function main($html)
{
global $baseurl;
if ($html) {
$dom = new DOMDocument();
$dom->preserveWhiteSpace = FALSE;
$dom->loadHTML($html);
$xpath = new DomXpath($dom);
$container = $xpath->query('//*[@id="maincontentcenter"]')->item(0);
$entries = $xpath->query('//*[@style="margin-left:15px;"]/ul', $container);
$jobs = array();
foreach ($entries as $entry) {
$department = $entry->previousSibling->previousSibling->nodeValue;
$li = $xpath->query('li', $entry);
foreach ($li as $job) {
$this_job = array();
$this_job['department'] = $department;
$a = $xpath->query('a', $job);
$href = $a->item(0)->getAttribute('href');
$title = $a->item(0)->nodeValue;
$this_job['title'] = $title;
$this_job['url'] = $baseurl . "/" . $href;
$details = jobDetails($this_job['url']);
$jobs[] = $this_job;
}
}
}
print_r($jobs);
}
private function CreateDomDocument()
{
$dom = new Domdocument();
@$dom->loadHTML($this->htmlStack);
$xpath = new DomXpath($dom);
// Find all the product blocks that contain beer attribute that we want to search for such as rating, title, price, etc.
$elements = $xpath->query('//article[contains(@class, "' . $this->class . '")]');
foreach ($elements as $element) {
// get the node value from the .title class (the title of the product)
$title = $xpath->query('descendant::div[@class="title"]', $element);
$title = $title->item(0)->nodeValue;
// get the attribute value from the .link-overlay class (the actual link to the product)
$link = $xpath->query('descendant::a[@class="link-overlay"]', $element);
$link = $link->item(0)->getAttribute('href');
$link = 'https://www.gall.nl' . $link;
// get the image src value to rip the image from the site
$image = $xpath->query('descendant::div[@class="image"]/node()/node()', $element);
$image = $image->item(1)->getAttribute('src');
// Dive in the link that a single product has
$results = $this->curlGetDeep($link);
// if a regex option has been supplied run it on the results
if ($this->regex) {
$title = preg_replace($this->regex, '', $title);
}
// remove the unwanted 6X 35x6(bulk beers) results with this regex
if (!preg_match('/\\dX(\\d+)?/', $title)) {
$this->elementsArray[] = ['title' => $title, 'link' => $link, 'image' => $image, 'description' => $results['description'], 'abv' => $results['abv']];
}
}
$this->json = json_encode(['beers' => $this->elementsArray]);
}
/**
* Extract a series of information from a twitter page.
*
* @see doc/Twitter.md for the detailed structure of the returned array
* @see Twitter::queryToMeta() for details about the $meta array
*
* @param string $page content of a twitter page
* @param array $meta
*
* @return array|bool false on error
*/
protected function parse($page, $meta)
{
$twitter['meta'] = $meta;
$tweet = '//ol[@id="stream-items-id"]/li[@data-item-type="tweet"]';
// foo_bar will be converted to foo['bar']
$tweet_details = ['user_avatar' => './/img[contains(@class, "avatar")]/@src', 'user_fullname' => './/@data-name', 'user_name' => './/@data-screen-name', 'stats_fav' => './/span[contains(@class, "ProfileTweet-action--favorite")]//@data-tweet-stat-count', 'stats_rt' => './/span[contains(@class, "ProfileTweet-action--retweet")]//@data-tweet-stat-count', 'id' => './/@data-tweet-id', 'retweetid' => './/@data-retweet-id', 'text' => './/p[contains(@class, "tweet-text")]', 'datetime' => './/@data-time'];
$user_stats = '//ul[contains(@class, "ProfileNav-list")]';
$user_stats_details = ['followers' => './/li[contains(@class, "ProfileNav-item--followers")]/a/@title', 'following' => './/li[contains(@class, "ProfileNav-item--following")]/a/@title', 'likes' => './/li[contains(@class, "ProfileNav-item--favorites")]/a/@title', 'lists' => './/li[contains(@class, "ProfileNav-item--lists")]/a/@title', 'tweets' => './/li[contains(@class, "ProfileNav-item--tweets")]/a/@title'];
libxml_use_internal_errors(true);
$dom = new \DOMDocument();
$dom->preserveWhiteSpace = false;
$dom->loadHTML($page);
$xpath = new \DomXpath($dom);
foreach ($xpath->query($tweet) as $item) {
foreach ($tweet_details as $key => $query) {
$value = @$xpath->query($query, $item)->item(0)->textContent;
if ($key == 'retweetid') {
$details['retweet'] = empty($value) ? 0 : 1;
continue;
}
if ($key == 'text') {
$value = $this->clean($value);
$value = htmlspecialchars($value, ENT_NOQUOTES);
$details['text']['raw'] = $value;
$details['text']['html'] = $this->html($value);
$details['reply'] = substr($value, 0, 1) === '@' ? 1 : 0;
continue;
}
if ($key == 'datetime') {
$value = date('c', $value);
}
if (strpos($key, '_') !== false) {
$subarray = explode('_', $key);
$subname = $subarray[0];
$subkey = $subarray[1];
$details[$subname][$subkey] = $value;
} else {
$details[$key] = $value;
}
}
$details['user']['url'] = 'https://twitter.com/' . $details['user']['name'];
$details['url'] = $details['user']['url'] . '/status/' . $details['id'];
// type casting
$details['id'] = (int) $details['id'];
$details['stats']['fav'] = (int) $details['stats']['fav'];
$details['stats']['rt'] = (int) $details['stats']['rt'];
ksort($details);
$twitter['tweets'][] = $details;
}
$twitter['tweets_count'] = count($twitter['tweets']);
if ($meta['type'] == 'user') {
$item = $xpath->query($user_stats)->item(0);
foreach ($user_stats_details as $key => $query) {
$value = @$xpath->query($query, $item)->item(0)->textContent;
$twitter['user_stats'][$key] = preg_match('~([\\d,]+)~', $value, $matches) ? (int) strtr($matches[1], [',' => '']) : 0;
}
}
ksort($twitter);
return $twitter;
}
function get_viber($owner)
{
$conn = curl_init();
if ($conn != false) {
curl_setopt($conn, CURLOPT_RETURNTRANSFER, true);
curl_setopt($conn, CURLOPT_SSL_VERIFYPEER, false);
curl_setopt($conn, CURLOPT_CONNECTTIMEOUT, 5);
$url = 'http://chats.viber.com/' . $owner;
curl_setopt($conn, CURLOPT_URL, $url);
$res = curl_exec($conn);
if ($res != false) {
curl_close($conn);
$doc = new DOMDocument();
@$doc->loadHTML($res);
$tags = $doc->getElementsByTagName('img');
$images = array();
$i = 0;
foreach ($tags as $tag) {
$images[$i] = $tag->getAttribute('src');
//echo $images[$i].'<br>';
$i++;
}
$xpath = new DomXpath($doc);
$classname = 'participant-name';
$par_names = array();
$results = $xpath->query("//*[@class and contains(concat(' ', normalize-space(@class), ' '), ' {$classname} ')]");
if ($results->length > 0) {
$par_names[0] = $results->item(0)->nodeValue;
if (strlen($par_names[0]) > 15) {
$par_names[0] = substr($par_names[0], 0, 12) . '...';
}
$par_names[1] = $results->item(1)->nodeValue;
if (strlen($par_names[1]) > 15) {
$par_names[1] = substr($par_names[1], 0, 12) . '...';
}
$par_names[2] = $results->item(2)->nodeValue;
if (strlen($par_names[2]) > 15) {
$par_names[2] = substr($par_names[2], 0, 12) . '...';
}
$par_names[3] = $results->item(3)->nodeValue;
if (strlen($par_names[3]) > 15) {
$par_names[3] = substr($par_names[3], 0, 12) . '...';
}
}
$classname2 = 'followers';
$results2 = $xpath->query("//*[@class and contains(concat(' ', normalize-space(@class), ' '), ' {$classname2} ')]");
if ($results2->length > 0) {
$followers = $results2->item(0)->nodeValue;
}
$html = gen_html($images, $par_names, $owner, $followers);
return $html;
} else {
return '<h2>Content not available!</h2>';
}
}
}
public function processRawContent($cr)
{
$processCR = array();
//Raw HTML to DOM/XPATH
$dom = new DOMDocument();
$searchPage = mb_convert_encoding($cr['content_raw'], 'HTML-ENTITIES', "UTF-8");
@$dom->loadHTML($searchPage);
@($xpath = new DomXpath($dom));
$processCR['content_raw'] = $cr['content_raw'];
//userID :
$divProfil = $xpath->query('//td[@class = "messCase2"]//a[contains(@href, "profil-")]');
if ($divProfil->length == 0) {
//pas de profil
$processCR['hfr_user_id'] = 1;
} else {
$processCR['hfr_user_id'] = $divProfil->item(0)->getAttribute('href');
$processCR['hfr_user_id'] = preg_replace("/[^0-9]+/", "", $processCR['hfr_user_id']);
}
//echo $processActu['hfr_user_id']."<br>";
//post ID
$processCR["hfr_post_id"] = $xpath->query('//a[contains(@href, "#t")]')->item(0)->getAttribute('href');
$processCR['hfr_post_id'] = str_replace("#t", "", $processCR["hfr_post_id"]);
//pseudo :
$processCR['username'] = $xpath->query('//td[@class = "messCase1"]//b[@class = "s2"]')->item(0)->textContent;
$processCR['username'] = preg_replace('/[^(\\x20-\\x7F)]*/', '', htmlentities($processCR['username']));
//date posted
$divDate = $xpath->query('//div[@class="toolbar"]/div[@class="left"]');
$oldate = preg_replace("/[^0-9:-]+/", "", $divDate->item(0)->nodeValue);
$processCR['date_posted'] = date('Y-m-d H:i', strtotime($oldate));
//echo $crDatePosted.' ';
//le contenu du CR
$divCR = $xpath->query('//div[@id = "para' . $processCR['hfr_post_id'] . '"]/node()');
//print_r($divCR);
$processHTMLstep1 = '';
foreach ($divCR as $node) {
//var_dump($actuNode);
//echo $dom->saveHTML($actuNode)."\n";
if ($node->nodeType == XML_ELEMENT_NODE && ($node->getAttribute('class') == 'edited' || $node->getAttribute('class') == 'signature' || $node->getAttribute('style') == 'clear: both;')) {
continue;
}
if ($node->nodeValue || $node->childNodes) {
$processHTMLstep1 .= htmlentities($dom->saveHTML($node));
}
}
$processHTMLfinal = $processHTMLstep1;
$processHTMLfinal = str_replace("\n", '', $processHTMLfinal);
$processHTMLfinal = str_replace("\r", '', $processHTMLfinal);
$processHTMLfinal = htmlentities(str_replace('href="/hfr/', 'target="_blank" href="http://forum.hardware.fr/hfr/', html_entity_decode($processHTMLfinal)));
$processHTMLfinal = htmlentities(preg_replace('/style="[a-zA-Z0-9:;\\.\\s\\(\\)\\-\\,]*"/i', '', html_entity_decode($processHTMLfinal)));
$processHTMLfinal = htmlentities(preg_replace('/onload="[^\\"]+"/i', 'onload="return;"', html_entity_decode($processHTMLfinal)));
$processHTMLfinal = htmlentities(preg_replace('/title="[^\\"]+"/i', '', html_entity_decode($processHTMLfinal)));
$processHTMLfinal = str_replace("hfr-rehost.net/", 'reho.st/', $processHTMLfinal);
$processHTMLfinal = htmlentities(str_replace("<p><br></p>", '', html_entity_decode($processHTMLfinal)));
$processCR['content'] = $processHTMLfinal;
return $processCR;
}
public function query($query, $options = [])
{
$this->result = file_get_contents($this->url . "select?" . $this->make_query($query, $options));
if ($this->result) {
$this->xml = new DOMDocument("1.0", "UTF-8");
$this->xml->formatOutput = true;
$this->xml->loadXML($this->result);
$xpath = new DomXpath($this->xml);
$resultNode = $xpath->query("//result");
if ($resultNode->length) {
$this->rows = $resultNode->item(0)->getAttribute("numFound");
$this->data = $xpath->query("//doc");
}
}
}
public function openGame($idx)
{
$file = $this->games[$idx]['name'];
$doc = new DOMDocument();
$doc->preserveWhiteSpace = true;
$doc->formatOutput = true;
// modify state
$libxml_previous_state = libxml_use_internal_errors(true);
// parse
$doc->loadHTMLFile("Games/" . $file);
// Usuwanie scriptow - bład z tagamiw scripcie
$domNodeList = $doc->getElementsByTagname('script');
$domElemsToRemove = array();
foreach ($domNodeList as $domElement) {
$domElemsToRemove[] = $domElement;
}
foreach ($domElemsToRemove as $domElement) {
$domElement->parentNode->removeChild($domElement);
}
$xpath = new DomXpath($doc);
// traverse all results
foreach ($xpath->query('//div[@tiddler]') as $tid) {
if ($tid->getAttribute('tiddler') != 'checkvars') {
$this->tiddlers[$tid->getAttribute('tiddler')] = $tid->nodeValue;
}
}
// handle errors
libxml_clear_errors();
// restore
libxml_use_internal_errors($libxml_previous_state);
}
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 run()
{
$document = jqm_use($this->node->_parentElement);
$doc = $document->_DOM;
if ($doc->doctype) {
$doc->removeChild($doc->doctype);
}
$pt = $this->node->getPathByID();
if ($pt) {
$path = new DomXpath($doc);
$get = $path->query($pt);
if ($get->length > 0) {
$this->node->_attributes = array();
$this->node->_innerHtml = NULL;
$this->node->_name = NULL;
$this->node->_localName = NULL;
$this->node->_length = false;
$save = $get->item(0)->parentNode->removeChild($get->item(0));
$document->offsetUnset($this->node->_token);
}
} else {
$document->offsetUnset($this->node->_token);
$this->node->_length = false;
}
}
public function run($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 $this;
}
public function convert($mailchimp)
{
$vars = array('*|MC:SUBJECT|*' => '<?php echo $this->getSubject() ?>', '*|ARCHIVE|*' => '<?php echo $this->getViewInBrowserUrl() ?>', '*|CURRENT_YEAR|*' => '<?php echo date("Y") ?>', '*|UNSUB|*' => '<?php echo $this->getUnsubscribeUrl() ?>', '*|FACEBOOK:PROFILEURL|*' => '<?php echo $this->getFacebookUrl() ?>', '*|TWITTER:PROFILEURL|*' => '<?php echo $this->getTwitterUrl() ?>', '*|LIST:COMPANY|*' => '<?php echo $this->getStoreName() ?>', '*|LIST:DESCRIPTION|*' => '', '*|HTML:LIST_ADDRESS_HTML|*' => '*****@*****.**', '*|IF:REWARDS|* *|HTML:REWARDS|*' => '', '*|FORWARD|*' => '', '/*@editable*/' => '', '*|UPDATE_PROFILE|*' => '', 'mc:repeatable' => '', 'mc:allowtext' => '', 'mc:hideable' => '', 'mc:allowdesigner' => '', '*|IFNOT:ARCHIVE_PAGE|*' => '', '*|END:IF|*' => '', 'mc:edit' => 'mcedit');
foreach ($vars as $old => $new) {
$mailchimp = str_replace($old, $new, $mailchimp);
}
$dom = new DOMDocument();
$dom->recover = true;
$dom->strictErrorChecking = false;
@$dom->loadHTML($mailchimp);
$xpath = new DomXpath($dom);
$items = $xpath->query('//*[@mcedit]');
foreach ($items as $mcedit) {
$area = 'area.' . $mcedit->getAttribute('mcedit');
if ($area == 'area.monkeyrewards') {
$mcedit->parentNode->removeChild($mcedit);
continue;
}
$mcedit->removeAttribute('mcedit');
$a = $dom->createComment('<?php echo $this->area("' . $area . '"); ?>');
// $b = $dom->createComment('{{/ivar}}');
$mcedit->parentNode->insertBefore($a, $mcedit);
// $mcedit->parentNode->insertBefore($b);
}
$dom->formatOutput = true;
$dom->preserveWhitespace = false;
$html = $dom->saveHTML();
$html = str_replace('%7B', '{', $html);
$html = str_replace('%7D', '}', $html);
$html = str_replace('%20', ' ', $html);
$html = preg_replace('/mc:label="[a-zA-Z0-9_]*"/', '', $html);
$html = preg_replace('/mc:variant="[^"]*"/', '', $html);
$html = str_replace('%24', '$', $html);
return $html;
}
/**
* @param $content
*
* @return \DOMElement
*/
private function getImage($content)
{
$dom = new \DOMDocument();
$dom->loadHTML($content);
$xpath = new \DomXpath($dom);
return $xpath->query('//img[@alt="Exception detected!"]')->item(0);
}
public function run($query = '*')
{
$path = explode('/', $this->node->_path);
$el = $this->__toDomElement();
$document = jqm_use($this->node->_parentElement);
if (count($path) > 0) {
if ($query === NULL || $query == '*') {
$query = $el->parentNode->tagName;
}
$t = count($path);
$doc = $document->_DOM;
if ($doc->doctype) {
$doc->removeChild($doc->doctype);
}
$xpath = new DomXpath($doc);
for ($i = 0; $t > $i;) {
array_pop($path);
++$i;
if (empty($path)) {
continue;
}
$find = $xpath->query(implode('/', $path));
if ($find->length > 0) {
$el = $this->createFragment($doc->saveHtml($find->item(0)), false);
$k = key($el->_domId);
$is = $el->is($query);
if ($is->get() === true) {
$el->_path = $find->item(0)->getNodepath();
if ($find->item(0)->nextSibling) {
$el->_next_path = $find->item(0)->nextSibling->getNodepath();
} else {
$el->_next_path = NULL;
}
if ($find->item(0)->previousSibling) {
$el->_prev_path = $find->item(0)->previousSibling->getNodepath();
} else {
$el->_prev_path = NULL;
}
if ($find->item(0)->parentNode) {
$el->_parent_path = $find->item(0)->parentNode->getNodepath();
} else {
$el->_parent_path = NULL;
}
$el->selector = $document->match_selector($query);
$this->node = $el;
$this->node->setDomID($el->_parentElement . $el->_token);
return $this;
break;
}
}
}
$this->node = new jqueryphp_abstracts_prevObject($this->node->_selector . ':parent(*)');
return $this;
} else {
return $this->node = new jqueryphp_abstracts_prevObject($this->node->_selector . ':parent(*)');
}
}
/**
* Calculates the read time of a post when created or updated.
*
* @param int $post_id Post id to calculate read time for.
* @param object $post Post object.
* @param boolean $update Is this an update or new.
*
* @return void
*/
function hh_calculate_and_update_post_read_time($post_id, $post, $update)
{
global $pagenow;
// Only those allowed.
if (!current_user_can('edit_posts', $post_id)) {
return;
}
// Resolve the issues with DOM reader with Quick Draft.
// https://github.com/Kenshino/HelpHub/commit/faba446e699893eb95c1eef4cf9f0e38c03a4680.
if ('post-new.php' === $pagenow) {
return;
}
// Store post content for raw usage.
$post_content = $post->post_content;
if (empty($post_content)) {
return;
}
// No post revisions.
if (wp_is_post_revision($post_id)) {
return;
}
// Get post types that need to have read time applied.
$calculate_for_posts = apply_filters('read_time_types', array('post'));
// If the post type is not found then return.
if (!in_array($post->post_type, $calculate_for_posts)) {
return;
}
// Average words per minute integer.
$average_word_per_minute = apply_filters('read_time_average', 175);
// Simple adjustment for pre tags.
libxml_use_internal_errors(true);
$data = new DOMDocument();
$data->loadHTML($post_content);
$xpath = new DomXpath($data);
libxml_use_internal_errors(false);
$pre_tags = array();
$word_count_offset = 0;
// Offset weight. Offset word count will be timed by this number.
$offset_weight = apply_filters('read_time_offset_weight', 1);
foreach ($xpath->query('//pre') as $node) {
$pre_tags[] = $node->nodeValue;
$word_count_offset = str_word_count($node->nodeValue) + $word_count_offset * $offset_weight;
}
// Word count.
$word_count = str_word_count(wp_strip_all_tags($post_content)) + $word_count_offset;
// Calculate basic read time.
$readtime = round($word_count / ($average_word_per_minute / 60));
// Grad and count all images.
preg_match_all('/(img|src)\\=(\\"|\')[^\\"\'\\>]+/i', $post->post_content, $media);
// Adjust read time given the number of images.
if ($image_count = count($media[0])) {
$readtime = $image_count * 12 - $image_count + $readtime;
}
// Update the post read time.
update_post_meta($post_id, '_read_time', $readtime);
}
public static function user_id_from_name($user)
{
$url = Thingiverse::BASE_URL . "/{$user}";
$dom = new DomDocument("1.0");
@$dom->loadHTMLFile($url);
// use @ to suppress parser warnings
$xp = new DomXpath($dom);
$rss_url = $xp->query("//link[@rel=\"alternate\"]")->item(0)->getAttribute("href");
$parts = explode(":", $rss_url);
return $parts[2];
}
private function templateSubsection($doc, $selector)
{
$tokenizer = new \Transphporm\Parser\Tokenizer($selector);
$xpathStr = $this->xPath->getXpath($tokenizer->getTokens());
$xpath = new \DomXpath($doc);
$nodes = $xpath->query($xpathStr);
$result = [];
foreach ($nodes as $node) {
$result[] = $node;
}
return $result;
}
function xselect($expression)
{
$xp = new DomXpath($this->getDocument());
$result = $xp->query($expression);
if ($result->length == 0) {
return null;
}
if ($result->length > 1) {
throw new Exception("More than one result returned.");
}
return $result->item(0)->nodeValue;
}
/**
* Deselect otion of a select given a select name/id and the option value to deselect.
*
* @param string $name select name/id
* @param string $option_value option value to be deselected
* @return idDoctrineTestBrowser
*/
public function deSelectOption($name, $option_value)
{
$position = 0;
$dom = $this->getResponseDom();
if (!$dom) {
throw new LogicException('Cannot select because there is no current page in the browser.');
}
$xpath = new DomXpath($dom);
if ($element = $xpath->query(sprintf('//select[(.="%s" or @id="%s" or @name="%s")]/option[@value="%s"]', $name, $name, $name, $option_value))->item($position)) {
$element->removeAttribute('selected');
} else {
throw new InvalidArgumentException(sprintf('Cannot find the select (%s) option value "%s" .', $name, $option_value));
}
return $this;
}
function translateLabelsOnPage($xpath_query)
{
global $nameTranslations;
foreach ($this->translate_langs as $lang) {
$doc = $this->fetchDocFromUrl($this->translatePageUrlTo($lang));
$xpath = new DomXpath($doc);
$graph = new SimpleGraph();
foreach ($xpath->query($xpath_query) as $a) {
$nodeId = $this->getNodeIdFromUrl($a->getAttribute('href'));
$uri = INST . 'institutions/' . $nodeId;
$this->graph->add_literal_triple($uri, RDFS_LABEL, $a->textContent, $lang);
$nameTranslations[$uri][$lang] = $a->textContent;
}
echo trim($graph->to_ntriples());
}
}
public function CreateDomDocument()
{
$dom = new Domdocument();
@$dom->loadHTML($this->htmlStack);
$xpath = new DomXpath($dom);
$elements = $xpath->query('//a[@class="' . $this->class . '"]');
foreach ($elements as $element) {
$element = $element->getAttribute('href');
// if a regex option has been supplied run it on the results
if ($this->regex) {
$element = preg_replace($this->regex, '', $element);
}
// remove the 6X 35x6(bulk beers) results with this regex
if (!preg_match('/\\dX(\\d+)?/', $element)) {
$this->elementsArray[] = ['title' => $element];
}
}
echo json_encode(['beers' => $this->elementsArray]);
}
public function run()
{
if (empty($this->node->_localName)) {
return $this;
}
$parent = $this->node->parent()->get();
if (is_a($parent, jqmel)) {
$path = $parent->_path;
$str = $parent->_innerHtml;
$document = jqm_use($this->node->_parentElement);
$dom = $document->_DOM;
if ($dom->doctype) {
$dom->removeChild($dom->doctype);
}
$xpath = new DomXpath($dom);
$find = $xpath->query($path);
if ($find->length > 0) {
$tp = new domdocument();
$tp->loadHtml($str);
if ($tp->doctype) {
$tp->removeChild($tp->doctype);
}
$frag = $dom->importNode($tp->firstChild->firstChild, true);
$find->item(0)->parentNode->replaceChild($frag, $find->item(0));
$find = $this->node->getPathByID();
if ($find) {
$this->node->_path = $find;
} else {
$pts = explode('/', $this->node->_path);
$ep = array_pop($pts);
$pts = explode('/', $frag->getNodepath());
array_pop($pts);
$pts[] = $ep;
$this->node->_path = implode('/', $pts);
}
}
}
}
function translateRolesTo($rolePositions, $lang)
{
global $RolesGraph;
$xpath = false;
if ($this->rolesNeedTranslating($rolePositions, $lang)) {
$doc = $this->fetchDocFromUrl($this->translatePageUrlTo($lang));
$xpath = new DomXpath($doc);
$translatedRoles = array();
foreach ($xpath->query(WHOISWHO_ROLE_XPATH) as $node) {
$roleLabels = $this->getRoleLabelsFromNode($node, $doc);
foreach ($roleLabels as $label) {
$translatedRoles[] = $label;
}
}
if (count($translatedRoles) !== count($rolePositions)) {
$this->log_message("different no of roles for {$lang} translation");
}
foreach ($rolePositions as $no => $roleUri) {
if (!$RolesGraph->roleHasLabelWithLang($roleUri, $lang, $RolesGraph) and isset($translatedRoles[$no])) {
$RolesGraph->add_literal_triple($roleUri, RDFS_LABEL, $translatedRoles[$no], $lang);
}
}
}
}
private function templateSubsection($css, $doc, $element)
{
$xpathStr = (new \Transphporm\Parser\CssToXpath($css, new \Transphporm\Parser\Value($this)))->getXpath();
$xpath = new \DomXpath($doc);
$nodes = $xpath->query($xpathStr);
$result = [];
foreach ($nodes as $node) {
$result[] = $element->ownerDocument->importNode($node, true);
}
return $result;
}
/**
* Looks for a link or a button in the response and submits the related request
*
* @param string The link/button value/href/alt
* @param array request parameters (associative array)
*
* @return sfWebBrowser The current browser object
*/
public function click($name, $arguments = array())
{
if (!$dom = $this->getResponseDom())
{
throw new Exception('Cannot click because there is no current page in the browser');
}
$xpath = new DomXpath($dom);
// text link, the name being in an attribute
if ($link = $xpath->query(sprintf('//a[@*="%s"]', $name))->item(0))
{
return $this->get($link->getAttribute('href'));
}
// text link, the name being the text value
if ($links = $xpath->query('//a[@href]'))
{
foreach($links as $link)
{
if(preg_replace(array('/\s{2,}/', '/\\r\\n|\\n|\\r/'), array(' ', ''), $link->nodeValue) == $name)
{
return $this->get($link->getAttribute('href'));
}
}
}
// image link, the name being the alt attribute value
if ($link = $xpath->query(sprintf('//a/img[@alt="%s"]/ancestor::a', $name))->item(0))
{
return $this->get($link->getAttribute('href'));
}
// form, the name being the button or input value
if (!$form = $xpath->query(sprintf('//input[((@type="submit" or @type="button") and @value="%s") or (@type="image" and @alt="%s")]/ancestor::form', $name, $name))->item(0))
{
throw new Exception(sprintf('Cannot find the "%s" link or button.', $name));
}
// form attributes
$url = $form->getAttribute('action');
$method = $form->getAttribute('method') ? strtolower($form->getAttribute('method')) : 'get';
// merge form default values and arguments
$defaults = array();
foreach ($xpath->query('descendant::input | descendant::textarea | descendant::select', $form) as $element)
{
$elementName = $element->getAttribute('name');
$nodeName = $element->nodeName;
$value = null;
if ($nodeName == 'input' && ($element->getAttribute('type') == 'checkbox' || $element->getAttribute('type') == 'radio'))
{
if ($element->getAttribute('checked'))
{
$value = $element->getAttribute('value');
}
}
else if (
$nodeName == 'input'
&&
(($element->getAttribute('type') != 'submit' && $element->getAttribute('type') != 'button') || $element->getAttribute('value') == $name)
&&
($element->getAttribute('type') != 'image' || $element->getAttribute('alt') == $name)
)
{
$value = $element->getAttribute('value');
}
else if ($nodeName == 'textarea')
{
$value = '';
foreach ($element->childNodes as $el)
{
$value .= $dom->saveXML($el);
}
}
else if ($nodeName == 'select')
{
if ($multiple = $element->hasAttribute('multiple'))
{
$elementName = str_replace('[]', '', $elementName);
$value = array();
}
else
{
$value = null;
}
$found = false;
foreach ($xpath->query('descendant::option', $element) as $option)
{
if ($option->getAttribute('selected'))
{
$found = true;
if ($multiple)
{
$value[] = $option->getAttribute('value');
}
else
{
$value = $option->getAttribute('value');
}
}
}
// if no option is selected and if it is a simple select box, take the first option as the value
if (!$found && !$multiple)
{
$value = $xpath->query('descendant::option', $element)->item(0)->getAttribute('value');
}
}
if (null !== $value)
{
$this->parseArgumentAsArray($elementName, $value, $defaults);
}
}
// create request parameters
$arguments = sfToolkit::arrayDeepMerge($defaults, $this->fields, $arguments);
if ('post' == $method)
{
return $this->post($url, $arguments);
}
else
{
return $this->get($url, $arguments);
}
}
public static function BUILD(Student $student, $file = null)
{
$zip = new \ZipArchive();
$dom = new \bloc\DOM\Document('data/student-site');
$xpath = new \DomXpath($dom);
$parse = function ($context) use(&$parse, $xpath, $dom) {
foreach ($xpath->query('./file', $context) as $file) {
if ($file->hasAttribute('pattern')) {
$pattern = strtolower(str_replace(' ', '-', sprintf($file->getAttribute('pattern'), $context->parentNode->getAttribute('title'))));
$file->setAttribute('name', $pattern);
}
}
foreach ($xpath->query('./directory', $context) as $directory) {
$name = $directory->getAttribute('name');
if ($directory->hasAttribute('clone')) {
$replacements = $xpath->query($directory->getAttribute('clone'));
foreach ($replacements as $replacement) {
$cloned = $directory->appendChild($replacement->cloneNode(!$replacement->hasAttribute('clone')));
foreach ($xpath->query('./file[@clone="ignore"]', $cloned) as $ignore) {
$cloned->removeChild($ignore);
}
}
$parent = $directory->parentNode;
$directory = $parent->removeChild($directory);
$pattern = $directory->getAttribute('pattern');
foreach (explode('|', $name) as $idx) {
$newnode = $parent->appendChild($directory->cloneNode(true));
$newnode->setAttribute('name', strtolower($idx));
$newnode->setAttribute('title', sprintf($pattern, $idx));
$parse($newnode);
}
} else {
if ($directory->childNodes->length > 0) {
$parse($directory);
}
}
}
};
call_user_func($parse, $dom->documentElement, '', './');
$filepath = 'views/student/site';
$cache = [];
$zip->open($file, \ZIPARCHIVE::OVERWRITE);
$zipper = function ($context, $path) use(&$zipper, $zip, $filepath, $student, &$cache) {
$path .= '/';
$title = $context->getAttribute('title') ?: '';
foreach ($context->childNodes as $level) {
$type = $level->nodeName;
$name = $level->getAttribute('name');
if ($type == 'file') {
$ext = substr($name, strpos($name, '.'));
$base = substr($name, 0, strpos($name, '.'));
if (!array_key_exists($name, $cache)) {
if (file_exists(PATH . $filepath . $path . $name)) {
$cache[$name] = $filepath . $path . $name;
} else {
if (file_exists(PATH . $filepath . $level->getAttribute('template'))) {
$cache[$name] = $filepath . $level->getAttribute('template');
}
}
}
if ($ext == '.html' || $ext == '.xml') {
$view = new \bloc\view($cache[$name]);
if ($context->hasAttribute('template')) {
$view->template = $filepath . $context->getAttribute('template');
}
$text = (string) $view->render(['id' => $name, 'empty' => null, 'student' => $student, 'title' => $title, 'datapath' => substr($path, 1, -6), 'embeds' => strtolower(str_replace(' ', '-', $title)), 'domain' => DOMAIN]);
if ($ext != '.xml') {
$text = substr($text, strlen('<?xml version="1.0"?>') + 1);
}
$zip->addFromString(substr($path . $name, 1), $text);
} else {
$zip->addFile(PATH . $cache[$name], substr($path . $name, 1));
}
} else {
$zip->addEmptyDir($path . '/');
$zipper($level, $path . $name);
}
}
};
$zipper($dom->documentElement, '', '');
$zip->close();
return $file;
}
public function gibbactuPostProcess()
{
$Actu = new Actu($this->db);
$NotProcessedActus = $Actu->find(array('processed = ?', 0));
//print_r($NotProcessedActus);
libxml_use_internal_errors(true);
ob_start();
foreach ($NotProcessedActus as $NPA) {
$processActu = array();
//echo $NPA->id."<br>";
//Raw HTML to DOM/XPATH
$dom = new DOMDocument();
$searchPage = mb_convert_encoding($NPA->content_raw, 'HTML-ENTITIES', "UTF-8");
@$dom->loadHTML($searchPage);
@($xpath = new DomXpath($dom));
//userID :
$processActu['hfr_user_id'] = $xpath->query('//td[@class = "messCase2"]//a[contains(@href, "profil-")]')->item(0)->getAttribute('href');
$processActu['hfr_user_id'] = preg_replace("/[^0-9]+/", "", $processActu['hfr_user_id']);
//echo $processActu['hfr_user_id']."<br>";
//pseudo :
$processActu['username'] = $xpath->query('//td[@class = "messCase1"]//b[@class = "s2"]')->item(0)->textContent;
$processActu['username'] = preg_replace('/[^(\\x20-\\x7F)]*/', '', htmlentities($processActu['username']));
//echo $processActu['username']."<br>";
//if ($processActu['username'] <> 'roswellentongues') {
//continue;
//}
//Check si le smiley est bien directement dans le message
$checkSmileyGAParent = $xpath->query('//img[contains(@title, "[:gibbactu]")]//..//..//..//div[starts-with(@id, "para")]');
if (!$checkSmileyGAParent->length) {
//pas de smiley, on skip
//echo "smiley pas à la racine<br/><br/><br/>";
$NPA->active = 0;
$NPA->save();
continue;
}
//post ID
$processActu["hfr_post_id"] = $xpath->query('//a[contains(@href, "#t")]')->item(0)->getAttribute('href');
$processActu['hfr_post_id'] = str_replace("#t", "", $processActu["hfr_post_id"]);
//echo $processActu['hfr_post_id'];
//if ($processActu['hfr_post_id'] <> '10764130') {
// continue;
//}
//date posted
$divDate = $xpath->query('//div[@class="toolbar"]/div[@class="left"]');
$oldate = preg_replace("/[^0-9:-]+/", "", $divDate->item(0)->nodeValue);
$processActu['date_posted'] = date('Y-m-d H:i', strtotime($oldate));
//echo $crDatePosted.' ';
//echo $processActu['username']."\n";
//le contenu au même niveau que le smiley
$actuNodesZero = $xpath->query('//img[contains(@title, "[:gibbactu]")]/../node()');
$processHTMLstep0 = '';
$startInsert = false;
foreach ($actuNodesZero as $actuNodeZ) {
if ($actuNodeZ->nodeType == XML_ELEMENT_NODE && $actuNodeZ->getAttribute('title') == '[:gibbactu]') {
$startInsert = true;
continue;
}
if ($actuNodeZ->nodeType == XML_ELEMENT_NODE && !strcasecmp(trim($actuNodeZ->nodeValue), 'gibbactu')) {
continue;
}
if (!$startInsert) {
continue;
}
//echo "len =".strlen(trim($actuNodeZ->nodeValue))."\n";
//echo "htmlen =".strlen(trim($processHTMLstep0))."\n";
//echo "html =".($dom->saveHTML($actuNodeZ))."\n";
if ($actuNodeZ->nodeType == XML_ELEMENT_NODE) {
//echo "tag =".$actuNodeZ->tagName."\n";
} else {
//echo "tag =pasnoded\n";
if (!strcasecmp(trim(htmlentities($dom->saveHTML($actuNodeZ))), 'gibbactu')) {
//echo "== gibbactu\n";
continue;
}
}
if (trim(htmlentities($dom->saveHTML($actuNodeZ))) == ' ') {
//echo ">> on skip\n";
continue;
}
if (!strlen(trim($processHTMLstep0)) && $actuNodeZ->nodeType == XML_ELEMENT_NODE && $actuNodeZ->tagName == 'br') {
//echo ">> on skip\n";
continue;
}
//echo "==\nNew HTML:\n".htmlentities($processHTMLstep0)."\n============\n";
//var_dump($actuNodeZ);
$processHTMLstep0 .= htmlentities($dom->saveHTML($actuNodeZ));
//echo htmlentities($dom->saveHTML($actuNodeZ))."\n".$actuNodeZ->nodeValue."\n\n";
}
//On va chercher tout le contenu APRES le smiley gibbactu
$actuNodes = $xpath->query('//img[contains(@title, "[:gibbactu]")]/../following-sibling::*');
//var_dump($actuNodes);
$processHTMLstep1 = $processHTMLstep0;
foreach ($actuNodes as $actuNode) {
//echo $dom->saveHTML($actuNode)."\n";
//echo "===\n".$actuNode->tagName." = ".$actuNode->getAttribute('class')."\n\n";
//var_dump($actuNode);
if ($actuNode->getAttribute('class') == 'edited' || $actuNode->getAttribute('class') == 'signature' || $actuNode->getAttribute('style') == 'clear: both;') {
continue;
}
if ($actuNode->nodeValue || $actuNode->childNodes) {
$processHTMLstep1 .= htmlentities($dom->saveHTML($actuNode));
} else {
var_dump($actuNode->childNodes);
echo "noNodeValue=" . $dom->saveHTML($actuNode) . "\n\n";
}
}
$processHTMLfinal = $processHTMLstep1;
$processHTMLfinal = str_replace("\n", '', $processHTMLfinal);
$processHTMLfinal = str_replace("\r", '', $processHTMLfinal);
$processHTMLfinal = htmlentities(str_replace("<p><br></p>", '', html_entity_decode($processHTMLfinal)));
$processHTMLfinal = htmlentities(str_replace('href="/hfr/', 'target="_blank" href="http://forum.hardware.fr/hfr/', html_entity_decode($processHTMLfinal)));
$processHTMLfinal = htmlentities(preg_replace('/style="[a-zA-Z0-9:;\\.\\s\\(\\)\\-\\,]*"/i', '', html_entity_decode($processHTMLfinal)));
$processHTMLfinal = htmlentities(preg_replace('/onload="[^\\"]+"/i', 'onload="return;"', html_entity_decode($processHTMLfinal)));
$processHTMLfinal = htmlentities(preg_replace('/title="[^\\"]+"/i', '', html_entity_decode($processHTMLfinal)));
if (!$processHTMLfinal) {
echo "aucun contenu<br/><br/><br/>";
$NPA->active = 0;
$NPA->save();
continue;
}
$NPA->hfr_cat_id = 5;
$NPA->hfr_subcat_id = 249;
$NPA->hfr_topic_id = 177180;
//$NPA->hfr_page_id
$NPA->hfr_post_id = $processActu['hfr_post_id'];
$NPA->hfr_user_id = $processActu['hfr_user_id'];
$NPA->username = $processActu['username'];
$NPA->date_modified = date('Y-m-d H:i:s');
$NPA->date_posted = $processActu['date_posted'];
$NPA->content = $processHTMLfinal;
$NPA->active = 1;
$NPA->processed = 1;
$NPA->save();
echo "\n\n\n\n\n";
}
$myStr = ob_get_contents();
ob_end_clean();
libxml_use_internal_errors(false);
echo $myStr;
//$this->f3->set('report',$myStr);
//echo Template::instance()->render('cron.htm');
}
public static function from_html_made_dom($dom)
{
$thing = new ThingiverseThing();
$d = new DOMDocument('1.0');
$b = $d->importNode($dom->cloneNode(true), true);
$d->appendChild($b);
$xp = new DomXpath($d);
$tmpelem = $xp->query("//div[@class=\"thing_info\"]/a")->item(0);
$thing->title = $tmpelem->childNodes->item(0)->wholeText;
$tmpelem = $xp->query("//div[@class=\"thing_info\"]/a")->item(1);
$thing->creator = $tmpelem->childNodes->item(0)->wholeText;
$thing->creator_url = Thingiverse::BASE_URL . "/{$thing->creator}";
// TODO: Get the time somehow :(
//$thing->created_at = "Unknown";
$tmpelem = $xp->query("//a")->item(2);
$thing->url = Thingiverse::BASE_URL . $tmpelem->getAttribute("href");
$thing->main_image = $tmpelem->childNodes->item(0)->getAttribute(src);
return $thing;
}
public function CLIassignment()
{
$doc = new \bloc\DOM\Document(self::SEMESTER);
$xml = new \DomXpath($doc);
$assignment = $xml->query('/course/assignments')->item(0)->appendChild($doc->createElement('assignment'));
$weeks = $xml->query('/course/classes/outline');
// title, points, id
echo "\nEnter Assignment Title: ";
$title = trim(fgets(STDIN));
$assignment->setAttribute('title', $title);
$id = preg_replace(['/\\s+/', '/[^a-z]/i'], ['-', ''], $title);
$assignment->setAttribute('id', $id);
echo "\nEnter number of points: ";
$points = trim(fgets(STDIN));
$assignment->setAttribute('points', (int) $points);
echo "\nWeek number: ";
$week = (int) trim(fgets(STDIN)) - 1;
if ($week >= 0 && $week < 15) {
$ref = $weeks->item($week)->appendChild($doc->createElement('assignment'));
$ref->setAttribute('ref', $id);
}
echo "\nAssignment Details: ";
$details = trim(fgets(STDIN));
if (!empty($details)) {
$assignment->setNodeValue($details);
}
$this->save($doc);
}
<?php
$result = 0;
passthru('cd ' . __DIR__ . ' && phpdoc --force', $result);
if ($result !== 0) {
die("phpdoc command not found.");
}
libxml_use_internal_errors(true);
$dom = new DomDocument();
$dom->loadHTMLFile("./docs/index.html");
$xpath = new DomXpath($dom);
// Remove "Global" namespace
$result = $xpath->query('//li/a[@href="namespaces/default.html"]');
if ($result->length > 0) {
$node = $result->item(0);
$node->parentNode->parentNode->removeChild($node->parentNode);
}
// Remove Packages
$result = $xpath->query('//div[@class="well"][2]');
if ($result->length > 0) {
$node = $result->item(0);
$node->parentNode->removeChild($node);
}
$dom->saveHTMLFile('./docs/index.html');
/**
* Check Result for Errors
*
* @param DomDocument $dom
* @throws Zend_Service_Exception
*/
protected static function _checkErrors(DomDocument $dom)
{
if ($dom->documentElement->getAttribute('stat') == 'fail') {
$xpath = new DomXpath($dom);
$err = $xpath->query('//err')->item(0);
throw new Zend_Service_Exception('Search failed due to error: ' . $err->getAttribute('msg') . ' (error #' . $err->getAttribute('code') . ')');
}
}