How to use a combination of XMLReader and SimpleXMLElement to parse large XML files in PHP.

big-xml.php

<?php
/**
 * an example of how to read huge XML files relatively quickly and efficiently
 * using a few core PHP libraries.
 *
 */

// Assume your file is very large, 140MB or somethig like that
$fn = __DIR__ . '/some_file.xml';

// The tag we want to extract from the file
$tag = 'item';

// we'll use XMLReader to "parse" the large XML file directly because it doesn't
// load the entire tree into memory, just "tokenizes" it enough to deal with
$reader = new \XMLReader();

// now open our file
if (!$reader->open($fn)) {
    throw new \RuntimeException("Could not open {$fn} with XMLReader");
}

// loop though the file, read just advances to the next node.
// XMLReader isn't aware of any the document tree, so nodes get
// iterated over as they appear in the file. We'll just read until
// the end of the file.
while ($reader->read()) {

    // XMLReader::$name will contain the current tab name, check to see if it
    // matches the tag you're looking for. If it does, we can just iterate
    // over those tags using XMLReader::next().
    while ($tag === $reader->name) {

        // since XMLReader doesn't really supply us with much of a usable
        // API, we can convert the current node to an instace of `SimpleXMLElement`
        $elem = new \SimpleXMLElement($reader->readOuterXML());

        // now use SimpleXMLElement as you normally would.
        foreach ($elem->children() as $child) {
            echo $child->getName(), ': ', $child, PHP_EOL;
        }

        // Children in a certain namespace even.
        foreach ($elem->children('http://purl.org/dc/elements/1.1/') as $child) {
            echo "{http://purl.org/dc/elements/1.1/}", $child->getName(), ': ', $child, PHP_EOL;
        }

        // move on to the next one
        $reader->next($tag);
    }
}

Sure, you're reading the current node only. But that node could be massive. Here's an example of an XML with a bunch of nodes, most of them tiny, and then one massive one towards the end:

https://home.treasury.gov/policy-issues/financial-sanctions/specially-designated-nationals-list-data-formats-data-schemas

In fact, most large XMLs will have large nodes.