/** * @param ParserQueue $pq * @param PendingNews $pn */ public function __construct(ParserQueue $pq, PendingNews $pn = null) { $pq->status = ParserQueue::STATUS_INPROCESS; $pq->save(); $this->source = $pq->source; $this->url = $this->prepareUrl($pq->url); $this->parserQueue = $pq; if ($pn) { $this->pendingNews = $pn; } }
public function actionCheck() { $pq = new ParserQueue(); $pq->source_id = 1; $pq->url = 'http://test.loc'; $pq->status = 'new'; $pq->created_at = new Expression("NOW()"); $pq->updated_at = new Expression("NOW()"); if ($pq->insert()) { echo "\nSAVED with id {$pq->id}\n"; } else { echo "\nFAILED\n"; var_dump($pq->getErrors()); } }
public function actionTest() { $data = []; if (Yii::$app->request->post()) { $data = Yii::$app->request->post(); if (isset($data['url'])) { $parser = new NewsParserComponent(ParserQueue::findOne(1)); $data['parserResult'] = $parser->parse(PageLoaderComponent::load($data['url']), $data['url'], Sources::findOne($data['Source'])); } } return $this->render('test', $data); }
public static function processMessage($msg) { // print_r($msg); try { $params = json_decode($msg->body); print_r($params); $pqItem = ParserQueue::findOne(["id" => $params->pq_id]); $pnItem = PendingNews::findOne(["id" => $params->pn_id]); if ($pqItem && $pnItem) { $newsParser = new NewsParserComponent($pqItem, $pnItem); $newsParser->run(); } } catch (Exception $e) { echo $e->getMessage(); if ($e->getCode() != 505) { $mq = new RabbitMQComponent(); $mq->postMessage("parse", "parse_rss", $msg->body); } } $msg->delivery_info['channel']->basic_ack($msg->delivery_info['delivery_tag']); // die(); }
public static function add(Sources $source, $title, $content, $image_src, $status = PendingNews::STATUS_NEW, ParserQueue $parser_queue = null, $data = []) { if ($searchContent = trim(strip_tags($content))) { $searchContent = preg_replace("/[^а-яa-z ]/ui", "", $searchContent); if (count(explode(" ", $searchContent)) >= Settings::findOne(['name' => 'news_min_length'])->value) { $pn = new PendingNews(); $pn->source_id = $source->id; $pn->title = $title; $pn->content = $content; $pn->search_content = $searchContent; $pn->status = $status; $pn->group_hash = md5(microtime()); $pn->thumb_src = $image_src; if ($parser_queue) { $pn->pq_id = $parser_queue->id; } if (!empty($data)) { $pn->additonal_data = json_encode($data); } $pn->created_at = new \yii\db\Expression("NOW()"); $pn->update_at = new \yii\db\Expression("NOW()"); if ($pn->save()) { if ($parser_queue) { $parser_queue->status = ParserQueue::STATUS_DONE; $parser_queue->save(); } } else { if ($parser_queue) { $parser_queue->status = ParserQueue::STATUS_FAIL; $parser_queue->save(); } } } } }
public function run() { $rss = PageLoaderComponent::load($this->source->url); $rss = trim($rss); $rss = preg_replace("/<rss([^>]+)>/mi", "<rss>", $rss); preg_match('/<\\?xml.*?encoding=(\'|")(.*)("|\\")/i', $rss, $matches); $charset = "utf-8"; if (isset($matches[2])) { $charset = $matches[2]; // $rss = mb_convert_encoding( $rss, "UTF-8", $charset ); } else { // $rss = mb_convert_encoding( $rss, "UTF-8" ); } $doc = new \DOMDocument("1.1", $charset); $doc->preserveWhiteSpace = false; libxml_use_internal_errors(true); $doc->loadXML($rss); $xpath = new \DOMXpath($doc); foreach ($this->itemPatterns as $pattern) { if ($newsList = $xpath->query($pattern->value)) { for ($i = 0; $i < $newsList->length; $i++) { $news = $newsList->item($i); $newsParams = array(); $newsParams['data'] = []; if ($this->source->category_id) { $newsParams['data']['category_id'] = $this->source->category_id; } $newsParams['source'] = $this->source->source; foreach ($news->childNodes as $node) { if ($this->titlePattern == $node->nodeName) { $newsParams['title'] = $node->nodeValue; } if ($this->contentPattern == $node->nodeName) { $newsParams['content'] = $node->nodeValue; } if ($this->linkPattern == $node->nodeName) { $newsParams['link'] = $node->nodeValue; } if ($this->categoryPattern == $node->nodeName) { $newsParams['data']['category'] = $node->nodeValue; } if ($this->imagePattern == $node->nodeName) { if (preg_match_all('/(https?:\\/\\/[a-z0-9\\/_а-я\\-\\.]*\\.(?:png|jpg))/i', $node->nodeValue, $images)) { $newsParams['image_src'] = $images[1][0]; } } if (!isset($newsParams['image_src'])) { if ($node->nodeName == 'enclosure') { if (preg_match_all('/(https?:\\/\\/[a-z0-9\\/_а-я\\-\\.]*\\.(?:png|jpg))/i', $node->getAttribute('url'), $images)) { $newsParams['image_src'] = $images[1][0]; } } } if (!isset($newsParams['image_src'])) { $newsParams['image_src'] = ""; } } try { $pqItem = ParserQueue::findOne(['url' => $newsParams['link']]); if (!$pqItem) { $pqItem = new ParserQueue(); $pqItem->source_id = $this->source->source_id; $pqItem->url = $newsParams['link']; $pqItem->status = ParserQueue::STATUS_INPROCESS; $pqItem->created_at = new \yii\db\Expression('NOW()'); $pqItem->updated_at = new \yii\db\Expression('NOW()'); if ($pqItem->save()) { PendingNews::add($newsParams['source'], $newsParams['title'], $newsParams['content'], isset($newsParams['image_src']) ? $newsParams['image_src'] : false, PendingNews::STATUS_NEW, $pqItem, $newsParams['data']); } } } catch (Exception $e) { // print_r( $e->getMessage() ); } } } } }
/** * @return \yii\db\ActiveQuery */ public function getParserQueues() { return $this->hasMany(ParserQueue::className(), ['source_id' => 'id']); }
private function processCombine($xpath) { $itemPatterns = SourcesSettings::findAll(['source_id' => $this->source->source_id, 'name' => 'rss_news_item_pattern']); if ($titlePattern = SourcesSettings::findOne(['source_id' => $this->source->source_id, 'name' => 'rss_title'])) { $titlePattern = $titlePattern->value; } if ($contentPattern = SourcesSettings::findOne(['source_id' => $this->source->source_id, 'name' => 'rss_content'])) { $contentPattern = $contentPattern->value; } if ($linkPattern = SourcesSettings::findOne(['source_id' => $this->source->source_id, 'name' => 'rss_link'])) { $linkPattern = $linkPattern->value; } if ($imagePattern = SourcesSettings::findOne(['source_id' => $this->source->source_id, 'name' => 'rss_image'])) { $imagePattern = $imagePattern->value; } if ($categoryPattern = SourcesSettings::findOne(['source_id' => $this->source->source_id, 'name' => 'rss_category'])) { $categoryPattern = $categoryPattern->value; } foreach ($itemPatterns as $pattern) { if ($newsList = $xpath->query($pattern->value)) { for ($i = 0; $i < $newsList->length; $i++) { $news = $newsList->item($i); $newsParams = array(); $newsParams['data'] = []; if ($this->source->category_id) { $newsParams['data']['category_id'] = $this->source->category_id; } $newsParams['source'] = $this->source->source; foreach ($news->childNodes as $node) { if ($titlePattern == $node->nodeName) { $newsParams['title'] = $node->nodeValue; } if ($categoryPattern) { if ($categoryPattern == $node->nodeName) { $newsParams['data']['category'] = $node->nodeValue; } } else { if ("category" == strtolower($node->nodeName)) { $newsParams['data']['category'] = $node->nodeValue; } } if ($linkPattern == $node->nodeName) { $newsParams['link'] = str_replace($this->source->source->url, '', $node->nodeValue); // $newsParams['link'] = $node->nodeValue; } if ($imagePattern == $node->nodeName) { if (preg_match_all('/(https?:\\/\\/[a-z0-9\\/_а-я\\-\\.]*\\.(?:png|jpg))/i', $node->nodeValue, $images)) { $newsParams['image_src'] = $images[1][0]; } } if (!isset($newsParams['image_src'])) { if ($node->nodeName == 'enclosure') { if (preg_match_all('/(https?:\\/\\/[a-z0-9\\/_а-я\\-\\.]*\\.(?:png|jpg))/i', $node->getAttribute('url'), $images)) { $newsParams['image_src'] = $images[1][0]; } } } } try { if (array_key_exists("link", $newsParams)) { $pqItem = new ParserQueue(); $pqItem->source_id = $this->source->source_id; $pqItem->url = $newsParams['link']; $pqItem->status = ParserQueue::STATUS_INPROCESS; $pqItem->created_at = new \yii\db\Expression('NOW()'); $pqItem->updated_at = new \yii\db\Expression('NOW()'); if ($pqItem->save()) { $pn = new PendingNews(); $pn->content = ' '; $pn->search_content = ' '; $pn->source_id = $this->source->source_id; if (isset($newsParams['title'])) { $pn->title = $newsParams['title']; } $pn->status = PendingNews::STATUS_SUSPENDED; $pn->group_hash = md5(time()); if (isset($newsParams['image_src'])) { $pn->thumb_src = $newsParams['image_src']; } if ($pqItem) { $pn->pq_id = $pqItem->id; } if (!empty($newsParams['data'])) { $pn->additonal_data = json_encode($newsParams['data']); } $pn->created_at = new \yii\db\Expression("NOW()"); $pn->update_at = new \yii\db\Expression("NOW()"); if ($pn->save()) { if ($pqItem) { $pqItem->status = ParserQueue::STATUS_DONE; $pqItem->save(); $this->mq->postMessage("parse", "parse_rss", json_encode(["pn_id" => $pn->id, "pq_id" => $pqItem->id])); } } else { if ($pqItem) { $pqItem->status = ParserQueue::STATUS_FAIL; $pqItem->save(); } } } else { // print_r( $pqItem->errors ); } } } catch (\yii\db\Exception $e) { // print_r( $e->getMessage() ); } } } } }