*/ define('CLI_SCRIPT', true); require dirname(dirname(dirname(dirname(__FILE__)))) . '/config.php'; require_once $CFG->libdir . '/clilib.php'; require_once $CFG->dirroot . '/local/linkchecker_robot/lib.php'; list($options, $unrecognized) = cli_get_params(array('help' => false, 'url' => null), array('h' => 'help', 'u' => 'url')); if ($unrecognized) { $unrecognized = implode("\n ", $unrecognized); cli_error(get_string('cliunknowoption', 'admin', $unrecognized)); } $help = "Scrape the url as the robot would see it, but do not process/queue it.\n\nOptions:\n-h, --help Print out this help\n-u, --url Url to scrape\n\nExample:\n\$sudo -u www-data php scrape-as.php --url=http://ford.com/\n"; if ($options['help']) { echo $help; die; } $robot = new \local_linkchecker_robot\robot\crawler(); $error = $robot->is_bot_valid(); if ($error) { print "Error: {$error}"; exit; } if (!$options['url']) { echo $help; die; } $url = $options['url']; $node = $robot->scrape($url); $dump = $node->contents; unset($node->contents); print $dump; var_dump($node);
* @copyright Catalyst IT * @license http://www.gnu.org/copyleft/gpl.html GNU GPL v3 or later */ define('CLI_SCRIPT', true); require dirname(dirname(dirname(dirname(__FILE__)))) . '/config.php'; require_once $CFG->libdir . '/clilib.php'; require_once $CFG->dirroot . '/local/linkchecker_robot/lib.php'; list($options, $unrecognized) = cli_get_params(array('help' => false, 'url' => null), array('h' => 'help', 'u' => 'url')); if ($unrecognized) { $unrecognized = implode("\n ", $unrecognized); cli_error(get_string('cliunknowoption', 'admin', $unrecognized)); } $help = "Crawl a url as the robot and parse it.\n\nUseful for when a page has been corrected and you want to instantly reflect this.\n\nOptions:\n-h, --help Print out this help\n-u, --url Url to crawl and process\n\nExample:\n\$sudo -u www-data php crawl-as.php --url=http://ford.com/\n"; if ($options['help']) { echo $help; die; } $robot = new \local_linkchecker_robot\robot\crawler(); $error = $robot->is_bot_valid(); if ($error) { print "Error: {$error}"; exit; } if (!$options['url']) { echo $help; die; } $url = $options['url']; $node = $robot->mark_for_crawl($url, $url); $res = $robot->scrape($url); $result = $robot->crawl($node, 2);