PHP Split XML based on multiple nodes

I honestly tried to find a solution for php, but a lot of threads sound similar, but are not applicable for me or are for completely different languages.

I want to split an xml file based on nodes. Ideally multiple nodes, but of course one is enough and could be applied multiple times.

e.g. I want to split this by the tag <thingy> and <othernode>:

<root>
   <stuff />
   <thingy><othernode>one</othernode></thingy>
   <thingy><othernode>two</othernode></thingy>
   <thingy>
      <othernode>three</othernode>
      <othernode>four</othernode>
   </thingy>
   <some other data/>
</root>

Ideally I want to have 4 xmlstrings of type:

<root>
   <stuff />
   <thingy><othernode>CONTENT</othernode></thingy>
   <some other data/>
</root>

With CONTENT being one, two, three and four. Plottwist: CONTENT can also be a whole subtree. Of course it all also can be filled with various namespaces and tag prefixes (like <q1:node/>. Formatting is irrelevant for me.

• I tried SimpleXml, but it lacks the possiblity to write into dom easily
• I tried DomDocument, but all what I do seems to destroy some links/relation of parent/child nodes in some way.
• I tried XmlReader/Writer, but that is extremely hard to maintain and combine (at least for me).

---

So far my best guess is something with DomDocument, node cloning and removing everything but one node?

Answer

Interesting question.

If I get it right, it is given that <othernode> is always a child of <thingy> and the split is for each <othernode> at the place of the first <thingy> in the original document.

DOMDocument appeared useful in this case, as it allows to easily move nodes around – including all its children.

Given the split on a node-list (from getElementsByTagName()):

echo "---n";
foreach ($split($doc->getElementsByTagName('othernode')) as $doc) {
    echo $doc->saveXML(), "---n";
}

When moving all <othernode> elements into a DOMDocumentFragement of its own while cleaning up <thingy> parent elements when emptied (unless the first anchor element) and then temporarily bring each of them back into the DOMDocument:

$split = static function (DOMNodeList $nodes): Generator {
    while (($element = $nodes->item(0)) && $element instanceof DOMElement) {
        $doc ??= $element->ownerDocument;
        $basin ??= $doc->createDocumentFragment();
        $anchor ??= $element->parentNode;
        [$parent] = [$element->parentNode, $basin->appendChild($element)];
        $parent->childElementCount || $parent === $anchor || $parent->parentNode->removeChild($parent);
    }

    if (empty($anchor)) {
        return;
    }

    assert(isset($basin, $doc));

    while ($element = $basin->childNodes->item(0)) {
        $element = $anchor->appendChild($element);
        yield $doc;
        $anchor->removeChild($element);
    }
};

This results in the following split:

---
<?xml version="1.0"?>
<root>
   <stuff/>
   <thingy><othernode>one</othernode></thingy>
   
   
   <some other="data"/>
</root>
---
<?xml version="1.0"?>
<root>
   <stuff/>
   <thingy><othernode>two</othernode></thingy>
   
   
   <some other="data"/>
</root>
---
<?xml version="1.0"?>
<root>
   <stuff/>
   <thingy><othernode>three</othernode></thingy>
   
   
   <some other="data"/>
</root>
---
<?xml version="1.0"?>
<root>
   <stuff/>
   <thingy><othernode>four</othernode></thingy>
   
   
   <some other="data"/>
</root>
---