function gaan($url) { global $done; echo "<h2>" . $url . "</h2>"; $spider = new WebSpider($url); $spider->run(); var_dump($spider->listFoundKeywords()); echo "\n\n"; foreach ($spider->listFoundURLs() as $furl) { if (!in_array($furl, $done)) { $done[] = $furl; gaan($furl); } } }
<?php require 'WebSpider.php'; require 'ImportWebsite.php'; $webSpider = new WebSpider($websiteUrl); $importWebsite = new ImportWebsite($websiteUrl); $pagesArray = array(); $websiteMap = $webSpider->getWebsiteMap(); foreach ($websiteMap as $websiteUrl => $level) { $content = $webSpider->getContent($websiteUrl); preg_match("/http:\\/\\/" . $websiteUrl . "\\/(.*)?\\.[html|php]/", $websiteUrl, $matches); if (empty($matches)) { $matches[1] = 'index'; } $pagesArray[] = array('title' => $matches[1], 'status' => 'y', 'slug' => $matches[1], 'content' => $content, 'date' => date('j M Y')); } $importWebsite->pages($pagesArray); $importWebsite->buildTemplate(); /* require 'Process/Manager.php'; require 'Process/Process.php'; $launchPath = rtrim(realpath(dirname(__FILE__)), '/').'/webSpider.php'; $manager = new Oara_Import_Process_Manager(); $manager->executable = "php -f"; $manager->root = $launchPath; $manager->processes = 10; $manager->sleep_time = 2; $manager->exec(); */