PHP Crawler::text方法代码示例

 public static function parse($contents, $id, $type)
 {
     $crawler = new Crawler();
     $crawler->addHTMLContent($contents, 'UTF-8');
     $rows = $crawler->filter('div[class="spaceit_pad"]');
     $title = preg_replace('/ (\\w+?) Details/', '$2', $crawler->filter('div[class="normal_header"]')->text());
     $result = array();
     if ($type === 'anime') {
         foreach ($rows as $historyItem) {
             $crawler = new Crawler($historyItem);
             $date = explode(' ', $crawler->text());
             $historyinfo['item'] = new Anime();
             $historyinfo['item']->setId((int) $id);
             $historyinfo['item']->setTitle($title);
             $historyinfo['item']->setWatchedEpisodes((int) $date[1]);
             $historyinfo['type'] = $type;
             $historyinfo['time_updated'] = Date::formatTime($date[4] . ' ' . $date[6]);
             $result[] = $historyinfo;
         }
     } else {
         foreach ($rows as $historyItem) {
             $crawler = new Crawler($historyItem);
             $date = explode(' ', $crawler->text());
             $historyinfo['item'] = new Manga();
             $historyinfo['item']->setId((int) $id);
             $historyinfo['item']->setTitle($title);
             $historyinfo['item']->setChaptersRead((int) $date[1]);
             $historyinfo['type'] = $type;
             $historyinfo['time_updated'] = Date::formatTime($date[4] . ' ' . $date[6]);
             $result[] = $historyinfo;
         }
     }
     return $result;
 }

 public static function parseDay($rows)
 {
     $result = array();
     foreach ($rows as $item) {
         $crawler = new Crawler($item);
         $anime = new Anime();
         $url = $crawler->filter('a[class="link-title"]')->attr('href');
         $id = preg_match('/\\/(anime|manga)\\/(\\d+)\\/.*?/', $url, $urlParts);
         if ($id !== false || $id !== 0) {
             $anime->setId((int) $urlParts[2]);
         }
         $anime->setTitle(trim($crawler->filter('a[class="link-title"]')->text()));
         $producer = $crawler->filter('span[class="producer"] a');
         if ($producer->count() > 0) {
             $anime->setProducers(explode(', ', $crawler->filter('span[class="producer"] a')->text()));
         }
         $anime->setEpisodes((int) str_replace(' eps', '', $crawler->filter('div[class="eps"] span')->text()));
         $genres = $crawler->filter('div[class="genres-inner js-genre-inner"] a');
         $genreArray = array();
         foreach ($genres as $genre) {
             $genreCrawler = new Crawler($genre);
             $genreArray[] = $genreCrawler->text();
         }
         $anime->setGenres($genreArray);
         $anime->setImageUrl($crawler->filter('div[class="image lazyload"]')->attr('data-bg'));
         $anime->setSynopsis(trim($crawler->filter('div[class="synopsis js-synopsis"]')->text()));
         $detail = explode('-', $crawler->filter('div[class="info"]')->text());
         $anime->setType(trim($detail[0]));
         $anime->setMembersCount((int) str_replace(',', '', trim($crawler->filter('span[class="member fl-r"]')->text())));
         $anime->setMembersScore((double) trim($crawler->filter('span[class="score"]')->text()));
         $result[] = $anime;
     }
     return $result;
 }

 /**
  * @param Crawler $crawler
  * @return string
  */
 private function parseTeachersName(Crawler $crawler)
 {
     $str = $crawler->text();
     $str = str_replace('Vyučující: ', '', $str);
     $str = str_replace('(email)', '', $str);
     return trim($str);
 }

 /**
  * Parse text from crawler node.
  *
  * @param \Symfony\Component\DomCrawler\Crawler $node
  * @param mixed                                 $default
  *
  * @return mixed
  */
 protected function parseTextFromNode($node, $default = null)
 {
     $text = trim($node->text());
     if (strlen($text) === 0) {
         return $default;
     }
     return $text;
 }

 /**
  * Нормализует значение цены на блюдо
  *
  * @param Crawler|null $element
  * @return float
  */
 public function normalisePrice(Crawler $element = null)
 {
     if ($element === null) {
         return 0.0;
     }
     $price = (string) $element->text();
     return (double) str_replace(',', '.', $price);
 }

 private function getPageData($html)
 {
     $crawler = new Crawler($html);
     $body = $crawler->filter('body');
     $header = $body->filter('.page-header');
     $breadcrumbs = $header->filter('.breadcrumb li span a')->each(function (Crawler $crawler) {
         return trim($crawler->text());
     });
     $descriptions = $body->filter('.container > .row .profile')->each(function (Crawler $crawler) {
         return trim($crawler->text());
     });
     $default = array_fill(0, 3, '');
     $descriptions = $descriptions + $default;
     $breadcrumbs = $breadcrumbs + $default;
     $data = ['city' => $breadcrumbs[1], 'category' => $breadcrumbs[2], 'title' => trim($header->filter('h1')->text()), 'salary' => substr(trim($header->filter('.profile-details-salary')->text()), 1), 'experience_year' => (int) $body->filter('.before-hint')->text(), 'experience_description' => $descriptions[0], 'achievement' => $descriptions[1], 'expect' => $descriptions[2], 'skills' => json_encode($this->getSkills($body))];
     return $data;
 }

 protected function extract(Crawler $crawler)
 {
     // mother fuckin callback nigga
     return $crawler->filter('table.ob_gBody tr')->each(function (Crawler $crawler, $i) {
         return $crawler->filter('.ob_gCc2')->each(function (Crawler $crawler, $i) {
             return $crawler->text();
         });
     });
 }

 /**
  * @param Crawler $crawler
  * @return Person
  */
 public static function parseFromDomCrawler(Crawler $crawler)
 {
     $content = $crawler->text();
     $content = StringHelper::removeEmptyLines($content);
     $contentItems = explode("\n", $content);
     $contentItems = array_map('trim', $contentItems);
     $name = trim(explode(',', $contentItems[1])[0]);
     $birthday = DateTimeParser::parseFromCzechDateString($contentItems[2]);
     return new Person($name, $birthday, $contentItems[3]);
 }

 private function getPageTags(Crawler $crawler)
 {
     return $crawler->filter('.item-info table tr')->each(function (Crawler $crawler) {
         $name = trim($crawler->filter('td')->first()->text());
         $list = $crawler->filter('.tag span')->each(function (Crawler $crawler) {
             return $crawler->text();
         });
         return compact('name', 'list');
     });
 }

 protected function looksLikeGroupTitle(Crawler $node)
 {
     $tags = ['h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'strong', 'dt'];
     if (in_array($node->nodeName(), $tags)) {
         return true;
     }
     if (preg_match('/[#\\*\\-_=\\+]{2,}/', $node->text())) {
         return true;
     }
     if (':' === substr($node->text(), -1)) {
         return true;
     }
     if (false !== strpos($node->attr('class'), 'header')) {
         return true;
     }
     if (false !== strpos($node->attr('class'), 'title')) {
         return true;
     }
     return false;
 }

 public function populateLeagues($crawler)
 {
     $types = [];
     foreach ($crawler->filterXPath('//*[@class="ismSecondary"]/*[@class="ismTableHeading"]') as $h) {
         $c = new Crawler($h);
         if (preg_match('~ leagues$~', $c->text())) {
             $types[] = strtolower(preg_replace('~ leagues$~', '', $c->text()));
         }
     }
     foreach ($crawler->filterXPath('//*[@class="ismTable ismLeagueTable"]') as $i => $t) {
         $table = new Crawler($t);
         $leagues = $table->filterXpath('//tbody/tr');
         foreach ($leagues as $l) {
             $c = new Crawler($l);
             $league = new UserLeague();
             $league->type = $types[$i];
             $league->populate($c);
             $this->leagues[] = $league;
         }
     }
 }

 public function testMessageRender()
 {
     $message = new FlashMessage();
     $message->title = 'Error!';
     $message->type = FlashMessage::TYPE_ERROR;
     $message->message = 'This is a test error';
     Yii::$app->session->setFlash('test', $message);
     $html = FlashMessages::widget(['id' => 'flash-test']);
     $dom = new Crawler($html);
     $this->assertContains('This is a test error', $dom->text());
     $this->assertContains('Error!', $dom->filter('h4')->text());
 }

 private function addToCart($mealName, $quantity, Crawler $crawler, Client $client)
 {
     $titles = $crawler->filter('h4')->reduce(function ($crawler) use($mealName) {
         return false !== strpos($crawler->text(), $mealName);
     });
     if (count($titles) !== 1) {
         throw new \RuntimeException(sprintf('Expected 1 title containing "%s", found %s.', $mealName, count($titles)));
     }
     $link = $titles->eq(0)->parents()->first()->filter('input[data-meal]');
     $mealId = $link->attr('data-meal');
     $client->request('POST', '/cart', array('meal' => $mealId, 'mode' => 'add', 'quantity' => $quantity));
     $this->assertEquals(200, $client->getResponse()->getStatusCode());
 }

 private function each(Crawler $node, $rule)
 {
     $elements = [];
     // '@idioms' => [':each' => '#relatedentries > dl > dd > ul > li > a > .arl8']
     if (is_string($rule)) {
         $node->filter($rule)->each(function (Crawler $node) use($elements) {
             $value = $node->text();
             if (null !== $value) {
                 $elements[] = $value;
             }
         });
     } else {
         foreach ($rule as $filter => $pattern) {
             $node->filter($filter)->each(function (Crawler $node, $i) use(&$elements, $pattern) {
                 $elements[$i] = [];
                 $this->parse($node, $pattern, $elements[$i]);
             });
         }
     }
     return $elements;
 }

 /**
  * Aucune annotation Route n'a été définie pour cette méthode de controller, 
  * la page ne sera pas directement accessible.
  * 
  * La méthode peut être appelée depuis un fichier de template 
  * {{ render(controller('MmiBlogBundle:Feed:last', {limit: 5})) }}
  */
 public function lastAction($limit)
 {
     //import d'un flux RSS
     try {
         $reader = Reader::import('http://etin.yourphototravel.com/fr/etins.rss');
     } catch (\Exception $e) {
         $reader = array();
     }
     //lecture du flux RSS et préparation des paramètres à envoyer à la vue
     $items = array();
     $count = 0;
     foreach ($reader as $item) {
         $crawler = new Crawler($item->getDescription());
         $items[] = array('title' => $item->getTitle(), 'link' => $item->getLink(), 'image' => $crawler->filter('img')->first()->attr('src'), 'description' => utf8_decode($crawler->text()));
         $count++;
         if ($count == $limit) {
             break;
         }
     }
     //création de la réponse
     return $this->get('templating')->renderResponse('MmiBlogBundle:Feed:list.html.twig', array('items' => $items));
 }