From 4ec9f73cdf00ee111bcfdec699cdfbd6adde0176 Mon Sep 17 00:00:00 2001
From: Will Pazner <will.e.p@gmail.com>
Date: Tue, 23 Mar 2021 15:38:42 -0700
Subject: [PATCH] Handle raw VTK appended data in XML parser

VTK files can have "raw binary data" in the AppendedData section. This means the
file may not be technically valid XML. The data can also have null characters
which requires care when parsing using C-style strings.
---
 general/tinyxml2.cpp | 35 +++++++++++++++++++++++++++++++----
 general/tinyxml2.h   |  8 ++++++++
 2 files changed, 39 insertions(+), 4 deletions(-)

diff --git a/general/tinyxml2.cpp b/general/tinyxml2.cpp
index 13b5ce03a0..43f9b4c0b8 100644
--- a/general/tinyxml2.cpp
+++ b/general/tinyxml2.cpp
@@ -1088,10 +1088,10 @@ char* XMLNode::ParseDeep( char* p, StrPair* parentEndTag, int* curLineNumPtr )
             // Declarations are only allowed at document level
             //
             // Multiple declarations are allowed but all declarations
-            // must occur before anything else. 
+            // must occur before anything else.
             //
-            // Optimized due to a security test case. If the first node is 
-            // a declaration, and the last node is a declaration, then only 
+            // Optimized due to a security test case. If the first node is
+            // a declaration, and the last node is a declaration, then only
             // declarations have so far been added.
             bool wellLocated = false;
 
@@ -1566,7 +1566,9 @@ void XMLAttribute::SetAttribute( float v )
 // --------- XMLElement ---------- //
 XMLElement::XMLElement( XMLDocument* doc ) : XMLNode( doc ),
     _closingType( OPEN ),
-    _rootAttribute( 0 )
+    _rootAttribute( 0 ),
+    _appendedData( NULL ),
+    _appendedDataSize( 0 )
 {
 }
 
@@ -1578,6 +1580,7 @@ XMLElement::~XMLElement()
         DeleteAttribute( _rootAttribute );
         _rootAttribute = next;
     }
+    delete [] _appendedData;
 }
 
 
@@ -2053,6 +2056,27 @@ char* XMLElement::ParseDeep( char* p, StrPair* parentEndTag, int* curLineNumPtr
         return p;
     }
 
+    if ( XMLUtil::StringEqual( _value.GetStr(), "AppendedData", 12 )
+            && _rootAttribute
+            && XMLUtil::StringEqual( _rootAttribute->Name(), "encoding", 8 )
+            && XMLUtil::StringEqual( _rootAttribute->Value(), "raw", 3 ) ) {
+        char *appendedDataStart = p;
+        while ( p - _document->_charBuffer < _document->_len )
+        {
+            if ( *p == '<' && XMLUtil::StringEqual( p, "</AppendedData>", 15 ) ) {
+                _appendedDataSize = p - appendedDataStart;
+                _appendedData = new char[_appendedDataSize];
+                memcpy(_appendedData, appendedDataStart, _appendedDataSize);
+                break;
+            }
+            ++p;
+        }
+        if (_appendedData == NULL)
+        {
+            _document->SetError(XML_ERROR_PARSING_APPENDED_DATA, _parseLineNum, 0);
+        }
+    }
+
     p = XMLNode::ParseDeep( p, parentEndTag, curLineNumPtr );
     return p;
 }
@@ -2147,6 +2171,7 @@ XMLDocument::XMLDocument( bool processEntities, Whitespace whitespaceMode ) :
     _errorStr(),
     _errorLineNum( 0 ),
     _charBuffer( 0 ),
+    _len( 0 ),
     _parseCurLineNum( 0 ),
 	_parsingDepth(0),
     _unlinked(),
@@ -2366,6 +2391,7 @@ XMLError XMLDocument::LoadFile( FILE* fp )
     }
 
     _charBuffer[size] = 0;
+    _len = size;
 
     Parse();
     return _errorID;
@@ -2417,6 +2443,7 @@ XMLError XMLDocument::Parse( const char* p, size_t len )
     _charBuffer = new char[ len+1 ];
     memcpy( _charBuffer, p, len );
     _charBuffer[len] = 0;
+    _len = len;
 
     Parse();
     if ( Error() ) {
diff --git a/general/tinyxml2.h b/general/tinyxml2.h
index 003ef43eaf..c1ca1cf7ae 100644
--- a/general/tinyxml2.h
+++ b/general/tinyxml2.h
@@ -533,6 +533,7 @@ enum XMLError {
     XML_ERROR_PARSING_COMMENT,
     XML_ERROR_PARSING_DECLARATION,
     XML_ERROR_PARSING_UNKNOWN,
+    XML_ERROR_PARSING_APPENDED_DATA,
     XML_ERROR_EMPTY_DOCUMENT,
     XML_ERROR_MISMATCHED_ELEMENT,
     XML_ERROR_PARSING,
@@ -1665,6 +1666,9 @@ public:
     /// See InsertNewChildElement()
     XMLUnknown* InsertNewUnknown(const char* text);
 
+    char* GetAppendedData() const { return _appendedData; }
+
+    size_t GetAppendedDataSize() const { return _appendedDataSize; }
 
     // internal:
     enum ElementClosingType {
@@ -1698,6 +1702,9 @@ private:
     // because the list needs to be scanned for dupes before adding
     // a new attribute.
     XMLAttribute* _rootAttribute;
+
+    char*  _appendedData;
+    size_t _appendedDataSize;
 };
 
 
@@ -1938,6 +1945,7 @@ private:
     mutable StrPair	_errorStr;
     int             _errorLineNum;
     char*			_charBuffer;
+    size_t          _len;
     int				_parseCurLineNum;
 	int				_parsingDepth;
 	// Memory tracking does add some overhead.