Searches for an node in the DOM tree for a given XPath expression or a CSS selector.
|
public find ( string $expression, string $type = Query::TYPE_CSS, boolean $wrapElement = true ) : |
||
| $expression | string | XPath expression or a CSS selector |
| $type | string | The type of the expression |
| $wrapElement | boolean | Returns array of \DiDom\Element if true, otherwise array of \DOMElement |
| Результат | ||
/**
* @dataProvider findProvider
*/
public function testFind($html, $selector, $type, $count)
{
$document = new Document($html, false);
$elements = $document->find($selector, $type);
$this->assertTrue(is_array($elements));
$this->assertEquals($count, count($elements));
}
protected function getGamesFromWeb($teamTournament)
{
//getting array of team aliases and participant id's
$aliases = ArrayHelper::map($teamTournament, 'alias', 'id');
$count = $this->tournament->num_tours;
//$html = new Document($this->tournament->autoProcessURL, true);
$html = new Document('italy.htm', true);
$table = $html->find('table.table.b-table-sortlist tbody')[0];
$j = 0;
$gamesFromWeb = [];
foreach ($table->find('tr') as $row) {
$time = $this->autoTimeToUnix($row->find('td.sport__calendar__table__date')[0]->text());
if ($time > time() - 60 * 60 * 24 * 7 * 2) {
$home = $row->find('td.sport__calendar__table__teams a.sport__calendar__table__team')[0]->text();
$guest = $row->find('td.sport__calendar__table__teams a.sport__calendar__table__team')[1]->text();
if (isset($aliases[$home]) && isset($aliases[$guest])) {
$gamesFromWeb[$j]['tour'] = $row->find('td.sport__calendar__table__tour')[0]->text();
$gamesFromWeb[$j]['date_time_game'] = $this->autoTimeToUnix($row->find('td.sport__calendar__table__date')[0]->text());
$gamesFromWeb[$j]['id_team_home'] = (int) $aliases[$home];
$gamesFromWeb[$j]['id_team_guest'] = (int) $aliases[$guest];
$homeScore = $row->find('td.sport__calendar__table__result span.sport__calendar__table__result__left')[0]->text();
$guestScore = $row->find('td.sport__calendar__table__result span.sport__calendar__table__result__right')[0]->text();
$gamesFromWeb[$j]['score_home'] = $this->calculateHomeScore($homeScore);
$gamesFromWeb[$j]['score_guest'] = $this->calculateHomeScore($guestScore);
$j++;
} else {
throw new \Exception('Error during alias parsing ' . $home . ' or ' . $guest);
}
}
}
$this->gamesFromWeb = $gamesFromWeb;
}
/**
* @param string $url
* @return array $this->data
*/
function parse_approvedevent($url)
{
$doc = new Document($url, true);
$title = $doc->find('.descr_approvedevent')[0]->text();
$this->data['title'] = trim(preg_replace('/\\s{2,}/', '', $title));
//remove whitespaces
foreach ($doc->find('div.caracs_approvedevent div') as $element) {
$div_id = $element->attr('id');
switch ($div_id) {
case 'ctl00_ContentPlaceHolder1_UFIApprovedEventsDetails1_OrganizingCompany_Panel':
$el = $element->find('a')[0]->text();
$this->data['orginizing_company'] = trim(preg_replace('/\\s{2,}/', '', $el));
break;
case 'ctl00_ContentPlaceHolder1_UFIApprovedEventsDetails1_OrganiserLocation_Panel':
$el = $element->find('span')[0]->text();
$this->data['orginiser_location'] = $el;
break;
case 'ctl00_ContentPlaceHolder1_UFIApprovedEventsDetails1_BusinessSectors_Panel':
$el = $element->find('span')[0]->text();
$this->data['business_sectors'] = $el;
break;
case 'ctl00_ContentPlaceHolder1_UFIApprovedEventsDetails1_Frequency_Panel':
$el = $element->find('span')[0]->text();
$this->data['frequency'] = $el;
break;
case 'ctl00_ContentPlaceHolder1_UFIApprovedEventsDetails1_EventOpenTo_Panel':
$el = $element->find('span')[0]->text();
$this->data['event_open_to'] = $el;
break;
case 'ctl00_ContentPlaceHolder1_UFIApprovedEventsDetails1_Email_Panel':
$el = $element->find('a')[0]->text();
$this->data['email'] = $el;
break;
case 'ctl00_ContentPlaceHolder1_UFIApprovedEventsDetails1_Website_Panel':
$el = $element->find('a')[0]->attr('href');
$this->data['website'] = $el;
break;
case 'ctl00_ContentPlaceHolder1_UFIApprovedEventsDetails1_Links_Panel':
foreach ($element->find('a') as $key => $href) {
$this->data['links'][$key] = $href->attr('href');
}
break;
}
}
$this->data['session'] = $this->parse_session($doc);
return $this->data;
}
/**
* @dataProvider findProvider
*/
public function testReturnDomElent($html, $selector, $type, $count)
{
$document = new Document($html, false);
$elements = $document->find($selector, $type, false);
$this->assertTrue(is_array($elements));
$this->assertEquals($count, count($elements));
if ($count > 0) {
$this->assertInstanceOf('DOMElement', $elements[0]);
}
}
/**
* Extract comments from DOM
*
* @param \DiDom\Document $dom
* @return array
*/
public function extractComments($dom)
{
$comments = [];
$nodes = $dom->find('.single-comment');
foreach ($nodes as $node) {
$comment = [];
$comment['text'] = $node->find('.text')[0]->text();
$comments[] = $comment;
}
return $comments;
}
public function testClass()
{
$html = '
<span class="odd first">Lorem ipsum dolor.</span>
<span class="even second">Tenetur totam, nostrum.</span>
<span class="odd third">Iste, doloremque, praesentium.</span>
';
$document = new Document($html);
$expected = ['Lorem ipsum dolor.', 'Iste, doloremque, praesentium.'];
$result = [];
foreach ($document->find('.odd') as $element) {
$result[] = $element->text();
}
$this->assertEquals($expected, $result);
$expected = ['Iste, doloremque, praesentium.'];
$result = [];
foreach ($document->find('.odd.third') as $element) {
$result[] = $element->text();
}
$this->assertEquals($expected, $result);
}
protected function getGamesFromWeb($teamTournament)
{
//getting array of team aliases and participant id's
$aliases = ArrayHelper::map($teamTournament, 'alias', 'id');
$count = $this->tournament->num_tours;
$j = 0;
$html = new Document($this->tournament->autoProcessURL, true);
//$html = new Document('pl.htm', true);
$results = $html->find('div.mainPart')[0];
$gamesFromWeb = [];
for ($i = 0; $i < $count; $i++) {
if (isset($results->find('div.stat.mB15 table.stat-table')[$i])) {
$tour = $html->find('h3.titleH3.bordered.mB10')[$i]->text();
$tour = $this->getTour($tour);
$resultTable = $results->find('div.stat.mB15 table.stat-table')[$i];
foreach ($resultTable->find('tbody tr') as $k => $one) {
if ($this->autoTimeToUnix($one->find('td.name-td')[0]->text()) > time() - 60 * 60 * 24 * 7 * 2 && $tour <= $count) {
if (isset($one->find('td.owner-td a.player')[0]) && isset($one->find('td.guests-td a.player')[0])) {
$owner = $one->find('td.owner-td a.player')[0]->text();
$guest = $one->find('td.guests-td a.player')[0]->text();
if (isset($aliases[$owner]) && isset($aliases[$guest])) {
$gamesFromWeb[$j]['id_team_home'] = (int) $aliases[$owner];
$gamesFromWeb[$j]['id_team_guest'] = (int) $aliases[$guest];
$gamesFromWeb[$j]['date_time_game'] = (int) $this->autoTimeToUnix($one->find('td.name-td')[0]->text());
$gamesFromWeb[$j]['tour'] = $tour;
$score = $one->find('td.score-td noindex')[0]->text();
$gamesFromWeb[$j]['score_home'] = $this->calculateHomeScore($score);
$gamesFromWeb[$j]['score_guest'] = $this->calculateGuestScore($score);
$j++;
} else {
throw new Exception('Error during alias parsing ' . $owner . ' or ' . $guest);
}
}
}
}
}
}
$this->gamesFromWeb = $gamesFromWeb;
}
/**
* @param $html
* @return array
*/
public function load($html)
{
$metaTags = ['canonical' => '', 'robots' => ''];
$this->crawler->loadHtml((string) $html);
foreach ($this->crawler->find('meta') as $meta) {
/** @var Element $meta */
$name = strtolower($meta->attr('name'));
$content = $meta->attr('content');
$metaTags[$name] = $content;
}
$links = [];
foreach ($this->crawler->find('a') as $link) {
/** @var Element $link */
$rel = $link->attr('rel');
$href = $link->attr('href');
if ('nofollow' === strtolower($rel)) {
continue;
}
$links[] = $href;
}
$this->links = array_unique($links);
$this->metaTags = $metaTags;
return ['links' => $links, 'meta' => $metaTags];
}
protected function execute(InputInterface $input, OutputInterface $output)
{
$container = $this->getContainer();
$em = $container->get('doctrine')->getManager();
//$utils = $container->get('sbranch.common.utils');
$items = $em->getRepository("AppBundle:Item")->findTimed();
$output->writeln('Total count items ' . sizeof($items));
$progress = new ProgressBar($output, sizeof($items));
$progress->start();
foreach ($items as $item) {
$progress->advance();
// if($item->getHtmlDocument()) {
// $document = new Document();
// $document->loadHtml($item->getHtmlDocument());
// } else {
//
// }
$document = new Document($item->getUrl(), true);
$title = $document->find('title');
//TODO: this parsing is shitcode
$piecesTitle = explode(":", $title[0]->text());
$piecesTitle2 = explode('-', trim(end($piecesTitle)));
$priceString = trim($piecesTitle2[0]);
$pricePieces = explode(' ', $priceString);
$currency = trim(end($pricePieces));
$price = trim(str_replace($currency, '', $priceString));
$price = trim(str_replace(' ', '', $price));
$title = $piecesTitle[0];
$item->setTitle($title);
if ($price != $item->getPrice()) {
$item->setPriceOld($item->getPrice());
}
$item->setPrice($price);
$item->setCurrency($currency);
$item->setLastCheck(new \DateTime());
$item->setHtmlDocument($document->html());
$em->persist($item);
}
$progress->finish();
$em->flush();
$output->writeln('Ok');
}
/**
* Parses a kat list (a search, user uploads, etc..)
* @param Document $html
* @param int $maxAge timestamp
* @return array
*/
public static function parseList($html, $maxAge = null)
{
$uploads = [];
foreach ($html->find("tr.odd, tr.even") as $i => $rawUpload) {
//echo $rawUpload->html();exit;
$elements = $rawUpload->find("td.center");
$date = strtotime($rawUpload->find(".center[title]")[0]->attr("title"));
if (!empty($maxAge) && $date < $maxAge) {
continue;
}
$link = $rawUpload->find("a.cellMainLink")[0];
// the uploader may be anonymous
$creator = null;
if ($rawUpload->has(".lightgrey.block a.plain")) {
$rawUser = $rawUpload->find(".lightgrey.block a.plain")[0];
$creator = new User(["nick" => $rawUser->text(), "link" => Kat::DOMAIN . $rawUser->attr("href")]);
}
$uploads[] = new BasicTorrent(["name" => $link->text(), "size" => $rawUpload->find(".nobr.center")[0]->text(), "created" => date("Y-m-d H:i:s", $date), "magnet" => $rawUpload->find("a[data-nop]")[0]->attr("href"), "torrent" => "https:" . $rawUpload->find("a[data-download]")[0]->attr("href"), "creator" => $creator, "files" => $elements[1]->text(), "seeders" => $elements[3]->text(), "leechers" => $elements[4]->text(), "link" => self::DOMAIN . $link->attr("href"), "comments" => (int) $rawUpload->find('div.iaconbox.center.floatright a')[0]->text()]);
}
return $uploads;
}
/**
* Searches for the element in the DOM tree.
*
* @param string $expression XPath expression or CSS selector
* @param string $type the type of the expression
* @param string $wrapList
*
* @return NodeList|DiDom\Element[]
*/
public function find($expression, $type = Query::TYPE_CSS, $wrapList = true)
{
$nodes = parent::find($expression, $type, false);
return $wrapList ? new NodeList($nodes) : $nodes;
}
function data($url)
{
$document = new Document($url, true);
$a = $document->find('.game-info p');
return $a[1]->text();
}
/**
* Gets the full movie crew divided in departments
* @return array
*/
public function getCastCredits()
{
$content = new Document($this->getCredits());
$titles = $content->find("#fullcredits_content h4");
$persons = $content->find(".simpleCreditsTable tbody");
$crew = [];
// skip useless h4s, (DiDom doesnt have :not pseudo class, so we make a foreach)
foreach ($titles as $pos => $h4) {
if ($h4->hasAttribute("id") || $h4->hasAttribute("name")) {
unset($titles[$pos]);
}
}
$titles = array_values($titles);
foreach ($titles as $pos => $h4) {
$title = trim($h4->text(), " \t\n\r\v ");
switch ($title) {
case "Directed by":
$name = "director";
break;
case "Music by":
$name = "music";
break;
case "Cinematography by":
$name = "cinematography";
break;
case "Film Editing by":
$name = "editing";
break;
case "Casting By":
$name = "casting";
break;
case "Production Design by":
$name = "production_design";
break;
case "Art Direction by":
$name = "art_direction";
break;
case "Set Decoration by":
$name = "set_decoration";
break;
case "Costume Design by":
$name = "costume_design";
break;
case "Makeup Department":
$name = "makeup_department";
break;
case "Production Management":
$name = "production_management";
break;
case "Art Department":
$name = "art_department";
break;
case "Sound Department":
$name = "sound_department";
break;
case "Special Effects by":
$name = "special_effects";
break;
case "Visual Effects by":
$name = "visual_effects";
break;
case "Stunts":
$name = "stunts";
break;
case "Camera and Electrical Department":
$name = "camera_department";
break;
case "Animation Department":
$name = "animation_department";
break;
case "Casting Department":
$name = "casting_department";
break;
case "Costume and Wardrobe Department":
$name = "wardrobe_department";
break;
case "Editorial Department":
$name = "editorial_department";
break;
case "Location Management":
$name = "location_management";
break;
case "Music Department":
$name = "music_department";
break;
case "Transportation Department":
$name = "transportation_department";
break;
case "Storyline":
$name = "storyline";
break;
case "Photo & Video":
$name = "photo";
break;
default:
continue 2;
break;
}
if (!isset($crew[$name])) {
$crew[$name] = [];
}
$regex = "/name\\/nm(\\d+)\\/(?:.*)/";
if (!isset($persons[$pos])) {
continue;
}
foreach ($persons[$pos]->find("a") as $person) {
preg_match($regex, $person->attr("href"), $matches);
if (!isset($matches[1]) || empty($matches[1])) {
continue;
}
$crew[$name][] = ["id" => $matches[1], "name" => trim($person->text())];
}
}
return $crew;
}
public function testFirst()
{
$html = '<ul><li>One</li><li>Two</li><li>Three</li></ul>';
$document = new Document($html, false);
$items = $document->find('ul > li');
$this->assertEquals($items[0]->getNode(), $document->first('ul > li')->getNode());
$this->assertEquals('One', $document->first('ul > li::text'));
$document = new Document();
$this->assertNull($document->first('ul > li'));
}
public function testReplace()
{
$html = '<ul><li>One</li><li>Two</li><li>Three</li></ul>';
$document = new Document($html, false);
$first = $document->find('li')[0];
$third = $document->find('li')[2];
$this->assertEquals($first->getNode(), $first->replace($third)->getNode());
$this->assertEquals($third->getNode(), $document->find('li')[0]->getNode());
$this->assertCount(3, $document->find('li'));
$document = new Document($html, false);
$first = $document->find('li')[0];
$third = $document->find('li')[2];
$this->assertEquals($first->getNode(), $first->replace($third, false)->getNode());
$this->assertEquals($third->getNode(), $document->find('li')[0]->getNode());
$this->assertCount(2, $document->find('li'));
}
public function testGetLineNo()
{
$element = new Element('div');
$this->assertEquals(0, $element->getLineNo());
$html = '<ul>
<li>One</li>
<li>Two</li>
<li>Three</li>
</ul>';
$document = new Document($html, false);
$this->assertEquals(4, $document->find('li')[2]->getLineNo());
}
<?php
require_once 'vendor/autoload.php';
use DiDom\Document;
$document = new Document('http://www.opencart.com/index.php?route=extension/extension&filter_license=0', true);
$extensions = $document->find('#content > div.extension-grid > div > div.name > a');
foreach ($extensions as $extension) {
echo $extension->text(), "<br>";
}
