|
|
|
|
<?php
|
|
|
|
|
|
|
|
|
|
// Debug
|
|
|
|
|
ini_set('display_errors', '1');
|
|
|
|
|
ini_set('display_startup_errors', '1');
|
|
|
|
|
error_reporting(E_ALL);
|
|
|
|
|
|
|
|
|
|
// Website
|
|
|
|
|
define('WEBSITE_DOMAIN', (issue($_SERVER['HTTP_HOST']) ? 'http://' . $_SERVER['HTTP_HOST'] : ''));
|
|
|
|
|
define('WEBSITE_PAGINATION_SEARCH_RESULTS_LIMIT', 100);
|
|
|
|
|
define('WEBSITE_IDENTICON_IMAGE_CACHE', true);
|
|
|
|
|
|
|
|
|
|
// Database
|
|
|
|
|
define('DB_HOST', 'localhost');
|
|
|
|
|
define('DB_PORT', 3306);
|
|
|
|
|
define('DB_NAME', '');
|
|
|
|
|
define('DB_USERNAME', '');
|
|
|
|
|
define('DB_PASSWORD', '');
|
|
|
|
|
|
|
|
|
|
// Sphinx
|
|
|
|
|
define('SPHINX_HOST', '127.0.0.1');
|
|
|
|
|
define('SPHINX_PORT', 9306);
|
|
|
|
|
|
|
|
|
|
// Crawl settings
|
|
|
|
|
define('CRAWL_PAGE_LIMIT', 10);
|
|
|
|
|
define('CRAWL_PAGE_SECONDS_OFFSET', 3600);
|
|
|
|
|
|
|
|
|
|
define('CRAWL_URL_REGEXP', '/^.*$/ui'); // ipv6 only '/^http:\/\/\[[\w:]+\].*$/ui'
|
|
|
|
|
|
|
|
|
|
define('CRAWL_HOST_DEFAULT_PAGES_LIMIT', 1000);
|
|
|
|
|
define('CRAWL_HOST_DEFAULT_STATUS', true);
|
|
|
|
|
define('CRAWL_HOST_DEFAULT_META_ONLY', false);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Default robots.txt rules on remote file not exists
|
|
|
|
|
* The crawler able to overwrite these rules
|
|
|
|
|
*
|
|
|
|
|
* Presets
|
|
|
|
|
* yggdrasil: /database/yggdrasil/host.robots.md
|
|
|
|
|
*
|
|
|
|
|
*/
|
|
|
|
|
define('CRAWL_ROBOTS_DEFAULT_RULES', null); // string|null
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Permanent rules that append to the robots.txt if exists else CRAWL_ROBOTS_DEFAULT_RULES
|
|
|
|
|
* The crawler does not overwrite these rules
|
|
|
|
|
*
|
|
|
|
|
* Presets
|
|
|
|
|
* yggdrasil: /database/yggdrasil/host.robotsPostfix.md
|
|
|
|
|
*
|
|
|
|
|
*/
|
|
|
|
|
define('CRAWL_ROBOTS_POSTFIX_RULES', null); // string|null
|
