tinyXML or rapidXML crash

Question

When parsing with tinyXML or rapidXML I have such an error when "<" character is put into XML file:

Process returned -1073741819 (0xC0000005)   execution time : 2.335 s
Press any key to continue.

Do you know how to avoid this behavior?

#include <iostream>
#include "tinyxml2.h"

using namespace std;

int main()
{
    tinyxml2::XMLDocument doc;
    doc.LoadFile("my.xml");

    tinyxml2::XMLElement* element;
    tinyxml2::XMLNode* node;

    node = doc.FirstChildElement("root") -> FirstChildElement("sample");

    while (node != NULL)
    {
        cout << "--START--" << endl;
        element = node -> FirstChildElement("field0");

        while(element != NULL)
        {
            const char* title = element -> GetText();

            if (title != NULL)
                cout << ":: " << title << endl;
            else
                cout << ":: NULL" << endl;

            element = element -> NextSiblingElement();

        }

        cout << "---END---" << endl << endl;

        node = node -> NextSibling();

    }

    return 0;

}

my.xml file is something like that - take a look on <crash> markup or put only < character instead of <crash> - it leads to crash anyway:

<root>
<sample>
    <field0><crash>1</field0>
    <field1>2</field1>
    <field2>3</field2>
    <field3>4</field3>
    <field4>5</field4>
    <field5>6</field5>
    <field6>7</field6>
    <field7>8</field7>
</sample>
</root>

gdb output:

Program received signal SIGSEGV, Segmentation fault.
0x0000000000402c5e in tinyxml2::XMLNode::FirstChildElement(char const*) const ()
(gdb) where
#0  0x0000000000402c5e in tinyxml2::XMLNode::FirstChildElement(char const*) const ()
#1  0x00000000004013f7 in tinyxml2::XMLNode::FirstChildElement(char const*) ()
#2  0x0000000000401236 in main ()

I want to add that I've had the same with rapidXML library.

@neutrino: You need to be more specific. If that XML isn't causing the crash, then show us the XML that is. — Robert Harvey, Oct 05 '14 at 15:08
No, it's not OK, but if there's any hope at all of us helping you, we need a better understanding of what's going on. Show us the actual XML that is causing the crash. — Robert Harvey, Oct 05 '14 at 15:09
this xml file crashes when run with code presented above. Just "<1" also crashes (only < character). — neutrino, Oct 05 '14 at 15:11
Then you either need to find a better XML parser, or debug this one and find out why it's crashing. You could also run the XML through a validator first, to make sure it is valid XML before you try to parse it. — Robert Harvey, Oct 05 '14 at 15:12
http://www.w3schools.com/xml/xml_validator.asp this xml file is of course broken and not validated when using this validator. — neutrino, Oct 05 '14 at 15:22
but I don't know why it leads to crash the whole app - sometimes I can get a broken file - it shouldn't lead to problems. — neutrino, Oct 05 '14 at 15:23

score 2 · Accepted Answer · answered Oct 13 '14 at 19:39

The problem is yours, not theirs:-

 node = doc.FirstChildElement("root") -> FirstChildElement("sample");

If presented with an invalid XML file, doc.FirstChildElement("root") is likely to return NULL. And you then dereference it...

Neither parser can be expected to 'partially' parse invalid XML in this way. Try this instead.

 node = doc.FirstChildElement("root");
 if (node == NULL) 
    throw something;
 else
  node = node -> FirstChildElement("sample");

score 0 · Answer 2 · answered Oct 05 '14 at 15:41

0

Your options are:

Debug TinyXML, find out why it's crashing, and fix it (if you have the source code),
Find an XML validator library and verify that the XML is valid before attempting to parse it, or
Find or write an XML parser that doesn't crash.

answered Oct 05 '14 at 15:41

Robert Harvey

178,213
47
333
501

The 2nd suggestion is the best option. Writing a stable, portable and bug-free parser is not an easy task. Re-inventing a wheel should be the last option if its not for learning purposes. – Sceptical Jule Oct 05 '14 at 21:50

tinyXML or rapidXML crash

2 Answers2