XML

• Usage
• Documentation
• Requirements
• Installation
• Performance
• License

Provides a developer friendly way of accessing nodes in very large XML documents.

Usage

<?php

use KHerGe\XML\FileReaderFactory;

// Open an existing and very large XML file.
$reader = (new FileReaderFactory())->open('/path/to/large.xml');

// Iterate through each node in the XML document.
foreach ($reader as $path => $node) {
    if ($node->isElement()) {
        echo $path, ' = ', $node->getValue(), "\n";
    }
}

Documentation

• FileReaderFactory - responsible for creating new instances of FileReader. This is the primary way of opening an XML file for reading. The Usage example is essentially what you need.
• NodeInterface - represents a node that was read from the XML document. You will be mostly interested in the methods and not the constants defined in the interface. Those constants are mainly for using an alternative implementation of the standard node representation class.

Requirements

• PHP 5.6+
  • libxml
  • xmlreader

Installation

composer require kherge/xml

Performance

Let's take the following example.

XMLReader

<?php

$reader = new XMLReader();
$reader->open('/path/to/large.xml');

while ($reader->read());

XML

<?php

use KHerGe\XML\FileReaderFactory;

$reader = (new FileReaderFactory())->open('/path/to/large.xml');

foreach ($reader as $element);

This library will perform roughly 25x slower than XMLReader if we simply stream through all of the elements without performing any work. There are a few important reasons why this is the case.

1. The path to each node is tracked.
2. The way data abstraction is handled.
3. The attributes are read for nodes that have them.

While 25x slower sounds a lot, you need to ask yourself how much of this library you would need to re-implement yourself anyway if you used XMLReader directly. The appeal of this library would be in how easy it can be to use it. In addition, the library is also designed so that various implementation details (e.g. the node representation class) can be replaced with your own implementation.

You will need to perform your own benchmarks, but you may end up breaking even if you use this library. However, if you do find any room for optimizations, please open a ticket, or better yet, submit a pull request!

License

This library is available under the MIT and Apache 2.0 licenses.