PHP XML reader benchmarking

Readme.md

Benchmarking PHP XML readers

Benchmarking memory usage for the purpose of reading large XML files.

Usage

make test

Results

	runtime (s)	memory usage (MiB)	memory usage, real (MiB)
test-xmlreader.php	3.3585588932037	0.62660217285156	2
test-simplexml.php	2.3316519260406	0.61759948730469	2
test-simplexml-string.php	2.3231160640717	24.296783447266	25.90625

Makefile

.PHONY: test setup

testdata.xml:
	curl -sSL -o testdata.xml http://aiweb.cs.washington.edu/research/projects/xmltk/xmldata/data/nasa/nasa.xml

result.log: testdata.xml
	php runner.php test-xmlreader.php 10 > results.log
	php runner.php test-simplexml.php 10 >> results.log
	php runner.php test-simplexml-string.php 10 >> results.log

test: result.log

runner.php

<?php

include($argv[1]);

echo "Testing: " . $argv[1] . "\n";

$start_time = microtime(TRUE);

for ($i = 0; $i < $argv[2]; $i++) {
    test();
}

$end_time = microtime(TRUE);

echo "-------------\n";

echo "runtime:             " . $end_time - $start_time . "s\n";
echo "memory usage:        " . (memory_get_peak_usage(false) / 1024 / 1024) . " MiB\n";
echo "memory usage (real): " . (memory_get_peak_usage(true) / 1024 / 1024) . " MiB\n";

test-simplexml-string.php

<?php

/**
 * Control test - this should have high memory usage.
 */
function test(): void
{
    $reader = new SimpleXMLElement(file_get_contents('testdata.xml'), dataIsURL: false);

    $fd = fopen('/dev/null', 'w');
    foreach ($reader->dataset as $dataset) {
        fwrite($fd, $dataset->asXML());
    }
    fclose($fd);
}

test-simplexml.php

<?php

function test(): void
{
    $reader = new SimpleXMLElement('file://' . dirname(__FILE__) . '/testdata.xml', dataIsURL: true);

    $fd = fopen('/dev/null', 'w');
    foreach ($reader->dataset as $dataset) {
        fwrite($fd, $dataset->asXML());
    }
    fclose($fd);
}

test-xmlreader.php

<?php

function test(): void {
    $reader = XMLReader::open('testdata.xml');

    $fd = fopen('/dev/null', 'w');
    try {
        // Read the XML document node by node. When <dataset> element is
        // reached, read only sibling nodes until the end of the document.
        while ($reader->read()) {
            // Check if the current node is an element
            if ($reader->nodeType != XMLReader::ELEMENT) {
                continue;
            }

            switch ($reader->name) {
                case 'dataset':
                    // Read all sibling projects.
                    do {
                        fwrite($fd, $reader->readOuterXml());
                    } while ($reader->next('dataset'));
                    break;

                default:
                    break;
            }
        }
    } finally {
        $reader->close();
        fclose($fd);
    }
}