.flowLevel != GetFlowLevel())
+			return;
+		
+		key.Invalidate();
+		m_simpleKeys.pop();
+	}
+
+	// VerifySimpleKey
+	// . Determines whether the latest simple key to be added is valid,
+	//   and if so, makes it valid.
+	bool Scanner::VerifySimpleKey()
+	{
+		if(m_simpleKeys.empty())
+			return false;
+
+		// grab top key
+		SimpleKey key = m_simpleKeys.top();
+
+		// only validate if we're in the correct flow level
+		if(key.flowLevel != GetFlowLevel())
+			return false;
+
+		m_simpleKeys.pop();
+
+		bool isValid = true;
+
+		// needs to be less than 1024 characters and inline
+		if(INPUT.line() != key.mark.line || INPUT.pos() - key.mark.pos > 1024)
+			isValid = false;
+
+		// invalidate key
+		if(isValid)
+			key.Validate();
+		else
+			key.Invalidate();
+
+		return isValid;
+	}
+
+	void Scanner::PopAllSimpleKeys()
+	{
+		while(!m_simpleKeys.empty())
+			m_simpleKeys.pop();
+	}
+}
+

src/singledocparser.cpp

@@ -0,0 +1,381 @@
+#include "singledocparser.h"
+#include "collectionstack.h"
+#include "directives.h"
+#include "yaml-cpp/eventhandler.h"
+#include "yaml-cpp/exceptions.h"
+#include "scanner.h"
+#include "tag.h"
+#include "token.h"
+#include <sstream>
+#include <cstdio>
+#include <algorithm>
+
+namespace YAML
+{
+	SingleDocParser::SingleDocParser(Scanner& scanner, const Directives& directives): m_scanner(scanner), m_directives(directives), m_pCollectionStack(new CollectionStack), m_curAnchor(0)
+	{
+	}
+
+	SingleDocParser::~SingleDocParser()
+	{
+	}
+
+	// HandleDocument
+	// . Handles the next document
+	// . Throws a ParserException on error.
+	void SingleDocParser::HandleDocument(EventHandler& eventHandler)
+	{
+		assert(!m_scanner.empty()); // guaranteed that there are tokens
+		assert(!m_curAnchor);
+
+		eventHandler.OnDocumentStart(m_scanner.peek().mark);
+		
+		// eat doc start
+		if(m_scanner.peek().type == Token::DOC_START)
+			m_scanner.pop();
+		
+		// recurse!
+		HandleNode(eventHandler);
+		
+		eventHandler.OnDocumentEnd();
+		
+		// and finally eat any doc ends we see
+		while(!m_scanner.empty() && m_scanner.peek().type == Token::DOC_END)
+			m_scanner.pop();
+	}
+	
+	void SingleDocParser::HandleNode(EventHandler& eventHandler)
+	{
+		// an empty node *is* a possibility
+		if(m_scanner.empty()) {
+			eventHandler.OnNull(Mark::null(), NullAnchor);
+			return;
+		}
+		
+		// save location
+		Mark mark = m_scanner.peek().mark;
+		
+		// special case: a value node by itself must be a map, with no header
+		if(m_scanner.peek().type == Token::VALUE) {
+			eventHandler.OnMapStart(mark, "", NullAnchor);
+			HandleMap(eventHandler);
+			eventHandler.OnMapEnd();
+			return;
+		}
+		
+		// special case: an alias node
+		if(m_scanner.peek().type == Token::ALIAS) {
+			eventHandler.OnAlias(mark, LookupAnchor(mark, m_scanner.peek().value));
+			m_scanner.pop();
+			return;
+		}
+		
+		std::string tag;
+		anchor_t anchor;
+		ParseProperties(tag, anchor);
+		
+		const Token& token = m_scanner.peek();
+		
+		// add non-specific tags
+		if(tag.empty())
+			tag = (token.type == Token::NON_PLAIN_SCALAR ? "!" : "?");
+		
+		// now split based on what kind of node we should be
+		switch(token.type) {
+			case Token::PLAIN_SCALAR:
+			case Token::NON_PLAIN_SCALAR:
+				eventHandler.OnScalar(mark, tag, anchor, token.value);
+				m_scanner.pop();
+				return;
+			case Token::FLOW_SEQ_START:
+			case Token::BLOCK_SEQ_START:
+				eventHandler.OnSequenceStart(mark, tag, anchor);
+				HandleSequence(eventHandler);
+				eventHandler.OnSequenceEnd();
+				return;
+			case Token::FLOW_MAP_START:
+			case Token::BLOCK_MAP_START:
+				eventHandler.OnMapStart(mark, tag, anchor);
+				HandleMap(eventHandler);
+				eventHandler.OnMapEnd();
+				return;
+			case Token::KEY:
+				// compact maps can only go in a flow sequence
+				if(m_pCollectionStack->GetCurCollectionType() == CollectionType::FlowSeq) {
+					eventHandler.OnMapStart(mark, tag, anchor);
+					HandleMap(eventHandler);
+					eventHandler.OnMapEnd();
+					return;
+				}
+				break;
+			default:
+				break;
+		}
+		
+		if(tag == "?")
+			eventHandler.OnNull(mark, anchor);
+		else
+			eventHandler.OnScalar(mark, tag, anchor, "");
+	}
+	
+	void SingleDocParser::HandleSequence(EventHandler& eventHandler)
+	{
+		// split based on start token
+		switch(m_scanner.peek().type) {
+			case Token::BLOCK_SEQ_START: HandleBlockSequence(eventHandler); break;
+			case Token::FLOW_SEQ_START: HandleFlowSequence(eventHandler); break;
+			default: break;
+		}
+	}
+	
+	void SingleDocParser::HandleBlockSequence(EventHandler& eventHandler)
+	{
+		// eat start token
+		m_scanner.pop();
+		m_pCollectionStack->PushCollectionType(CollectionType::BlockSeq);
+		
+		while(1) {
+			if(m_scanner.empty())
+				throw ParserException(Mark::null(), ErrorMsg::END_OF_SEQ);
+			
+			Token token = m_scanner.peek();
+			if(token.type != Token::BLOCK_ENTRY && token.type != Token::BLOCK_SEQ_END)
+				throw ParserException(token.mark, ErrorMsg::END_OF_SEQ);
+			
+			m_scanner.pop();
+			if(token.type == Token::BLOCK_SEQ_END)
+				break;
+			
+			// check for null
+			if(!m_scanner.empty()) {
+				const Token& token = m_scanner.peek();
+				if(token.type == Token::BLOCK_ENTRY || token.type == Token::BLOCK_SEQ_END) {
+					eventHandler.OnNull(token.mark, NullAnchor);
+					continue;
+				}
+			}
+			
+			HandleNode(eventHandler);
+		}
+		
+		m_pCollectionStack->PopCollectionType(CollectionType::BlockSeq);
+	}
+	
+	void SingleDocParser::HandleFlowSequence(EventHandler& eventHandler)
+	{
+		// eat start token
+		m_scanner.pop();
+		m_pCollectionStack->PushCollectionType(CollectionType::FlowSeq);
+		
+		while(1) {
+			if(m_scanner.empty())
+				throw ParserException(Mark::null(), ErrorMsg::END_OF_SEQ_FLOW);
+			
+			// first check for end
+			if(m_scanner.peek().type == Token::FLOW_SEQ_END) {
+				m_scanner.pop();
+				break;
+			}
+			
+			// then read the node
+			HandleNode(eventHandler);
+			
+			// now eat the separator (or could be a sequence end, which we ignore - but if it's neither, then it's a bad node)
+			Token& token = m_scanner.peek();
+			if(token.type == Token::FLOW_ENTRY)
+				m_scanner.pop();
+			else if(token.type != Token::FLOW_SEQ_END)
+				throw ParserException(token.mark, ErrorMsg::END_OF_SEQ_FLOW);
+		}
+		
+		m_pCollectionStack->PopCollectionType(CollectionType::FlowSeq);
+	}
+	
+	void SingleDocParser::HandleMap(EventHandler& eventHandler)
+	{
+		// split based on start token
+		switch(m_scanner.peek().type) {
+			case Token::BLOCK_MAP_START: HandleBlockMap(eventHandler); break;
+			case Token::FLOW_MAP_START: HandleFlowMap(eventHandler); break;
+			case Token::KEY: HandleCompactMap(eventHandler); break;
+			case Token::VALUE: HandleCompactMapWithNoKey(eventHandler); break;
+			default: break;
+		}
+	}
+	
+	void SingleDocParser::HandleBlockMap(EventHandler& eventHandler)
+	{
+		// eat start token
+		m_scanner.pop();
+		m_pCollectionStack->PushCollectionType(CollectionType::BlockMap);
+		
+		while(1) {
+			if(m_scanner.empty())
+				throw ParserException(Mark::null(), ErrorMsg::END_OF_MAP);
+			
+			Token token = m_scanner.peek();
+			if(token.type != Token::KEY && token.type != Token::VALUE && token.type != Token::BLOCK_MAP_END)
+				throw ParserException(token.mark, ErrorMsg::END_OF_MAP);
+			
+			if(token.type == Token::BLOCK_MAP_END) {
+				m_scanner.pop();
+				break;
+			}
+			
+			// grab key (if non-null)
+			if(token.type == Token::KEY) {
+				m_scanner.pop();
+				HandleNode(eventHandler);
+			} else {
+				eventHandler.OnNull(token.mark, NullAnchor);
+			}
+			
+			// now grab value (optional)
+			if(!m_scanner.empty() && m_scanner.peek().type == Token::VALUE) {
+				m_scanner.pop();
+				HandleNode(eventHandler);
+			} else {
+				eventHandler.OnNull(token.mark, NullAnchor);
+			}
+		}
+		
+		m_pCollectionStack->PopCollectionType(CollectionType::BlockMap);
+	}
+	
+	void SingleDocParser::HandleFlowMap(EventHandler& eventHandler)
+	{
+		// eat start token
+		m_scanner.pop();
+		m_pCollectionStack->PushCollectionType(CollectionType::FlowMap);
+		
+		while(1) {
+			if(m_scanner.empty())
+				throw ParserException(Mark::null(), ErrorMsg::END_OF_MAP_FLOW);
+			
+			Token& token = m_scanner.peek();
+			// first check for end
+			if(token.type == Token::FLOW_MAP_END) {
+				m_scanner.pop();
+				break;
+			}
+			
+			// grab key (if non-null)
+			if(token.type == Token::KEY) {
+				m_scanner.pop();
+				HandleNode(eventHandler);
+			} else {
+				eventHandler.OnNull(token.mark, NullAnchor);
+			}
+			
+			// now grab value (optional)
+			if(!m_scanner.empty() && m_scanner.peek().type == Token::VALUE) {
+				m_scanner.pop();
+				HandleNode(eventHandler);
+			} else {
+				eventHandler.OnNull(token.mark, NullAnchor);
+			}
+			
+			// now eat the separator (or could be a map end, which we ignore - but if it's neither, then it's a bad node)
+			Token& nextToken = m_scanner.peek();
+			if(nextToken.type == Token::FLOW_ENTRY)
+				m_scanner.pop();
+			else if(nextToken.type != Token::FLOW_MAP_END)
+				throw ParserException(nextToken.mark, ErrorMsg::END_OF_MAP_FLOW);
+		}
+		
+		m_pCollectionStack->PopCollectionType(CollectionType::FlowMap);
+	}
+	
+	// . Single "key: value" pair in a flow sequence
+	void SingleDocParser::HandleCompactMap(EventHandler& eventHandler)
+	{
+		m_pCollectionStack->PushCollectionType(CollectionType::CompactMap);
+		
+		// grab key
+		Mark mark = m_scanner.peek().mark;
+		m_scanner.pop();
+		HandleNode(eventHandler);
+		
+		// now grab value (optional)
+		if(!m_scanner.empty() && m_scanner.peek().type == Token::VALUE) {
+			m_scanner.pop();
+			HandleNode(eventHandler);
+		} else {
+			eventHandler.OnNull(mark, NullAnchor);
+		}
+		
+		m_pCollectionStack->PopCollectionType(CollectionType::CompactMap);
+	}
+	
+	// . Single ": value" pair in a flow sequence
+	void SingleDocParser::HandleCompactMapWithNoKey(EventHandler& eventHandler)
+	{
+		m_pCollectionStack->PushCollectionType(CollectionType::CompactMap);
+		
+		// null key
+		eventHandler.OnNull(m_scanner.peek().mark, NullAnchor);
+		
+		// grab value
+		m_scanner.pop();
+		HandleNode(eventHandler);
+		
+		m_pCollectionStack->PopCollectionType(CollectionType::CompactMap);
+	}
+	
+	// ParseProperties
+	// . Grabs any tag or anchor tokens and deals with them.
+	void SingleDocParser::ParseProperties(std::string& tag, anchor_t& anchor)
+	{
+		tag.clear();
+		anchor = NullAnchor;
+		
+		while(1) {
+			if(m_scanner.empty())
+				return;
+			
+			switch(m_scanner.peek().type) {
+				case Token::TAG: ParseTag(tag); break;
+				case Token::ANCHOR: ParseAnchor(anchor); break;
+				default: return;
+			}
+		}
+	}
+	
+	void SingleDocParser::ParseTag(std::string& tag)
+	{
+		Token& token = m_scanner.peek();
+		if(!tag.empty())
+			throw ParserException(token.mark, ErrorMsg::MULTIPLE_TAGS);
+		
+		Tag tagInfo(token);
+		tag = tagInfo.Translate(m_directives);
+		m_scanner.pop();
+	}
+	
+	void SingleDocParser::ParseAnchor(anchor_t& anchor)
+	{
+		Token& token = m_scanner.peek();
+		if(anchor)
+			throw ParserException(token.mark, ErrorMsg::MULTIPLE_ANCHORS);
+		
+		anchor = RegisterAnchor(token.value);
+		m_scanner.pop();
+	}
+	
+	anchor_t SingleDocParser::RegisterAnchor(const std::string& name)
+	{
+		if(name.empty())
+			return NullAnchor;
+		
+		return m_anchors[name] = ++m_curAnchor;
+	}
+	
+	anchor_t SingleDocParser::LookupAnchor(const Mark& mark, const std::string& name) const
+	{
+		Anchors::const_iterator it = m_anchors.find(name);
+		if(it == m_anchors.end())
+			throw ParserException(mark, ErrorMsg::UNKNOWN_ANCHOR);
+		
+		return it->second;
+	}
+}

src/singledocparser.h

@@ -0,0 +1,65 @@
+#ifndef SINGLEDOCPARSER_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define SINGLEDOCPARSER_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "yaml-cpp/anchor.h"
+#include "yaml-cpp/noncopyable.h"
+#include <string>
+#include <map>
+#include <memory>
+
+namespace YAML
+{
+	struct Directives;
+	struct Mark;
+	struct Token;
+	class CollectionStack;
+	class EventHandler;
+	class Node;
+	class Scanner;
+	
+	class SingleDocParser: private noncopyable
+	{
+	public:
+		SingleDocParser(Scanner& scanner, const Directives& directives);
+		~SingleDocParser();
+
+		void HandleDocument(EventHandler& eventHandler);
+
+	private:
+		void HandleNode(EventHandler& eventHandler);
+		
+		void HandleSequence(EventHandler& eventHandler);
+		void HandleBlockSequence(EventHandler& eventHandler);
+		void HandleFlowSequence(EventHandler& eventHandler);
+		
+		void HandleMap(EventHandler& eventHandler);
+		void HandleBlockMap(EventHandler& eventHandler);
+		void HandleFlowMap(EventHandler& eventHandler);
+		void HandleCompactMap(EventHandler& eventHandler);
+		void HandleCompactMapWithNoKey(EventHandler& eventHandler);
+		
+		void ParseProperties(std::string& tag, anchor_t& anchor);
+		void ParseTag(std::string& tag);
+		void ParseAnchor(anchor_t& anchor);
+		
+		anchor_t RegisterAnchor(const std::string& name);
+		anchor_t LookupAnchor(const Mark& mark, const std::string& name) const;
+		
+	private:
+		Scanner& m_scanner;
+		const Directives& m_directives;
+		std::auto_ptr<CollectionStack> m_pCollectionStack;
+		
+		typedef std::map<std::string, anchor_t> Anchors;
+		Anchors m_anchors;
+		
+		anchor_t m_curAnchor;
+	};
+}
+
+#endif // SINGLEDOCPARSER_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/stream.cpp

@@ -0,0 +1,454 @@
+#include "stream.h"
+#include <iostream>
+#include "exp.h"
+
+#ifndef YAML_PREFETCH_SIZE
+#define YAML_PREFETCH_SIZE 2048
+#endif
+
+#define S_ARRAY_SIZE( A ) (sizeof(A)/sizeof(*(A)))
+#define S_ARRAY_END( A ) ((A) + S_ARRAY_SIZE(A))
+
+#define CP_REPLACEMENT_CHARACTER (0xFFFD)
+
+namespace YAML
+{
+	enum UtfIntroState {
+		uis_start,
+		uis_utfbe_b1,
+		uis_utf32be_b2,
+		uis_utf32be_bom3,
+		uis_utf32be,
+		uis_utf16be,
+		uis_utf16be_bom1,
+		uis_utfle_bom1,
+		uis_utf16le_bom2,
+		uis_utf32le_bom3,
+		uis_utf16le,
+		uis_utf32le,
+		uis_utf8_imp,
+		uis_utf16le_imp,
+		uis_utf32le_imp3,
+		uis_utf8_bom1,
+		uis_utf8_bom2,
+		uis_utf8,
+		uis_error
+	};
+
+	enum UtfIntroCharType {
+		uict00,
+		uictBB,
+		uictBF,
+		uictEF,
+		uictFE,
+		uictFF,
+		uictAscii,
+		uictOther,
+		uictMax
+	};
+
+	static bool s_introFinalState[] = {
+		false, //uis_start
+		false, //uis_utfbe_b1
+		false, //uis_utf32be_b2
+		false, //uis_utf32be_bom3
+		true,  //uis_utf32be
+		true,  //uis_utf16be
+		false, //uis_utf16be_bom1
+		false, //uis_utfle_bom1
+		false, //uis_utf16le_bom2
+		false, //uis_utf32le_bom3
+		true,  //uis_utf16le
+		true,  //uis_utf32le
+		false, //uis_utf8_imp
+		false, //uis_utf16le_imp
+		false, //uis_utf32le_imp3
+		false, //uis_utf8_bom1
+		false, //uis_utf8_bom2
+		true,  //uis_utf8
+		true,  //uis_error
+	};
+
+	static UtfIntroState s_introTransitions[][uictMax] = {
+		// uict00,           uictBB,           uictBF,           uictEF,           uictFE,           uictFF,           uictAscii,        uictOther
+		  {uis_utfbe_b1,     uis_utf8,         uis_utf8,         uis_utf8_bom1,    uis_utf16be_bom1, uis_utfle_bom1,   uis_utf8_imp,     uis_utf8},
+		  {uis_utf32be_b2,   uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf16be,      uis_utf8},
+		  {uis_utf32be,      uis_utf8,         uis_utf8,         uis_utf8,         uis_utf32be_bom3, uis_utf8,         uis_utf8,         uis_utf8},
+		  {uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf32be,      uis_utf8,         uis_utf8},
+		  {uis_utf32be,      uis_utf32be,      uis_utf32be,      uis_utf32be,      uis_utf32be,      uis_utf32be,      uis_utf32be,      uis_utf32be},
+		  {uis_utf16be,      uis_utf16be,      uis_utf16be,      uis_utf16be,      uis_utf16be,      uis_utf16be,      uis_utf16be,      uis_utf16be},
+		  {uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf16be,      uis_utf8,         uis_utf8},
+		  {uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf16le_bom2, uis_utf8,         uis_utf8,         uis_utf8},
+		  {uis_utf32le_bom3, uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le},
+		  {uis_utf32le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le},
+		  {uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le},
+		  {uis_utf32le,      uis_utf32le,      uis_utf32le,      uis_utf32le,      uis_utf32le,      uis_utf32le,      uis_utf32le,      uis_utf32le},
+		  {uis_utf16le_imp,  uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8},
+		  {uis_utf32le_imp3, uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le},
+		  {uis_utf32le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le,      uis_utf16le},
+		  {uis_utf8,         uis_utf8_bom2,    uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8},
+		  {uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8},
+		  {uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8,         uis_utf8},
+	};
+
+	static char s_introUngetCount[][uictMax] = {
+		// uict00, uictBB, uictBF, uictEF, uictFE, uictFF, uictAscii, uictOther
+		  {0,      1,      1,      0,      0,      0,      0,         1},
+		  {0,      2,      2,      2,      2,      2,      2,         2},
+		  {3,      3,      3,      3,      0,      3,      3,         3},
+		  {4,      4,      4,      4,      4,      0,      4,         4},
+		  {1,      1,      1,      1,      1,      1,      1,         1},
+		  {1,      1,      1,      1,      1,      1,      1,         1},
+		  {2,      2,      2,      2,      2,      0,      2,         2},
+		  {2,      2,      2,      2,      0,      2,      2,         2},
+		  {0,      1,      1,      1,      1,      1,      1,         1},
+		  {0,      2,      2,      2,      2,      2,      2,         2},
+		  {1,      1,      1,      1,      1,      1,      1,         1},
+		  {1,      1,      1,      1,      1,      1,      1,         1},
+		  {0,      2,      2,      2,      2,      2,      2,         2},
+		  {0,      3,      3,      3,      3,      3,      3,         3},
+		  {4,      4,      4,      4,      4,      4,      4,         4},
+		  {2,      0,      2,      2,      2,      2,      2,         2},
+		  {3,      3,      0,      3,      3,      3,      3,         3},
+		  {1,      1,      1,      1,      1,      1,      1,         1},
+	};
+
+	inline UtfIntroCharType IntroCharTypeOf(std::istream::int_type ch)
+	{
+		if (std::istream::traits_type::eof() == ch) {
+			return uictOther;
+		}
+
+		switch (ch) {
+		case 0: return uict00;
+		case 0xBB: return uictBB;
+		case 0xBF: return uictBF;
+		case 0xEF: return uictEF;
+		case 0xFE: return uictFE;
+		case 0xFF: return uictFF;
+		}
+
+		if ((ch > 0) && (ch < 0xFF)) {
+			return uictAscii;
+		}
+
+		return uictOther;
+	}
+
+	inline char Utf8Adjust(unsigned long ch, unsigned char lead_bits, unsigned char rshift)
+	{
+		const unsigned char header = ((1 << lead_bits) - 1) << (8 - lead_bits);
+		const unsigned char mask = (0xFF >> (lead_bits + 1));
+		return static_cast<char>(static_cast<unsigned char>(
+			header | ((ch >> rshift) & mask)
+			));
+	}
+
+	inline void QueueUnicodeCodepoint(std::deque<char>& q, unsigned long ch)
+	{
+		// We are not allowed to queue the Stream::eof() codepoint, so
+		// replace it with CP_REPLACEMENT_CHARACTER
+		if (static_cast<unsigned long>(Stream::eof()) == ch)
+		{
+			ch = CP_REPLACEMENT_CHARACTER;
+		}
+
+		if (ch < 0x80)
+		{
+			q.push_back(Utf8Adjust(ch, 0, 0));
+		}
+		else if (ch < 0x800)
+		{
+			q.push_back(Utf8Adjust(ch, 2, 6));
+			q.push_back(Utf8Adjust(ch, 1, 0));
+		}
+		else if (ch < 0x10000)
+		{
+			q.push_back(Utf8Adjust(ch, 3, 12));
+			q.push_back(Utf8Adjust(ch, 1, 6));
+			q.push_back(Utf8Adjust(ch, 1, 0));
+		}
+		else
+		{
+			q.push_back(Utf8Adjust(ch, 4, 18));
+			q.push_back(Utf8Adjust(ch, 1, 12));
+			q.push_back(Utf8Adjust(ch, 1, 6));
+			q.push_back(Utf8Adjust(ch, 1, 0));
+		}
+	}
+
+	Stream::Stream(std::istream& input)
+		: m_input(input), m_nPushedBack(0),
+		m_pPrefetched(new unsigned char[YAML_PREFETCH_SIZE]), 
+		m_nPrefetchedAvailable(0), m_nPrefetchedUsed(0)
+	{
+		typedef std::istream::traits_type char_traits;
+
+		if(!input)
+			return;
+
+		// Determine (or guess) the character-set by reading the BOM, if any.  See
+		// the YAML specification for the determination algorithm.
+		char_traits::int_type intro[4];
+		int nIntroUsed = 0;
+		UtfIntroState state = uis_start;
+		for (; !s_introFinalState[state]; ) {
+			std::istream::int_type ch = input.get();
+			intro[nIntroUsed++] = ch;
+			UtfIntroCharType charType = IntroCharTypeOf(ch);
+			UtfIntroState newState = s_introTransitions[state][charType];
+			int nUngets = s_introUngetCount[state][charType];
+			if (nUngets > 0) {
+				for (; nUngets > 0; --nUngets) {
+					if (char_traits::eof() != intro[--nIntroUsed]) {
+						m_bufPushback[m_nPushedBack++] = 
+							char_traits::to_char_type(intro[nIntroUsed]);
+					}
+				}
+			}
+			state = newState;
+		}
+
+		switch (state) {
+		case uis_utf8: m_charSet = utf8; break;
+		case uis_utf16le: m_charSet = utf16le; break;
+		case uis_utf16be: m_charSet = utf16be; break;
+		case uis_utf32le: m_charSet = utf32le; break;
+		case uis_utf32be: m_charSet = utf32be; break;
+		default: m_charSet = utf8; break;
+		}
+
+		ReadAheadTo(0);
+	}
+
+	Stream::~Stream()
+	{
+		delete[] m_pPrefetched;
+	}
+
+	char Stream::peek() const
+	{
+		if (m_readahead.empty())
+		{
+			return Stream::eof();
+		}
+
+		return m_readahead[0];
+	}
+	
+	Stream::operator bool() const
+	{
+		return m_input.good() || (!m_readahead.empty() && m_readahead[0] != Stream::eof());
+	}
+
+	// get
+	// . Extracts a character from the stream and updates our position
+	char Stream::get()
+	{
+		char ch = peek();
+		AdvanceCurrent();
+		m_mark.column++;
+		
+		if(ch == '\n') {
+			m_mark.column = 0;
+			m_mark.line++;
+		}
+		
+		return ch;
+	}
+
+	// get
+	// . Extracts 'n' characters from the stream and updates our position
+	std::string Stream::get(int n)
+	{
+		std::string ret;
+		ret.reserve(n);
+		for(int i=0;i<n;i++)
+			ret += get();
+		return ret;
+	}
+
+	// eat
+	// . Eats 'n' characters and updates our position.
+	void Stream::eat(int n)
+	{
+		for(int i=0;i<n;i++)
+			get();
+	}
+
+	void Stream::AdvanceCurrent()
+	{
+		if (!m_readahead.empty())
+		{
+			m_readahead.pop_front();
+			m_mark.pos++;
+		}
+
+		ReadAheadTo(0);
+	}
+
+	bool Stream::_ReadAheadTo(size_t i) const
+	{
+		while (m_input.good() && (m_readahead.size() <= i))
+		{
+			switch (m_charSet)
+			{
+			case utf8: StreamInUtf8(); break;
+			case utf16le: StreamInUtf16(); break;
+			case utf16be: StreamInUtf16(); break;
+			case utf32le: StreamInUtf32(); break;
+			case utf32be: StreamInUtf32(); break;
+			}
+		}
+		
+		// signal end of stream
+		if(!m_input.good())
+			m_readahead.push_back(Stream::eof());
+
+		return m_readahead.size() > i;
+	}
+
+	void Stream::StreamInUtf8() const
+	{
+		unsigned char b = GetNextByte();
+		if (m_input.good())
+		{
+			m_readahead.push_back(b);
+		}
+	}
+
+	void Stream::StreamInUtf16() const
+	{
+		unsigned long ch = 0;
+		unsigned char bytes[2];
+		int nBigEnd = (m_charSet == utf16be) ? 0 : 1;
+
+		bytes[0] = GetNextByte();
+		bytes[1] = GetNextByte();
+		if (!m_input.good())
+		{
+			return;
+		}
+		ch = (static_cast<unsigned long>(bytes[nBigEnd]) << 8) |
+			static_cast<unsigned long>(bytes[1 ^ nBigEnd]);
+
+		if (ch >= 0xDC00 && ch < 0xE000)
+		{
+			// Trailing (low) surrogate...ugh, wrong order
+			QueueUnicodeCodepoint(m_readahead, CP_REPLACEMENT_CHARACTER);
+			return;
+		}
+		else if (ch >= 0xD800 && ch < 0xDC00)
+		{
+			// ch is a leading (high) surrogate
+
+			// Four byte UTF-8 code point
+
+			// Read the trailing (low) surrogate
+			for (;;)
+			{
+				bytes[0] = GetNextByte();
+				bytes[1] = GetNextByte();
+				if (!m_input.good())
+				{
+					QueueUnicodeCodepoint(m_readahead, CP_REPLACEMENT_CHARACTER);
+					return;
+				}
+				unsigned long chLow = (static_cast<unsigned long>(bytes[nBigEnd]) << 8) |
+					static_cast<unsigned long>(bytes[1 ^ nBigEnd]);
+				if (chLow < 0xDC00 || ch >= 0xE000)
+				{
+					// Trouble...not a low surrogate.  Dump a REPLACEMENT CHARACTER into the stream.
+					QueueUnicodeCodepoint(m_readahead, CP_REPLACEMENT_CHARACTER);
+
+					// Deal with the next UTF-16 unit
+					if (chLow < 0xD800 || ch >= 0xE000)
+					{
+						// Easiest case: queue the codepoint and return
+						QueueUnicodeCodepoint(m_readahead, ch);
+						return;
+					}
+					else
+					{
+						// Start the loop over with the new high surrogate
+						ch = chLow;
+						continue;
+					}
+				}
+
+				// Select the payload bits from the high surrogate
+				ch &= 0x3FF;
+				ch <<= 10;
+
+				// Include bits from low surrogate
+				ch |= (chLow & 0x3FF);
+
+				// Add the surrogacy offset
+				ch += 0x10000;
+			}
+		}
+
+		QueueUnicodeCodepoint(m_readahead, ch);
+	}
+
+	inline char* ReadBuffer(unsigned char* pBuffer)
+	{
+		return reinterpret_cast<char*>(pBuffer);
+	}
+
+	unsigned char Stream::GetNextByte() const
+	{
+		if (m_nPushedBack)
+		{
+			return m_bufPushback[--m_nPushedBack];
+		}
+
+		if (m_nPrefetchedUsed >= m_nPrefetchedAvailable)
+		{
+			std::streambuf *pBuf = m_input.rdbuf();
+			m_nPrefetchedAvailable = pBuf->sgetn(ReadBuffer(m_pPrefetched), 
+				YAML_PREFETCH_SIZE);
+			m_nPrefetchedUsed = 0;
+			if (!m_nPrefetchedAvailable)
+			{
+				m_input.setstate(std::ios_base::eofbit);
+			}
+
+			if (0 == m_nPrefetchedAvailable)
+			{
+				return 0;
+			}
+		}
+
+		return m_pPrefetched[m_nPrefetchedUsed++];
+	}
+
+	void Stream::StreamInUtf32() const
+	{
+		static int indexes[2][4] = {
+			{3, 2, 1, 0},
+			{0, 1, 2, 3}
+		};
+
+		unsigned long ch = 0;
+		unsigned char bytes[4];
+		int* pIndexes = (m_charSet == utf32be) ? indexes[1] : indexes[0];
+
+		bytes[0] = GetNextByte();
+		bytes[1] = GetNextByte();
+		bytes[2] = GetNextByte();
+		bytes[3] = GetNextByte();
+		if (!m_input.good())
+		{
+			return;
+		}
+
+		for (int i = 0; i < 4; ++i)
+		{
+			ch <<= 8;
+			ch |= bytes[pIndexes[i]];
+		}
+
+		QueueUnicodeCodepoint(m_readahead, ch);
+	}
+}

src/stream.h

@@ -0,0 +1,83 @@
+#ifndef STREAM_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define STREAM_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "yaml-cpp/noncopyable.h"
+#include "yaml-cpp/mark.h"
+#include <cstddef>
+#include <deque>
+#include <ios>
+#include <iostream>
+#include <set>
+#include <string>
+
+namespace YAML
+{
+	static const size_t MAX_PARSER_PUSHBACK = 8;
+
+	class Stream: private noncopyable
+	{
+	public:
+		friend class StreamCharSource;
+		
+		Stream(std::istream& input);
+		~Stream();
+
+		operator bool() const;
+		bool operator !() const { return !static_cast <bool>(*this); }
+
+		char peek() const;
+		char get();
+		std::string get(int n);
+		void eat(int n = 1);
+
+		static char eof() { return 0x04; }
+		
+		const Mark mark() const { return m_mark; }
+		int pos() const { return m_mark.pos; }
+		int line() const { return m_mark.line; }
+		int column() const { return m_mark.column; }
+		void ResetColumn() { m_mark.column = 0; }
+
+	private:
+		enum CharacterSet {utf8, utf16le, utf16be, utf32le, utf32be};
+
+		std::istream& m_input;
+		Mark m_mark;
+		
+		CharacterSet m_charSet;
+		unsigned char m_bufPushback[MAX_PARSER_PUSHBACK];
+		mutable size_t m_nPushedBack;
+		mutable std::deque<char> m_readahead;
+		unsigned char* const m_pPrefetched;
+		mutable size_t m_nPrefetchedAvailable;
+		mutable size_t m_nPrefetchedUsed;
+		
+		void AdvanceCurrent();
+		char CharAt(size_t i) const;
+		bool ReadAheadTo(size_t i) const;
+		bool _ReadAheadTo(size_t i) const;
+		void StreamInUtf8() const;
+		void StreamInUtf16() const;
+		void StreamInUtf32() const;
+		unsigned char GetNextByte() const;
+	};
+
+	// CharAt
+	// . Unchecked access
+	inline char Stream::CharAt(size_t i) const {
+		return m_readahead[i];
+	}
+	
+	inline bool Stream::ReadAheadTo(size_t i) const {
+		if(m_readahead.size() > i)
+			return true;
+		return _ReadAheadTo(i);
+	}	
+}
+
+#endif // STREAM_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/streamcharsource.h

@@ -0,0 +1,48 @@
+#ifndef STREAMCHARSOURCE_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define STREAMCHARSOURCE_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "yaml-cpp/noncopyable.h"
+#include <cstddef>
+
+namespace YAML
+{
+	class StreamCharSource
+	{
+	public:
+		StreamCharSource(const Stream& stream): m_offset(0), m_stream(stream) {}
+		StreamCharSource(const StreamCharSource& source): m_offset(source.m_offset), m_stream(source.m_stream) {}
+		~StreamCharSource() {}
+
+		operator bool() const;
+		char operator [] (std::size_t i) const { return m_stream.CharAt(m_offset + i); }
+		bool operator !() const { return !static_cast<bool>(*this); }
+
+		const StreamCharSource operator + (int i) const;
+			
+	private:
+		std::size_t m_offset;
+		const Stream& m_stream;
+
+		StreamCharSource& operator = (const StreamCharSource&); // non-assignable
+	};
+	
+	inline StreamCharSource::operator bool() const {
+		return m_stream.ReadAheadTo(m_offset);
+	}
+	
+	inline const StreamCharSource StreamCharSource::operator + (int i) const {
+		StreamCharSource source(*this);
+		if(static_cast<int> (source.m_offset) + i >= 0)
+			source.m_offset += i;
+		else
+			source.m_offset = 0;
+		return source;
+	}
+}
+
+#endif // STREAMCHARSOURCE_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/stringsource.h

@@ -0,0 +1,47 @@
+#ifndef STRINGSOURCE_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define STRINGSOURCE_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include <cstddef>
+
+namespace YAML
+{
+	class StringCharSource
+	{
+	public:
+		StringCharSource(const char *str, std::size_t size): m_str(str), m_size(size), m_offset(0) {}
+
+		operator bool() const { return m_offset < m_size; }
+		char operator [] (std::size_t i) const { return m_str[m_offset + i]; }
+		bool operator !() const { return !static_cast<bool>(*this); }
+
+		const StringCharSource operator + (int i) const {
+			StringCharSource source(*this);
+			if(static_cast<int> (source.m_offset) + i >= 0)
+				source.m_offset += i;
+			else
+				source.m_offset = 0;
+			return source;
+		}
+			
+		StringCharSource& operator ++ () {
+			++m_offset;
+			return *this;
+		}
+		
+		StringCharSource& operator += (std::size_t offset) {
+			m_offset += offset;
+			return *this;
+		}
+	private:
+		const char *m_str;
+		std::size_t m_size;
+		std::size_t m_offset;
+	};
+}
+
+#endif // STRINGSOURCE_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/tag.cpp

@@ -0,0 +1,52 @@
+#include "tag.h"
+#include "directives.h"
+#include "token.h"
+#include <cassert>
+#include <stdexcept>
+
+namespace YAML
+{
+	Tag::Tag(const Token& token): type(static_cast<TYPE>(token.data))
+	{
+		switch(type) {
+			case VERBATIM:
+				value = token.value;
+				break;
+			case PRIMARY_HANDLE:
+				value = token.value;
+				break;
+			case SECONDARY_HANDLE:
+				value = token.value;
+				break;
+			case NAMED_HANDLE:
+				handle = token.value;
+				value = token.params[0];
+				break;
+			case NON_SPECIFIC:
+				break;
+			default:
+				assert(false);
+		}
+	}
+
+	const std::string Tag::Translate(const Directives& directives)
+	{
+		switch(type) {
+			case VERBATIM:
+				return value;
+			case PRIMARY_HANDLE:
+				return directives.TranslateTagHandle("!") + value;
+			case SECONDARY_HANDLE:
+				return directives.TranslateTagHandle("!!") + value;
+			case NAMED_HANDLE:
+				return directives.TranslateTagHandle("!" + handle + "!") + value;
+			case NON_SPECIFIC:
+				// TODO:
+				return "!";
+			default:
+				assert(false);
+		}
+		throw std::runtime_error("yaml-cpp: internal error, bad tag type");
+	}
+}
+

src/tag.h

@@ -0,0 +1,28 @@
+#ifndef TAG_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define TAG_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+#include <string>
+
+namespace YAML
+{
+	struct Token;
+	struct Directives;
+
+	struct Tag {
+		enum TYPE {
+			VERBATIM, PRIMARY_HANDLE, SECONDARY_HANDLE, NAMED_HANDLE, NON_SPECIFIC
+		};
+		
+		Tag(const Token& token);
+		const std::string Translate(const Directives& directives);
+		
+		TYPE type;
+		std::string handle, value;
+	};
+}
+
+#endif // TAG_H_62B23520_7C8E_11DE_8A39_0800200C9A66

src/token.h

@@ -0,0 +1,85 @@
+#ifndef TOKEN_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+#define TOKEN_H_62B23520_7C8E_11DE_8A39_0800200C9A66
+
+#if !defined(__GNUC__) || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4) || (__GNUC__ >= 4) // GCC supports "pragma once" correctly since 3.4
+#pragma once
+#endif
+
+
+#include "yaml-cpp/mark.h"
+#include <iostream>
+#include <string>
+#include <vector>
+
+namespace YAML
+{
+	const std::string TokenNames[] = {
+		"DIRECTIVE",
+		"DOC_START",
+		"DOC_END",
+		"BLOCK_SEQ_START",
+		"BLOCK_MAP_START",
+		"BLOCK_SEQ_END",
+		"BLOCK_MAP_END",
+		"BLOCK_ENTRY",
+		"FLOW_SEQ_START",
+		"FLOW_MAP_START",
+		"FLOW_SEQ_END",
+		"FLOW_MAP_END",
+		"FLOW_MAP_COMPACT",
+		"FLOW_ENTRY",
+		"KEY",
+		"VALUE",
+		"ANCHOR",
+		"ALIAS",
+		"TAG",
+		"SCALAR"
+	};
+
+	struct Token {
+		// enums
+		enum STATUS { VALID, INVALID, UNVERIFIED };
+		enum TYPE {
+			DIRECTIVE,
+			DOC_START,
+			DOC_END,
+			BLOCK_SEQ_START,
+			BLOCK_MAP_START,
+			BLOCK_SEQ_END,
+			BLOCK_MAP_END,
+			BLOCK_ENTRY,
+			FLOW_SEQ_START,
+			FLOW_MAP_START,
+			FLOW_SEQ_END,
+			FLOW_MAP_END,
+			FLOW_MAP_COMPACT,
+			FLOW_ENTRY,
+			KEY,
+			VALUE,
+			ANCHOR,
+			ALIAS,
+			TAG,
+			PLAIN_SCALAR,
+			NON_PLAIN_SCALAR
+		};
+
+		// data
+		Token(TYPE type_, const Mark& mark_): status(VALID), type(type_), mark(mark_), data(0) {}
+
+		friend std::ostream& operator << (std::ostream& out, const Token& token) {
+			out << TokenNames[token.type] << std::string(": ") << token.value;
+			for(std::size_t i=0;i<token.params.size();i++)
+				out << std::string(" ") << token.params[i];
+			return out;
+		}
+
+		STATUS status;
+		TYPE type;
+		Mark mark;
+		std::string value;
+		std::vector <std::string> params;
+		int data;
+	};
+}
+
+#endif // TOKEN_H_62B23520_7C8E_11DE_8A39_0800200C9A66