HTML with in tags

[steven smethurst]

Hello

Im an parsing the output from a database
one of the fields contain html data
When i try to read the data in that node i only get up to the first tag.

for example:
<node>
   <path>index</path>
   <htmldata>hello world this is a test of a bug in tinyXML <BR/>This part will never get shown
   </htmldata>
</node>

when i print out htmldata node i get "hello world this is a test of a bug in tinyXML"

instead of what i expected
"hello world this is a test of a bug in tinyXML <BR/>This part will never get shown"

I searched thou the forms but did not find anyone else with the same problem as me.

is this a bug?
is html tags with in side of nodes not supported by TinyXML ?
am doing else something wrong ?

Thank you for your time

[Lee Thomason]

At first I thought that wasn't valid XML...but I think it is. Odd looking, but syntactically correct.

Please file a bug!
Thanks, lee

[steven smethurst]

Done and i included an example of the problem

http://sourceforge.net/tracker/index.php?func=detail&aid=1003662&group_id=13559&atid=113559

[Ellers]

Works fine for me. Running the program below this is the output I get (it has indenting but sourceforge doesn't display it):

Loaded OK. Now, traverse...
Document
Element Named [node]
  Element Named [path]
   Text [index]
  Element Named [htmldata]
   Text [hello world this is a test of a bug in tinyXML]
   Element Named [BR]
   Text [This part will never get shown]
Press any key to continue

Within the element "htmldata" you have 3 nodes: a text node, a element node, and another text node.

Are you sure you're traversing the loaded document correctly?

#include "tinyxml.h"

void traverse( TiXmlNode * pNode, int indent = 0 )
{
    static const char* indentString = "                           ";
    static int indentStringLen = strlen( indentString );

    printf( "%s", &indentString[(indent<indentStringLen)?indentStringLen-indent:0] );

    switch ( pNode->Type( )) {
    case TiXmlNode::DOCUMENT:
        printf( "Document" );
        break;
    case TiXmlNode::ELEMENT:
        printf( "Element" );
        printf( " Named [%s]", pNode->ToElement( )->Value( ));
        break;
    case TiXmlNode::COMMENT:
        printf( "Comment" );
        break;
    case TiXmlNode::UNKNOWN:
        printf( "Unknown" );
        break;
    case TiXmlNode::TEXT:
        printf( "Text" );
        printf( " [" );
        printf( pNode->ToText( )->Value( ));
        printf( "]" );
        break;
    case TiXmlNode::DECLARATION:
        printf( "Declaration" );
        break;
    default:
        printf( "UNRECOGNISED TYPE" );
        break;
    }
    printf( "\n" );

    for ( TiXmlNode* pChild = pNode->FirstChild( ); 0 != pChild; pChild = pChild->NextSibling( )) {

        traverse( pChild, indent+1 );
    }
}

void test2( )
{
    TiXmlDocument doc("html_prob.xml");
    bool bLoad=doc.LoadFile();

    if(! bLoad) {
        printf( "FAILED to load\n" );
        return;
    }
    printf( "Loaded OK. Now, traverse...\n" );

    traverse( &doc );
}

int main(int argc, char* argv[])
{
    test2( );
    return 0;
}

[steven smethurst]

when I print out the html_data node i only get the start of the node up until the first xhtml tag <br/>

this is proven by my test and by yours

"Text [hello world this is a test of a bug in tinyXML]
Element Named [BR]
Text [This part will never get shown]"

There is some Text
Then a Element Named BR
Then there is more Text

What I expected to happen is that it would print all the data between the start and end tag of html_data

"Text [hello world this is a test of a bug in tinyXML <BR/> this part will never get shown]"

In other words this is what I was thinking
=== XML file ===
<node>
<record>
    <si_>1058</si_>
    <notes>Something</notes>
</record>
</node>
=== ===

Getting the value of the node record would give me
Text [<si_>1058</si_>\n<notes>Something</notes>]

But that is wrong, and I was mistaken in my understanding of TinyXML

I guess my question should have been
How do I get the text and sub tags beneath a node as text?

For example:
How could I get the this as an output if I had a cursor on the record node?
<si_>1058</si_>\n<notes>Something</notes>

Thank you for your time.

[Ellers]

While I'm thinking of it, "embedding" html within xml can be troublesome. Browsers (and people that write html) tend to write html in its original malform-able way. If the html is strict xhtml then you'll be ok, but...

it may be easier/safer to put the whole html into the xml as one text string, with "&lt" instead of the opening tag char etc. The end-user should never know the difference and you won't have to worry about the html corrupting your xml stream.

[steven smethurst]

i agree is would be a lot safer if i COULD change all the preexisting tags to "&lt"
but thats not the case.

i get this output from a database in XML form
i do not have control over how it is formed.

I do know that the output will be xhtml strict thou, so i do not have to worry about malformed tags.

[Ellers]

If you're stuck with the embedding approach then your best (only?) option is the way I described. A traversal of the revelant parts of the three, converting the DOM back to an xml char string/stream. If you're expecting huge amounts of data you could look into using a SAX parser. You'll still need to be careful with your coding, but you'll basically do

- read nodes until start of html block
- during html just accumulate the text
- at the end of the html section go back to parsing as normal.

Personally I prefer DOM, not SAX, but its partly personal preference, and partly problem domain...

[Ellers]

TinyXml is a DOM processing XML parser. It is doing exactly what it should as a DOM processor - building a tree of nodes that exactly describe the XML document you provided.

The 'record' element contains:
- an element named 'si_' which itself contains one text node
- an element called 'nodes' which also contains one text node.

What I think you are asking is how to write code to:
- "beginning at node 'record', traverse the tree and CONVERT all contained nodes into a std::string (or stream) re-writing the DOM into character XML data"

This can be done by modifying my traverse() function from above.

I haven't compiled this, but it will be something like:

void traverseAppend( TiXmlNode * pNode, std::string & s )
{
    switch ( pNode->Type( )) {

    case TiXmlNode::DOCUMENT:
        // NOP
        break;
    case TiXmlNode::ELEMENT:
        s.append( "<" );
        s.append( pNode->ToElement( )->Value( ));
        s.append( ">" );
        break;
    case TiXmlNode::COMMENT:
        s.append( "<!--" );
        s.append( pNode->ToComment()->Value());
        s.append( "-->" );
        break;
    case TiXmlNode::UNKNOWN:
        //printf( "Unknown" );
        break;
    case TiXmlNode::TEXT:
        s.append( pNode->ToText( )->Value( ));
        break;
    case TiXmlNode::DECLARATION:
        //printf( "Declaration" );
        break;
    default:
        //printf( "UNRECOGNISED TYPE" );
        break;
    }

    for ( TiXmlNode* pChild = pNode->FirstChild( ); 0 != pChild; pChild = pChild->NextSibling( ))
    {
        traverseAppend( pChild, s );
    }
}

std::string astring;
traverseAppend( recordNodeHandle, astring )
// astring is now "<si_>1058</si_><notes>Something</nodes>"

You should understand that this is not a TinyXml thing, it is an *XML* thing, and it will be the same problem with any DOM processing library. With a SAX parser it will be slightly different, but essentially the same problem. True, MSXML or something like that might provide a convenience method like

astring = recordNode->asXml()

which does the traversal exactly as I've shown, writing the node into a string.

Personally, I don't like embedding one XML grammar in another because of these problems which keep coming up. I read some articles about embedding xhtml in xml some time ago for work and the general consensus seemed to be either:

- put the xhtml in as a single text code, e.g. escaping special chars like &lt;html&gt; etc
or
- MIME/Base64 encode it (simple and effective)

HTH