interblue.pl/modules/pshowimporter/classes/PShow_XMLReaderToSimpleXmlObjects.php

<?php

/**
 * File from http://PrestaShow.pl
 *
 * DISCLAIMER
 * Do not edit or add to this file if you wish to upgrade this module to newer
 * versions in the future.
 *
 *  @authors     PrestaShow.pl <kontakt@prestashow.pl>
 *  @copyright   2015 PrestaShow.pl
 *  @license     http://PrestaShow.pl/license
 */
class PShow_XMLReaderToSimpleXmlObjects
{

    /**
     * @var PShow_XMLReaderToSimpleXmlObjects
     */
    private static $instance = null;

    /**
     * @var string
     */
    public $filepath;

    /**
     * @var XMLReader
     */
    private $xml;

    /**
     * Cache for function xpath()
     * 
     * @var array 
     */
    protected $cache_xpath = array();

    /**
     * @param string $filename
     * @return PShow_XMLReaderToSimpleXmlObjects
     * @throws Exception
     */
    public static function getInstance($filename = null)
    {
        if (self::$instance === null || self::$instance->filepath != $filename) {
            if ($filename === null) {
                throw new Exception("SimpleXMLParser: filename not specified");
            }

            self::$instance = new PShow_XMLReaderToSimpleXmlObjects($filename);
        }

        return self::$instance;
    }

    /**
     * @param string $filename
     * @throws Exception
     */
    public function __construct($filename)
    {
        if (!file_exists($filename)) {
            throw new Exception("SimpleXMLParser: file not exists");
        }

        $this->filepath = $filename;
    }

    /**
     *
     * @param string $xpath
     * @param int $offset
     * @param int $limit
     * @param bool $onlyCount
     * @return int
     * @throws UnexpectedValueException
     */
    public function xpath($xpath, $offset = 0, $limit = null, $onlyCount = false)
    {
        $cache_key = "filepath:" . $this->filepath . ";xpath:" . $xpath . ";offset:" . $offset . ";limit:" . $limit . ";onlyCount:" . $onlyCount;
        $cache_key_md5 = md5($cache_key);
        if (in_array($cache_key_md5, $this->cache_xpath)) {
            return $this->cache_xpath[$cache_key_md5];
        }

        $results = array();
        $xpath = explode('/', $xpath);
        $objectTag = end($xpath);
        $n = 0;

        $config = PShow_Config::getFileConfig(pathinfo($this->filepath, PATHINFO_FILENAME));

        // re-encode file to prevent encoding errors
        if (isset($config['primary']['re_encode_utf8']) && $config['primary']['re_encode_utf8']) {
            PShow_File::re_encode_utf8_file($this->filepath);
            PShow_Log::add(pathinfo($this->filepath, PATHINFO_FILENAME) . '.log', 're-encoding file with UTF8');
        }

        $this->xml = new XMLReader();
        $this->xml->open($this->filepath, $this->getXMLEncoding($this->filepath), (LIBXML_PARSEHUGE | LIBXML_COMPACT | LIBXML_NOBLANKS |
            LIBXML_NOEMPTYTAG | LIBXML_NSCLEAN | LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NOCDATA));

        while ($this->xml->read() && $this->xml->name !== $objectTag);

        while (--$offset >= 0 && $this->xml->next($objectTag));

        while ($this->xml->name == $objectTag) {

            ++$n;

            if (!$onlyCount) {

                $node = @$this->xml->expand();

                if (!$node instanceof DOMElement) {
                    $size = memory_get_usage();
                    $size = max(0, (int) $size);
                    $units = array('B', 'KB', 'MB', 'GB', 'TB');
                    $power = $size > 0 ? floor(log($size, 1024)) : 0;
                    $memory_usage = number_format($size / pow(1024, $power), 2, '.', ',') . $units[$power];
                    $cpu_usage = sys_getloadavg();

                    PShow_Log::add(pathinfo($this->filepath, PATHINFO_FILENAME) . ".log", "Error in expanding XML; Elements count " . count($results) . ";  "
                        . "Tag: " . $this->xml->name . "; Memory usage: " . $memory_usage . "; "
                        . "CPU usage: " . ($cpu_usage[0]) . '%');

                    break;
                } else {
                    $doc = new DomDocument();
                    $node = $doc->importNode($node, true);

                    $results[] = simplexml_import_dom($node);

                    unset($node);
                }
            }

            $this->xml->next($objectTag);

            if ($limit !== null && --$limit <= 0) {
                break;
            }
        }

        $this->xml->close();

        if ($error = libxml_get_last_error()) {
            PShow_Log::add(pathinfo($this->filepath, PATHINFO_FILENAME) . ".log", "Error in XML reading: [" . $error->line . "] " . trim($error->message));
        }

        if (!$onlyCount) {
            $this->cache_xpath[$cache_key_md5] = $results;
        } else {
            $this->cache_xpath[$cache_key_md5] = $n;
        }

        unset($results);

        return $this->cache_xpath[$cache_key_md5];
    }

    /**
     * 
     * @param string $filepath
     * @return string|null
     */
    public function getXMLEncoding($filepath)
    {
        $output = array();
        @exec('file -i \'' . $filepath . '\'', $output);
        if (isset($output[0])) {
            $ex = explode('charset=', $output[0]);
            if (isset($ex[1])) {
                return $ex[1];
            }
        }

        return null;
    }

    /**
     * @return array
     */
    public function getAllTags()
    {
        $results = array();
        $limit = 100;
        $searchLimit = 100000;

        $actPath = array();

        $config = PShow_Config::getFileConfig(pathinfo($this->filepath, PATHINFO_FILENAME));

        // re-encode file to prevent encoding errors
        if (isset($config['primary']['re_encode_utf8']) && $config['primary']['re_encode_utf8']) {
            PShow_File::re_encode_utf8_file($this->filepath);
            PShow_Log::add(pathinfo($this->filepath, PATHINFO_FILENAME) . '.log', 're-encoded file with utf-8');
        }

        $this->xml = new XMLReader();
        $this->xml->open(
            $this->filepath, $this->getXMLEncoding($this->filepath), (LIBXML_PARSEHUGE | LIBXML_COMPACT | LIBXML_NOBLANKS |
            LIBXML_NOEMPTYTAG | LIBXML_NSCLEAN | LIBXML_NOCDATA | LIBXML_BIGLINES | LIBXML_NSCLEAN | LIBXML_PEDANTIC)
        );

        while ($this->xml->read()) {
            $actPath_str = (count($actPath)) ? implode(',', $actPath) . ',' : '';

            if ($this->xml->nodeType == XMLReader::ELEMENT && !in_array($actPath_str . $this->xml->name . ',', $results)) {
                array_push($results, $actPath_str . $this->xml->name . ',');

                if (--$limit <= 0) {
                    break;
                }
            }

            if ($this->xml->nodeType == XMLReader::ELEMENT && !$this->xml->isEmptyElement) {
                array_push($actPath, $this->xml->name);
            }

            if ($this->xml->nodeType == XMLReader::END_ELEMENT) {
                array_pop($actPath);
            }

            if (--$searchLimit <= 0) {
                break;
            }
        }

        $this->xml->close();

        return $results;
    }
}