PHP DomXPath类代码示例

 /**
  * @param string $accountNumber
  * @param int|null $maxItems
  * @param \DateTime|null $dateFrom
  * @param \DateTime|null $dateTo
  * @return \mixed[]
  */
 public function getTransfers($accountNumber, $maxItems = null, DateTime $dateFrom = null, DateTime $dateTo = null)
 {
     $data = $this->getRawData($accountNumber, $dateFrom, $dateTo);
     $transferData = substr($data, strpos($data, '<table class=\'main\'>'));
     $transferData = substr($transferData, 0, strpos($transferData, '</table>') + 8);
     $dom = new \DOMDocument();
     $dom->loadHTML($transferData);
     $finder = new \DomXPath($dom);
     $nodes = $finder->query('//table[@class="main"]//tbody//tr');
     $transfers = array();
     /** @var \DOMElement $node */
     $i = 1;
     $length = $nodes->length;
     foreach ($nodes as $node) {
         $value = str_replace(',', '.', trim($node->childNodes->item(2)->nodeValue));
         $value = (double) str_replace(' ', '', $value);
         $transfers[] = array(\DateTime::createFromFormat('d.m.Y', $node->childNodes->item(0)->nodeValue), $value, $node->childNodes->item(12)->nodeValue, $node->childNodes->item(14)->nodeValue);
         if ($maxItems !== null && $i >= $maxItems) {
             break;
         }
         $i++;
         if ($i >= $length) {
             break;
         }
     }
     $stateData = substr($data, strpos($data, '<table class=\'summary\'>'));
     $stateData = substr($stateData, 0, strpos($stateData, '</table>') + 8);
     $dom = new \DOMDocument();
     $dom->loadHTML($stateData);
     $finder = new \DomXPath($dom);
     $nodes = $finder->query('//table[@class="summary"]//tbody//tr//td[2]');
     $value = trim($nodes->item(0)->nodeValue);
     $value = str_replace(',', '.', $value);
     return array('state' => (double) str_replace(' ', '', $value), 'transfers' => $transfers);
 }

 /**
  * Fetch and Parse a form for input fields
  *
  */
 public function getInputs($type = 'text')
 {
     $content = preg_replace("/&(?!(?:apos|quot|[gl]t|amp);|#)/", '&', $this->source);
     $doc = new DOMDocument();
     @$doc->loadHTML($content);
     $xpath = new DomXPath($doc);
     $list = array();
     if ($type == 'text') {
         $items = $xpath->query('//form//input | //form//select | //form//textarea');
         foreach ($items as $item) {
             if ($item->getAttribute('type') != 'submit' and $item->getAttribute('name') != 'Submit-button') {
                 if ($item->getAttribute('type') != 'hidden') {
                     $field = str_replace('[]', '', $this->has_attribute($item, 'name'));
                     $field = trim($field);
                     array_push($list, $field);
                 }
             }
         }
     } elseif ($type == 'hidden') {
         $items = $xpath->query('//input | //select');
         foreach ($items as $item) {
             if ($item->getAttribute('type') == 'hidden') {
                 array_push($list, array('name' => trim($this->has_attribute($item, 'name')), 'value' => trim($this->has_attribute($item, 'value'))));
             }
         }
     }
     return $list;
 }

 /**
  * 
  * Enter description here ...
  * @param unknown_type $url
  */
 public static function getProductComments($url)
 {
     header('Content-type: text/html; charset=utf-8');
     $url = "http://ormatek.com/products/980";
     $doc = file_get_contents($url);
     $doc = mb_convert_encoding($doc, 'HTML-ENTITIES', "UTF-8");
     $query = ".//*[@class='comment']";
     $dom = new DomDocument();
     libxml_use_internal_errors(true);
     $dom->loadHTML($doc);
     $xpath = new DomXPath($dom);
     $nodes = $xpath->query($query);
     $i = 0;
     if (!is_array($nodes)) {
         return null;
     }
     foreach ($nodes as $node) {
         $name = $node->getElementsByTagName("b")->item(0)->nodeValue;
         $text = $node->getElementsByTagName("p")->item(0)->nodeValue;
         $date = $node->getElementsByTagName("span")->item(0)->nodeValue;
         $rating = 0;
         $i++;
         $param[] = array('id' => $i, 'name' => $name, 'date' => $date, 'rating' => $rating, 'text' => $text);
     }
     return $param;
 }

 /**
  * Handle provided directory
  *
  * Optionally an existing result file can be provided
  *
  * If a valid file could be generated the file name is supposed to be
  * returned, otherwise return null.
  *
  * @param Project $project
  * @param string  $existingResult
  *
  * @return string
  */
 public function handle(Project $project, $existingResult = null)
 {
     if (!isset($project->analyzers['pdepend'])) {
         return;
     }
     $pdependResultFile = $project->dataDir . '/' . $project->analyzers['pdepend'];
     $document = new \DomDocument();
     $document->load($pdependResultFile);
     $xPath = new \DomXPath($document);
     foreach ($xPath->query('//package') as $packageNode) {
         $packageCommits = 0;
         foreach ($xPath->query('./class', $packageNode) as $classNode) {
             $fileNode = $xPath->query('./file', $classNode)->item(0);
             $file = $fileNode->getAttribute('name');
             $classCommits = $this->countGitChangesPerFileRange($project, $file, $classNode->getAttribute('start'), $classNode->getAttribute('end'));
             $packageCommits += $classCommits;
             $classNode->setAttribute('commits', $classCommits);
             foreach ($xPath->query('./method', $classNode) as $methodNode) {
                 $methodCommits = $this->countGitChangesPerFileRange($project, $file, $methodNode->getAttribute('start'), $methodNode->getAttribute('end'));
                 $methodNode->setAttribute('commits', $methodCommits);
             }
         }
         $packageNode->setAttribute('commits', $packageCommits);
     }
     $document->save($pdependResultFile);
     return null;
 }

 /**
  * Collects all translateable strings(from tags with 'trans' attribute) from html templates
  *
  * @param string $file absoulute path to file
  */
 function extractFromHtml($file)
 {
     $dom = new DomDocument();
     $dom->loadHTMLFile($file);
     $xpath = new DomXPath($dom);
     $nodes = $xpath->query("//*[@trans]");
     foreach ($nodes as $i => $node) {
         //            $info = "<{$node->tagName}> at $file";
         $info = "";
         if ($node->hasAttribute('context')) {
             $info = 'Context: ' . $node->getAttribute('context') . '.  ' . $info;
         }
         if ($node->hasAttribute('placeholder')) {
             $this->addTranslationStr($node->getAttribute('placeholder'), $info);
         }
         if ($node->hasAttribute('title')) {
             $this->addTranslationStr($node->getAttribute('title'), $info);
         }
         if ($node->hasAttribute('data-intro')) {
             $this->addTranslationStr($node->getAttribute('data-intro'), $info);
         }
         if ($node->nodeValue != '' && $node->getAttribute('trans') != 'attr-only') {
             $this->addTranslationStr($node->nodeValue, $info);
         }
     }
 }

 private static function _createBlockCode($code)
 {
     if (substr($code, 0, 5) == '<span') {
         $dom = new DomDocument();
         $dom->loadXML($code);
         $xpath = new DomXPath($dom);
         $parentSpan = $xpath->query('/span')->item(0);
         $block_code = self::$_dom->createElement('fo:inline');
         $block_code->setAttribute('color', substr($parentSpan->getAttributeNode('style')->value, 7, 7));
         $nodes = $xpath->query('/span/node()');
         foreach ($nodes as $node) {
             if ($node->nodeType == XML_ELEMENT_NODE) {
                 $child = self::$_dom->createElement('fo:inline', $node->nodeValue);
                 $child->setAttribute('color', substr($node->getAttributeNode('style')->value, 7, 7));
             } else {
                 $child = self::$_dom->importNode($node, true);
             }
             $block_code->appendChild($child);
         }
         if (preg_match("/^\\s+\$/", $block_code->firstChild->textContent)) {
             $block_code->removeChild($block_code->firstChild);
         }
     } else {
         $block_code = self::$_dom->createElement('fo:inline', $code);
     }
     return $block_code;
 }

    /**
     * test xml generation for IPhone
     */
    public function testSearch()
    {
        $doc = new DOMDocument();
        $doc->loadXML('<?xml version="1.0" encoding="utf-8"?>
			<!DOCTYPE AirSync PUBLIC "-//AIRSYNC//DTD AirSync//EN" "http://www.microsoft.com/">
			<Search xmlns="uri:Search">
				<Store>
					<Name>GAL</Name>
					<Query>Lars</Query>
					<Options>
						<Range>0-3</Range>
						<RebuildResults/>
						<DeepTraversal/>
					</Options>
				</Store>
			</Search>
		');
        $search = new Syncroton_Command_Search($doc, $this->_device, null);
        $search->handle();
        $responseDoc = $search->getResponse();
        #$responseDoc->formatOutput = true; echo $responseDoc->saveXML();
        $xpath = new DomXPath($responseDoc);
        $xpath->registerNamespace('Search', 'uri:Search');
        $nodes = $xpath->query('//Search:Search/Search:Status');
        $this->assertEquals(1, $nodes->length, $responseDoc->saveXML());
        $this->assertEquals(Syncroton_Command_Search::STATUS_SUCCESS, $nodes->item(0)->nodeValue, $responseDoc->saveXML());
        $nodes = $xpath->query('//Search:Search/Search:Response/Search:Store/Search:Total');
        $this->assertEquals(1, $nodes->length, $responseDoc->saveXML());
        $this->assertEquals(5, $nodes->item(0)->nodeValue, $responseDoc->saveXML());
        $nodes = $xpath->query('//Search:Search/Search:Response/Search:Store/Search:Result');
        $this->assertEquals(4, $nodes->length, $responseDoc->saveXML());
        #$this->assertEquals(Syncroton_Command_Search::STATUS_SUCCESS, $nodes->item(0)->nodeValue, $responseDoc->saveXML());
    }

 /**
  * test testGetProperties method
  */
 public function testGetProperties()
 {
     $body = '<?xml version="1.0" encoding="utf-8"?>
              <propfind xmlns="DAV:">
                 <prop>
                     <getlastmodified xmlns="DAV:"/>
                     <getcontentlength xmlns="DAV:"/>
                     <resourcetype xmlns="DAV:"/>
                     <getetag xmlns="DAV:"/>
                     <id xmlns="http://owncloud.org/ns"/>
                 </prop>
              </propfind>';
     $request = new Sabre\HTTP\Request(array('REQUEST_METHOD' => 'PROPFIND', 'REQUEST_URI' => '/remote.php/webdav/' . Tinebase_Core::getUser()->accountDisplayName, 'HTTP_DEPTH' => '0'));
     $request->setBody($body);
     $this->server->httpRequest = $request;
     $this->server->exec();
     //var_dump($this->response->body);
     $this->assertEquals('HTTP/1.1 207 Multi-Status', $this->response->status);
     $responseDoc = new DOMDocument();
     $responseDoc->loadXML($this->response->body);
     //$responseDoc->formatOutput = true; echo $responseDoc->saveXML();
     $xpath = new DomXPath($responseDoc);
     $xpath->registerNamespace('owncloud', 'http://owncloud.org/ns');
     $nodes = $xpath->query('//d:multistatus/d:response/d:propstat/d:prop/owncloud:id');
     $this->assertEquals(1, $nodes->length, $responseDoc->saveXML());
     $this->assertNotEmpty($nodes->item(0)->nodeValue, $responseDoc->saveXML());
 }

 /**
  * @param string $html
  * @return TorrentsCollection
  */
 public function parse(string $html) : TorrentsCollection
 {
     libxml_use_internal_errors(true);
     $doc = new \DOMDocument();
     $doc->loadHTML($html);
     $finder = new \DomXPath($doc);
     $torrents = $finder->query("//*[contains(concat(' ', normalize-space(@class), ' '), ' hl-tr ')]");
     $links = $finder->query("//*[contains(@class, 'med tLink hl-tags bold')]");
     $result = new TorrentsCollection();
     for ($i = 0; $i < $torrents->length; $i++) {
         if (is_null($links->item($i))) {
             continue;
         }
         $childNodes = $torrents->item($i)->childNodes;
         $status = self::clean($childNodes->item(2)->attributes->getNamedItem('title')->textContent);
         $category = self::clean($childNodes->item(4)->textContent);
         $title = self::clean($childNodes->item(6)->textContent);
         $torrentId = (int) self::clean($links->item($i)->attributes->getNamedItem('data-topic_id')->textContent);
         $detailUrl = self::DETAIL_URL . $torrentId;
         $authorUrl = parse_url($childNodes->item(8)->firstChild->firstChild->attributes->getNamedItem('href')->textContent);
         $authorId = (int) explode('=', $authorUrl['query'])[1];
         $authorName = self::clean($childNodes->item(8)->firstChild->firstChild->textContent);
         $downloadUrl = self::clean($childNodes->item(10)->childNodes->item(3)->attributes->getNamedItem('href')->textContent);
         $downloadUrl = self::DOWNLOAD_URL . $downloadUrl;
         $size = self::clean(str_replace(' ↓', '', $childNodes->item(10)->childNodes->item(3)->textContent));
         $seeders = (int) self::clean($childNodes->item(12)->firstChild->textContent);
         $leachers = (int) self::clean($childNodes->item(14)->textContent);
         $downloads = (int) self::clean($childNodes->item(16)->textContent);
         $added = (new \DateTimeImmutable())->setTimestamp((int) self::clean($childNodes->item(18)->childNodes->item(1)->textContent));
         $torrent = (new Torrent())->setId($torrentId)->setTitle($title)->setCategory($category)->setStatus($status)->setDetailUrl($detailUrl)->setAuthorId($authorId)->setAuthorName($authorName)->setSize($size)->setDownloadUrl($downloadUrl)->setSeeders($seeders)->setLeachers($leachers)->setDownloads($downloads)->setAdded($added);
         $result->add($torrent);
     }
     return $result;
 }

function leistungsschutzrecht_fetchsites()
{
    require_once "include/network.php";
    $sites = array();
    $url = "http://www.vg-media.de/lizenzen/digitale-verlegerische-angebote/wahrnehmungsberechtigte-digitale-verlegerische-angebote.html";
    $site = fetch_url($url);
    $doc = new DOMDocument();
    @$doc->loadHTML($site);
    $xpath = new DomXPath($doc);
    $list = $xpath->query("//td/a");
    foreach ($list as $node) {
        $attr = array();
        if ($node->attributes->length) {
            foreach ($node->attributes as $attribute) {
                $attr[$attribute->name] = $attribute->value;
            }
        }
        if (isset($attr["href"])) {
            $urldata = parse_url($attr["href"]);
            if (isset($urldata["host"]) and !isset($urldata["path"])) {
                $cleanedurlpart = explode("%", $urldata["host"]);
                $hostname = explode(".", $cleanedurlpart[0]);
                $site = $hostname[sizeof($hostname) - 2] . "." . $hostname[sizeof($hostname) - 1];
                $sites[$site] = $site;
            }
        }
    }
    if (sizeof($sites)) {
        set_config('leistungsschutzrecht', 'sites', $sites);
    }
}

 protected function xpathQuery($html, $xpath)
 {
     $dom = new \DomDocument();
     $dom->loadHtml($html);
     $domXpath = new \DomXPath($dom);
     return $domXpath->query($xpath);
 }

 public function ovfImport($url, $xpath = false)
 {
     $this->url = $url;
     $dom = new DomDocument();
     $loaded = $dom->load($url);
     if (!$loaded) {
         return $loaded;
     }
     $xpath = new DomXPath($dom);
     // register the namespace
     $xpath->registerNamespace('envl', 'http://schemas.dmtf.org/ovf/envelope/1');
     $vs_nodes = $xpath->query('//envl:VirtualSystem');
     // selects all name element
     $refs_nodes = $xpath->query('//envl:References');
     // selects all name element
     $disk_nodes = $xpath->query('//envl:DiskSection');
     // selects all name element
     $refs_nodes_0 = $refs_nodes->item(0);
     $disk_nodes_0 = $disk_nodes->item(0);
     $vs_nodes_0 = $vs_nodes->item(0);
     if ($refs_nodes_0 && $disk_nodes_0 && $vs_nodes_0) {
         $this->buildReferences($refs_nodes_0);
         $this->buildDiskSection($disk_nodes_0);
         $this->buildVirtualSystem($vs_nodes_0);
         return true;
     } else {
         return false;
     }
 }

 function DataAsHTML($xslFileName)
 {
     $xmlData = new SimpleXMLElement('<page/>');
     $this->generateXML($xmlData, $this->data);
     $xmlfilemodule = simplexml_load_file("modules/" . $this->data['name'] . "/elements.xml");
     $xmlData = dom_import_simplexml($xmlData)->ownerDocument;
     foreach (dom_import_simplexml($xmlfilemodule)->childNodes as $child) {
         $child = $xmlData->importNode($child, TRUE);
         $xmlData->documentElement->appendChild($child);
     }
     $xslfile = "templates/" . $this->template_name . "/" . $xslFileName . ".xsl";
     $xslfilemodule = "modules/" . $this->data['name'] . "/elements.xsl";
     if (in_array($this->data['name'], $this->selfPages)) {
         $xslfile = "templates/" . $this->template_name . "/" . $this->data['name'] . ".xsl";
     }
     $domXSLobj = new DOMDocument();
     $domXSLobj->load($xslfile);
     $xpath = new DomXPath($domXSLobj);
     $next = $xpath->query('//xsl:include');
     $next = $next->item(0);
     $next->setAttribute('href', $xslfilemodule);
     $next->setAttribute('xml:base', ROOT_SYSTEM);
     $xsl = new XSLTProcessor();
     $xsl->importStyleSheet($domXSLobj);
     return $xsl->transformToXml($xmlData);
 }

function getProducts()
{
    include "../classes/httpFile.php";
    include "../functions/crawler_functions.php";
    error_reporting(E_ALL ^ E_WARNING);
    $http = new HttpConnection();
    $http->setCookiePath("cookies/");
    $http->init();
    $DOM = new DOMDocument();
    $url_array = split(PHP_EOL, file_get_contents('../file/cats_list.ccd'));
    while (count($url_array) > 0) {
        $url = array_shift($url_array);
        $http->get($url, true);
        get_dom($url, $DOM, $http);
        $finder = new DomXPath($DOM);
        $classname = "product-container";
        $product = $finder->query("//*[contains( normalize-space( @class ), ' {$classname} ' )\r\n\t\t  \t\t\tor substring( normalize-space( @class ), 1, string-length( '{$classname}' ) + 1 ) = '{$classname} '\r\n\t\t  \t\t\tor substring( normalize-space( @class ), string-length( @class ) - string-length( '{$classname}' ) ) = ' {$classname}'\r\n\t\t  \t\t\tor @class = '{$classname}']");
        foreach ($product as $p) {
            $enlaces = $p->getElementsByTagName('a');
            $enlace = $enlaces->item(0)->getAttribute('href');
            echo $enlace . '<br/>';
            file_put_contents("../file/url_list.ccd", $enlace . PHP_EOL, FILE_APPEND);
        }
    }
    $http->close();
}

 function bold()
 {
     $this->CI =& get_instance();
     $content = "";
     $output = $this->CI->output->get_output();
     $dom = new DOMDocument();
     $dom->loadHTML($output);
     $finder = new DomXPath($dom);
     $classname = "lead";
     $nodes = $finder->query("//*[contains(concat(' ', normalize-space(@class), ' '), ' {$classname} ')]");
     foreach ($nodes as $node) {
         $words = explode(' ', $node->nodeValue);
         $temp = $dom->createElement("p");
         $temp->setAttribute("class", "lead");
         foreach ($words as $word) {
             if (ctype_upper($word[0])) {
                 //First char is uppercase
                 $newNode = $dom->createElement('strong', $word . " ");
                 $temp->appendChild($newNode);
             } else {
                 $newNode = $dom->createTextNode($word . " ");
                 $temp->appendChild($newNode);
             }
         }
         $node->parentNode->replaceChild($temp, $node);
     }
     echo $dom->saveHTML();
 }