How to locate homescreen variables within XML text blocks

[Guest sullivanpt]

I know almost nothing about the MS XML document object model. I knew that a homescreen plugin was passed two IXMLDOMNode as parameters, and that every parsing example I could find used get_text to grab the text inside a label. All fine and dandy, except I wanted to implement variables within the label just like the calendar and messaging plug-ins do. (e.g. , etc.). The get_text function just strips these tags out, not the desired result.

I spent a good many hours on this one and finally came to a solution. It may not be the "correct" solution. I'm posting this here in the hopes of saving someone else a few dozen hours.

The trick (obvious to most maybe?) is to parse the tree all the way down, to the "text" node's children. There I found the text broken up and ordered, with the parsed XML variables in between.

Example:

SMS ()

actually parses as:

lv 1, nd 1: label :: SMS ()

lv 2, nd 0: text :: SMS ()

lv 3, nd 0: (null) :: SMS (

lv 3, nd 1: unreadSMS ::

lv 3, nd 2: (null) :: )

You can see that applying get_text to the "label" or "text" IXMLDOMNode (as they do in the samples) is useless. I needed to apply get_text to the unnamed nodes beneath the "text" attribute.

I wrote the following function to help me figure this out:

void ParseAndDumpTree(IXMLDOMNode* pnodeParams,int level)

{

int node = 0;

BSTR bst,text;

IXMLDOMNode* child = NULL;

while(pnodeParams)

{

if(!FAILED(pnodeParams->get_baseName(&bst)))

{

if(!FAILED(pnodeParams->get_text(&text)))

{

DebugOut (TEXT("lv %d, nd %d: %s :: %sn"),level,node,bst,text);

SysFreeString(text); // can free NULL safely

}

else

DebugOut (TEXT("lv %d, nd %d: %sn"),level,node,bst);

SysFreeString(bst); // can free NULL safely

}

if(!FAILED(pnodeParams->get_firstChild(&child)))

ParseAndDumpTree(child,level+1);

if(FAILED(pnodeParams->get_nextSibling(&pnodeParams))) // memory hole

break;

node++;

}

}

[Guest bretto]

I had the same hassle so I just ended up grabbing the raw xml for the text node, stripping off the tags and doing a simple search and replace on the left over string. Not ideal I guess but effective enough the serve my purposes.

Cheers

Bretto

[Guest DocBru]

I took the same approach as you. Parse the text node's child nodes.

Another thing I was also struggling with is that get_text() trims the string, i.e. leadind and trailing spaces will be removed, which is annoying when you want to do variable substitution inside a string. I used get_nodeValue() instead.

Thanks

Bruno