PHP phpQuery类代码示例

 function createSpots()
 {
     // TODO :: Some caching ??
     $this->pq = $pq = new phpQuery();
     $this->dom = $dom = $pq->newDocument($this->owner->template->template_source);
     if (!$this->owner instanceof \Frontend) {
         $pq->pq($dom)->attr('xepan-page-content', 'true');
         $pq->pq($dom)->addClass('xepan-page-content');
     }
     foreach ($dom['.xepan-component'] as $d) {
         $d = $pq->pq($d);
         if (!$d->hasClass('xepan-serverside-component')) {
             continue;
         }
         $i = $this->spots++;
         $inner_html = $d->html();
         $with_spot = '{' . $this->owner->template->name . '_' . $i . '}' . $inner_html . '{/}';
         $d->html($with_spot);
     }
     $content = $this->updateBaseHrefForTemplates();
     $content = str_replace('<!--xEpan-ATK-Header-Start', '', $content);
     $content = str_replace('xEpan-ATK-Header-End-->', '', $content);
     $this->owner->template->loadTemplateFromString($content);
     $this->owner->template->trySet($this->app->page . '_active', 'active');
 }

 function init()
 {
     parent::init();
     if (!$this->api->auth->isLoggedIn()) {
         $this->js()->univ()->errorMessage('You Are Not Logged In')->execute();
     }
     if ($_POST['length'] != strlen($_POST['body_html'])) {
         $this->js()->univ()->successMessage('Length send ' . $_POST['length'] . " AND Length calculated again is " . strlen($_POST['body_html']))->execute();
     }
     if ($_POST['crc32'] != sprintf("%u", crc32($_POST['body_html']))) {
         $this->js()->univ()->successMessage('CRC send ' . $_POST['crc32'] . " AND CRC calculated again is " . sprintf("%u", crc32($_POST['body_html'])))->execute();
     }
     try {
         $content = $_POST['body_html'];
         include_once getcwd() . '/lib/phpQuery.php';
         $pq = new \phpQuery();
         $doc =& $pq->newDocument(trim($content));
         // include_once getcwd().'/lib/phpQuery.php';
         // $doc = \phpQuery::newDocument( $content );
         $server = $doc['[data-is-serverside-component=true]'];
         foreach ($doc['[data-is-serverside-component=true]'] as $ssc) {
             $pq->pq($ssc)->html("")->append($html);
         }
         $content = $doc->htmlOuter();
         $this->api->current_page['content'] = urldecode(trim($content));
         $this->api->current_page['body_attributes'] = urldecode($_POST['body_attributes']);
         $this->api->exec_plugins('epan-page-before-save', $this->api->current_page);
         $this->api->current_page->save();
         $this->api->exec_plugins('epan-page-after-save', $this->api->current_page);
         if ($_POST['take_snapshot'] == 'Y') {
             // $this->api->exec_plugins('epan-page-before-snapshot',$this->api->current_page);
             $new_version = $this->api->current_page->ref('EpanPageSnapshots');
             $new_version['title'] = $this->api->current_page['title'];
             $new_version['keywords'] = $this->api->current_page['keywords'];
             $new_version['description'] = $this->api->current_page['description'];
             $new_version['body_attributes'] = $this->api->current_page['body_attributes'];
             $new_version['content'] = $this->api->current_page['content'];
             $new_version->save();
             // $this->api->exec_plugins('epan-page-after-snapshot',$this->api->current_page);
         }
     } catch (Exception_StopInit $e) {
     } catch (Exception $e) {
         throw $e;
         $this->js()->univ()->errorMessage('Error... Could not save your page ' . $e->getMEssage())->excute();
         exit;
     }
     echo "saved";
     exit;
 }

 public static function hyphenate($strBuffer)
 {
     global $objPage;
     $arrSkipPages = \Config::get('hyphenator_skipPages');
     if (is_array($arrSkipPages) && in_array($objPage->id, $arrSkipPages)) {
         return $strBuffer;
     }
     $o = new \Org\Heigl\Hyphenator\Options();
     $o->setHyphen(\Config::get('hyphenator_hyphen'))->setDefaultLocale(static::getLocaleFromLanguage($objPage->language))->setRightMin(\Config::get('hyphenator_rightMin'))->setLeftMin(\Config::get('hyphenator_leftMin'))->setWordMin(\Config::get('hyphenator_wordMin'))->setFilters(\Config::get('hyphenator_filter'))->setQuality(\Config::get('hyphenator_quality'))->setTokenizers(\Config::get('hyphenator_tokenizers'));
     $h = new \Org\Heigl\Hyphenator\Hyphenator();
     $h->setOptions($o);
     $doc = \phpQuery::newDocumentHTML($strBuffer);
     foreach (pq('body')->find(\Config::get('hyphenator_tags')) as $n => $item) {
         $strText = pq($item)->html();
         // ignore html tags, otherwise ­ will be added to links for example
         if ($strText != strip_tags($strText)) {
             continue;
         }
         $strText = str_replace('­', '', $strText);
         // remove manual ­ html entities before
         $strText = $h->hyphenate($strText);
         if (is_array($strText)) {
             $strText = current($strText);
         }
         pq($item)->html($strText);
     }
     return $doc->htmlOuter();
 }

function search_query($kw)
{
    $wf = new Workflows();
    $url = 'http://www.bilibili.tv/search?keyword=' . urlencode($kw) . '&orderby=&formsubmit=';
    $content = $wf->request($url, array(CURLOPT_ENCODING => 1));
    $doc = phpQuery::newDocumentHTML($content);
    $list = $doc->find('li.l');
    $i = 0;
    foreach ($list as $item) {
        $link = pq($item)->children('a:first')->attr('href');
        if (strpos($link, 'http') !== 0) {
            $link = 'http://www.bilibili.tv' . $link;
        }
        $info = pq($item)->find('div.info > i');
        $author = pq($item)->find('a.upper:first')->text();
        $view = $info->eq(0)->text();
        $comment = $info->eq(1)->text();
        $bullet = $info->eq(2)->text();
        $save = $info->eq(3)->text();
        $date = $info->eq(4)->text();
        $subtitle = 'UP主:' . $author . ' 播放:' . $view . ' 评论:' . $comment . ' 弹幕:' . $bullet . ' 收藏:' . $save . ' 日期:' . $date;
        $wf->result($i, $link, pq($item)->find('div.t:first')->text(), $subtitle, 'icon.png', 'yes');
        $i++;
    }
    if (count($wf->results()) == 0) {
        $wf->result('0', $url, '在bilibili.tv中搜索', $kw, 'icon.png', 'yes');
    }
    return $wf->toxml();
}

 public function finish()
 {
     phpQuery::newDocumentFileHTML($this->downloadUrl, $this->charset);
     $nodeCount = pq("#list2")->count();
     $articleList = pq("#list2 tr");
     echo "Start\n";
     for ($i = 0; $i < $this->loop; $i++) {
         echo '=';
     }
     echo "\n";
     $i = 0;
     foreach ($articleList as $tr) {
         $i++;
         if ($i <= 1) {
             # Filter the first node which has no meaning.
             continue;
         }
         $className = $tr->nodeValue;
         # This class order and its name.
         $classNameCombine = explode(' ', $className);
         $j = 0;
         $this->content[$i - 1]['chapterName'] = '';
         foreach ($classNameCombine as $k => $v) {
             # Put all class order and its name into $this->content array.
             $v = trim($v);
             if (empty($v)) {
                 continue;
             }
             $j++;
             if ($j == 1) {
                 $this->content[$i - 1]['chapterId'] = $v;
             } else {
                 $this->content[$i - 1]['chapterName'] .= $v . " ";
             }
         }
         $child = pq("td", $tr)->html();
         $trueContent = explode(' ', $child);
         $linkString = '';
         foreach ($trueContent as $k => $v) {
             $v = trim($v);
             if (empty($v)) {
                 continue;
             }
             if ($v == "<a") {
                 $linkString = $v . ' ' . $trueContent[$k + 1];
                 break;
             }
         }
         $urls = explode("\"", $linkString);
         $this->content[$i - 1]['link'] = $urls[1];
         //            $name = explode("<", $urls[2]);
         //            $name = substr($name[0], 1);
         //            $this->content[$i-1]['name'] = $name;
     }
     var_dump($this->content);
     for ($i = 0; $i < $this->loop; $i++) {
         echo '=';
     }
     echo "\nFinished! \n";
 }

function getAllSubCategories($filename)
{
    $categories = unserialize(file_get_contents($filename));
    echo "---- run getAllSubCategories() ----\n\r";
    foreach ($categories as $j => $category) {
        if ($category['childrens']) {
            foreach ($category['childrens'] as $k => $children) {
                echo "---- crawling childrens for {$children['name']} ----\n\r";
                $doc = phpQuery::newDocumentHTML(fetch('http://www.walmart.com' . $children['link']));
                phpQuery::selectDocument($doc);
                foreach (pq('.shop-by-category li') as $el) {
                    echo "---- " . pq($el)->find('a')->attr('href') . "} ----\n\r";
                    $childrens[] = array('name' => pq($el)->find('a')->data('name'), 'link' => pq($el)->find('a')->attr('href'));
                }
                $categories[$j]['childrens'][$k]['childrens'] = $childrens;
            }
        }
    }
    echo "---- creating deparment file ----\n\r";
    $file = fopen($filename, 'w+');
    echo "---- writing deparment file ----\n\r";
    fputs($file, serialize($categories));
    echo "---- closing deparment file ----\n\r";
    fclose($file);
}

 public function first_image($htmldata)
 {
     $pq = phpQuery::newDocumentHTML($htmldata);
     $img = $pq->find('img:first');
     $src = $img->attr('src');
     return $src;
 }

 public function parse_archive(Archive $archive)
 {
     if ($archive->cover_mode) {
         return [$this->get_image_by_cover($archive->cover)];
     }
     $result = curl($archive->url, curl_options($this->domain));
     $archivePQ = \phpQuery::newDocumentHTML($result['data']);
     $content = $archivePQ['#content'];
     $title = $content['.entry-title']->html();
     $imgs = $content['.entry-content']->html();
     if (preg_match_all('#' . $this->domain . 'wp-content/[^"]*#', $imgs, $matches)) {
         $archive->images = array_merge(array_unique($matches[0]));
     }
     if (!$archive->images) {
         $title = $content['.main-title']->html();
         $imgs = $content['.main-body']->html();
         if (preg_match_all('#<a[^>]+href="(' . $this->domain . 'wp-content/[^"]*)">#', $imgs, $matches)) {
             $archive->images = array_merge(array_unique($matches[1]));
         }
     }
     if (strtolower($this->charset) != strtolower($GLOBALS['app_config']['charset'])) {
         $archive->title = iconv(strtoupper($this->charset), strtoupper($GLOBALS['app_config']['charset']) . '//IGNORE', $title);
     }
     $archive->title = $title;
 }

 public function find_product_url($product_name)
 {
     $url = 'http://ek.ua/';
     $data = array('search_' => $product_name);
     $response = Request::factory($url)->query($data)->execute();
     //echo Debug::vars($response->body());
     //echo Debug::vars($response->headers());
     $response = $response->body();
     try {
         $doc = phpQuery::newDocument($response);
     } catch (Exception $ex) {
         $errors[] = $ex->getMessage();
         echo Debug::vars($errors);
     }
     if (!isset($errors)) {
         /* Нужно взять заголовок и найти в нем слово найдено */
         /*
          * что я заметил, удивительно но на разных машинах этот заголовои идет с разным классом
          * на данный момент этот класс oth
          */
         $str = $doc->find('h1.oth')->html();
         if (preg_match('/найдено/', $str)) {
             echo Debug::vars('алилуя !!!');
             die;
         }
         return $str;
     }
 }

 /**
  * Query the response for a JQuery compatible CSS selector
  *
  * @link https://code.google.com/p/phpquery/wiki/Selectors
  * @param $selector string
  * @return phpQueryObject
  */
 public function queryHTML($selector)
 {
     if (is_null($this->pq)) {
         $this->pq = phpQuery::newDocument($this->content);
     }
     return $this->pq->find($selector);
 }

 function theContent($content)
 {
     $doc = phpQuery::newDocumentHTML($content);
     phpQuery::selectDocument($doc);
     foreach (pq('a') as $link) {
         $linkurl = pq($link)->attr('href');
         $linkHostname = $this->getUrlHostname($linkurl);
         if ($linkHostname != FALSE) {
             if (!in_array($linkHostname, $this->blockedDomains)) {
                 $linkHash = md5($linkurl);
                 $query = "SELECT * FROM {$this->tableBitlyExternal} WHERE id='{$linkHash}'LIMIT 1;";
                 $linkData = $this->db->get_row($query, ARRAY_A);
                 if (empty($linkData)) {
                     $bitly = new Bitly($this->bitlyUsername, $this->bitlyApikey);
                     $shortURL = $bitly->shorten($linkurl);
                     $shortURLData = get_object_vars($bitly->getData());
                     if (!empty($shortURLData)) {
                         $linkData = array('id' => $linkHash, 'url' => $linkurl, 'short_url' => $shortURLData['shortUrl'], 'hash' => $shortURLData['userHash'], 'created' => current_time('mysql'));
                         $this->db->insert($this->tableBitlyExternal, $linkData);
                     }
                 }
                 if (!empty($linkData)) {
                     pq($link)->attr('href', $linkData['short_url']);
                 }
             }
         }
     }
     return $doc->htmlOuter();
 }

 public function getLinkContent()
 {
     require_once './ThinkPHP/Library/Vendor/Collection/phpQuery.php';
     $link = op_t(I('post.url'));
     $content = get_content_by_url($link);
     $charset = preg_match("/<meta.+?charset=[^\\w]?([-\\w]+)/i", $content, $temp) ? strtolower($temp[1]) : "utf-8";
     \phpQuery::$defaultCharset = $charset;
     \phpQuery::newDocument($content);
     $title = pq("meta[name='title']")->attr('content');
     if (empty($title)) {
         $title = pq("title")->html();
     }
     $title = iconv($charset, "UTF-8", $title);
     $keywords = pq("meta[name='keywords'],meta[name='Keywords']")->attr('content');
     $description = pq("meta[name='description'],meta[name='Description']")->attr('content');
     $url = parse_url($link);
     $img = pq("img")->eq(0)->attr('src');
     if (is_bool(strpos($img, 'http://'))) {
         $img = 'http://' . $url['host'] . $img;
     }
     $title = text($title);
     $description = text($description);
     $keywords = text($keywords);
     $return['title'] = $title;
     $return['img'] = $img;
     $return['description'] = empty($description) ? $title : $description;
     $return['keywords'] = empty($keywords) ? $title : $keywords;
     exit(json_encode($return));
 }

 public function addclass($html, $selector, $class)
 {
     require __DIR__ . '/phpQuery.php';
     $pq = \phpQuery::newDocument($html);
     $pq[$selector]->addClass($class);
     return new \Twig_Markup($pq, 'UTF-8');
 }

 /**
  * ___render___
  *
  * @method ___render___
  * @param {string} $content
  */
 protected function ___render___($content)
 {
     if ($content && is_string($content)) {
         if ($this->functions_for_render && (is_array($this->functions_for_render) || $this->functions_for_render instanceof ArrayObject)) {
             $before = $this->cache_read === true ? array() : array_filter((array) $this->functions_for_render, create_function('$v', 'return $v[\'after_cache\'] !== true;'));
             $after = array_filter((array) $this->functions_for_render, create_function('$v', 'return $v[\'after_cache\'] === true;'));
         } else {
             $before = $after = array();
         }
         if (0 < count($before) || 0 < count($after)) {
             $pq = phpQuery::newDocument($content);
             //before cache
             foreach ($before as $fn) {
                 $pq = call_user_func($fn, $pq);
             }
             if ($this->cache_file && !$this->cache_read) {
                 file_put_contents($this->cache_file, $pq);
             }
             //after cache
             foreach ($before as $fn) {
                 $pq = call_user_func($fn, $pq);
             }
             $content = $pq;
         } else {
             if ($this->cache_file) {
                 file_put_contents($this->cache_file, $content);
             }
         }
     }
     echo $content;
 }

    public function go()
    {
        require_once $_SERVER["DOCUMENT_ROOT"] . "/classes/XML.php";
        $xml = file_get_contents("http://alfabank.by/a-blog.xml");
        $rss = new \XML($xml);
        foreach ($rss->rss->channel->item as $it) {
            $dom = \phpQuery::newDocumentHTML($it->childByName("content:encoded"));
            $date = date_create($it->pubDate->getData());
            $arResult = array(
                "IBLOCK_ID" => self::ABLOG_IBLOCK_ID,
                "NAME" => $it->title->getData(),
                "DATE_ACTIVE_FROM" => date_format($date, 'd.m.Y'),
                "PREVIEW_TEXT" => $it->description->getData(),
                "PREVIEW_PICTURE" => \CFile::MakeFileArray($dom->find('img')->attr('src')),
                "DETAIL_TEXT" => $it->childByName("content:encoded"),
                "CODE" => \Ns\Bitrix\Helper::Create('iblock')->useVariant('text')->translite($it->title->getData()),
                "PROPERTY_VALUES" => array("ORIGINAL_LINK" => $it->link->getData())
            );
            $this->objElement->Add($arResult);
            if ($this->objElement->LAST_ERROR) {
                prentExpection($this->objElement->LAST_ERROR);
            }
        }

        return true;
	}