XML parser with a functional approach

1.0.1 2015-01-13 22:36 UTC

This package is auto-updated.

Last update: 2024-06-25 19:12:18 UTC



Build Status

Scrutinizer Code Quality

Code Coverage

Functional XML-reader for PHP 5.3+.

A somewhat different approach to reading/parsing XML files with PHP, using a hierarchy of anonymous functions (closures) reflecting the hierarchy of the XML document itself.

This is useful when reading structured XML documents - e.g. XML documents with a predictable structure. It's probably less than enjoyable when reading unstructured documents, such as XHTML documents.

Parsing happens on-the-fly, e.g. avoiding the overhead of loading an entire document into memory and performing repetitive queries against it. This approach is memory efficient, enabling you to parse very large documents in a streaming fashion - it is not super fast (throughput ~500 KB/sec on my laptop) but XML parsing is never truly fast, so you should definitely always cache the parsed results.


Let's say you wish to read the following XML file:

<?xml version="1.0" encoding="UTF-8"?>
    <cat name="whiskers">
        <kitten name="mittens"/>
    <cat name="tinker">
        <kitten name="binky"/>

Your reader might look something like this:

$doc = new Parser();

$doc['cats/cat'] = function (Visitor $cat, $name) {
    echo "a cat named: {$name}\n";

    $cat['kitten'] = function ($name) {
        echo "a kitten named: {$name}\n";


The output would be this:

a cat named: whiskers
a kitten named: mittens
a cat named: tinker
a kitten named: binky

If it's not obvious, the path cats/cat designates a <cat> node inside a <cats> node.

You can also match text-nodes, e.g. a path like foo/bar#text will match YO in <foo><bar>YO</bar></foo>.

And finally, you can use #end to match closing tags, if needed.

Incidentally, I don't actually have cats - but if I did, you can bet those would be their names.

See "test.php" and "example/cd_catalog.php" for more examples of how to use this.