euroset_home_parser.php

<?php

libxml_use_internal_errors(true);

class Spider {
    private $basicUrl = 'http://euroset.ru/';
    private $restrictedSectionTitles = [];
    const SECTION_MOBILE = 1;
    const SECTION_PAD = 2;
    const SECTION_MOBILE_TITLE = 'Мобильные телефоны';
    const SECTION_PAD_TITLE = 'Планшеты, Ноутбуки';
    const RESTRICTED_CLASS = 'section-title';
    const PRODUCTLIST_XPATH = './div[2]/div/div[1]/ul[1]/li/div/ul[1]/li/a[1]';
    const XML_PRICE_TAG = 'price';
    const XML_LINK_TAG = 'link';
    const XML_TITLE_TAG = 'model';
    const XML_ROOT_TAG = 'rrpmon';
    const DATE_FORMAT = 'Y-m-d H:i:s';

    public function __construct() {
        $this->restrictedSectionTitles = [
            self::SECTION_MOBILE => self::SECTION_MOBILE_TITLE,
            self::SECTION_PAD => self::SECTION_PAD_TITLE
        ];
    }

    private function getClassXpath($className) {
        return '//*[contains(concat(" ", normalize-space(@class), " "), "' . $className . '")]';
    }

    private function domainize($url) {
        return $this->basicUrl . ltrim($url, '/');
    }

    private function createXMLFromDict($data) {
        // create dom and root
        $dom = new DOMDocument('1.0', 'utf-8');
        $root = $dom->createElement(self::XML_ROOT_TAG);
        $root->setAttribute('shop', $this->basicUrl);
        $root->setAttribute('date', date(self::DATE_FORMAT));

        // create items
        foreach($data as $element) {
            $item = $dom->createElement('item');
            foreach($element as $key => $value) {
                // beautify data
                $value = trim($value);
                switch($key) {
                    case self::XML_LINK_TAG:
                        $value = $this->domainize($value);
                        break;

                    case self::XML_PRICE_TAG:
                        $value = floatval($value);
                        break;
                }
                $subItem = $dom->createElement($key, $value);
                $item->appendChild($subItem);
            }
            $root->appendChild($item);
        }

        // append root to xml and return
        $dom->appendChild($root);
        return $dom->saveXML();
    }

    public function createXMLFromUrl($pageUrl) {
        // create dom and xpath objects
        $dom = new DOMDocument();
        try {
            $dom->loadHTMLFile($this->domainize($pageUrl));
            $xpathDom = new DOMXPath($dom);
        }
        catch(Exception $e) {
            return 'Sorry, raised exception: ' . $e->getMessage();
        }

        // find restricted sections
        $elements = $xpathDom->query($this->getClassXpath(self::RESTRICTED_CLASS));
        $findedSections = [];
        $grabbedData = [];
        foreach($elements as $i => $element) {
            // go through restricted section titles and check
            foreach($this->restrictedSectionTitles as $type => $title) {
                if(strpos($element->nodeValue, $title) > -1) {
                    $findedSections[$type] = $element;
                    // find wrapper of title and product sections
                    $productParent = $element
                        ->parentNode
                        ->parentNode
                        ->parentNode;
                    // find product list items (first a tag actually)
                    $productItems = $xpathDom->query(self::PRODUCTLIST_XPATH, $productParent);
                    foreach($productItems as $_ => $item) {
                        $grabbedData[] = [
                            self::XML_PRICE_TAG => $item->getAttribute('data-product-price'),
                            self::XML_LINK_TAG => $item->getAttribute('href'),
                            self::XML_TITLE_TAG => $item->getAttribute('title')
                        ];
                    }
                }
            }
        }

        // hello, gc
        unset($dom);
        unset($xpathDom);

        // prepare xml and return it
        return $this->createXMLFromDict($grabbedData);
    }
}

$spiderInstance = new Spider();
print_r($spiderInstance->createXMLFromUrl('/'));