Merged aliases branch into trunk, changes r100:150

include/exceptions.h

@@ -38,6 +38,8 @@ namespace YAML
 		const std::string CHAR_IN_ANCHOR       = "illegal character found while scanning anchor";
 		const std::string ZERO_INDENT_IN_BLOCK = "cannot set zero indentation for a block scalar";
 		const std::string CHAR_IN_BLOCK        = "unexpected character in block scalar";
+		const std::string AMBIGUOUS_ANCHOR     = "cannot assign the same alias to multiple nodes";
+		const std::string UNKNOWN_ANCHOR       = "the referenced anchor is not defined";
 
 		const std::string INVALID_SCALAR       = "invalid scalar";
 		const std::string KEY_NOT_FOUND        = "key not found";

include/node.h

@@ -66,6 +66,11 @@ namespace YAML
 		const Node& operator [] (unsigned u) const;
 		const Node& operator [] (int i) const;
 
+		// for anchors/aliases
+		const Node *Identity() const { return m_pIdentity; }
+		bool IsAlias() const { return m_alias; }
+		bool IsReferenced() const { return m_referenced; }
+
 		// insertion
 		friend std::ostream& operator << (std::ostream& out, const Node& node);
 
@@ -89,6 +94,8 @@ namespace YAML
 		std::string m_anchor, m_tag;
 		Content *m_pContent;
 		bool m_alias;
+		const Node *m_pIdentity;
+		mutable bool m_referenced;
 	};
 
 	// templated things we need to keep inline in the header

src/CMakeLists.txt

-set(FILES content.cpp iterator.cpp node.cpp parserstate.cpp
+set(FILES alias.cpp content.cpp iterator.cpp node.cpp parserstate.cpp
           scalar.cpp scanscalar.cpp sequence.cpp stream.cpp
           exp.cpp map.cpp parser.cpp regex.cpp scanner.cpp
           scantoken.cpp simplekey.cpp)

src/alias.cpp

+#include "crt.h"
+#include "alias.h"
+#include <iostream>
+
+namespace YAML
+{
+	Alias::Alias(Content* pNodeContent)
+		: m_pRef(pNodeContent)
+	{
+	}
+
+	void Alias::Parse(Scanner *pScanner, const ParserState& state)
+	{
+	}
+
+	void Alias::Write(std::ostream& out, int indent, bool startedLine, bool onlyOneCharOnLine)
+	{
+		out << "\n";
+	}
+
+	bool Alias::GetBegin(std::vector <Node *>::const_iterator& i) const
+	{
+		return m_pRef->GetBegin(i);
+	}
+
+	bool Alias::GetBegin(std::map <Node *, Node *, ltnode>::const_iterator& i) const
+	{
+		return m_pRef->GetBegin(i);
+	}
+
+	bool Alias::GetEnd(std::vector <Node *>::const_iterator& i) const
+	{
+		return m_pRef->GetEnd(i);
+	}
+
+	bool Alias::GetEnd(std::map <Node *, Node *, ltnode>::const_iterator& i) const
+	{
+		return m_pRef->GetEnd(i);
+	}
+
+	Node* Alias::GetNode(unsigned n) const
+	{
+		return m_pRef->GetNode(n);
+	}
+
+	unsigned Alias::GetSize() const
+	{
+		return m_pRef->GetSize();
+	}
+
+	bool Alias::IsScalar() const
+	{
+		return m_pRef->IsScalar();
+	}
+
+	bool Alias::IsMap() const
+	{
+		return m_pRef->IsMap();
+	}
+
+	bool Alias::IsSequence() const
+	{
+		return m_pRef->IsSequence();
+	}
+
+	bool Alias::Read(std::string& v) const
+	{
+		return m_pRef->Read(v);
+	}
+
+	bool Alias::Read(int& v) const
+	{
+		return m_pRef->Read(v);
+	}
+
+	bool Alias::Read(unsigned& v) const
+	{
+		return m_pRef->Read(v);
+	}
+
+	bool Alias::Read(long& v) const
+	{
+		return m_pRef->Read(v);
+	}
+
+	bool Alias::Read(float& v) const
+	{
+		return m_pRef->Read(v);
+	}
+
+	bool Alias::Read(double& v) const
+	{
+		return m_pRef->Read(v);
+	}
+
+	bool Alias::Read(char& v) const
+	{
+		return m_pRef->Read(v);
+	}
+
+	bool Alias::Read(bool& v) const
+	{
+		return m_pRef->Read(v);
+	}
+
+	int Alias::Compare(Content *pContent)
+	{
+		return m_pRef->Compare(pContent);
+	}
+
+	int Alias::Compare(Scalar *pScalar)
+	{
+		return m_pRef->Compare(pScalar);
+	}
+
+	int Alias::Compare(Sequence *pSequence)
+	{
+		return m_pRef->Compare(pSequence);
+	}
+
+	int Alias::Compare(Map *pMap)
+	{
+		return m_pRef->Compare(pMap);
+	}
+}

src/alias.h

+#pragma once
+
+#include "content.h"
+
+namespace YAML
+{
+	class Alias : public Content
+	{
+	public:
+		Alias(Content *pNodeContent);
+
+		virtual void Parse(Scanner* pScanner, const ParserState& state);
+		virtual void Write(std::ostream& out, int indent, bool startedLine, bool onlyOneCharOnLine);
+
+		virtual bool GetBegin(std::vector <Node *>::const_iterator&) const;
+		virtual bool GetBegin(std::map <Node *, Node *, ltnode>::const_iterator&) const;
+		virtual bool GetEnd(std::vector <Node *>::const_iterator&) const;
+		virtual bool GetEnd(std::map <Node *, Node *, ltnode>::const_iterator&) const;
+		virtual Node* GetNode(unsigned) const;
+		virtual unsigned GetSize() const;
+		virtual bool IsScalar() const;
+		virtual bool IsMap() const;
+		virtual bool IsSequence() const;
+
+		virtual bool Read(std::string&) const;
+		virtual bool Read(int&) const;
+		virtual bool Read(unsigned&) const;
+		virtual bool Read(long&) const;
+		virtual bool Read(float&) const;
+		virtual bool Read(double&) const;
+		virtual bool Read(char&) const;
+		virtual bool Read(bool&) const;
+
+		virtual int Compare(Content *);
+		virtual int Compare(Scalar *);
+		virtual int Compare(Sequence *);
+		virtual int Compare(Map *);
+
+	private:
+		Content* m_pRef;
+	};
+}

src/node.cpp

@@ -7,7 +7,9 @@
 #include "scalar.h"
 #include "sequence.h"
 #include "map.h"
+#include "alias.h"
 #include "iterpriv.h"
+#include <iostream>
 
 namespace YAML
 {
@@ -17,7 +19,7 @@ namespace YAML
 		return *pNode1 < *pNode2;
 	}
 
-	Node::Node(): m_pContent(0), m_alias(false)
+	Node::Node(): m_pContent(0), m_alias(false), m_pIdentity(this), m_referenced(true)
 	{
 	}
 
@@ -31,6 +33,9 @@ namespace YAML
 		delete m_pContent;
 		m_pContent = 0;
 		m_alias = false;
+		m_referenced = false;
+		m_anchor.clear();
+		m_tag.clear();
 	}
 
 	void Node::Parse(Scanner *pScanner, const ParserState& state)
@@ -47,30 +52,50 @@ namespace YAML
 
 		ParseHeader(pScanner, state);
 
-		// is this an alias? if so, it can have no content
-		if(m_alias)
+		// is this an alias? if so, its contents are an alias to
+		// a previously defined anchor
+		if(m_alias) {
+			// the scanner throws an exception if it doesn't know this anchor name
+			const Node *pReferencedNode = pScanner->Retrieve(m_anchor);
+			m_pIdentity = pReferencedNode;
+
+			// mark the referenced node for the sake of the client code
+			pReferencedNode->m_referenced = true;
+
+			// use of an Alias object keeps the referenced content from
+			// being deleted twice
+			Content *pAliasedContent = pReferencedNode->m_pContent;
+			if(pAliasedContent)
+				m_pContent = new Alias(pAliasedContent);
+			
 			return;
+		}
 
 		// now split based on what kind of node we should be
 		switch(pScanner->peek().type) {
 			case TT_SCALAR:
 				m_pContent = new Scalar;
-				m_pContent->Parse(pScanner, state);
 				break;
 			case TT_FLOW_SEQ_START:
 			case TT_BLOCK_SEQ_START:
 			case TT_BLOCK_ENTRY:
 				m_pContent = new Sequence;
-				m_pContent->Parse(pScanner, state);
 				break;
 			case TT_FLOW_MAP_START:
 			case TT_BLOCK_MAP_START:
 				m_pContent = new Map;
-				m_pContent->Parse(pScanner, state);
 				break;
 			default:
 				break;
 		}
+
+		// Have to save anchor before parsing to allow for aliases as
+		// contained node (recursive structure)
+		if(!m_anchor.empty())
+			pScanner->Save(m_anchor, this);
+
+		if(m_pContent)
+			m_pContent->Parse(pScanner, state);
 	}
 
 	// ParseHeader
@@ -129,33 +154,46 @@ namespace YAML
 
 	void Node::Write(std::ostream& out, int indent, bool startedLine, bool onlyOneCharOnLine) const
 	{
+		// If using an anchor or tag for the whole document, document start
+		// must be explicit
+		bool indicateDocStart = (indent == 0);
+
 		// write anchor/alias
 		if(m_anchor != "") {
+			if (indicateDocStart) {
+				out << "--- ";
+				indicateDocStart = false;
+			}
+
 			if(m_alias)
-				out << std::string("*");
+				out << "*";
 			else
-				out << std::string("&");
-			out << m_anchor << std::string(" ");
+				out << "&";
+			out << m_anchor << " ";
 			startedLine = true;
 			onlyOneCharOnLine = false;
 		}
 
 		// write tag
 		if(m_tag != "") {
+			if (indicateDocStart) {
+				out << "--- ";
+				indicateDocStart = false;
+			}
+
 			// put the tag in the "proper" brackets
-			if(m_tag.substr(0, 2) == "!<" && m_tag.substr(m_tag.size() - 1) == ">")
-				out << m_tag;
+			if(m_tag.substr(0, 2) == std::string("!<") && m_tag.substr(m_tag.size() - 1) == std::string(">"))
+				out << m_tag << " ";
 			else
-				out << std::string("!<") << m_tag << std::string("> ");
+				out << "!<" << m_tag << "> ";
 			startedLine = true;
 			onlyOneCharOnLine = false;
 		}
 
-		if(!m_pContent) {
-			out << std::string("\n");
-		} else {
+		if(!m_pContent)
+			out << "\n";
+		else
 			m_pContent->Write(out, indent, startedLine, onlyOneCharOnLine);
-		}
 	}
 
 	CONTENT_TYPE Node::GetType() const

src/parser.cpp

@@ -55,6 +55,9 @@ namespace YAML
 		// and finally eat any doc ends we see
 		while(!m_pScanner->empty() && m_pScanner->peek().type == TT_DOC_END)
 			m_pScanner->pop();
+
+		// clear anchors from the scanner, which are no longer relevant
+		m_pScanner->ClearAnchors();
 	}
 
 	// ParseDirectives

src/scanner.cpp

@@ -29,8 +29,13 @@ namespace YAML
 	void Scanner::pop()
 	{
 		EnsureTokensInQueue();
-		if(!m_tokens.empty())
+		if(!m_tokens.empty()) {
+			// Saved anchors shouldn't survive popping the document end marker
+			if (m_tokens.front().type == TT_DOC_END) {
+				ClearAnchors();
+			}
 			m_tokens.pop();
+		}
 	}
 
 	// peek
@@ -217,6 +222,7 @@ namespace YAML
 		m_startedStream = true;
 		m_simpleKeyAllowed = true;
 		m_indents.push(-1);
+		m_anchors.clear();
 	}
 
 	// EndStream
@@ -273,4 +279,47 @@ namespace YAML
 			m_tokens.push(Token(TT_BLOCK_END, INPUT.line, INPUT.column));
 		}
 	}
+
+	// Save
+	// . Saves a pointer to the Node object referenced by a particular anchor
+	//   name.
+	void Scanner::Save(const std::string& anchor, Node* value)
+	{
+		m_anchors[anchor] = value;
+	}
+
+	// Retrieve
+	// . Retrieves a pointer previously saved for an anchor name.
+	// . Throws an exception if the anchor has not been defined.
+	const Node *Scanner::Retrieve(const std::string& anchor) const
+	{
+		typedef std::map<std::string, const Node *> map;
+
+		map::const_iterator itNode = m_anchors.find(anchor);
+
+		if(m_anchors.end() == itNode)
+			ThrowParserException(ErrorMsg::UNKNOWN_ANCHOR);
+
+		return itNode->second;
+	}
+
+	// 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
+	{
+		int line = -1, column = -1;
+		if(!m_tokens.empty()) {
+			const Token& token = m_tokens.front();
+			line = token.line;
+			column = token.column;
+		}
+		throw ParserException(line, column, msg);
+	}
+
+	void Scanner::ClearAnchors()
+	{
+		m_anchors.clear();
+	}
 }

src/scanner.h

@@ -5,11 +5,14 @@
 #include <queue>
 #include <stack>
 #include <set>
+#include <map>
 #include "stream.h"
 #include "token.h"
 
 namespace YAML
 {
+	class Node;
+
 	class Scanner
 	{
 	public:
@@ -21,6 +24,11 @@ namespace YAML
 		void pop();
 		Token& peek();
 
+		// anchor management
+		void Save(const std::string& anchor, Node* value);
+		const Node *Retrieve(const std::string& anchor) const;
+		void ClearAnchors();
+
 	private:
 		// scanning
 		void EnsureTokensInQueue();
@@ -35,6 +43,7 @@ namespace YAML
 		void InsertSimpleKey();
 		bool VerifySimpleKey(bool force = false);
 		void VerifyAllSimpleKeys();
+		void ThrowParserException(const std::string& msg) const;
 
 		bool IsWhitespaceToBeEaten(char ch);
 
@@ -81,5 +90,6 @@ namespace YAML
 		bool m_isLastKeyValid;
 		std::stack <SimpleKey> m_simpleKeys;
 		std::stack <int> m_indents;
+		std::map <std::string, const Node *> m_anchors;
 	};
 }

src/stream.cpp

@@ -55,6 +55,7 @@ namespace YAML
 	std::string Stream::get(int n)
 	{
 		std::string ret;
+		ret.reserve(n);
 		for(int i=0;i<n;i++)
 			ret += get();
 		return ret;

yaml-reader/tests.cpp

@@ -16,6 +16,7 @@ namespace Test
 		files.push_back("tests/mixed.yaml");
 		files.push_back("tests/scalars.yaml");
 		files.push_back("tests/directives.yaml");
+		files.push_back("tests/aliased.yaml");
 
 		bool passed = true;
 		for(unsigned i=0;i<files.size();i++) {

yaml-reader/tests/aliased.yaml

+--- &list
+- This document contains a recursive list.
+- *list
+...

yamlcpp.vcproj

@@ -167,6 +167,10 @@
 			<Filter
 				Name="Representation"
 				>
+			        <File
+				        RelativePath=".\src\alias.cpp"
+				        >
+				</File>
 				<File
 					RelativePath=".\src\content.cpp"
 					>
@@ -257,6 +261,10 @@
 			<Filter
 				Name="Representation"
 				>
+			        <File
+				        RelativePath=".\include\alias.h"
+					>
+				</File>
 				<File
 					RelativePath=".\src\content.h"
 					>