synth_tools

Back

6298fd8aa3 Gunnerkrigg and cleanup of alt-tag-only sites elseif (strpos($article['link'], 'wondermark.com/c') !== FALSE) { $doc = new DOMXpath($doc); $imgs = $xpath->query('//img'); //doesn't get simpler than this foreach($imgs as $img){ // Questionable Content (cleanup // only keep everything starting at the first // Least I Could Do (wtf image size?) $xpath = $this->get_xpath_dealie($article['link']); elseif (strpos($article['content'], 'thedoghousediaries.com/dhdcomics/') !== FALSE){ $xpath = $this->get_xpath_dealie($article['link']); $article['content'] = $this->get_img_tags($xpath, "//div[@class='comic-wrap']", $article); //Sites that provide images and just need alt tags elseif (strpos($article['link'], 'http://www.geekculture.com/joyoftech/') !== FALSE) { $article['content'] = $this->get_img_tags($xpath, '(//div[@id="comic"]//img)', $article); } // Scenes From A Multiverse (to get alt tag elseif (strpos($article['link'], 'girlswithslingshots.com/comic/') !== FALSE) { $xpath = $this->get_xpath_dealie($article['link']); $article['content'] = $this->get_img_tags($xpath, '(//div[@id="comic"]//img)', $article) . $article['content']; // $article['content'] = $this->get_img_tags($xpath, '//td/img[contains(@src, "/comics/images/")]', $article); // $article['content'] = $this->get_img_tags($xpath, '//td/img[contains(@src, "/comics/images/")]', $article); // } // Joy of Tech elseif (strpos($article['link'], 'paintraincomic.com/comic/') !== FALSE) { $xpath = $this->get_xpath_dealie($article['link']); $article['content'] = $this->get_img_tags($xpath, "//div[@class='singleImage']/img[@class='magicfields']", $article); $article['content'] = $this->get_img_tags($xpath, "//div[@id='comic-img']//img", $article); } // Dinosaur Comics Cleanup $entries = $xpath->query("//div[@class='entry']"); if (preg_match("@.*(.*)@", $article['content'], $matches)) { $article['content'] = preg_replace('#(/[0-9-]+)-150x150\.gif#', '$1.gif', $article['content']); $article['content'] = preg_replace('#(/[0-9-]+)-150x150\.gif#', '$1.gif', $article['content']); $article['content'] = $this->get_img_tags($xpath, "//div[@id='comic']/img", $article); } // Girls with Slingshots elseif (strpos($article['content'], 'imgs.xkcd.com/comics/') !== FALSE) { // Achewood (alt tag already present) elseif (strpos($article['content'], 'thedoghousediaries.com/dhdcomics/') !== FALSE){ $article['content'] = $this->get_img_tags($xpath, "//div[@id='comic']//img", $article); } // Timothy Winchester (People I Know foreach ($imgs as $img) { $article['content'] = $this->get_img_tags($xpath, '(//div[@id="main"]//img)', $article); elseif (strpos($article['content'], 'invisiblebread.com/2') !== FALSE) { $xpath = $this->get_xpath_dealie($article['link']); $article['content'] = $this->get_img_tags($xpath, "//div[@class='comicpage']//img[contains(@src, 'Strip')]", $article); .

New Pull Request

Labels Milestones