void TiXmlText::Print( FILE* cfile, int /*depth*/ ) const
{
TIXML_STRING buffer;
PutString( value, &buffer );
fprintf( cfile, "%s", buffer.c_str() );
}
bool TiXmlDocument::LoadBuffer( char* buffer, long size, TiXmlEncoding encoding )
{
// Delete the existing data:
Clear();
location.Clear();
// Get the file size, so we can pre-allocate the string. HUGE speed impact.
long length = size;
// Strange case, but good to handle up front.
if ( length <= 0 )
{
SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, TIXML_ENCODING_UNKNOWN );
return false;
}
// If we have a file, assume it is all one big XML file, and read it in.
// The document parser may decide the document ends sooner than the entire file, however.
TIXML_STRING data;
data.reserve( length );
// Subtle bug here. TinyXml did use fgets. But from the XML spec:
// 2.11 End-of-Line Handling
// <snip>
// <quote>
// ...the XML processor MUST behave as if it normalized all line breaks in external
// parsed entities (including the document entity) on input, before parsing, by translating
// both the two-character sequence #xD #xA and any #xD that is not followed by #xA to
// a single #xA character.
// </quote>
//
// It is not clear fgets does that, and certainly isn't clear it works cross platform.
// Generally, you expect fgets to translate from the convention of the OS to the c/unix
// convention, and not work generally.
/*
while( fgets( buf, sizeof(buf), file ) )
{
data += buf;
}
*/
char* buf = new char[ length+1 ];
buf[0] = 0;
memcpy(buf, buffer, length);
const char* lastPos = buf;
const char* p = buf;
buf[length] = 0;
while( *p ) {
assert( p < (buf+length) );
if ( *p == 0xa ) {
// Newline character. No special rules for this. Append all the characters
// since the last string, and include the newline.
data.append( lastPos, (p-lastPos+1) ); // append, include the newline
++p; // move past the newline
lastPos = p; // and point to the new buffer (may be 0)
assert( p <= (buf+length) );
}
else if ( *p == 0xd ) {
// Carriage return. Append what we have so far, then
// handle moving forward in the buffer.
if ( (p-lastPos) > 0 ) {
data.append( lastPos, p-lastPos ); // do not add the CR
}
data += (char)0xa; // a proper newline
if ( *(p+1) == 0xa ) {
// Carriage return - new line sequence
p += 2;
lastPos = p;
assert( p <= (buf+length) );
}
else {
// it was followed by something else...that is presumably characters again.
++p;
lastPos = p;
assert( p <= (buf+length) );
}
}
else {
++p;
}
}
// Handle any left over characters.
if ( p-lastPos ) {
data.append( lastPos, p-lastPos );
}
delete [] buf;
buf = 0;
Parse( data.c_str(), 0, encoding );
if ( Error() )
return false;
else
return true;
}
for( node = sentinel.next; node != &sentinel; node = node->next )
{
if ( node->name == name )
return node;
}
return 0;
}
#ifdef TIXML_USE_STL
TIXML_ISTREAM & operator >> (TIXML_ISTREAM & in, TiXmlNode & base)
{
TIXML_STRING tag;
tag.reserve( 8 * 1000 );
base.StreamIn( &in, &tag );
base.Parse( tag.c_str() );
return in;
}
#endif
TIXML_OSTREAM & operator<< (TIXML_OSTREAM & out, const TiXmlNode & base)
{
base.StreamOut (& out);
return out;
}
void TiXmlBase::PutString( const TIXML_STRING& str, TIXML_STRING* outString )
{
int i=0;
while( i<(int)str.length() )
{
int c = str[i];
if ( c == '&'
&& i < ( (int)str.length() - 2 )
&& str[i+1] == '#'
&& str[i+2] == 'x' )
{
// Hexadecimal character reference.
// Pass through unchanged.
// © -- copyright symbol, for example.
while ( i<(int)str.length() )
{
outString->append( str.c_str() + i, 1 );
++i;
if ( str[i] == ';' )
break;
}
}
else if ( c == '&' )
{
outString->append( entity[0].str, entity[0].strLength );
++i;
}
else if ( c == '<' )
{
outString->append( entity[1].str, entity[1].strLength );
++i;
}
else if ( c == '>' )
{
outString->append( entity[2].str, entity[2].strLength );
++i;
}
else if ( c == '\"' )
{
outString->append( entity[3].str, entity[3].strLength );
++i;
}
else if ( c == '\'' )
{
outString->append( entity[4].str, entity[4].strLength );
++i;
}
else if ( c < 32 || c > 126 )
{
// Easy pass at non-alpha/numeric/symbol
// 127 is the delete key. Below 32 is symbolic.
char buf[ 32 ];
sprintf( buf, "&#x%02X;", (unsigned) ( c & 0xff ) );
outString->append( buf, strlen( buf ) );
++i;
}
else
{
char realc = (char) c;
outString->append( &realc, 1 );
++i;
}
}
}
void TiXmlBase::PutString( const TIXML_STRING& str, TIXML_STRING* outString )
{
int i=0;
while( i<(int)str.length() )
{
unsigned char c = (unsigned char) str[i];
if ( c == '&'
&& i < ( (int)str.length() - 2 )
&& str[i+1] == '#'
&& str[i+2] == 'x' )
{
// Hexadecimal character reference.
// Pass through unchanged.
// © -- copyright symbol, for example.
while ( i<(int)str.length() )
{
outString->append( str.c_str() + i, 1 );
++i;
if ( str[i] == ';' )
break;
}
}
else if ( c == '&' )
{
outString->append( entity[0].str, entity[0].strLength );
++i;
}
else if ( c == '<' )
{
outString->append( entity[1].str, entity[1].strLength );
++i;
}
else if ( c == '>' )
{
outString->append( entity[2].str, entity[2].strLength );
++i;
}
else if ( c == '\"' )
{
outString->append( entity[3].str, entity[3].strLength );
++i;
}
else if ( c == '\'' )
{
outString->append( entity[4].str, entity[4].strLength );
++i;
}
else if ( c < 32 )
{
// Easy pass at non-alpha/numeric/symbol
// Below 32 is symbolic.
char buf[ 32 ];
sprintf( buf, "&#x%02X;", (unsigned) ( c & 0xff ) );
outString->append( buf, strlen( buf ) );
++i;
}
else
{
// Assume everthing else is unicode. c should never actually
// be out of the range of 0-255. Else something has gone strange.
//char realc = (char) c;
//outString->append( &realc, 1 );
*outString += (char) c; // somewhat more efficient function call.
++i;
}
}
}
const char* TiXmlElement::Parse( const char* p )
{
p = SkipWhiteSpace( p );
TiXmlDocument* document = GetDocument();
if ( !p || !*p || *p != '<' )
{
if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT );
return false;
}
p = SkipWhiteSpace( p+1 );
// Read the name.
p = ReadName( p, &value );
if ( !p || !*p )
{
if ( document ) document->SetError( TIXML_ERROR_FAILED_TO_READ_ELEMENT_NAME );
return false;
}
TIXML_STRING endTag ("</");
endTag += value;
endTag += ">";
// Check for and read attributes. Also look for an empty
// tag or an end tag.
while ( p && *p )
{
p = SkipWhiteSpace( p );
if ( !p || !*p )
{
if ( document ) document->SetError( TIXML_ERROR_READING_ATTRIBUTES );
return 0;
}
if ( *p == '/' )
{
++p;
// Empty tag.
if ( *p != '>' )
{
if ( document ) document->SetError( TIXML_ERROR_PARSING_EMPTY );
return 0;
}
return (p+1);
}
else if ( *p == '>' )
{
// Done with attributes (if there were any.)
// Read the value -- which can include other
// elements -- read the end tag, and return.
++p;
p = ReadValue( p ); // Note this is an Element method, and will set the error if one happens.
if ( !p || !*p )
return 0;
// We should find the end tag now
if ( StringEqual( p, endTag.c_str(), false ) )
{
p += endTag.length();
return p;
}
else
{
if ( document ) document->SetError( TIXML_ERROR_READING_END_TAG );
return 0;
}
}
else
{
// Try to read an element:
TiXmlAttribute attrib;
attrib.SetDocument( document );
p = attrib.Parse( p );
if ( !p || !*p )
{
if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT );
return 0;
}
SetAttribute( attrib.Name(), attrib.Value() );
}
}
return p;
}
bool TiXmlDocument::LoadFile( const char* filename, TiXmlEncoding encoding )
{
// Delete the existing data:
Clear();
location.Clear();
// There was a really terrifying little bug here. The code:
// value = filename
// in the STL case, cause the assignment method of the std::string to
// be called. What is strange, is that the std::string had the same
// address as it's c_str() method, and so bad things happen. Looks
// like a bug in the Microsoft STL implementation.
// See STL_STRING_BUG above.
// Fixed with the StringToBuffer class.
value = filename;
//SECURITY-UPDATE:2/3/07
//FILE* file = fopen( value.c_str (), "r" );
//if ( file )
//{
FILE *file=NULL;
errno_t err=0;
err=fopen_s(&file,value.c_str (), "r");
if(file && err==0)
{
// Get the file size, so we can pre-allocate the string. HUGE speed impact.
long length = 0;
fseek( file, 0, SEEK_END );
length = ftell( file );
fseek( file, 0, SEEK_SET );
// Strange case, but good to handle up front.
if ( length == 0 )
{
fclose( file );
return false;
}
// If we have a file, assume it is all one big XML file, and read it in.
// The document parser may decide the document ends sooner than the entire file, however.
TIXML_STRING data;
data.reserve( length );
const int BUF_SIZE = 2048;
char buf[BUF_SIZE];
while( fgets( buf, BUF_SIZE, file ) )
{
data += buf;
}
fclose( file );
Parse( data.c_str(), 0, encoding );
if ( Error() )
return false;
else
return true;
}
SetError( TIXML_ERROR_OPENING_FILE, 0, 0, TIXML_ENCODING_UNKNOWN );
return false;
}
bool TiXmlDocument::LoadFile( FILE* file, TiXmlEncoding encoding )
{
if ( !file )
{
SetError( TIXML_ERROR_OPENING_FILE, 0, 0, TIXML_ENCODING_UNKNOWN );
return false;
}
// Delete the existing data:
Clear();
location.Clear();
// Get the file size, so we can pre-allocate the string. HUGE speed impact.
long length = 0;
fseek( file, 0, SEEK_END );
length = ftell( file );
fseek( file, 0, SEEK_SET );
// Strange case, but good to handle up front.
if ( length <= 0 )
{
SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, TIXML_ENCODING_UNKNOWN );
return false;
}
// If we have a file, assume it is all one big XML file, and read it in.
// The document parser may decide the document ends sooner than the entire file, however.
TIXML_STRING data;
data.reserve( length );
// Subtle bug here. TinyXml did use fgets. But from the XML spec:
// 2.11 End-of-Line Handling
// <snip>
// <quote>
// ...the XML processor MUST behave as if it normalized all line breaks in external
// parsed entities (including the document entity) on input, before parsing, by translating
// both the two-character sequence #xD #xA and any #xD that is not followed by #xA to
// a single #xA character.
// </quote>
//
// It is not clear fgets does that, and certainly isn't clear it works cross platform.
// Generally, you expect fgets to translate from the convention of the OS to the c/unix
// convention, and not work generally.
/*
while( fgets( buf, sizeof(buf), file ) )
{
data += buf;
}
*/
#define BLANK_LINE_COMMENT_MAGIC "##BLANK-LINE##"
bool bUseBlankLineMagic = true;
char* buf = new char[ length+1 ];
buf[0] = 0;
if ( fread( buf, length, 1, file ) != 1 ) {
delete [] buf;
SetError( TIXML_ERROR_OPENING_FILE, 0, 0, TIXML_ENCODING_UNKNOWN );
return false;
}
const char* lastPos = buf;
const char* p = buf;
bool bInComment = false;
bool bInTag = false;
bool bEmptyLine = false;
int iNewLineCount = 0;
bool bOnlyWhiteSpaceChars = false;
buf[length] = 0;
while( *p ) {
assert( p < (buf+length) );
if ( *p == 0xa || *p == 0xd )
{
if ( bEmptyLine && !bInTag && !bInComment && bUseBlankLineMagic )
iNewLineCount++;
bEmptyLine = true;
}
if ( *p == 0xa ) {
// Newline character. No special rules for this. Append all the characters
// since the last string, and include the newline.
data.append( lastPos, (p-lastPos+1) ); // append, include the newline
++p; // move past the newline
lastPos = p; // and point to the new buffer (may be 0)
assert( p <= (buf+length) );
}
else if ( *p == 0xd ) {
// Carriage return. Append what we have so far, then
// handle moving forward in the buffer.
if ( (p-lastPos) > 0 ) {
data.append( lastPos, p-lastPos ); // do not add the CR
}
data += (char)0xa; // a proper newline
if ( *(p+1) == 0xa ) {
// Carriage return - new line sequence
p += 2;
lastPos = p;
assert( p <= (buf+length) );
}
else {
// it was followed by something else...that is presumably characters again.
++p;
lastPos = p;
assert( p <= (buf+length) );
}
}
else
if ( *p == ' ' || *p == '\t' )
{
// White space
++p;
}
else {
if ( strncmp ( p, "<!--", 4 ) == 0 )
bInComment = true; // Entering comment
else
if ( strncmp ( p, "-->", 3 ) == 0 )
bInComment = false; // Leaving comment
if ( strncmp ( p, "<", 1 ) == 0 )
{
bInTag = true; // Entering tag
// If preceeding text contains only white space, save the blank lines as comments
if ( bOnlyWhiteSpaceChars )
{
for ( int i = 0 ; i < iNewLineCount ; i++ )
{
data.append( "<!--" BLANK_LINE_COMMENT_MAGIC "-->" );
}
bOnlyWhiteSpaceChars = false;
iNewLineCount = 0;
}
}
else
if ( strncmp ( p, ">", 1 ) == 0 )
{
bInTag = false; // Leaving tag
// Start of possible white space area containing blank lines
bOnlyWhiteSpaceChars = true;
iNewLineCount = 0;
}
else
bOnlyWhiteSpaceChars = false;
bEmptyLine = false;
++p;
}
}
// Handle any left over characters.
if ( p-lastPos ) {
data.append( lastPos, p-lastPos );
}
delete [] buf;
buf = 0;
Parse( data.c_str(), 0, encoding );
if ( Error() )
return false;
else
return true;
}
void TiXmlBase::PutString( const TIXML_STRING& str, TIXML_STRING* outString )
{
int i=0;
while( i<(int)str.length() )
{
unsigned char c = (unsigned char) str[i];
if ( c == '&'
&& i < ( (int)str.length() - 2 )
&& str[i+1] == '#'
&& str[i+2] == 'x' )
{
// Hexadecimal character reference.
// Pass through unchanged.
// © -- copyright symbol, for example.
//
// The -1 is a bug fix from Rob Laveaux. It keeps
// an overflow from happening if there is no ';'.
// There are actually 2 ways to exit this loop -
// while fails (error case) and break (semicolon found).
// However, there is no mechanism (currently) for
// this function to return an error.
while ( i<(int)str.length()-1 )
{
outString->append( str.c_str() + i, 1 );
++i;
if ( str[i] == ';' )
break;
}
}
else if ( c == '&' )
{
outString->append( entity[0].str, entity[0].strLength );
++i;
}
else if ( c == '<' )
{
outString->append( entity[1].str, entity[1].strLength );
++i;
}
else if ( c == '>' )
{
outString->append( entity[2].str, entity[2].strLength );
++i;
}
else if ( c == '\"' )
{
outString->append( entity[3].str, entity[3].strLength );
++i;
}
else if ( c == '\'' )
{
outString->append( entity[4].str, entity[4].strLength );
++i;
}
else if ( c == 0x0d && ( (int)str.length() - 1 )
&& str[i+1] == 0x0a)
{
// insert \n
*outString += '\n';
++i;
++i;
}
else if (isspace (c))
{
// pass through unchagend
*outString += (char) c; // somewhat more efficient function call.
++i;
}
else if (iscntrl (c)) // ( c < 32)
{
std::cerr << "WARNING: control character 0x"
<< std::hex << std::setw(2) << std::setfill('0') << (unsigned)c
<< " in text input at character "
<< std::dec << std::setw(0) << i
<< std::endl;
++i;
}
else
{
// just pass these through, since we've
// declared an encoding that presumably allows
// them
*outString += (char) c; // somewhat more efficient function call.
//char realc = (char) c;
//outString->append( &realc, 1 );
++i;
}
}
}
bool TiXmlDocument::LoadFile( const char* filename, TiXmlEncoding encoding )
{
// Delete the existing data:
Clear();
location.Clear();
// There was a really terrifying little bug here. The code:
// value = filename
// in the STL case, cause the assignment method of the std::string to
// be called. What is strange, is that the std::string had the same
// address as it's c_str() method, and so bad things happen. Looks
// like a bug in the Microsoft STL implementation.
// See STL_STRING_BUG above.
// Fixed with the StringToBuffer class.
value = filename;
// reading in binary mode so that tinyxml can normalize the EOL
FILE* file = fopen( value.c_str (), "rb" );
if ( file )
{
// Get the file size, so we can pre-allocate the string. HUGE speed impact.
long length = 0;
fseek( file, 0, SEEK_END );
length = ftell( file );
fseek( file, 0, SEEK_SET );
// Strange case, but good to handle up front.
if ( length == 0 )
{
fclose( file );
return false;
}
// If we have a file, assume it is all one big XML file, and read it in.
// The document parser may decide the document ends sooner than the entire file, however.
TIXML_STRING data;
data.reserve( length );
// Subtle bug here. TinyXml did use fgets. But from the XML spec:
// 2.11 End-of-Line Handling
// <snip>
// <quote>
// ...the XML processor MUST behave as if it normalized all line breaks in external
// parsed entities (including the document entity) on input, before parsing, by translating
// both the two-character sequence #xD #xA and any #xD that is not followed by #xA to
// a single #xA character.
// </quote>
//
// It is not clear fgets does that, and certainly isn't clear it works cross platform.
// Generally, you expect fgets to translate from the convention of the OS to the c/unix
// convention, and not work generally.
/*
while( fgets( buf, sizeof(buf), file ) )
{
data += buf;
}
*/
char* buf = new char[ length+1 ];
buf[0] = 0;
if ( fread( buf, length, 1, file ) != 1 ) {
//if ( fread( buf, 1, length, file ) != (size_t)length ) {
SetError( TIXML_ERROR_OPENING_FILE, 0, 0, TIXML_ENCODING_UNKNOWN );
fclose( file );
return false;
}
fclose( file );
const char* lastPos = buf;
const char* p = buf;
buf[length] = 0;
while( *p ) {
assert( p < (buf+length) );
if ( *p == 0xa ) {
// Newline character. No special rules for this. Append all the characters
// since the last string, and include the newline.
data.append( lastPos, p-lastPos+1 ); // append, include the newline
++p; // move past the newline
lastPos = p; // and point to the new buffer (may be 0)
assert( p <= (buf+length) );
}
else if ( *p == 0xd ) {
// Carriage return. Append what we have so far, then
// handle moving forward in the buffer.
if ( (p-lastPos) > 0 ) {
data.append( lastPos, p-lastPos ); // do not add the CR
}
data += (char)0xa; // a proper newline
if ( *(p+1) == 0xa ) {
// Carriage return - new line sequence
p += 2;
lastPos = p;
assert( p <= (buf+length) );
}
else {
// it was followed by something else...that is presumably characters again.
++p;
lastPos = p;
assert( p <= (buf+length) );
}
}
else {
++p;
}
}
// Handle any left over characters.
if ( p-lastPos ) {
data.append( lastPos, p-lastPos );
}
delete [] buf;
buf = 0;
Parse( data.c_str(), 0, encoding );
if ( Error() )
return false;
else
return true;
}
SetError( TIXML_ERROR_OPENING_FILE, 0, 0, TIXML_ENCODING_UNKNOWN );
return false;
}
bool TiXmlDocument::LoadFile( FILE* file, TiXmlEncoding encoding )
{
if ( !file )
{
SetError( TIXML_ERROR_OPENING_FILE, 0, 0, TIXML_ENCODING_UNKNOWN );
return false;
}
Clear();
location.Clear();
long length = 0;
fseek( file, 0, SEEK_END );
length = ftell( file );
fseek( file, 0, SEEK_SET );
if ( length <= 0 )
{
SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, TIXML_ENCODING_UNKNOWN );
return false;
}
TIXML_STRING data;
data.reserve( length );
char* buf = new char[ length+1 ];
buf[0] = 0;
if ( fread( buf, length, 1, file ) != 1 ) {
delete [] buf;
SetError( TIXML_ERROR_OPENING_FILE, 0, 0, TIXML_ENCODING_UNKNOWN );
return false;
}
const char* lastPos = buf;
const char* p = buf;
buf[length] = 0;
while( *p ) {
assert( p < (buf+length) );
if ( *p == 0xa ) {
data.append( lastPos, (p-lastPos+1) );
++p;
lastPos = p;
assert( p <= (buf+length) );
}
else if ( *p == 0xd ) {
if ( (p-lastPos) > 0 ) {
data.append( lastPos, p-lastPos );
}
data += (char)0xa;
if ( *(p+1) == 0xa ) {
p += 2;
lastPos = p;
assert( p <= (buf+length) );
}
else {
++p;
lastPos = p;
assert( p <= (buf+length) );
}
}
else {
++p;
}
}
if ( p-lastPos ) {
data.append( lastPos, p-lastPos );
}
delete [] buf;
buf = 0;
Parse( data.c_str(), 0, encoding );
if ( Error() )
return false;
else
return true;
}
void TiXmlBase::EncodeString( const TIXML_STRING& str, TIXML_STRING* outString )
{
int i=0;
while( i<(int)str.length() )
{
unsigned char c = (unsigned char) str[i];
if ( c == '&'
&& i < ( (int)str.length() - 2 )
&& str[i+1] == '#'
&& str[i+2] == 'x' )
{
while ( i<(int)str.length()-1 )
{
outString->append( str.c_str() + i, 1 );
++i;
if ( str[i] == ';' )
break;
}
}
else if ( c == '&' )
{
outString->append( entity[0].str, entity[0].strLength );
++i;
}
else if ( c == '<' )
{
outString->append( entity[1].str, entity[1].strLength );
++i;
}
else if ( c == '>' )
{
outString->append( entity[2].str, entity[2].strLength );
++i;
}
else if ( c == '\"' )
{
outString->append( entity[3].str, entity[3].strLength );
++i;
}
else if ( c == '\'' )
{
outString->append( entity[4].str, entity[4].strLength );
++i;
}
else if ( c < 32 )
{
char buf[ 32 ];
#if defined(TIXML_SNPRINTF)
TIXML_SNPRINTF( buf, sizeof(buf), "&#x%02X;", (unsigned) ( c & 0xff ) );
#else
sprintf( buf, "&#x%02X;", (unsigned) ( c & 0xff ) );
#endif
outString->append( buf, (int)strlen( buf ) );
++i;
}
else
{
*outString += (char) c;
++i;
}
}
}
bool TiXmlDocument::ReadFromMemory( const char* pBuf, size_t sz, TiXmlEncoding encoding)
{
// Delete the existing data:
Clear();
location.Clear();
// Get the file size, so we can pre-allocate the string. HUGE speed impact.
long length = (long) sz;
// Strange case, but good to handle up front.
if ( length == 0 )
{
SetError( TIXML_ERROR_DOCUMENT_EMPTY, 0, 0, TIXML_ENCODING_UNKNOWN );
return false;
}
// If we have a file, assume it is all one big XML file, and read it in.
// The document parser may decide the document ends sooner than the entire file, however.
TIXML_STRING data;
data.reserve( length );
char* buf = new char[ length+1 ];
memset(buf,0,length+1);
memcpy(buf, pBuf, length);
const char* lastPos = buf;
const char* p = buf;
buf[length] = 0;
while( *p ) {
assert( p < (buf+length) );
if ( *p == 0xa ) {
// Newline character. No special rules for this. Append all the characters
// since the last string, and include the newline.
data.append( lastPos, (p-lastPos+1) ); // append, include the newline
++p; // move past the newline
lastPos = p; // and point to the new buffer (may be 0)
assert( p <= (buf+length) );
}
else if ( *p == 0xd ) {
// Carriage return. Append what we have so far, then
// handle moving forward in the buffer.
if ( (p-lastPos) > 0 ) {
data.append( lastPos, p-lastPos ); // do not add the CR
}
data += (char)0xa; // a proper newline
if ( *(p+1) == 0xa ) {
// Carriage return - new line sequence
p += 2;
lastPos = p;
assert( p <= (buf+length) );
}
else {
// it was followed by something else...that is presumably characters again.
++p;
lastPos = p;
assert( p <= (buf+length) );
}
}
else {
++p;
}
}
// Handle any left over characters.
if ( p-lastPos ) {
data.append( lastPos, p-lastPos );
}
delete [] buf;
buf = 0;
Parse( data.c_str(), 0, encoding );
if ( Error() )
return false;
else
return true;
}
{
if ( strcmp( node->name.c_str(), name ) == 0 )
return node;
}
return 0;
}
*/
#ifdef TIXML_USE_STL
std::istream& operator>> (std::istream & in, TiXmlNode & base)
{
TIXML_STRING tag;
tag.reserve( 8 * 1000 );
base.StreamIn( &in, &tag );
base.Parse( tag.c_str(), 0, TIXML_DEFAULT_ENCODING );
return in;
}
#endif
#ifdef TIXML_USE_STL
std::ostream& operator<< (std::ostream & out, const TiXmlNode & base)
{
TiXmlPrinter printer;
printer.SetStreamPrinting();
base.Accept( &printer );
out << printer.Str();
return out;
}
{
if ( node->name == name )
return node;
}
return 0;
}
#ifdef TIXML_USE_STL
TIXML_ISTREAM & operator >> (TIXML_ISTREAM & in, TiXmlNode & base)
{
TIXML_STRING tag;
tag.reserve( 8 * 1000 );
base.StreamIn( &in, &tag );
base.Parse( tag.c_str(), 0 );
return in;
}
#endif
TIXML_OSTREAM & operator<< (TIXML_OSTREAM & out, const TiXmlNode & base)
{
base.StreamOut (& out);
return out;
}
#ifdef TIXML_USE_STL
std::generic_string & operator<< (std::generic_string& out, const TiXmlNode& base )
{
void TiXmlBase::EncodeString( const TIXML_STRING& str, TIXML_STRING* outString )
{
int i=0;
while( i<(int)str.length() )
{
unsigned char c = (unsigned char) str[i];
if ( c == '&'
&& i < ( (int)str.length() - 2 )
&& str[i+1] == '#'
&& str[i+2] == 'x' )
{
// Hexadecimal character reference.
// Pass through unchanged.
// © -- copyright symbol, for example.
//
// The -1 is a bug fix from Rob Laveaux. It keeps
// an overflow from happening if there is no ';'.
// There are actually 2 ways to exit this loop -
// while fails (error case) and break (semicolon found).
// However, there is no mechanism (currently) for
// this function to return an error.
while ( i<(int)str.length()-1 )
{
outString->append( str.c_str() + i, 1 );
++i;
if ( str[i] == ';' )
break;
}
}
else if ( c == '&' )
{
outString->append( entity[0].str, entity[0].strLength );
++i;
}
else if ( c == '<' )
{
outString->append( entity[1].str, entity[1].strLength );
++i;
}
else if ( c == '>' )
{
outString->append( entity[2].str, entity[2].strLength );
++i;
}
else if ( c == '\"' )
{
outString->append( entity[3].str, entity[3].strLength );
++i;
}
else if ( c == '\'' )
{
outString->append( entity[4].str, entity[4].strLength );
++i;
}
else if ( c < 32 )
{
// Easy pass at non-alpha/numeric/symbol
// Below 32 is symbolic.
char buf[ 32 ];
#if defined(TIXML_SNPRINTF)
TIXML_SNPRINTF( buf, sizeof(buf), "&#x%02X;", (unsigned) ( c & 0xff ) );
#else
sprintf( buf, "&#x%02X;", (unsigned) ( c & 0xff ) );
#endif
//*ME: warning C4267: convert 'size_t' to 'int'
//*ME: Int-Cast to make compiler happy ...
outString->append( buf, (int)strlen( buf ) );
++i;
}
else
{
//char realc = (char) c;
//outString->append( &realc, 1 );
*outString += (char) c; // somewhat more efficient function call.
++i;
}
}
}
bool TiXmlDocument::LoadFile( const TCHAR* filename )
{
// Delete the existing data:
Clear();
location.Clear();
// There was a really terrifying little bug here. The code:
// value = filename
// in the STL case, cause the assignment method of the std::generic_string to
// be called. What is strange, is that the std::generic_string had the same
// address as it's c_str() method, and so bad things happen. Looks
// like a bug in the Microsoft STL implementation.
// See STL_STRING_BUG above.
// Fixed with the StringToBuffer class.
value = filename;
FILE* file = generic_fopen( value.c_str (), TEXT("r") );
if ( file )
{
// Get the file size, so we can pre-allocate the generic_string. HUGE speed impact.
long length = 0;
fseek( file, 0, SEEK_END );
length = ftell( file );
fseek( file, 0, SEEK_SET );
// Strange case, but good to handle up front.
if ( length == 0 )
{
fclose( file );
return false;
}
// If we have a file, assume it is all one big XML file, and read it in.
// The document parser may decide the document ends sooner than the entire file, however.
TIXML_STRING data;
data.reserve( length );
const int BUF_SIZE = 2048;
TCHAR buf[BUF_SIZE];
while( generic_fgets( buf, BUF_SIZE, file ) )
{
data += buf;
}
fclose( file );
//input is in UTF-8, so transformation is needed to UTF-16 used by windows for TCHAR in unicode mode
std::vector<char> inputdataInUTF8(data.size()+1); //+1 for the null termination
size_t datalength = wcstombs(inputdataInUTF8.data(), data.c_str(), data.size());
int transformedDataCharCount = MultiByteToWideChar(CP_UTF8, 0, (LPCSTR)inputdataInUTF8.data(), -1, nullptr, 0);
std::vector<wchar_t> transformedData(transformedDataCharCount+1); //+1 for the null termination
transformedDataCharCount = MultiByteToWideChar(CP_UTF8, 0, (LPCSTR)inputdataInUTF8.data(), -1, transformedData.data(), transformedDataCharCount);
if(transformedDataCharCount > 0)
{
//replace the original data with the new tranformed one, on success ot transformation otherwise go with old style data
data.clear();
data = transformedData.data();
}
Parse( data.c_str(), 0 );
if ( Error() )
return false;
else
return true;
}
SetError( TIXML_ERROR_OPENING_FILE, 0, 0 );
return false;
}
const char* TiXmlElement::Parse( const char* p, TiXmlParsingData* data, TiXmlEncoding encoding )
{
p = SkipWhiteSpace( p, encoding );
TiXmlDocument* document = GetDocument();
if ( !p || !*p )
{
if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT, 0, 0, encoding );
return 0;
}
if ( data )
{
data->Stamp( p, encoding );
location = data->Cursor();
}
if ( *p != '<' )
{
if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT, p, data, encoding );
return 0;
}
p = SkipWhiteSpace( p+1, encoding );
// Read the name.
const char* pErr = p;
p = ReadName( p, &value, encoding );
if ( !p || !*p )
{
if ( document ) document->SetError( TIXML_ERROR_FAILED_TO_READ_ELEMENT_NAME, pErr, data, encoding );
return 0;
}
TIXML_STRING endTag ("</");
endTag += value;
endTag += ">";
// Check for and read attributes. Also look for an empty
// tag or an end tag.
while ( p && *p )
{
pErr = p;
p = SkipWhiteSpace( p, encoding );
if ( !p || !*p )
{
if ( document ) document->SetError( TIXML_ERROR_READING_ATTRIBUTES, pErr, data, encoding );
return 0;
}
if ( *p == '/' )
{
++p;
// Empty tag.
if ( *p != '>' )
{
if ( document ) document->SetError( TIXML_ERROR_PARSING_EMPTY, p, data, encoding );
return 0;
}
return (p+1);
}
else if ( *p == '>' )
{
// Done with attributes (if there were any.)
// Read the value -- which can include other
// elements -- read the end tag, and return.
++p;
p = ReadValue( p, data, encoding ); // Note this is an Element method, and will set the error if one happens.
if ( !p || !*p ) {
// We were looking for the end tag, but found nothing.
// Fix for [ 1663758 ] Failure to report error on bad XML
if ( document ) document->SetError( TIXML_ERROR_READING_END_TAG, p, data, encoding );
return 0;
}
// We should find the end tag now
if ( StringEqual( p, endTag.c_str(), false, encoding ) )
{
p += endTag.length();
return p;
}
else
{
if ( document ) document->SetError( TIXML_ERROR_READING_END_TAG, p, data, encoding );
return 0;
}
}
else
{
// Try to read an attribute:
TiXmlAttribute* attrib = new TiXmlAttribute();
if ( !attrib )
{
if ( document ) document->SetError( TIXML_ERROR_OUT_OF_MEMORY, pErr, data, encoding );
return 0;
}
attrib->SetDocument( document );
pErr = p;
p = attrib->Parse( p, data, encoding );
if ( !p || !*p )
{
if ( document ) document->SetError( TIXML_ERROR_PARSING_ELEMENT, pErr, data, encoding );
delete attrib;
return 0;
}
// Handle the strange case of double attributes:
#ifdef TIXML_USE_STL
TiXmlAttribute* node = attributeSet.Find( attrib->NameTStr() );
#else
TiXmlAttribute* node = attributeSet.Find( attrib->Name() );
#endif
if ( node )
{
node->SetValue( attrib->Value() );
delete attrib;
return 0;
}
attributeSet.Add( attrib );
}
}
return p;
}
// <-- Strange class for a bug fix. Search for STL_STRING_BUG
TiXmlBase::StringToBuffer::StringToBuffer( const TIXML_STRING& str )
: buffer(new char[str.length() + 1]) {
if ( buffer ) {
strcpy( buffer, str.c_str() );
}
}
