www.sourceforge.net/projects/tinyxml
Original code (2.0 and earlier )copyright (c) 2000-2006 Lee Thomason (www.grinninglizard.com)
This software is provided 'as-is', without any express or implied
warranty. In no event will the authors be held liable for any
damages arising from the use of this software.
Permission is granted to anyone to use this software for any
purpose, including commercial applications, and to alter it and
redistribute it freely, subject to the following restrictions:
1. The origin of this software must not be misrepresented; you must
not claim that you wrote the original software. If you use this
software in a product, an acknowledgment in the product documentation
would be appreciated but is not required.
2. Altered source versions must be plainly marked as such, and
must not be misrepresented as being the original software.
3. This notice may not be removed or altered from any source
distribution.
*/
#ifndef TINYXML_INCLUDED
#define TINYXML_INCLUDED
#ifdef _MSC_VER
#pragma warning( push )
#pragma warning( disable : 4530 )
#pragma warning( disable : 4786 )
#endif
#include <ctype.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <assert.h>
#if defined( _DEBUG ) && !defined( DEBUG )
#define DEBUG
#endif
#ifdef TIXML_USE_STL
#include <string>
#include <iostream>
#include <sstream>
#define TIXML_STRING std::string
#else
#include "tinystr.h"
#define TIXML_STRING TiXmlString
#endif
#define TIXML_SAFE
#ifdef TIXML_SAFE
#if defined(_MSC_VER) && (_MSC_VER >= 1400 )
#define TIXML_SNPRINTF _snprintf_s
#define TIXML_SSCANF sscanf_s
#elif defined(_MSC_VER) && (_MSC_VER >= 1200 )
#define TIXML_SNPRINTF _snprintf
#define TIXML_SSCANF sscanf
#elif defined(__GNUC__) && (__GNUC__ >= 3 )
#define TIXML_SNPRINTF snprintf
#define TIXML_SSCANF sscanf
#else
#define TIXML_SNPRINTF snprintf
#define TIXML_SSCANF sscanf
#endif
#endif
class TiXmlDocument;
class TiXmlElement;
class TiXmlComment;
class TiXmlUnknown;
class TiXmlAttribute;
class TiXmlText;
class TiXmlDeclaration;
class TiXmlParsingData;
const int TIXML_MAJOR_VERSION = 2;
const int TIXML_MINOR_VERSION = 6;
const int TIXML_PATCH_VERSION = 1;
in the XML file.
*/
struct TiXmlCursor
{
TiXmlCursor() { Clear(); }
void Clear() { row = col = -1; }
int row;
int col;
};
Implements the interface to the "Visitor pattern" (see the Accept() method.)
If you call the Accept() method, it requires being passed a TiXmlVisitor
class to handle callbacks. For nodes that contain other nodes (Document, Element)
you will get called with a VisitEnter/VisitExit pair. Nodes that are always leaves
are simply called with Visit().
If you return 'true' from a Visit method, recursive parsing will continue. If you return
false, <b>no children of this node or its sibilings</b> will be Visited.
All flavors of Visit methods have a default implementation that returns 'true' (continue
visiting). You need to only override methods that are interesting to you.
Generally Accept() is called on the TiXmlDocument, although all nodes suppert Visiting.
You should never change the document from a callback.
@sa TiXmlNode::Accept()
*/
class TiXmlVisitor
{
public:
virtual ~TiXmlVisitor() {}
virtual bool VisitEnter( const TiXmlDocument& ) { return true; }
virtual bool VisitExit( const TiXmlDocument& ) { return true; }
virtual bool VisitEnter( const TiXmlElement& , const TiXmlAttribute* ) { return true; }
virtual bool VisitExit( const TiXmlElement& ) { return true; }
virtual bool Visit( const TiXmlDeclaration& ) { return true; }
virtual bool Visit( const TiXmlText& ) { return true; }
virtual bool Visit( const TiXmlComment& ) { return true; }
virtual bool Visit( const TiXmlUnknown& ) { return true; }
};
enum
{
TIXML_SUCCESS,
TIXML_NO_ATTRIBUTE,
TIXML_WRONG_TYPE
};
enum TiXmlEncoding
{
TIXML_ENCODING_UNKNOWN,
TIXML_ENCODING_UTF8,
TIXML_ENCODING_LEGACY
};
const TiXmlEncoding TIXML_DEFAULT_ENCODING = TIXML_ENCODING_UNKNOWN;
It does little except to establish that TinyXml classes
can be printed and provide some utility functions.
In XML, the document and elements can contain
other elements and other types of nodes.
@verbatim
A Document can contain: Element (container or leaf)
Comment (leaf)
Unknown (leaf)
Declaration( leaf )
An Element can contain: Element (container or leaf)
Text (leaf)
Attributes (not on tree)
Comment (leaf)
Unknown (leaf)
A Decleration contains: Attributes (not on tree)
@endverbatim
*/
class TiXmlBase
{
friend class TiXmlNode;
friend class TiXmlElement;
friend class TiXmlDocument;
public:
TiXmlBase() : userData(0) {}
virtual ~TiXmlBase() {}
or the string class (TiXmlString in non-STL mode, std::string
in STL mode.) Either or both cfile and str can be null.
This is a formatted print, and will insert
tabs and newlines.
(For an unformatted stream, use the << operator.)
*/
virtual void Print( FILE* cfile, int depth ) const = 0;
not. In order to make everyone happy, these global, static functions
are provided to set whether or not TinyXml will condense all white space
into a single space or not. The default is to condense. Note changing this
value is not thread safe.
*/
static void SetCondenseWhiteSpace( bool condense ) { condenseWhiteSpace = condense; }
static bool IsWhiteSpaceCondensed() { return condenseWhiteSpace; }
The row and column are 1-based. (That is the first row and first column is
1,1). If the returns values are 0 or less, then the parser does not have
a row and column value.
Generally, the row and column value will be set when the TiXmlDocument::Load(),
TiXmlDocument::LoadFile(), or any TiXmlNode::Parse() is called. It will NOT be set
when the DOM was created from operator>>.
The values reflect the initial load. Once the DOM is modified programmatically
(by adding or changing nodes and attributes) the new values will NOT update to
reflect changes in the document.
There is a minor performance cost to computing the row and column. Computation
can be disabled if TiXmlDocument::SetTabSize() is called with 0 as the value.
@sa TiXmlDocument::SetTabSize()
*/
int Row() const { return location.row + 1; }
int Column() const { return location.col + 1; }
void SetUserData( void* user ) { userData = user; }
void* GetUserData() { return userData; }
const void* GetUserData() const { return userData; }
static const int utf8ByteTable[256];
virtual const char* Parse( const char* p,
TiXmlParsingData* data,
TiXmlEncoding encoding ) = 0;
or they will be transformed into entities!
*/
static void EncodeString( const TIXML_STRING& str, TIXML_STRING* out );
enum
{
TIXML_NO_ERROR = 0,
TIXML_ERROR,
TIXML_ERROR_OPENING_FILE,
TIXML_ERROR_PARSING_ELEMENT,
TIXML_ERROR_FAILED_TO_READ_ELEMENT_NAME,
TIXML_ERROR_READING_ELEMENT_VALUE,
TIXML_ERROR_READING_ATTRIBUTES,
TIXML_ERROR_PARSING_EMPTY,
TIXML_ERROR_READING_END_TAG,
TIXML_ERROR_PARSING_UNKNOWN,
TIXML_ERROR_PARSING_COMMENT,
TIXML_ERROR_PARSING_DECLARATION,
TIXML_ERROR_DOCUMENT_EMPTY,
TIXML_ERROR_EMBEDDED_NULL,
TIXML_ERROR_PARSING_CDATA,
TIXML_ERROR_DOCUMENT_TOP_ONLY,
TIXML_ERROR_STRING_COUNT
};
protected:
static const char* SkipWhiteSpace( const char*, TiXmlEncoding encoding );
inline static bool IsWhiteSpace( char c )
{
return ( isspace( (unsigned char) c ) || c == '\n' || c == '\r' );
}
inline static bool IsWhiteSpace( int c )
{
if ( c < 256 )
return IsWhiteSpace( (char) c );
return false;
}
#ifdef TIXML_USE_STL
static bool StreamWhiteSpace( std::istream * in, TIXML_STRING * tag );
static bool StreamTo( std::istream * in, int character, TIXML_STRING * tag );
#endif
a pointer just past the last character of the name,
or 0 if the function has an error.
*/
static const char* ReadName( const char* p, TIXML_STRING* name, TiXmlEncoding encoding );
Wickedly complex options, but it keeps the (sensitive) code in one place.
*/
static const char* ReadText( const char* in,
TIXML_STRING* text,
bool ignoreWhiteSpace,
const char* endTag,
bool ignoreCase,
TiXmlEncoding encoding );
static const char* GetEntity( const char* in, char* value, int* length, TiXmlEncoding encoding );
inline static const char* GetChar( const char* p, char* _value, int* length, TiXmlEncoding encoding )
{
assert( p );
if ( encoding == TIXML_ENCODING_UTF8 )
{
*length = utf8ByteTable[ *((const unsigned char*)p) ];
assert( *length >= 0 && *length < 5 );
}
else
{
*length = 1;
}
if ( *length == 1 )
{
if ( *p == '&' )
return GetEntity( p, _value, length, encoding );
*_value = *p;
return p+1;
}
else if ( *length )
{
for( int i=0; p[i] && i<*length; ++i ) {
_value[i] = p[i];
}
return p + (*length);
}
else
{
return 0;
}
}
static bool StringEqual( const char* p,
const char* endTag,
bool ignoreCase,
TiXmlEncoding encoding );
static const char* errorString[ TIXML_ERROR_STRING_COUNT ];
TiXmlCursor location;
void* userData;
static int IsAlpha( unsigned char anyByte, TiXmlEncoding encoding );
static int IsAlphaNum( unsigned char anyByte, TiXmlEncoding encoding );
inline static int ToLower( int v, TiXmlEncoding encoding )
{
if ( encoding == TIXML_ENCODING_UTF8 )
{
if ( v < 128 ) return tolower( v );
return v;
}
else
{
return tolower( v );
}
}
static void ConvertUTF32ToUTF8( unsigned long input, char* output, int* length );
private:
TiXmlBase( const TiXmlBase& );
void operator=( const TiXmlBase& base );
struct Entity
{
const char* str;
unsigned int strLength;
char chr;
};
enum
{
NUM_ENTITY = 5,
MAX_ENTITY_LENGTH = 6
};
static Entity entity[ NUM_ENTITY ];
static bool condenseWhiteSpace;
};
(Except for attributes).
Nodes have siblings, a parent, and children. A node can be
in a document, or stand on its own. The type of a TiXmlNode
can be queried, and it can be cast to its more defined type.
*/
class TiXmlNode : public TiXmlBase
{
friend class TiXmlDocument;
friend class TiXmlElement;
public:
#ifdef TIXML_USE_STL
formatting, but doesn't expect them.
*/
friend std::istream& operator >> (std::istream& in, TiXmlNode& base);
without any newlines or formatting, as opposed to Print(), which
includes tabs and new lines.
The operator<< and operator>> are not completely symmetric. Writing
a node to a stream is very well defined. You'll get a nice stream
of output, without any extra whitespace or newlines.
But reading is not as well defined. (As it always is.) If you create
a TiXmlElement (for example) and read that from an input stream,
the text needs to define an element or junk will result. This is
true of all input streams, but it's worth keeping in mind.
A TiXmlDocument will read nodes until it reads a root element, and
all the children of that root element.
*/
friend std::ostream& operator<< (std::ostream& out, const TiXmlNode& base);
friend std::string& operator<< (std::string& out, const TiXmlNode& base );
#endif
unsupported types are picked up by UNKNOWN.)
*/
enum NodeType
{
TINYXML_DOCUMENT,
TINYXML_ELEMENT,
TINYXML_COMMENT,
TINYXML_UNKNOWN,
TINYXML_TEXT,
TINYXML_DECLARATION,
TINYXML_TYPECOUNT
};
virtual ~TiXmlNode();
TiXmlNode.
@verbatim
Document: filename of the xml file
Element: name of the element
Comment: the comment text
Unknown: the tag contents
Text: the text string
@endverbatim
The subclasses will wrap this function.
*/
const char *Value() const { return value.c_str (); }
#ifdef TIXML_USE_STL
this is more efficient than calling Value().
Only available in STL mode.
*/
const std::string& ValueStr() const { return value; }
#endif
const TIXML_STRING& ValueTStr() const { return value; }
@verbatim
Document: filename of the xml file
Element: name of the element
Comment: the comment text
Unknown: the tag contents
Text: the text string
@endverbatim
*/
void SetValue(const char * _value) { value = _value;}
#ifdef TIXML_USE_STL
void SetValue( const std::string& _value ) { value = _value; }
#endif
void Clear();
TiXmlNode* Parent() { return parent; }
const TiXmlNode* Parent() const { return parent; }
const TiXmlNode* FirstChild() const { return firstChild; }
TiXmlNode* FirstChild() { return firstChild; }
const TiXmlNode* FirstChild( const char * value ) const;
TiXmlNode* FirstChild( const char * _value ) {
return const_cast< TiXmlNode* > ((const_cast< const TiXmlNode* >(this))->FirstChild( _value ));
}
const TiXmlNode* LastChild() const { return lastChild; }
TiXmlNode* LastChild() { return lastChild; }
const TiXmlNode* LastChild( const char * value ) const;
TiXmlNode* LastChild( const char * _value ) {
return const_cast< TiXmlNode* > ((const_cast< const TiXmlNode* >(this))->LastChild( _value ));
}
#ifdef TIXML_USE_STL
const TiXmlNode* FirstChild( const std::string& _value ) const { return FirstChild (_value.c_str ()); }
TiXmlNode* FirstChild( const std::string& _value ) { return FirstChild (_value.c_str ()); }
const TiXmlNode* LastChild( const std::string& _value ) const { return LastChild (_value.c_str ()); }
TiXmlNode* LastChild( const std::string& _value ) { return LastChild (_value.c_str ()); }
#endif
One way to iterate over nodes is:
@verbatim
for( child = parent->FirstChild(); child; child = child->NextSibling() )
@endverbatim
IterateChildren does the same thing with the syntax:
@verbatim
child = 0;
while( child = parent->IterateChildren( child ) )
@endverbatim
IterateChildren takes the previous child as input and finds
the next one. If the previous child is null, it returns the
first. IterateChildren will return null when done.
*/
const TiXmlNode* IterateChildren( const TiXmlNode* previous ) const;
TiXmlNode* IterateChildren( const TiXmlNode* previous ) {
return const_cast< TiXmlNode* >( (const_cast< const TiXmlNode* >(this))->IterateChildren( previous ) );
}
const TiXmlNode* IterateChildren( const char * value, const TiXmlNode* previous ) const;
TiXmlNode* IterateChildren( const char * _value, const TiXmlNode* previous ) {
return const_cast< TiXmlNode* >( (const_cast< const TiXmlNode* >(this))->IterateChildren( _value, previous ) );
}
#ifdef TIXML_USE_STL
const TiXmlNode* IterateChildren( const std::string& _value, const TiXmlNode* previous ) const { return IterateChildren (_value.c_str (), previous); }
TiXmlNode* IterateChildren( const std::string& _value, const TiXmlNode* previous ) { return IterateChildren (_value.c_str (), previous); }
#endif
Returns a pointer to the new object or NULL if an error occured.
*/
TiXmlNode* InsertEndChild( const TiXmlNode& addThis );
NOTE: the node to be added is passed by pointer, and will be
henceforth owned (and deleted) by tinyXml. This method is efficient
and avoids an extra copy, but should be used with care as it
uses a different memory model than the other insert functions.
@sa InsertEndChild
*/
TiXmlNode* LinkEndChild( TiXmlNode* addThis );
Returns a pointer to the new object or NULL if an error occured.
*/
TiXmlNode* InsertBeforeChild( TiXmlNode* beforeThis, const TiXmlNode& addThis );
Returns a pointer to the new object or NULL if an error occured.
*/
TiXmlNode* InsertAfterChild( TiXmlNode* afterThis, const TiXmlNode& addThis );
Returns a pointer to the new object or NULL if an error occured.
*/
TiXmlNode* ReplaceChild( TiXmlNode* replaceThis, const TiXmlNode& withThis );
bool RemoveChild( TiXmlNode* removeThis );
const TiXmlNode* PreviousSibling() const { return prev; }
TiXmlNode* PreviousSibling() { return prev; }
const TiXmlNode* PreviousSibling( const char * ) const;
TiXmlNode* PreviousSibling( const char *_prev ) {
return const_cast< TiXmlNode* >( (const_cast< const TiXmlNode* >(this))->PreviousSibling( _prev ) );
}
#ifdef TIXML_USE_STL
const TiXmlNode* PreviousSibling( const std::string& _value ) const { return PreviousSibling (_value.c_str ()); }
TiXmlNode* PreviousSibling( const std::string& _value ) { return PreviousSibling (_value.c_str ()); }
const TiXmlNode* NextSibling( const std::string& _value) const { return NextSibling (_value.c_str ()); }
TiXmlNode* NextSibling( const std::string& _value) { return NextSibling (_value.c_str ()); }
#endif
const TiXmlNode* NextSibling() const { return next; }
TiXmlNode* NextSibling() { return next; }
const TiXmlNode* NextSibling( const char * ) const;
TiXmlNode* NextSibling( const char* _next ) {
return const_cast< TiXmlNode* >( (const_cast< const TiXmlNode* >(this))->NextSibling( _next ) );
}
Calls NextSibling and ToElement. Will skip all non-Element
nodes. Returns 0 if there is not another element.
*/
const TiXmlElement* NextSiblingElement() const;
TiXmlElement* NextSiblingElement() {
return const_cast< TiXmlElement* >( (const_cast< const TiXmlNode* >(this))->NextSiblingElement() );
}
Calls NextSibling and ToElement. Will skip all non-Element
nodes. Returns 0 if there is not another element.
*/
const TiXmlElement* NextSiblingElement( const char * ) const;
TiXmlElement* NextSiblingElement( const char *_next ) {
return const_cast< TiXmlElement* >( (const_cast< const TiXmlNode* >(this))->NextSiblingElement( _next ) );
}
#ifdef TIXML_USE_STL
const TiXmlElement* NextSiblingElement( const std::string& _value) const { return NextSiblingElement (_value.c_str ()); }
TiXmlElement* NextSiblingElement( const std::string& _value) { return NextSiblingElement (_value.c_str ()); }
#endif
const TiXmlElement* FirstChildElement() const;
TiXmlElement* FirstChildElement() {
return const_cast< TiXmlElement* >( (const_cast< const TiXmlNode* >(this))->FirstChildElement() );
}
const TiXmlElement* FirstChildElement( const char * _value ) const;
TiXmlElement* FirstChildElement( const char * _value ) {
return const_cast< TiXmlElement* >( (const_cast< const TiXmlNode* >(this))->FirstChildElement( _value ) );
}
#ifdef TIXML_USE_STL
const TiXmlElement* FirstChildElement( const std::string& _value ) const { return FirstChildElement (_value.c_str ()); }
TiXmlElement* FirstChildElement( const std::string& _value ) { return FirstChildElement (_value.c_str ()); }
#endif
The possible types are: DOCUMENT, ELEMENT, COMMENT,
UNKNOWN, TEXT, and DECLARATION.
*/
int Type() const { return type; }
Returns null if not in a document.
*/
const TiXmlDocument* GetDocument() const;
TiXmlDocument* GetDocument() {
return const_cast< TiXmlDocument* >( (const_cast< const TiXmlNode* >(this))->GetDocument() );
}
bool NoChildren() const { return !firstChild; }
virtual const TiXmlDocument* ToDocument() const { return 0; }
virtual const TiXmlElement* ToElement() const { return 0; }
virtual const TiXmlComment* ToComment() const { return 0; }
virtual const TiXmlUnknown* ToUnknown() const { return 0; }
virtual const TiXmlText* ToText() const { return 0; }
virtual const TiXmlDeclaration* ToDeclaration() const { return 0; }
virtual TiXmlDocument* ToDocument() { return 0; }
virtual TiXmlElement* ToElement() { return 0; }
virtual TiXmlComment* ToComment() { return 0; }
virtual TiXmlUnknown* ToUnknown() { return 0; }
virtual TiXmlText* ToText() { return 0; }
virtual TiXmlDeclaration* ToDeclaration() { return 0; }
by the caller.
*/
virtual TiXmlNode* Clone() const = 0;
XML tree will be conditionally visited and the host will be called back
via the TiXmlVisitor interface.
This is essentially a SAX interface for TinyXML. (Note however it doesn't re-parse
the XML for the callbacks, so the performance of TinyXML is unchanged by using this
interface versus any other.)
The interface has been based on ideas from:
- http://www.saxproject.org/
- http://c2.com/cgi/wiki?HierarchicalVisitorPattern
Which are both good references for "visiting".
An example of using Accept():
@verbatim
TiXmlPrinter printer;
tinyxmlDoc.Accept( &printer );
const char* xmlcstr = printer.CStr();
@endverbatim
*/
virtual bool Accept( TiXmlVisitor* visitor ) const = 0;
protected:
TiXmlNode( NodeType _type );
void CopyTo( TiXmlNode* target ) const;
#ifdef TIXML_USE_STL
virtual void StreamIn( std::istream* in, TIXML_STRING* tag ) = 0;
#endif
TiXmlNode* Identify( const char* start, TiXmlEncoding encoding );
TiXmlNode* parent;
NodeType type;
TiXmlNode* firstChild;
TiXmlNode* lastChild;
TIXML_STRING value;
TiXmlNode* prev;
TiXmlNode* next;
private:
TiXmlNode( const TiXmlNode& );
void operator=( const TiXmlNode& base );
};
number of attributes, each with a unique name.
@note The attributes are not TiXmlNodes, since they are not
part of the tinyXML document object model. There are other
suggested ways to look at this problem.
*/
class TiXmlAttribute : public TiXmlBase
{
friend class TiXmlAttributeSet;
public:
TiXmlAttribute() : TiXmlBase()
{
document = 0;
prev = next = 0;
}
#ifdef TIXML_USE_STL
TiXmlAttribute( const std::string& _name, const std::string& _value )
{
name = _name;
value = _value;
document = 0;
prev = next = 0;
}
#endif
TiXmlAttribute( const char * _name, const char * _value )
{
name = _name;
value = _value;
document = 0;
prev = next = 0;
}
const char* Name() const { return name.c_str(); }
const char* Value() const { return value.c_str(); }
#ifdef TIXML_USE_STL
const std::string& ValueStr() const { return value; }
#endif
int IntValue() const;
double DoubleValue() const;
const TIXML_STRING& NameTStr() const { return name; }
IntValue() method with richer error checking.
If the value is an integer, it is stored in 'value' and
the call returns TIXML_SUCCESS. If it is not
an integer, it returns TIXML_WRONG_TYPE.
A specialized but useful call. Note that for success it returns 0,
which is the opposite of almost all other TinyXml calls.
*/
int QueryIntValue( int* _value ) const;
int QueryDoubleValue( double* _value ) const;
void SetName( const char* _name ) { name = _name; }
void SetValue( const char* _value ) { value = _value; }
void SetIntValue( int _value );
void SetDoubleValue( double _value );
#ifdef TIXML_USE_STL
void SetName( const std::string& _name ) { name = _name; }
void SetValue( const std::string& _value ) { value = _value; }
#endif
const TiXmlAttribute* Next() const;
TiXmlAttribute* Next() {
return const_cast< TiXmlAttribute* >( (const_cast< const TiXmlAttribute* >(this))->Next() );
}
const TiXmlAttribute* Previous() const;
TiXmlAttribute* Previous() {
return const_cast< TiXmlAttribute* >( (const_cast< const TiXmlAttribute* >(this))->Previous() );
}
bool operator==( const TiXmlAttribute& rhs ) const { return rhs.name == name; }
bool operator<( const TiXmlAttribute& rhs ) const { return name < rhs.name; }
bool operator>( const TiXmlAttribute& rhs ) const { return name > rhs.name; }
returns: the next char after the value end quote
*/
virtual const char* Parse( const char* p, TiXmlParsingData* data, TiXmlEncoding encoding );
virtual void Print( FILE* cfile, int depth ) const {
Print( cfile, depth, 0 );
}
void Print( FILE* cfile, int depth, TIXML_STRING* str ) const;
void SetDocument( TiXmlDocument* doc ) { document = doc; }
private:
TiXmlAttribute( const TiXmlAttribute& );
void operator=( const TiXmlAttribute& base );
TiXmlDocument* document;
TIXML_STRING name;
TIXML_STRING value;
TiXmlAttribute* prev;
TiXmlAttribute* next;
};
It is only used internally, both by the ELEMENT and the DECLARATION.
The set can be changed transparent to the Element and Declaration
classes that use it, but NOT transparent to the Attribute
which has to implement a next() and previous() method. Which makes
it a bit problematic and prevents the use of STL.
This version is implemented with circular lists because:
- I like circular lists
- it demonstrates some independence from the (typical) doubly linked list.
*/
class TiXmlAttributeSet
{
public:
TiXmlAttributeSet();
~TiXmlAttributeSet();
void Add( TiXmlAttribute* attribute );
void Remove( TiXmlAttribute* attribute );
const TiXmlAttribute* First() const { return ( sentinel.next == &sentinel ) ? 0 : sentinel.next; }
TiXmlAttribute* First() { return ( sentinel.next == &sentinel ) ? 0 : sentinel.next; }
const TiXmlAttribute* Last() const { return ( sentinel.prev == &sentinel ) ? 0 : sentinel.prev; }
TiXmlAttribute* Last() { return ( sentinel.prev == &sentinel ) ? 0 : sentinel.prev; }
TiXmlAttribute* Find( const char* _name ) const;
TiXmlAttribute* FindOrCreate( const char* _name );
# ifdef TIXML_USE_STL
TiXmlAttribute* Find( const std::string& _name ) const;
TiXmlAttribute* FindOrCreate( const std::string& _name );
# endif
private:
TiXmlAttributeSet( const TiXmlAttributeSet& );
void operator=( const TiXmlAttributeSet& );
TiXmlAttribute sentinel;
};
and can contain other elements, text, comments, and unknowns.
Elements also contain an arbitrary number of attributes.
*/
class TiXmlElement : public TiXmlNode
{
public:
TiXmlElement (const char * in_value);
#ifdef TIXML_USE_STL
TiXmlElement( const std::string& _value );
#endif
TiXmlElement( const TiXmlElement& );
void operator=( const TiXmlElement& base );
virtual ~TiXmlElement();
for the attribute of that name, or null if none exists.
*/
const char* Attribute( const char* name ) const;
for the attribute of that name, or null if none exists.
If the attribute exists and can be converted to an integer,
the integer value will be put in the return 'i', if 'i'
is non-null.
*/
const char* Attribute( const char* name, int* i ) const;
for the attribute of that name, or null if none exists.
If the attribute exists and can be converted to an double,
the double value will be put in the return 'd', if 'd'
is non-null.
*/
const char* Attribute( const char* name, double* d ) const;
Attribute() method with richer error checking.
If the attribute is an integer, it is stored in 'value' and
the call returns TIXML_SUCCESS. If it is not
an integer, it returns TIXML_WRONG_TYPE. If the attribute
does not exist, then TIXML_NO_ATTRIBUTE is returned.
*/
int QueryIntAttribute( const char* name, int* _value ) const;
int QueryDoubleAttribute( const char* name, double* _value ) const;
int QueryFloatAttribute( const char* name, float* _value ) const {
double d;
int result = QueryDoubleAttribute( name, &d );
if ( result == TIXML_SUCCESS ) {
*_value = (float)d;
}
return result;
}
#ifdef TIXML_USE_STL
int QueryStringAttribute( const char* name, std::string* _value ) const {
const char* cstr = Attribute( name );
if ( cstr ) {
*_value = std::string( cstr );
return TIXML_SUCCESS;
}
return TIXML_NO_ATTRIBUTE;
}
attribute into the specified type. Very easy, very powerful, but
be careful to make sure to call this with the correct type.
NOTE: This method doesn't work correctly for 'string' types that contain spaces.
@return TIXML_SUCCESS, TIXML_WRONG_TYPE, or TIXML_NO_ATTRIBUTE
*/
template< typename T > int QueryValueAttribute( const std::string& name, T* outValue ) const
{
const TiXmlAttribute* node = attributeSet.Find( name );
if ( !node )
return TIXML_NO_ATTRIBUTE;
std::stringstream sstream( node->ValueStr() );
sstream >> *outValue;
if ( !sstream.fail() )
return TIXML_SUCCESS;
return TIXML_WRONG_TYPE;
}
int QueryValueAttribute( const std::string& name, std::string* outValue ) const
{
const TiXmlAttribute* node = attributeSet.Find( name );
if ( !node )
return TIXML_NO_ATTRIBUTE;
*outValue = node->ValueStr();
return TIXML_SUCCESS;
}
#endif
will be created if it does not exist, or changed if it does.
*/
void SetAttribute( const char* name, const char * _value );
#ifdef TIXML_USE_STL
const std::string* Attribute( const std::string& name ) const;
const std::string* Attribute( const std::string& name, int* i ) const;
const std::string* Attribute( const std::string& name, double* d ) const;
int QueryIntAttribute( const std::string& name, int* _value ) const;
int QueryDoubleAttribute( const std::string& name, double* _value ) const;
void SetAttribute( const std::string& name, const std::string& _value );
void SetAttribute( const std::string& name, int _value );
void SetDoubleAttribute( const std::string& name, double value );
#endif
will be created if it does not exist, or changed if it does.
*/
void SetAttribute( const char * name, int value );
will be created if it does not exist, or changed if it does.
*/
void SetDoubleAttribute( const char * name, double value );
*/
void RemoveAttribute( const char * name );
#ifdef TIXML_USE_STL
void RemoveAttribute( const std::string& name ) { RemoveAttribute (name.c_str ()); }
#endif
const TiXmlAttribute* FirstAttribute() const { return attributeSet.First(); }
TiXmlAttribute* FirstAttribute() { return attributeSet.First(); }
const TiXmlAttribute* LastAttribute() const { return attributeSet.Last(); }
TiXmlAttribute* LastAttribute() { return attributeSet.Last(); }
and concise, GetText() is limited compared to getting the TiXmlText child
and accessing it directly.
If the first child of 'this' is a TiXmlText, the GetText()
returns the character string of the Text node, else null is returned.
This is a convenient method for getting the text of simple contained text:
@verbatim
<foo>This is text</foo>
const char* str = fooElement->GetText();
@endverbatim
'str' will be a pointer to "This is text".
Note that this function can be misleading. If the element foo was created from
this XML:
@verbatim
<foo><b>This is text</b></foo>
@endverbatim
then the value of str would be null. The first child node isn't a text node, it is
another element. From this XML:
@verbatim
<foo>This is <b>text</b></foo>
@endverbatim
GetText() will return "This is ".
WARNING: GetText() accesses a child node - don't become confused with the
similarly named TiXmlHandle::Text() and TiXmlNode::ToText() which are
safe type casts on the referenced node.
*/
const char* GetText() const;
virtual TiXmlNode* Clone() const;
virtual void Print( FILE* cfile, int depth ) const;
returns: next char past '>'
*/
virtual const char* Parse( const char* p, TiXmlParsingData* data, TiXmlEncoding encoding );
virtual const TiXmlElement* ToElement() const { return this; }
virtual TiXmlElement* ToElement() { return this; }
*/
virtual bool Accept( TiXmlVisitor* visitor ) const;
protected:
void CopyTo( TiXmlElement* target ) const;
void ClearThis();
#ifdef TIXML_USE_STL
virtual void StreamIn( std::istream * in, TIXML_STRING * tag );
#endif
Reads the "value" of the element -- another element, or text.
This should terminate with the current end tag.
*/
const char* ReadValue( const char* in, TiXmlParsingData* prevData, TiXmlEncoding encoding );
private:
TiXmlAttributeSet attributeSet;
};
*/
class TiXmlComment : public TiXmlNode
{
public:
TiXmlComment() : TiXmlNode( TiXmlNode::TINYXML_COMMENT ) {}
TiXmlComment( const char* _value ) : TiXmlNode( TiXmlNode::TINYXML_COMMENT ) {
SetValue( _value );
}
TiXmlComment( const TiXmlComment& );
void operator=( const TiXmlComment& base );
virtual ~TiXmlComment() {}
virtual TiXmlNode* Clone() const;
virtual void Print( FILE* cfile, int depth ) const;
returns: next char past '>'
*/
virtual const char* Parse( const char* p, TiXmlParsingData* data, TiXmlEncoding encoding );
virtual const TiXmlComment* ToComment() const { return this; }
virtual TiXmlComment* ToComment() { return this; }
*/
virtual bool Accept( TiXmlVisitor* visitor ) const;
protected:
void CopyTo( TiXmlComment* target ) const;
#ifdef TIXML_USE_STL
virtual void StreamIn( std::istream * in, TIXML_STRING * tag );
#endif
private:
};
and CDATA. It will default to the mode it was parsed from the XML file and
you generally want to leave it alone, but you can change the output mode with
SetCDATA() and query it with CDATA().
*/
class TiXmlText : public TiXmlNode
{
friend class TiXmlElement;
public:
normal, encoded text. If you want it be output as a CDATA text
element, set the parameter _cdata to 'true'
*/
TiXmlText (const char * initValue ) : TiXmlNode (TiXmlNode::TINYXML_TEXT)
{
SetValue( initValue );
cdata = false;
}
virtual ~TiXmlText() {}
#ifdef TIXML_USE_STL
TiXmlText( const std::string& initValue ) : TiXmlNode (TiXmlNode::TINYXML_TEXT)
{
SetValue( initValue );
cdata = false;
}
#endif
TiXmlText( const TiXmlText& copy ) : TiXmlNode( TiXmlNode::TINYXML_TEXT ) { copy.CopyTo( this ); }
void operator=( const TiXmlText& base ) { base.CopyTo( this ); }
virtual void Print( FILE* cfile, int depth ) const;
bool CDATA() const { return cdata; }
void SetCDATA( bool _cdata ) { cdata = _cdata; }
virtual const char* Parse( const char* p, TiXmlParsingData* data, TiXmlEncoding encoding );
virtual const TiXmlText* ToText() const { return this; }
virtual TiXmlText* ToText() { return this; }
*/
virtual bool Accept( TiXmlVisitor* content ) const;
protected :
virtual TiXmlNode* Clone() const;
void CopyTo( TiXmlText* target ) const;
bool Blank() const;
#ifdef TIXML_USE_STL
virtual void StreamIn( std::istream * in, TIXML_STRING * tag );
#endif
private:
bool cdata;
};
@verbatim
<?xml version="1.0" standalone="yes"?>
@endverbatim
TinyXml will happily read or write files without a declaration,
however. There are 3 possible attributes to the declaration:
version, encoding, and standalone.
Note: In this version of the code, the attributes are
handled as special cases, not generic attributes, simply
because there can only be at most 3 and they are always the same.
*/
class TiXmlDeclaration : public TiXmlNode
{
public:
TiXmlDeclaration() : TiXmlNode( TiXmlNode::TINYXML_DECLARATION ) {}
#ifdef TIXML_USE_STL
TiXmlDeclaration( const std::string& _version,
const std::string& _encoding,
const std::string& _standalone );
#endif
TiXmlDeclaration( const char* _version,
const char* _encoding,
const char* _standalone );
TiXmlDeclaration( const TiXmlDeclaration& copy );
void operator=( const TiXmlDeclaration& copy );
virtual ~TiXmlDeclaration() {}
const char *Version() const { return version.c_str (); }
const char *Encoding() const { return encoding.c_str (); }
const char *Standalone() const { return standalone.c_str (); }
virtual TiXmlNode* Clone() const;
virtual void Print( FILE* cfile, int depth, TIXML_STRING* str ) const;
virtual void Print( FILE* cfile, int depth ) const {
Print( cfile, depth, 0 );
}
virtual const char* Parse( const char* p, TiXmlParsingData* data, TiXmlEncoding encoding );
virtual const TiXmlDeclaration* ToDeclaration() const { return this; }
virtual TiXmlDeclaration* ToDeclaration() { return this; }
*/
virtual bool Accept( TiXmlVisitor* visitor ) const;
protected:
void CopyTo( TiXmlDeclaration* target ) const;
#ifdef TIXML_USE_STL
virtual void StreamIn( std::istream * in, TIXML_STRING * tag );
#endif
private:
TIXML_STRING version;
TIXML_STRING encoding;
TIXML_STRING standalone;
};
unknown. It is a tag of text, but should not be modified.
It will be written back to the XML, unchanged, when the file
is saved.
DTD tags get thrown into TiXmlUnknowns.
*/
class TiXmlUnknown : public TiXmlNode
{
public:
TiXmlUnknown() : TiXmlNode( TiXmlNode::TINYXML_UNKNOWN ) {}
virtual ~TiXmlUnknown() {}
TiXmlUnknown( const TiXmlUnknown& copy ) : TiXmlNode( TiXmlNode::TINYXML_UNKNOWN ) { copy.CopyTo( this ); }
void operator=( const TiXmlUnknown& copy ) { copy.CopyTo( this ); }
virtual TiXmlNode* Clone() const;
virtual void Print( FILE* cfile, int depth ) const;
virtual const char* Parse( const char* p, TiXmlParsingData* data, TiXmlEncoding encoding );
virtual const TiXmlUnknown* ToUnknown() const { return this; }
virtual TiXmlUnknown* ToUnknown() { return this; }
*/
virtual bool Accept( TiXmlVisitor* content ) const;
protected:
void CopyTo( TiXmlUnknown* target ) const;
#ifdef TIXML_USE_STL
virtual void StreamIn( std::istream * in, TIXML_STRING * tag );
#endif
private:
};
XML pieces. It can be saved, loaded, and printed to the screen.
The 'value' of a document node is the xml file name.
*/
class TiXmlDocument : public TiXmlNode
{
public:
TiXmlDocument();
TiXmlDocument( const char * documentName );
#ifdef TIXML_USE_STL
TiXmlDocument( const std::string& documentName );
#endif
TiXmlDocument( const TiXmlDocument& copy );
void operator=( const TiXmlDocument& copy );
virtual ~TiXmlDocument() {}
Returns true if successful. Will delete any existing
document data before loading.
*/
bool LoadFile( TiXmlEncoding encoding = TIXML_DEFAULT_ENCODING );
bool SaveFile() const;
bool LoadFile( const char * filename, TiXmlEncoding encoding = TIXML_DEFAULT_ENCODING );
bool SaveFile( const char * filename ) const;
doesn't stream - the entire object pointed at by the FILE*
will be interpreted as an XML file. TinyXML doesn't stream in XML from the current
file location. Streaming may be added in the future.
*/
bool LoadFile( FILE*, TiXmlEncoding encoding = TIXML_DEFAULT_ENCODING );
bool SaveFile( FILE* ) const;
#ifdef TIXML_USE_STL
bool LoadFile( const std::string& filename, TiXmlEncoding encoding = TIXML_DEFAULT_ENCODING )
{
return LoadFile( filename.c_str(), encoding );
}
bool SaveFile( const std::string& filename ) const
{
return SaveFile( filename.c_str() );
}
#endif
method (either TIXML_ENCODING_LEGACY or TIXML_ENCODING_UTF8 will force TinyXml
to use that encoding, regardless of what TinyXml might otherwise try to detect.
*/
virtual const char* Parse( const char* p, TiXmlParsingData* data = 0, TiXmlEncoding encoding = TIXML_DEFAULT_ENCODING );
In well formed XML, there should only be one. TinyXml is tolerant of
multiple elements at the document level.
*/
const TiXmlElement* RootElement() const { return FirstChildElement(); }
TiXmlElement* RootElement() { return FirstChildElement(); }
- The ErrorId() will contain the integer identifier of the error (not generally useful)
- The ErrorDesc() method will return the name of the error. (very useful)
- The ErrorRow() and ErrorCol() will return the location of the error (if known)
*/
bool Error() const { return error; }
const char * ErrorDesc() const { return errorDesc.c_str (); }
prefer the ErrorId, this function will fetch it.
*/
int ErrorId() const { return errorId; }
and the first row is row 1. A value of 0 means the row and column wasn't applicable
(memory errors, for example, have no row/column) or the parser lost the error. (An
error in the error reporting, in that case.)
@sa SetTabSize, Row, Column
*/
int ErrorRow() const { return errorLocation.row+1; }
int ErrorCol() const { return errorLocation.col+1; }
to report the correct values for row and column. It does not change the output
or input in any way.
By calling this method, with a tab size
greater than 0, the row and column of each node and attribute is stored
when the file is loaded. Very useful for tracking the DOM back in to
the source file.
The tab size is required for calculating the location of nodes. If not
set, the default of 4 is used. The tabsize is set per document. Setting
the tabsize to 0 disables row/column tracking.
Note that row and column tracking is not supported when using operator>>.
The tab size needs to be enabled before the parse or load. Correct usage:
@verbatim
TiXmlDocument doc;
doc.SetTabSize( 8 );
doc.Load( "myfile.xml" );
@endverbatim
@sa Row, Column
*/
void SetTabSize( int _tabsize ) { tabsize = _tabsize; }
int TabSize() const { return tabsize; }
state is automatically cleared if you Parse a new XML block.
*/
void ClearError() { error = false;
errorId = 0;
errorDesc = "";
errorLocation.row = errorLocation.col = 0;
}
void Print() const { Print( stdout, 0 ); }
will allocate a character array (new char[]) and return it as a pointer. The
calling code pust call delete[] on the return char* to avoid a memory leak.
*/
virtual void Print( FILE* cfile, int depth = 0 ) const;
void SetError( int err, const char* errorLocation, TiXmlParsingData* prevData, TiXmlEncoding encoding );
virtual const TiXmlDocument* ToDocument() const { return this; }
virtual TiXmlDocument* ToDocument() { return this; }
*/
virtual bool Accept( TiXmlVisitor* content ) const;
protected :
virtual TiXmlNode* Clone() const;
#ifdef TIXML_USE_STL
virtual void StreamIn( std::istream * in, TIXML_STRING * tag );
#endif
private:
void CopyTo( TiXmlDocument* target ) const;
bool error;
int errorId;
TIXML_STRING errorDesc;
int tabsize;
TiXmlCursor errorLocation;
bool useMicrosoftBOM;
};
A TiXmlHandle is a class that wraps a node pointer with null checks; this is
an incredibly useful thing. Note that TiXmlHandle is not part of the TinyXml
DOM structure. It is a separate utility class.
Take an example:
@verbatim
<Document>
<Element attributeA = "valueA">
<Child attributeB = "value1" />
<Child attributeB = "value2" />
</Element>
<Document>
@endverbatim
Assuming you want the value of "attributeB" in the 2nd "Child" element, it's very
easy to write a *lot* of code that looks like:
@verbatim
TiXmlElement* root = document.FirstChildElement( "Document" );
if ( root )
{
TiXmlElement* element = root->FirstChildElement( "Element" );
if ( element )
{
TiXmlElement* child = element->FirstChildElement( "Child" );
if ( child )
{
TiXmlElement* child2 = child->NextSiblingElement( "Child" );
if ( child2 )
{
// Finally do something useful.
@endverbatim
And that doesn't even cover "else" cases. TiXmlHandle addresses the verbosity
of such code. A TiXmlHandle checks for null pointers so it is perfectly safe
and correct to use:
@verbatim
TiXmlHandle docHandle( &document );
TiXmlElement* child2 = docHandle.FirstChild( "Document" ).FirstChild( "Element" ).Child( "Child", 1 ).ToElement();
if ( child2 )
{
// do something useful
@endverbatim
Which is MUCH more concise and useful.
It is also safe to copy handles - internally they are nothing more than node pointers.
@verbatim
TiXmlHandle handleCopy = handle;
@endverbatim
What they should not be used for is iteration:
@verbatim
int i=0;
while ( true )
{
TiXmlElement* child = docHandle.FirstChild( "Document" ).FirstChild( "Element" ).Child( "Child", i ).ToElement();
if ( !child )
break;
// do something
++i;
}
@endverbatim
It seems reasonable, but it is in fact two embedded while loops. The Child method is
a linear walk to find the element, so this code would iterate much more than it needs
to. Instead, prefer:
@verbatim
TiXmlElement* child = docHandle.FirstChild( "Document" ).FirstChild( "Element" ).FirstChild( "Child" ).ToElement();
for( child; child; child=child->NextSiblingElement() )
{
// do something
}
@endverbatim
*/
class TiXmlHandle
{
public:
TiXmlHandle( TiXmlNode* _node ) { this->node = _node; }
TiXmlHandle( const TiXmlHandle& ref ) { this->node = ref.node; }
TiXmlHandle operator=( const TiXmlHandle& ref ) { this->node = ref.node; return *this; }
TiXmlHandle FirstChild() const;
TiXmlHandle FirstChild( const char * value ) const;
TiXmlHandle FirstChildElement() const;
TiXmlHandle FirstChildElement( const char * value ) const;
The first child is 0, the second 1, etc.
*/
TiXmlHandle Child( const char* value, int index ) const;
The first child is 0, the second 1, etc.
*/
TiXmlHandle Child( int index ) const;
The first child element is 0, the second 1, etc. Note that only TiXmlElements
are indexed: other types are not counted.
*/
TiXmlHandle ChildElement( const char* value, int index ) const;
The first child element is 0, the second 1, etc. Note that only TiXmlElements
are indexed: other types are not counted.
*/
TiXmlHandle ChildElement( int index ) const;
#ifdef TIXML_USE_STL
TiXmlHandle FirstChild( const std::string& _value ) const { return FirstChild( _value.c_str() ); }
TiXmlHandle FirstChildElement( const std::string& _value ) const { return FirstChildElement( _value.c_str() ); }
TiXmlHandle Child( const std::string& _value, int index ) const { return Child( _value.c_str(), index ); }
TiXmlHandle ChildElement( const std::string& _value, int index ) const { return ChildElement( _value.c_str(), index ); }
#endif
*/
TiXmlNode* ToNode() const { return node; }
*/
TiXmlElement* ToElement() const { return ( ( node && node->ToElement() ) ? node->ToElement() : 0 ); }
*/
TiXmlText* ToText() const { return ( ( node && node->ToText() ) ? node->ToText() : 0 ); }
*/
TiXmlUnknown* ToUnknown() const { return ( ( node && node->ToUnknown() ) ? node->ToUnknown() : 0 ); }
Return the handle as a TiXmlNode. This may return null.
*/
TiXmlNode* Node() const { return ToNode(); }
Return the handle as a TiXmlElement. This may return null.
*/
TiXmlElement* Element() const { return ToElement(); }
Return the handle as a TiXmlText. This may return null.
*/
TiXmlText* Text() const { return ToText(); }
Return the handle as a TiXmlUnknown. This may return null.
*/
TiXmlUnknown* Unknown() const { return ToUnknown(); }
private:
TiXmlNode* node;
};
-# Print to memory (especially in non-STL mode)
-# Control formatting (line endings, etc.)
When constructed, the TiXmlPrinter is in its default "pretty printing" mode.
Before calling Accept() you can call methods to control the printing
of the XML document. After TiXmlNode::Accept() is called, the printed document can
be accessed via the CStr(), Str(), and Size() methods.
TiXmlPrinter uses the Visitor API.
@verbatim
TiXmlPrinter printer;
printer.SetIndent( "\t" );
doc.Accept( &printer );
fprintf( stdout, "%s", printer.CStr() );
@endverbatim
*/
class TiXmlPrinter : public TiXmlVisitor
{
public:
TiXmlPrinter() : depth( 0 ), simpleTextPrint( false ),
buffer(), indent( " " ), lineBreak( "\n" ) {}
virtual bool VisitEnter( const TiXmlDocument& doc );
virtual bool VisitExit( const TiXmlDocument& doc );
virtual bool VisitEnter( const TiXmlElement& element, const TiXmlAttribute* firstAttribute );
virtual bool VisitExit( const TiXmlElement& element );
virtual bool Visit( const TiXmlDeclaration& declaration );
virtual bool Visit( const TiXmlText& text );
virtual bool Visit( const TiXmlComment& comment );
virtual bool Visit( const TiXmlUnknown& unknown );
but tab (\t) is also useful, or null/empty string for no indentation.
*/
void SetIndent( const char* _indent ) { indent = _indent ? _indent : "" ; }
const char* Indent() { return indent.c_str(); }
Some operating systems prefer other characters, or can be
set to the null/empty string for no indenation.
*/
void SetLineBreak( const char* _lineBreak ) { lineBreak = _lineBreak ? _lineBreak : ""; }
const char* LineBreak() { return lineBreak.c_str(); }
linebreaks. Common when the XML is needed for network transmission.
*/
void SetStreamPrinting() { indent = "";
lineBreak = "";
}
const char* CStr() { return buffer.c_str(); }
size_t Size() { return buffer.size(); }
#ifdef TIXML_USE_STL
const std::string& Str() { return buffer; }
#endif
private:
void DoIndent() {
for( int i=0; i<depth; ++i )
buffer += indent;
}
void DoLineBreak() {
buffer += lineBreak;
}
int depth;
bool simpleTextPrint;
TIXML_STRING buffer;
TIXML_STRING indent;
TIXML_STRING lineBreak;
};
#ifdef _MSC_VER
#pragma warning( pop )
#endif
#endif