private function getDomObject()
{
if ($this->_verbose > 2) {
$this->log('IN:[' . __FUNCTION__ . ']', 0, "purple");
}
if ($this->_verbose > 1) {
$this->log("Content length: " . strlen($this->_content), 0, "light_purple");
}
$DOM = new \DomDocument();
switch ($this->_domType) {
case "xml":
if ($this->_verbose > 2) {
$this->log('<' . __LINE__ . '> DOMDocument->loadXML', 0, "purple");
}
@$DOM->loadXML($this->_content);
break;
case "html":
if ($this->_verbose > 2) {
$this->log('<' . __LINE__ . '> DOMDocument->loadXML');
}
@$DOM->loadHtml($this->_content);
break;
default:
throw new \Exception("Unknwon DOM type used to load content \"" . $this->_domType . "\"");
break;
}
return $DOM;
}
public function run($str = NULL)
{
if ($str == NULL) {
return $this;
}
$document = jqm_use($this->node->_parentElement);
$dom = $document->_DOM;
if ($dom->doctype) {
$dom->removeChild($dom->doctype);
}
$find = $this->node->getPathById($dom);
if (!$find) {
$find = $this->node->_path;
}
$xpath = new DomXpath($dom);
$find = $xpath->query($find);
if ($find->length > 0) {
$child = new DomDocument();
$child->loadHtml($str);
if ($child->doctype) {
$child->removeChild($child->doctype);
}
$child->normalize();
$frag = $dom->importNode($child->firstChild->firstChild->firstChild, true);
$save = $find->item(0)->parentNode->insertBefore($frag, $find->item(0));
$this->node->_path = $save->nextSibling->getNodePath();
$document->_DOM = $dom;
}
return $this;
}
protected function xpathQuery($html, $xpath)
{
$dom = new \DomDocument();
$dom->loadHtml($html);
$domXpath = new \DomXPath($dom);
return $domXpath->query($xpath);
}
public function run($str = NULL)
{
if ($str == NULL) {
return $this;
}
$dom = $this->node->dom()->get();
$this->node->_lastDom = $dom->lastdom . $str;
$rec = new DomDocument();
$rec->loadHtml($str);
if ($rec->doctype) {
$rec->removeChild($rec->doctype);
}
$document = jqm_use($this->node->_parentElement);
$dom = $document->_DOM;
if ($dom->doctype) {
$dom->removeChild($dom->doctype);
}
$body = $rec->childNodes->item(0)->firstChild;
$frag = $dom->importNode($body, true);
$xpath = new DomXpath($dom);
$find = $this->node->getPathByID($dom);
$find = $xpath->query($find);
if ($find->length > 0) {
$save = $find->item(0)->parentNode->insertBefore($frag->firstChild, $find->item(0)->nextSibling);
$this->node->_path = $save->previousSibling->getNodePath();
$document->_DOM = $dom;
}
return $this;
}
public function testConvertsBoldTag()
{
$html = '<b>some text</b>';
$doc = new \DomDocument();
$doc->loadHtml($html);
$tag = $doc->getElementsByTagName('b')->item(0);
$parser = new \Markdownable\Tag\B();
$this->assertSame($parser->parse($tag), '**some text**');
}
public function testConvertsPTag()
{
$html = '<p>A paragraph of text</p>';
$doc = new \DomDocument();
$doc->loadHtml($html);
$tag = $doc->getElementsByTagName('p')->item(0);
$parser = new \Markdownable\Tag\P();
$this->assertSame($parser->parse($tag), PHP_EOL . PHP_EOL . 'A paragraph of text' . PHP_EOL);
}
public function testConvertsEmTag()
{
$html = '<em>text</em>';
$doc = new \DomDocument();
$doc->loadHtml($html);
$tag = $doc->getElementsByTagName('em')->item(0);
$parser = new \Markdownable\Tag\Em();
$this->assertSame($parser->parse($tag), '*text*');
}
public function testConvertsHtmlBrTag()
{
$html = '<br>';
$doc = new \DomDocument();
$doc->loadHtml($html);
$tag = $doc->getElementsByTagName('br')->item(0);
$parser = new \Markdownable\Tag\Br();
$this->assertSame($parser->parse($tag), PHP_EOL);
}
public function testConvertsAnchorTagWithTitle()
{
$html = '<a href="http://example.com" title="A Link">link</a>';
$doc = new \DomDocument();
$doc->loadHtml($html);
$tag = $doc->getElementsByTagName('a')->item(0);
$parser = new \Markdownable\Tag\A();
$this->assertSame($parser->parse($tag), '[link](http://example.com "A Link")');
}
public function testConvertsImgTagWithTitle()
{
$html = '<img src="img.png" alt="alt text" title="An Image" />';
$doc = new \DomDocument();
$doc->loadHtml($html);
$tag = $doc->getElementsByTagName('img')->item(0);
$parser = new \Markdownable\Tag\Img();
$this->assertSame($parser->parse($tag), '');
}
public function testConvertsH6Tag()
{
$html = '<h6>Header 6</h6>';
$doc = new \DomDocument();
$doc->loadHtml($html);
$tag = $doc->getElementsByTagName('h6')->item(0);
$parser = new \Markdownable\Tag\H6();
$this->assertSame($parser->parse($tag), PHP_EOL . PHP_EOL . '###### Header 6' . PHP_EOL . PHP_EOL);
}
/**
* Parse te given XML attributes into an array
*
* @author troelskn
* @link http://stackoverflow.com/a/1083821/172068
* @param $input
* @return array
*/
public function parseAttributes($input)
{
$dom = new DomDocument();
$dom->loadHtml("<html {$input} />");
$attributes = array();
foreach ($dom->documentElement->attributes as $name => $attr) {
$attributes[$name] = $attr->value;
}
return $attributes;
}
function getDataFromFeed($url)
{
$data = apc_fetch($url);
if (!$data) {
error_log('CACHE MISS: ' . $url);
$data = file_get_contents($url);
apc_store($url, $data, 600);
}
$doc = new DomDocument();
@$doc->loadHtml($data);
return $doc;
}
/**
* Extracts information from the file $fileName associated with the url $url.
*
* The document type for this document is given in $type, and the images on
* disk should be in the directory named $imagePath. The urls where the
* images link to should be in $imageUrlPath.
*
* @param string $fileName
* @param string $type
* @param string $url
* @param string $imagePath
* @param string $imageUrlPath
* @return array(ezcSearchDocument)
*/
public static function extract($fileName, $type, $url, $imagePath = null, $imageUrlPath = null)
{
$published = filemtime($fileName);
$converted = file_get_contents($fileName);
$dom = new DomDocument();
@$dom->loadHtml($converted);
$tbody = $dom->getElementsByTagName('div')->item(0);
$xpath = new DOMXPath($dom);
$tocElem = $xpath->evaluate("//h1[@class='title']", $tbody)->item(0);
$title = $tocElem ? $tocElem->nodeValue : 'no title';
$docs = array();
$body = $urls = array();
$currentUrl = $url;
$lastUrl = $url;
$currentBody = '';
// child::*[self::p or self::h1]
$xpath = new DOMXPath($dom);
$tbody = $xpath->evaluate("//p|//h1|//ol|//ul|//dl|//img|//a", $tbody);
$body = '';
foreach ($tbody as $item) {
switch ($item->tagName) {
case 'a':
$name = $item->getAttribute('name');
if (strlen($name)) {
$currentUrl = $url . '#' . $name;
}
break;
case 'img':
$alt = $item->getAttribute('alt');
$src = $item->getAttribute('src');
$location = $imagePath == null ? dirname($fileName) . '/' . $src : $imagePath . '/' . preg_replace('@(\\.\\./)+@', '', $src);
$imgurl = $src[0] == '/' ? $src : ($imageUrlPath === null ? $url . '/' . $src : $imageUrlPath . '/' . preg_replace('@(\\.\\./)+@', '', $src));
echo " - {$src} => {$imgurl}\n";
$docs[] = self::extractImage($alt, $location, $imgurl);
break;
case 'p':
case 'h1':
case 'dl':
if ($lastUrl !== $currentUrl) {
$docs[] = new ezcSearchSimpleArticle(null, $title, $currentBody, $published, $lastUrl, $type);
$currentBody = '';
$lastUrl = $currentUrl;
}
$currentBody .= strip_tags($dom->saveXml($item)) . "\n\n";
break;
}
}
if ($currentBody != '') {
$docs[] = new ezcSearchSimpleArticle(null, $title, $currentBody, $published, $lastUrl, $type);
}
return $docs;
}
function getDataFromHtml($url)
{
$ignoreCache = isset($_GET['ignore_cache']);
$data = apc_fetch($url);
if (!$data || $ignoreCache) {
error_log('CACHE MISS: ' . $url);
$data = file_get_contents($url);
apc_store($url, $data, 600);
}
$doc = new DomDocument();
@$doc->loadHtml($data);
return $doc;
}
public function testConvertsMultilineBlockquoteTag()
{
$html = <<<EOD
<blockquote>
Some line of text
Another line of text
</blockquote>
EOD;
$doc = new \DomDocument();
$doc->loadHtml($html);
$tag = $doc->getElementsByTagName('blockquote')->item(0);
$parser = new \Markdownable\Tag\Blockquote();
$this->assertSame($parser->parse($tag), PHP_EOL . PHP_EOL . '> Some line of text' . PHP_EOL . '> Another line of text' . PHP_EOL . PHP_EOL);
}
/**
* Filter the page html and look for an <a><img> element added by the chooser
* or an <a> element added by the moodle file picker
*
*
* @param string $html
* @param array $options
* @return string
*/
public function filter($html, array $options = array())
{
global $COURSE;
$courseid = isset($COURSE->id) ? $COURSE->id : null;
if (empty($html) || !is_string($html) || strpos($html, $this->_binumi_client->get_host()) === false) {
return $html;
}
$dom = new DomDocument();
$sanitized_html = mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8');
if (defined('LIBXML_HTML_NOIMPLIED') && defined('LIBXML_HTML_NODEFDTD')) {
@$dom->loadHtml($sanitized_html, LIBXML_HTML_NOIMPLIED | LIBXML_HTML_NODEFDTD);
} else {
@$dom->loadHtml($sanitized_html);
}
$xpath = new DOMXPath($dom);
foreach ($xpath->query('//a') as $node) {
$href = $node->getAttribute('href');
$class = $node->getAttribute('class');
if (empty($href)) {
continue;
}
if ($class != 'binumi-embed') {
continue;
}
if ((bool) preg_match($this->_re_embed_url, $href)) {
$newnode = $dom->createDocumentFragment();
$imgnode = $node->firstChild;
$href = htmlspecialchars($href);
extract($this->_get_image_elem_dimensions($imgnode));
$html = $this->_get_iframe_embed_html($href, $width, $height);
$newnode->appendXML($html);
$node->parentNode->replaceChild($newnode, $node);
}
}
return $dom->saveHTML();
}
function getUrlsFromHtml($html)
{
$urls = array();
$xml = new \DomDocument();
@$xml->loadHtml($html);
$xml = $xml->saveXml();
$xml = str_replace('xmlns="http://www.w3.org/1999/xhtml" xmlns="http://www.w3.org/1999/xhtml"', 'xmlns="http://www.w3.org/1999/xhtml"', $xml);
$xml = str_replace('xml:lang="en" lang="en" xml:lang="en"', 'xml:lang="en" lang="en"', $xml);
$xml = simplexml_load_string(trim($xml));
$xml->registerXpathNamespace('e', 'http://www.w3.org/1999/xhtml');
$nodes = $xml->xpath('//e:a');
foreach ($nodes as $node) {
$urls[] = (string) $node['href'];
}
return $urls;
}
function bady2cdata($html, $xml)
{
//Get XML params
$dom = new DomDocument();
$dom->preserveWhiteSpace = FALSE;
$dom->loadHtml($html);
$params = $dom->getElementsByTagName('body');
foreach ($params as $k => $v) {
$html = $v->nodeValue;
}
$params = $dom->getElementsByTagName('title');
foreach ($params as $k => $v) {
$title = $v->nodeValue;
}
$params = $dom->getElementsByTagName('meta');
foreach ($params as $k => $v) {
if ($v->getAttribute('name') == 'description') {
$description = $v->getAttribute('content');
}
if ($v->getAttribute('name') == 'Author') {
$author = $v->getAttribute('content');
}
}
//Write to XML
$dom = new DomDocument();
$dom->preserveWhiteSpace = FALSE;
$dom->loadXML($xml);
$ModulePrefs = $dom->getElementsByTagName('ModulePrefs');
foreach ($ModulePrefs as $prefs) {
$prefs->setAttribute('title', $title);
$prefs->setAttribute('description', $description);
$prefs->setAttribute('author', $author);
}
$params = $dom->getElementsByTagName('Content');
foreach ($params as $k => $v) {
//echo $v->nodeValue;
$v->nodeValue = $html;
}
$s = '<?xml version="1.0" encoding="UTF-8" ?>';
$s .= $dom->saveHTML();
return $s;
}
static function render_template_example($post)
{
$html = self::render_html_base_by_post($post);
$return = spnl_do_content_tags($html, $post->ID, $post->ID, 0, true);
$return = spnl_do_email_tags($return, $post->ID, $post->ID, 0, true);
$return = spnl_do_subscriber_tags($return, $post->ID, $post->ID, 0, true);
//$body_html = preg_replace( $pattern , site_url() ."?sendpress=link&fxti=".$subscriber_key."&spreport=". $this->id ."&spurl=$0", $body_html );
if (class_exists("DomDocument")) {
$dom = new DomDocument();
$dom->strictErrorChecking = false;
@$dom->loadHtml($return);
$pTags = $dom->getElementsByTagName('p');
foreach ($pTags as $pElement) {
$px = $pElement->getAttribute('style');
$pElement->setAttribute('style', $px . ' margin-top:0;margin-bottom:10px;');
}
$return = $dom->saveHtml();
}
return $return;
}
public function testConvertsOlListTag()
{
$html = <<<EOD
<ol>
<li>Item 1</li>
<li>Item 2</li>
<li>Item 3</li>
</ol>
EOD;
$doc = new \DomDocument();
$doc->loadHtml($html);
$tag = $doc->getElementsByTagName('ol')->item(0);
$parser = new \Markdownable\Tag\Li();
$output = '';
foreach ($tag->childNodes as $child) {
if ($child instanceof \DOMElement) {
$output .= $parser->parse($child);
}
}
$this->assertSame($output, '1. Item 1' . PHP_EOL . '2. Item 2' . PHP_EOL . '3. Item 3' . PHP_EOL);
}
<?php
include "inc/main.inc";
include "inc/header.inc";
$doc = new DomDocument();
// We need to validate our document before refering to the id
$doc->validateOnParse = false;
$doc->loadHtml(file_get_contents($_GET['page']));
?>
<div data-role="page" id="content" data-add-back-btn="true">
<div data-role="header">
<h1><?echo $_GET['title'] ?></h1>
</div><!-- /header -->
<div data-role="content">
<p>
<?
$links = $doc->getElementsByTagName("a");
foreach($links as $link) {
$href = $link->getAttribute("href");
$newLink = "content.php?page=" . $href;
$link->setAttribute("href",$newLink);
}
$elemContent = $doc->getElementById('content');
$elemText = $elemContent->ownerDocument->saveXML($elemContent);
echo $elemText;
?>
</p>
function onAfterDispatch()
{
global $offlajnParams, $offlajnDashboard;
$app = JFactory::getApplication();
if (!defined('OFFLAJNADMIN') || isset($_REQUEST['output']) && $_REQUEST['output'] == 'json') {
return;
}
$doc = JFactory::getDocument();
$c = $doc->getBuffer('component');
$dom = new DomDocument();
if (function_exists("mb_convert_encoding")) {
@$dom->loadHtml('<?xml encoding="UTF-8"><div>' . mb_convert_encoding($c, 'HTML-ENTITIES', "UTF-8") . '</div>');
} else {
@$dom->loadHtml('<?xml encoding="UTF-8"><div>' . htmlspecialchars_decode(utf8_decode(htmlentities($c, ENT_COMPAT, 'utf-8', false))) . '</div>');
}
$lis = array();
$moduleparams = "";
$advanced = JRequest::getCmd('option') == 'com_advancedmodules';
if (version_compare(JVERSION, '3.0.0', 'ge') && !$this->getElementById($dom, 'module-sliders')) {
// Joomla 3.0.3 fix
if (version_compare(JVERSION, '3.1.99', 'ge')) {
$moduleparams = $this->getElementByClass($dom, 'span9');
} elseif (version_compare(JVERSION, '3.0.3', 'ge')) {
$moduleparams = $this->getElementById($dom, 'collapse0');
} else {
$moduleparams = $this->getElementById($dom, 'options-basic');
}
if ($advanced) {
$moduleparams = version_compare(JVERSION, '3.2.2', 'ge') ? $this->getElementByClass($dom, 'span9') : $this->getElementByClass($dom, 'span6', 1);
}
if ($moduleparams) {
$element = $dom->createElement('div');
$element->setAttribute('id', 'content-box');
$moduleparams->appendChild($element);
$moduleparams = $element;
$element = $dom->createElement('div');
$element->setAttribute('id', 'module-sliders');
$element->setAttribute('class', 'pane-sliders');
$moduleparams->appendChild($element);
$moduleparams = $element;
}
} elseif (version_compare(JVERSION, '1.6.0', 'ge')) {
$moduleparams = $this->getElementById($dom, 'module-sliders');
} else {
$moduleparams = $this->getElementById($dom, 'menu-pane');
}
if ($moduleparams) {
$removed = array();
while ($cNode = $moduleparams->firstChild) {
$removed[] = $moduleparams->removeChild($cNode);
}
if (version_compare(JVERSION, '1.6.0', 'ge')) {
array_splice($removed, 0, 2);
} else {
array_splice($removed, 0, 1);
}
$html = '<div>';
$html .= isset($offlajnDashboard) ? $offlajnDashboard : '';
$html .= isset($offlajnParams['first']) && is_array($offlajnParams['first']) ? implode("\n", $offlajnParams['first']) : '';
$html .= isset($offlajnParams['last']) && is_array($offlajnParams['last']) ? implode("\n", $offlajnParams['last']) : '';
$html .= '</div>';
$tabsDom = new DomDocument();
if (function_exists("mb_convert_encoding")) {
@$tabsDom->loadHtml('<?xml encoding="UTF-8">' . mb_convert_encoding($html, 'HTML-ENTITIES', "UTF-8"));
} else {
@$tabsDom->loadHtml('<?xml encoding="UTF-8">' . htmlspecialchars_decode(utf8_decode(htmlentities($html, ENT_COMPAT, 'utf-8', false))));
}
$node = $dom->importNode($tabsDom->getElementsByTagName('div')->item(0), true);
while ($cNode = $node->firstChild) {
if (@$cNode->tagName == 'div') {
$moduleparams->appendChild($cNode);
} else {
$node->removeChild($cNode);
}
}
if (count($removed) > 0) {
foreach ($removed as $r) {
if ($r instanceof DOMElement) {
$r->setAttribute("class", $r->getAttribute("class") . " legacy");
$moduleparams->appendChild($r);
}
}
}
if (!version_compare(JVERSION, '1.6.0', 'ge')) {
$tables = $dom->getElementsByTagName('table');
foreach ($tables as $table) {
$table->setAttribute("cellspacing", "0");
}
}
$params = $moduleparams->getElementsByTagName('h3');
foreach ($params as $param) {
$span = $param->getElementsByTagName('span')->item(0);
$titleWords = explode(" ", $span->textContent);
$titleWords[count($titleWords) - 1] = "<b>" . $titleWords[count($titleWords) - 1] . "</b>";
$newTitle = implode(' ', $titleWords);
$span->removeChild($span->firstChild);
$newText = $dom->createCDATASection($newTitle);
$span->appendChild($newText);
}
$j = 0;
foreach ($moduleparams->childNodes as $param) {
$param->setAttribute("id", "offlajnpanel-" . $j);
$j++;
}
}
if (!isset($doc->_script['text/javascript'])) {
$doc->_script['text/javascript'] = array();
}
$doc->_script['text/javascript'] = preg_replace("/window.addEvent.*?pane-toggler.*?\\}\\);.*?\\}\\);/i", '', $doc->_script['text/javascript']);
$doc->_script['text/javascript'] .= '
window.addEvent("domready", function(){
if(document.formvalidator)
document.formvalidator.isValid = function() {return true;};
});';
if (version_compare(JVERSION, '3.0.0', 'ge')) {
if ($moduleparams && $moduleparams->parentNode) {
function getInnerHTML($Node)
{
$Document = new DOMDocument();
$Document->appendChild($Document->importNode($Node, true));
return $Document->saveHTML();
}
$nc = getInnerHTML($moduleparams->parentNode);
} else {
$nc = $dom->saveHTML();
}
$nc = preg_replace("/.*?<body>/si", '', $nc, 1);
$nc = preg_replace("/<\\/body>.*/si", '', $nc, 1);
$pattern = '/<div\\s*class="tab-pane"\\s*id="options-basic".*?>/';
if (version_compare(JVERSION, '3.1.99', 'ge')) {
$pattern = '/<div\\s*class="span9".*?>/';
} elseif (version_compare(JVERSION, '3.0.3', 'ge')) {
$pattern = '/<div\\s*class="accordion-body collapse in"\\s*id="collapse0".*?>/';
}
if ($advanced) {
$pattern = version_compare(JVERSION, '3.2.2', 'ge') ? '/<div\\s*class="span9".*?>/' : '/<\\/div>\\s*<div\\s*class="span6".*?>/';
}
preg_match($pattern, $c, $matches);
if (count($matches) > 0) {
$c = str_replace($matches[0], $matches[0] . $nc, $c);
} else {
$c = $nc;
}
} else {
$c = $dom->saveHtml();
$c = preg_replace("/.*?<body><div>/si", '', $c, 1);
$c = preg_replace("/<\\/div><\\/body>.*/si", '', $c, 1);
}
$doc->setBuffer($c, 'component');
}
/**
* Filter the page html and look for an <a><img> element added by the chooser
* or an <a> element added by the moodle file picker
*
* NOTE: Thumbnail html from the Chooser and a link from the old filepicker
* are of the same form (see $_re_api1_public_urls).
* A thumbnail link from the new repository filepicker plugin is
* different (see $_re_api2_public_urls).
* The latest version of the local lib and rich text editor plugins
* use both a trusted and a regular embed url as the href value of
* the thumbnail html (this is the preferred route going forward).
* Both types of embed_urls will need a user to be authenticated
* before they can view the embed.
*
* @param string $html
* @param array $options
* @return string
*/
public function filter($html, array $options = array())
{
global $COURSE;
$courseid = isset($COURSE->id) ? $COURSE->id : null;
if (empty($html) || !is_string($html) || strpos($html, $this->_mcore_client->get_host()) === false) {
return $html;
}
$dom = new DomDocument();
@$dom->loadHtml(mb_convert_encoding($html, 'HTML-ENTITIES', 'UTF-8'));
$xpath = new DOMXPath($dom);
foreach ($xpath->query('//a') as $node) {
$href = $node->getAttribute('href');
if (empty($href)) {
continue;
}
if ((bool) preg_match($this->_re_embed_url, $href)) {
$newnode = $dom->createDocumentFragment();
$imgnode = $node->firstChild;
if ($this->_mcore_client->has_lti_config() && !is_null($courseid)) {
$href = $this->_generate_embed_url($href, $courseid);
} else {
$href = htmlspecialchars($href);
}
extract($this->_get_image_elem_dimensions($imgnode));
$html = $this->_get_iframe_embed_html($href, $width, $height);
$newnode->appendXML($html);
$node->parentNode->replaceChild($newnode, $node);
} else {
if ((bool) preg_match($this->_re_api1_public_urls, $href)) {
$newnode = $dom->createDocumentFragment();
$imgnode = $node->firstChild;
extract($this->_get_image_elem_dimensions($imgnode));
$html = $this->_get_embed_html_from_api1_public_url($href, $width, $height, $courseid);
$newnode->appendXML($html);
$node->parentNode->replaceChild($newnode, $node);
} else {
if ((bool) preg_match($this->_re_api2_public_urls, $href)) {
$newnode = $dom->createDocumentFragment();
$width = $this->_default_thumb_width;
$height = $this->_default_thumb_height;
$html = $this->_get_embed_html_from_api2_public_url($href, $width, $height, $courseid);
$newnode->appendXML($html);
$node->parentNode->replaceChild($newnode, $node);
}
}
}
}
return $dom->saveHTML();
}
foreach ($searchpages as $searchpage) {
$original_file = file_get_contents("{$searchpage}");
$stripped_file = strip_tags($original_file, "<a>");
preg_match_all("/<a(?:[^>]*)href=\"([^\"]*)\"(?:[^>]*)>(?:[^<]*)<\\/a>/is", $stripped_file, $matches);
//extract pages
foreach ($matches[1] as $key => $value) {
if (strstr($value, "/company/")) {
$companypages[] = $value;
}
}
}
#print_r($companypages);
foreach ($companypages as $key => $value) {
$html = file_get_contents($value);
$dom = new DomDocument();
$dom->loadHtml($html);
$xpath = new DomXpath($dom);
$companies[$key]['name'] = trim($xpath->query('//*[@id="or-name"]')->item(0)->nodeValue);
$companies[$key]['title'] = trim($xpath->query('//*[@id="or-job-title"]')->item(0)->nodeValue);
#$companies[$key]['company']= $xpath->query('//*[@title]')->item(0);
$companies[$key]['company'] = ucwords(str_replace('-', ' ', substr($value, strrpos($value, '/') + 1)));
$companies[$key]['address'] = trim($xpath->query('//*[@class="street-address"]')->item(0)->nodeValue);
$companies[$key]['city'] = trim($xpath->query('//*[@class="locality"]')->item(0)->nodeValue);
$companies[$key]['state'] = trim($xpath->query('//*[@class="region"]')->item(0)->nodeValue);
$companies[$key]['zip'] = trim($xpath->query('//*[@class="postal-code"]')->item(0)->nodeValue);
$companies[$key]['country'] = trim($xpath->query('//*[@class="country-name"]')->item(0)->nodeValue);
}
foreach ($companies as $value) {
scraperwiki::save(array('name', 'title', 'company', 'address', 'city', 'state', 'zip', 'country'), $value);
echo implode(',', $value) . PHP_EOL;
}
<?php
header('Access-Control-Allow-Origin: *');
$xml = new DomDocument("1.0", "UTF-8");
$container = $xml->createElement("holder");
$xml->appendChild($container);
$html = file_get_contents('http://imgur.com/t/pizza');
$tidy = new tidy();
$html = $tidy->repairString($html);
$d = new DomDocument();
$d->loadHtml($html);
$p = new DomXPath($d);
$images = array();
foreach ($p->query('img') as $i) {
$images[] = $i;
$container->appendChild($xml->createTextNode($i));
}
foreach ($images as $i) {
echo $out;
}
/*
for($i=1; $i<10;$i++){
$hold = $xml->createElement("image");
$src = $xml->createAttribute("src");
$tsrc = $xml->createTextNode($images[$i]);
$src->appendChild($tsrc);
$hold->appendChild($src);
$container->appendChild($hold);
}
function html()
{
$post_template = $this->id();
global $wpdb;
//$email = $this->email();
// Get any existing copy of our transient data
if (SendPress_Email_Cache::get($this->id()) != null) {
$body_html = SendPress_Email_Cache::get($this->id());
$post_template = get_post_meta($this->id(), '_sendpress_template', true);
$body_html = spnl_do_email_tags($body_html, $post_template, $this->id(), $this->subscriber_id(), true);
} else {
if (false === ($body_html = get_transient('sendpress_report_body_html_' . $this->id())) || $this->purge() == true) {
// It wasn't there, so regenerate the data and save the transient
if (!$this->post_info) {
$this->post_info = get_post($this->id());
}
if ($this->cache() !== false) {
$body_html = $this->cache();
} else {
$body_html = SendPress_Template::get_instance()->render($this->id(), false, false, $this->remove_links());
$this->cache($body_html);
}
set_transient('sendpress_report_body_html_' . $this->id(), $body_html, 60 * 60 * 2);
}
}
$subscriber = SendPress_Data::get_subscriber($this->subscriber_id());
if (!is_null($subscriber)) {
$body_html = str_replace("*|FNAME|*", $subscriber->firstname, $body_html);
$body_html = str_replace("*|LNAME|*", $subscriber->lastname, $body_html);
$body_html = str_replace("*|EMAIL|*", $subscriber->email, $body_html);
$body_html = str_replace("*|ID|*", $subscriber->subscriberID, $body_html);
}
$open_info = array("id" => $this->subscriber_id(), "report" => $this->id(), "view" => "open");
$code = SendPress_Data::encrypt($open_info);
$link = SendPress_Manager::public_url($code);
$tracker = "<img src='" . $link . "' width='1' height='1'/></body>";
$body_html = str_replace("</body>", $tracker, $body_html);
$body_link = get_post_meta($this->id(), 'body_link', true);
$body_html = spnl_do_subscriber_tags($body_html, $post_template, $this->id(), $this->subscriber_id(), true);
//$pattern ="/(?<=href=(\"|'))[^\"']+(?=(\"|'))/";
//$body_html = preg_replace( $pattern , site_url() ."?sendpress=link&fxti=".$subscriber_key."&spreport=". $this->id ."&spurl=$0", $body_html );
if (class_exists("DomDocument")) {
$dom = new DomDocument();
$dom->strictErrorChecking = false;
@$dom->loadHtml($body_html);
$pTags = $dom->getElementsByTagName('p');
foreach ($pTags as $pElement) {
$px = $pElement->getAttribute('style');
$pElement->setAttribute('style', $px . ' margin-top:0;margin-bottom:10px;');
}
if ($this->tracker()) {
$aTags = $dom->getElementsByTagName('a');
foreach ($aTags as $aElement) {
$href = $aElement->getAttribute('href');
/*
$style = $aElement->getAttribute('style');
if($style == ""){
$aElement->setAttribute('style');
}
*/
//ADD TO DB?
if (strrpos($href, "*|") === false && strrpos($href, "#") !== 0) {
if (SendPress_Option::get('skip_mailto', false) == true && strrpos($href, "mailto") !== false) {
continue;
}
/*
$urlinDB = SendPress_Data::get_url_by_report_url( $this->id(), $href );
if(!isset($urlinDB[0])){
$urlData = array(
'url' => trim($href),
'reportID' => $this->id(),
);
$urlID = SendPress_Data::insert_report_url( $urlData );
} else {
$urlID = $urlinDB[0]->urlID;
}
$link = array(
"id"=>$this->subscriber_id(),
"report"=> $this->id(),
"urlID"=> $urlID,
"view"=>"link"
);
*/
$link = array("id" => $this->subscriber_id(), "report" => $this->id(), "view" => "tracker", "url" => $href);
$code = SendPress_Data::encrypt($link);
$link = SendPress_Manager::public_url($code);
$href = $link;
$aElement->setAttribute('href', $href);
}
}
}
$body_html = $dom->saveHtml();
}
$link_data = array("id" => $this->subscriber_id(), "report" => $this->id(), "urlID" => '0', "view" => "manage", "listID" => $this->list_id(), "action" => "unsubscribe");
$code = SendPress_Data::encrypt($link_data);
$link = SendPress_Manager::public_url($code);
if (SendPress_Option::get('old_unsubscribe_link', false) === true) {
$start_text = __("Not interested anymore?", "sendpress");
$unsubscribe = __("Unsubscribe", "sendpress");
$instantly = __("Instantly", "sendpress");
$remove_me_old = $start_text . ' <a href="' . $link . '" style="color: ' . $body_link . ';" >' . $unsubscribe . '</a> ' . $instantly . '.';
$body_html = str_replace("*|SP:UNSUBSCRIBE|*", $remove_me_old, $body_html);
$body_html = str_replace("*|SP:MANAGE|*", '', $body_html);
} else {
$link_data = array("id" => $this->subscriber_id(), "report" => $this->id(), "urlID" => '0', "view" => "manage", "listID" => $this->list_id(), "action" => "");
$code = SendPress_Data::encrypt($link_data);
$manage_link = SendPress_Manager::public_url($code);
$unsubscribe = __("Unsubscribe", "sendpress");
$manage = __("Manage Subscription", "sendpress");
$remove_me = ' <a href="' . $link . '" style="color: ' . $body_link . ';" >' . $unsubscribe . '</a> | ';
$manage = ' <a href="' . $manage_link . '" style="color: ' . $body_link . ';" >' . $manage . '</a> ';
$body_html = str_replace("*|SP:UNSUBSCRIBE|*", $remove_me, $body_html);
$body_html = str_replace("*|SP:MANAGE|*", $manage, $body_html);
}
if (!is_null($subscriber)) {
$body_html = str_replace("*|FNAME|*", $subscriber->firstname, $body_html);
$body_html = str_replace("*|LNAME|*", $subscriber->lastname, $body_html);
$body_html = str_replace("*|EMAIL|*", $subscriber->email, $body_html);
$body_html = str_replace("*|ID|*", $subscriber->subscriberID, $body_html);
}
//$body_html = apply_filters('sendpress_post_render_email', $body_html);
//echo $body_html;
//print_r($email);
return $body_html;
}
static function link_style($color, $content)
{
if (class_exists("DomDocument")) {
$dom = new DomDocument('1.0', 'UTF-8');
//$content = str_replace (' ', '@nbsp;', $content);
if (function_exists('mb_convert_encoding')) {
$content = mb_convert_encoding($content, 'HTML-ENTITIES', 'UTF-8');
// htmlspecialchars($content);
}
$dom->strictErrorChecking = false;
@$dom->loadHtml($content);
$aTags = $dom->getElementsByTagName('a');
foreach ($aTags as $aElement) {
$style = $aElement->getAttribute('style');
$style .= ' color: ' . $color . '; ';
$aElement->setAttribute('style', $style);
}
//$content = $dom->saveHTML();
$content = preg_replace(array("/^\\<\\!DOCTYPE.*?<html><body>/si", "!</body></html>\$!si"), "", $dom->saveHTML());
$content = str_replace("%7B", "{", $content);
$content = str_replace("%7D", "}", $content);
return $content;
}
return $content;
}
/**
* Extract a single review into a review instance
* @param DomElement $node
* @return Review
*/
public function extractReview(DomElement $node)
{
$review = new Review();
$doc = new DomDocument();
$html = $node->ownerDocument->saveHtml($node);
$doc->loadHtml('<?xml encoding="utf-8" ?>' . $html);
$xpath = new DOMXpath($doc);
/**
* Author Username
* // member_info/ * /username/span
*/
$nodes = $xpath->query("//div[contains(@class, 'member_info')]/*/div[contains(@class, 'username')]/span");
if ($nodes->length === 1) {
$review->setAuthor(trim($nodes->item(0)->nodeValue));
}
/**
* Author Location
*/
$nodes = $xpath->query("//div[contains(@class, 'member_info')]/div[contains(@class, 'location')]");
if ($nodes->length === 1) {
$review->setAuthorLocation(trim($nodes->item(0)->nodeValue));
}
/**
* Review Permalink
*/
$nodes = $xpath->query("//div[contains(@class, 'quote')]/a");
if ($nodes->length === 1) {
// URL linked to from the title is an absolute path without host/scheme
// Also, URL contains a fragment that should be removed
$path = trim($nodes->item(0)->getAttribute('href'));
$url = new Uri($this->getOptions()->getUrl());
$new = sprintf('%s://%s/%s', $url->getScheme(), $url->getHost(), ltrim($path, '/'));
$url = new Uri($new);
$url->setFragment(null);
$review->setUrl((string) $url);
}
/**
* Title Quote
*/
$nodes = $xpath->query("//div[contains(@class, 'quote')]/a/span[contains(@class, 'noQuotes')]");
if ($nodes->length === 1) {
$review->setTitle(trim($nodes->item(0)->nodeValue));
}
/**
* Rating as an alt tag in an image
*/
$nodes = $xpath->query("//div[contains(@class, 'rating')]/span[contains(@class, 'rate')]/img");
if ($nodes->length === 1) {
$img = $nodes->item(0);
$alt = $img->getAttribute('alt');
if (preg_match('/([0-9\\.]+)\\s[\\w]+\\s([0-9\\.]+)/', $alt, $match)) {
$review->setStarRating((double) $match[1]);
$review->setMaxStarRating((double) $match[2]);
}
}
/**
* Rating Date
*/
$nodes = $xpath->query("//div[contains(@class, 'rating')]/span[contains(@class, 'ratingDate')]");
if ($nodes->length === 1) {
$span = $nodes->item(0);
if ($span->hasAttribute('title')) {
$dateString = $span->getAttribute('title');
$date = DateTime::createFromFormat('d F Y', $dateString);
if (false !== $date) {
$review->setDate($date);
}
} else {
if (preg_match('/([0-9]{1,2}\\s[\\w]+\\s[0-9]{4})/i', trim($span->nodeValue), $match)) {
$date = DateTime::createFromFormat('d F Y', $match[1]);
if (false !== $date) {
$review->setDate($date);
}
}
}
}
/**
* Review Excerpt
*/
$nodes = $xpath->query("//div[contains(@class, 'entry')]/p[contains(@class, 'partial_entry')]");
if ($nodes->length === 1) {
// Use first child node->nodeValue to skip links for 'more...'
$review->setExcerpt(trim($nodes->item(0)->childNodes->item(0)->nodeValue));
}
return $review;
}
/**
* Converts a raw solr result into a document using the definition $def
*
* @param ezcSearchDocumentDefinition $def
* @param mixed $response
* @return ezcSearchResult
*/
private function createResponseFromData(ezcSearchDocumentDefinition $def, $response)
{
if (is_string($response)) {
// try to find the error message and return that
$s = new ezcSearchResult();
$dom = new DomDocument();
@$dom->loadHtml($response);
$tbody = $dom->getElementsByTagName('body')->item(0);
$xpath = new DOMXPath($dom);
$tocElem = $xpath->evaluate('//pre', $tbody)->item(0);
$error = $tocElem->nodeValue;
$s->error = $error;
return $s;
}
$s = new ezcSearchResult();
$s->status = $response->responseHeader->status;
$s->queryTime = $response->responseHeader->QTime;
$s->resultCount = $response->response->numFound;
$s->start = $response->response->start;
foreach ($response->response->docs as $document) {
$resultDocument = $this->createDataForHit($document, $def);
$idProperty = $def->idProperty;
$s->documents[$resultDocument->document->{$idProperty}] = $resultDocument;
}
// process highlighting
if (isset($response->highlighting) && count($s->documents)) {
foreach ($s->documents as $id => $document) {
$document->highlight = array();
if (isset($response->highlighting->{$id})) {
foreach ($def->fields as $field) {
$fieldName = $this->mapFieldType($field->field, $field->type);
if ($field->highlight && isset($response->highlighting->{$id}->{$fieldName})) {
$document->highlight[$field->field] = $response->highlighting->{$id}->{$fieldName};
}
}
}
}
}
// process facets
if (isset($response->facet_counts) && isset($response->facet_counts->facet_fields)) {
$facets = $response->facet_counts->facet_fields;
foreach ($def->fields as $field) {
$fieldName = $this->mapFieldType($field->field, $field->type);
if (isset($facets->{$fieldName})) {
// sigh, stupid array format needs fixing
$facetValues = array();
$facet = $facets->{$fieldName};
for ($i = 0; $i < count($facet); $i += 2) {
$facetValues[$facet[$i]] = $facet[$i + 1];
}
$s->facets[$field->field] = $facetValues;
}
}
}
return $s;
}
