I'm serious, this is exactly what QueryPath was written for. The QueryPath code for what you are asking would be: foreach (qp('text.xml', 'cn') as $cn) { // print an attr dpm('doc attr: ' . $cn->attr('doc')); // print the PCDATA text dpm('text: ' . $cn->text()); } Since QueryPath takes any old CSS3 selectors, you could do more complicated queries if you wanted. For example, if you only wanted CN elements where the doc attr was true, you'd just do this: foreach (qp('text.xml', 'cn[doc=true]') as $cn) { dpm('text: ' . $cn->text()); } Not to mention that QueryPath is actually faster than most SimpleXML operations because it traverse the DOM tree more efficiently. And you can download QueryPath just like any other Drupal module. Honestly, if you're not keen on the QueryPath idea, you should just use PHP's DOM interface. It's more efficient than SimpleXML, which is a HORRIBLE data structure for working with XML anyway. It's inefficient and it's full of the sorts of oddities that you are describing right now. On Thu, Mar 24, 2011 at 8:58 PM, <jeff@ayendesigns.com> wrote:

Thanks for the replies...but no joy yet. The XML data looks like this (I'm minimizing it)

<?xml version="1.0" encoding="UTF-8" standalone="no"?> <document>     <channels>         <channel>             <episode>                 <episode-content>                     <doc>                         <cncpts>                             <cn doc="true" er="164" kw="false" ne="true" sc="9.647706" sr="164" t="unk">here I am</cn>

and I have

    $t = simplexml_load_file('text.xml','SimpleXMLElement', LIBXML_NOCDATA);     dsm($t);     foreach ($t->channels->channel->nlp->episode->{episode-content}->doc->cncpts->cn as $k) {         $dsm($k);     }

The first dsm immediately after loading the file does not show the seven 'cn' elements that I have, and the second is empty because the foreach complains.

-- http://technosophos.com http://querypath.org

Oh, good. I hate to see anyone agonize over SimpleXML. Matt On Thu, Mar 24, 2011 at 9:39 PM, <jeff@ayendesigns.com> wrote:

Matt, perfecto! Worked like a charm the first time. I had a chuckle...your second message came while I was changing the code to qp. It wasn't any discomfort or raised eyebrow on my part, I tried the simplexml stuff again first since it was already coded that way.

Thanks!!

On 03/24/2011 10:20 PM, Matt wrote:

...

I'm serious, this is exactly what QueryPath was written for. The QueryPath code for what you are asking would be:

-- http://technosophos.com http://querypath.org

On 3/24/11 10:58 PM, David Metzler wrote:

Regarding introspection..... Not quite true. I use print_r on simplexml objects all the time with dsm.

drupal_set_message('<pre>' . print_r($xml,1) .'</pre>');

Hmm. That's never worked for me, and the note on php.net suggests it doesn't work for others. But if it works for you, great. I do want to note, though, that dsm() is not a shorthand for drupal_set_message(); it's a completely different function in the devel module.

although it sounds like the drupal community is pretty "down" on its use.

For my part, I didn't mean to suggest I'm "down" on SimpleXML, rather that it may be working fine (though it sounds like it wasn't for Jeff) even if inspection suggests otherwise.

I'll start giving qp a look , but this community has frowned on bundling third party libraries with drupal contrib modules in the past.

Kevin O has already pointed out QueryPath is okay on licensing, but that's not enough to make it okay for Drupal.org. The policy against hosting third party work on Drupal.org goes beyond licensing concerns. See: http://drupal.org/node/422996 That said, QueryPath is _not_ third party. The maintainer of the module on Drupal.org is the lead developer of the QueryPath project at querypath.org. It's first party. -- Scott Reynen MakeDataMakeSense.com

If you wanted, you can avoid the children() calls by writing a slightly more complex selector in qp(). You can use CSS3-style selectors to get the right relationships. Example: $xml = '<doc> <a> <b> <c>foo</c> </b> <c>another c</c> </a> <c>a third c</c> </doc>' // This will be 3, since there are 3 <c> elements qp($xml, 'c')->size() // This will match both of the <c> tags underneath the <a>, so it will be 2. qp($xml 'a c')->size() // This will match just the <c>'s that are direct descendants of <b>'s that are direct descendants of <a>'s. qp($xml, 'a>b>c')->size(); The selector syntax is exactly the same as what you would use if you were writing a stylesheet: a>b>c { color: red; } Matt On Fri, Mar 25, 2011 at 8:29 AM, <jeff@ayendesigns.com> wrote:

I hadn't been concluding I had trouble with simplexml only because dsm() didn't show anything, more that dsm reinforced the situation. I was referring to what should have been the element in code and coming up empty, or with in error, in the case of foreach.

Xpath doesn't work for some reason. I presume there could be something unusual with the file layout, since it's full of cpath stuff and I'm excerpting it here. It might, in part, have something to do with some of the path names being hyphenated. I know php has a problem with that when creating an object reference unless it is encased in {}.

Not only does querypath work with the tag name, but it worked too well at first. There are, apparently, other such tags elsewhere in the data, and I was receiving output for all of them. It turns out you can do some wonderful things with querypath. First, I provided further context by using:

       foreach (qp('text.xml', parent-tag-name)->children as $item)

but the problem with that was I received all the children, which included the tags I needed, but also others. So, you can also do this:

       foreach (qp('text.xml', parent-tag-name)->children(child-tag-name) as $item)

and then only receive the correct tags in the correct concept, such as xpath would have done with parent-tag/child-tag had it worked in this case.

Jeff

-- http://technosophos.com http://querypath.org