PHP Crawler::parents方法代码示例

 /**
  * get itemprop given value && set the couple in Card
  * 
  * if isItemref is set to true, then only itemprop who have no direct itemscope parent in node will be recorded
  * 
  * @param Crawler $node
  * @param iCard $card
  * @param boolean $isItemref
  * @return boolean
  */
 private function setCardProperty(Crawler $node, CardInterface $card, $isItemref = false)
 {
     $property = $node->attr('itemprop');
     $nd = $node;
     if ($node->parents()->attr('itemscope') !== null) {
         return false;
     }
     if ($this->nestedScope($node, $card, $property, $isItemref)) {
         return true;
     }
     if ($this->manageRawProperty($node, $property, $card)) {
         return true;
     }
     if ($this->manageImgProperty($node, $property, $card)) {
         return true;
     }
     if ($this->manageLinkProperty($node, $property, $card)) {
         return true;
     }
     if ($this->manageNumericProperty($node, $property, $card)) {
         return true;
     }
     if ($isItemref) {
         while (count($nd)) {
             if ($nd->attr('itemscope') !== null) {
                 return true;
             }
             $nd = $nd->parents();
         }
     }
     $card->{$property} = trim($node->text());
 }

 /**
  * @param \Symfony\Component\DomCrawler\Crawler $node
  *
  * @return \Symfony\Component\DomCrawler\Form|\Symfony\Component\DomCrawler\Field\ChoiceFormField[]|\Symfony\Component\DomCrawler\Field\FileFormField[]
  */
 protected function getFormFor($node)
 {
     $form = $node->parents()->filter('form')->first();
     if (!$form) {
         $this->fail('The selected node does not have a form ancestor.');
     }
     $action = $this->getFormUrl($form);
     if (isset($this->forms[$action])) {
         return $this->forms[$action];
     }
     /** @var \DOMElement $autoSubmit */
     $autoSubmit = new \DOMElement('input');
     $autoSubmit = $form->current()->appendChild($autoSubmit);
     $autoSubmit->setAttribute('type', 'submit');
     // for forms with no submits
     $autoSubmit->setAttribute('name', 'codeception_added_auto_submit');
     // Symfony2.1 DOM component requires name for each field.
     $formSubmits = $form->filter('*[type=submit]');
     $values = null;
     // If there are more than one submit (+1 auto_added) in one form we should add value of actually clicked one
     if ($formSubmits->count() > 2) {
         $nodeItem = $node->getNode(0);
         foreach ($formSubmits as $formSubmit) {
             if ($formSubmit === $nodeItem) {
                 $values = array($nodeItem->getAttribute('name') => $nodeItem->getAttribute('value'));
                 break;
             }
         }
     }
     $form = $formSubmits->form($values);
     $this->forms[$action] = $form;
     return $this->forms[$action];
 }

 /**
  * Returns the DomCrawler\Form object for the form pointed to by
  * $node or its closes form parent.
  *
  * @param \Symfony\Component\DomCrawler\Crawler $node
  * @return \Symfony\Component\DomCrawler\Form
  */
 protected function getFormFor(Crawler $node)
 {
     if (strcasecmp($node->first()->getNode(0)->tagName, 'form') === 0) {
         $form = $node->first();
     } else {
         $form = $node->parents()->filter('form')->first();
     }
     if (!$form) {
         $this->fail('The selected node is not a form and does not have a form ancestor.');
     }
     $action = (string) $this->getFormUrl($form);
     if (!isset($this->forms[$action])) {
         $this->forms[$action] = $this->getFormFromCrawler($form, $action);
     }
     return $this->forms[$action];
 }

 /**
  * @param Crawler $block
  * @return array
  */
 public function getSimilarSiblingsInfo(Crawler $block)
 {
     $all_siblings = $block->parents()->first()->children();
     $siblings = [$block];
     $sample_node = $block->getNode(0);
     $sample_tag = $block->nodeName();
     $sample_classes = explode(' ', $block->extract(['class'])[0]);
     foreach ($all_siblings as $sibling) {
         if ($sample_node->isSameNode($sibling)) {
             continue;
         }
         $to_add = false;
         if ($sibling->nodeName == $sample_tag) {
             if ($sample_classes === FALSE) {
                 $to_add = true;
             } else {
                 $sibling_classes = explode(' ', $sibling->getAttribute('class'));
                 $matched_classes = array_intersect($sample_classes, $sibling_classes);
                 if (!empty($matched_classes)) {
                     if ($matched_classes != $sample_classes) {
                         $sample_classes = $matched_classes;
                     }
                     $to_add = true;
                 }
             }
         }
         if ($to_add) {
             $siblings[] = new Crawler($sibling);
         }
     }
     return [$siblings, ['tag' => $sample_tag, 'classes' => $sample_classes]];
 }

 /**
  * @param \Symfony\Component\DomCrawler\Crawler $node
  *
  * @return \Symfony\Component\DomCrawler\Form|\Symfony\Component\DomCrawler\Field\ChoiceFormField[]|\Symfony\Component\DomCrawler\Field\FileFormField[]
  */
 protected function getFormFor($node)
 {
     $form = $node->parents()->filter('form')->first();
     if (!$form) {
         $this->fail('The selected node does not have a form ancestor.');
     }
     $action = $this->getFormUrl($form);
     if (isset($this->forms[$action])) {
         return $this->forms[$action];
     }
     $formSubmits = $form->filter('*[type=submit]');
     // Inject a submit button if there isn't one.
     if ($formSubmits->count() == 0) {
         $autoSubmit = new \DOMElement('input');
         $form->rewind();
         $autoSubmit = $form->current()->appendChild($autoSubmit);
         $autoSubmit->setAttribute('type', 'submit');
         // for forms with no submits
         $autoSubmit->setAttribute('name', 'codeception_added_auto_submit');
     }
     // Retrieve the store the Form object.
     $this->forms[$action] = $form->form();
     return $this->forms[$action];
 }

 public function indexGames()
 {
     $crawler = new Crawler();
     //--------------------extract all product details----------------------
     global $products;
     $products = array();
     $html = file_get_contents("http://www.lelong.com.my/toys-and-games/game-console/");
     $crawler->addContent($html);
     //------------------extract retailer logo------------------------
     //$retailer_logo = $crawler->filter('div#top1Logo img')->attr('src');
     //---------------------------------------------------------------
     //---------------------------filter category-------------------------------
     $category = $crawler->filter('a[href="/toys-and-games/game-console/"]')->text();
     //-------------------------------------------------------------------------
     $crawler->filter('div.item4inline')->each(function ($crawler) {
         for ($i = 2; $i <= 5;) {
             $url = 'http://www.lelong.com.my/toys-and-games/game-console/?D=' . $i;
             $html = file_get_contents($url);
             $crawler->addContent($html);
             global $products;
             global $rank;
             $rank = $crawler->filter('span.catalogTitle')->each(function ($crawler, $i) use(&$products) {
                 $products[$i]['title'] = $crawler->text();
                 $products[$i]['url'] = str_replace('//', '', $crawler->parents()->attr('href'));
             });
             $rank = $crawler->filter('div.catalogPrice b')->each(function ($crawler, $i) use(&$products) {
                 $toReplace = array('RM', ',');
                 $with = array('', '');
                 $products[$i]['price'] = str_replace($toReplace, $with, $crawler->text());
             });
             $rank = $crawler->filter('div.catalog-wrap')->each(function ($crawler, $i) use(&$products) {
                 $products[$i]['image'] = $crawler->parents()->attr('id');
             });
             $rank = $crawler->filter('div.catalogIcon')->each(function ($crawler, $i) use(&$products) {
                 $products[$i]['shipping'] = $crawler->children()->text();
             });
             ++$rank;
             $i++;
             //print_r($products);
         }
     });
     //--------------insert data using model-----------------
     foreach ($products as $pro) {
         $product = new Products();
         if ($category == 'Game Console') {
             $product->category_id = 6;
             $product->condition_id = 3;
         }
         $arrProduct = explode(' ', $pro['title']);
         $brands = \DB::table('brand')->whereIn('brand_title', $arrProduct)->get();
         if ($brands) {
             foreach ($brands as $brand) {
                 $product->brand_id = $brand->id;
             }
         } else {
             $product->brand_id = 204;
         }
         $product->product_name = $pro['title'];
         $product->shopper_link = $pro['url'];
         $product->product_price = $pro['price'];
         $product->picture_link = $pro['image'];
         $product->product_shipping = $pro['shipping'];
         $product->save();
     }
     //-------------------------------------------------------
 }

 private function fillAnnonceRepo($manager, &$annonces)
 {
     $numeroPage = 1;
     $bddUpToDate = false;
     $count = count($annonces);
     $initRepo = false;
     if ($count == 0) {
         $initRepo = true;
     }
     $countNew = 0;
     $newAnnonces = array();
     // Atualise ou rempli pour la premiere fois la BDD
     while (!$bddUpToDate) {
         // Url contenant les informations a récuperer
         $lbcUrl = "http://www.leboncoin.fr/animaux/offres/rhone_alpes/?o=" . $numeroPage;
         $content = file_get_contents($lbcUrl);
         $crawler = new Crawler($content);
         $classListLbc = "list-lbc";
         // Parsing de chaque div contenant chaque annonce
         foreach ($crawler->filter('.lbc') as $lbcAnnonce) {
             $tmpCrawler = new Crawler($lbcAnnonce);
             $infoNeeded = array("title" => "", "placement" => "", "price" => "", "date" => "");
             foreach ($infoNeeded as $key => $value) {
                 $tmpNode = $tmpCrawler->filter('.' . $key);
                 if (count($tmpNode) > 0) {
                     $infoNeeded[$key] = $tmpCrawler->filter('.' . $key)->text();
                 }
             }
             $lbcUrl = $tmpCrawler->parents()->attr("href");
             $newAnnonce = new Annonce($infoNeeded['title'], $infoNeeded['placement'], $infoNeeded['price'], $infoNeeded['date'], $lbcUrl, $this->getImage($tmpCrawler));
             // Si actualisation de la BDD, verification de l'existence ou non de l'annonce, si oui, fin de la boucle
             if (!$initRepo) {
                 foreach ($annonces as $annonce) {
                     if ($annonce->Compare($newAnnonce)) {
                         $bddUpToDate = true;
                         break;
                     }
                 }
             }
             if ($bddUpToDate == true) {
                 break;
             }
             // Sinnon stockage de l'annonce
             $newAnnonces[] = $newAnnonce;
             $count++;
             $countNew++;
             if ($countNew > 100) {
                 $bddUpToDate = true;
                 break;
             }
             $manager->persist($newAnnonce);
         }
         $numeroPage++;
     }
     // Si actualisation de la BDD, vérification du nombre d'article, si celui-ci est supérieur a 100 suppresion des plus anciens
     if ($initRepo == false) {
         $annonces = array_merge($newAnnonces, $annonces);
         $count = count($annonces);
         if ($count > 100) {
             $count = 100;
             while (isset($annonces[$count])) {
                 $manager->remove($annonces[$count++]);
             }
         }
     }
 }