get_articles.php

#!/usr/bin/php
<?php

gc_enable();
ini_set('display_errors', 1);
error_reporting(E_ALL);
if (file_exists('rb.php')) {
    require_once 'rb.php';
    try {
        if (R::setup('mysql:host=localhost;dbname=', '', '')) {

        } else {
            throw new Exception('ERROR: DB not connected' . PHP_EOL);
        }
    } catch (Exception $e) {
        echo $e->getMessage();
    }
}

class FeedParseBase {

    protected $_url;
    protected $_data;
    protected $_dom;
    protected $_article;
    protected $_options;
    protected $_context;
    protected $_og_data;

    public function __construct($url) {
        $this->_url = $url;
        $this->_options = array(
            'http' => array(
                'method' => "GET",
                'header' => "Accept-Language: en-US,en;q=0.8,ja;q=0.6" .
                'Accept: text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,*/*;q=0.8' .
                'Cookie: querylyvid=1147956133'
            )
        );
        $this->_context = stream_context_create($this->_options);

        try {
            if (!$this->_data = file_get_contents($this->_url, false, $this->_context)) {
                throw new Exception("ERROR: 403/404 or otherwise not found" . PHP_EOL);
            } else {
                $dom = new DOMDocument();
                $dom->load($this->_url);
                $this->_dom = $dom;
            }
        } catch (Exception $e) {
            echo $e->getMessage();
        }

        $this->get_article();
    }

    public function __destruct() {
        unset($this->_article);
        unset($this->_context);
        unset($this->_data);
        unset($this->_dom);
        unset($this->_og_data);
        gc_collect_cycles();
    }

    public function get_article() {
        try {
            if (!$this->_dom) {
                throw new Exception('ERROR: Dom not loaded' . PHP_EOL);
            } else {
                $entries = $this->_dom->getElementsByTagName('entry');
                foreach ($entries as $entry) {
                    if ($entry->hasChildNodes()) {
                        foreach ($entry->childNodes as $childNode) {
                            $this->_article[$childNode->nodeName] = $childNode->nodeValue;

                            if ($childNode->nodeName == 'link') {
                                $href = $childNode->getAttribute('href');
                                $query_str = parse_url($href, PHP_URL_QUERY);
                                parse_str($query_str, $query_params);
                                $query_params['url'] = str_replace('http://m.', 'http://', $query_params['url']);
                                $this->_article['link'] = $query_params['url'];
                            }
                            if ($childNode->nodeName == 'content') {
                                $this->_article['description'] = $childNode->nodeValue;
                            }
                        }
                    }
                    $this->get_article_og($this->_article['link']);
                    $this->save_to_db('news');
                }
            }
        } catch (Exception $e) {
            echo $e->getMessage();
        }
    }

    protected function get_article_og($url) {
        $url = $this->_article['link'];
        try {
            if ($html = @file_get_contents($url, false, $this->_context)) {
                $doc = new DomDocument();
                @$doc->loadHTML($html);
                $xpath = new DOMXPath($doc);
                $query = '//*/meta[starts-with(@property, \'og:\')]';
                $metas = $xpath->query($query);
                $rmetas = array();
                foreach ($metas as $meta) {
                    $property = str_replace(':', '_', $meta->getAttribute('property'));
                    $content = $meta->getAttribute('content');
                    $rmetas[$property] = $content;
                }
                $this->_og_data = $rmetas;
                sleep(1);
                return true;
            } else {
                throw new Exception('ERROR: OpenGraph = ' . $url . PHP_EOL);
                if (R::find('nog', ' link = ?', $url) == NULL) {
                    $nog = R::dispense('nog');
                    $nog->url = $url;
                    R::store($nog);
                }
            }
        } catch (Exception $e) {
            echo $e->getMessage();
        }
    }

    protected function save_to_db($category) {
        try {
            if (isset($this->_article['link'])) {
                if (R::find('article', ' link = ?', array($this->_article['link'])) == NULL) {
                    $article = R::dispense('article');
                    $article->title = $this->_article['title'];
                    $article->description = $this->_article['description'];
                    $article->link = $this->_article['link'];
                    date_default_timezone_set('America/Chicago');

                    $date = $this->_article['published'];
                    $time = strtotime($date);
                    $fixed = date('Y-m-d H:i:s', $time);

                    $article->published = $fixed;
                    $article->setMeta('cast.published','datetime');
                    $article->category = $category;
                    if (isset($this->_og_data['og_image'])) {
                        $article->image_link = $this->_og_data['og_image'];
                        $filename = md5($this->_og_data['og_image']) . ".jpg";
                        $article->image_file = $filename;
                        if (!file_exists('cache/' . $filename)) {
                            if ($file = @file_get_contents($this->_og_data['og_image'])) {
                                file_put_contents('cache/' . $filename, $file);
                                exec('convert cache' . DIRECTORY_SEPARATOR . $filename . ' -background none -resize 480x360 -quality 65 -sharpen 2x4 cache' . DIRECTORY_SEPARATOR . 'thumb' . DIRECTORY_SEPARATOR . $filename);
                                R::store($article);
                            }
                        }
                    }
                } else {
                    new Exception('ERROR: Not Stored - ' . $this->_article['link'] . PHP_EOL);
                }
            } else {
                throw new Exception('ERROR: Already Exists - ' . $this->_article['link'] . PHP_EOL);
            }
        } catch (Exception $e) {
            echo $e->getMessage();
        }
    }

}

class GoogleFeed extends FeedParseBase {

}


$feeds = array(
            // Cannabis
            "https://www.google.com/alerts/feeds/05402891751147148632/10666330494545493357"
        );


foreach($feeds as $feed) {
    $google = new GoogleFeed($feed);
    unset($google);
}