Remember the other day, while talking about JSON, I mentioned that parsing XML in JavaScript cross browser was a tedious nightmare. Well, I've found a simple way to get the information you want from Domino's ?ReadViewEntries XML using JavaScript.

Take a URL like this:

Here's a cut-down snippet of the XML (restricted to 2 columns for the sake of brevity):

<?xml version="1.0" encoding="UTF-8"?>
 <viewentries toplevelentries="290">
  <viewentry position="113" unid="061...926" noteid="956">
   <entrydata columnnumber="0" name="$1">
    <text>Jake Howlett/ROCKALL</text>
   </entrydata>
   <entrydata columnnumber="1" name="$0">
    <text>Jake</text>
   </entrydata>
  </viewentry>
  <viewentry position="114" unid="791...7DC" noteid="9AE">
   <entrydata columnnumber="0" name="$1">
    <text>James Bond/ROCKALL</text>
   </entrydata>

Let's say we want to go through each of the ten rows and find the Notes username for each person. Here's the JavaScript to do that:

//assume root is the XML document object returned by an Ajax method!
var entries = root.getElementsByTagName("viewentry");
for (var i=0; i<entries.length; i++) {
 columns = entries[i].getElementsByTagName("text");
 alert(columns.item(0).firstChild.nodeValue);
}

Hopefully the code explains itself. It's worth noting though that this looks for the first text column. It wouldn't find the first date-time column as this is not returned as a <text> element but as a <datetime> element. If you're working with number columns or date columns you'd have to alter the getElementsByTagName() part. Note that, to get the third column, you'd go on to look for .item(2).

What might not make sense from this code is the .firstChild.nodeValue part that follows the getElementsByTagName bit. Surely the <text> element we're working on doesn't have a child and in turn this doesn't have a node to get a value of. Au contraire. This is where it all gets really confusing.

In the DOM everything is a node. Even when you've got a hold of the <text> node you're looking for the text within it is still another node. The problem herein is the way different browsers deals with this. Mozilla infamously deals with all whitespace as nodes. Because of these differences it's a nightmare to try and directly access a certain node within each result in the following way, which, in this case, would work, but only in IE:

Trust me. For the sake of your sanity you don't want to try messing with childNodes!! Stick with getElementsByTagName() if you know the type of the column you're looking for.

As a result of this eureka-like moment I've updated the NamePicker demo code and it now works in all browsers. You don't know how long I spent trying to work all this out. Now I prefer JSON more than ever. It's just that using ReadViewEntries XML makes a lot more sense some times.

Finally. While looking for a solution to this I managed to find this dubious Googlewhack, which will probably only last a day or so. Get it while it's hot.