settype($test_exists['last_modified'], 'string');
$exists_spider_id = $test_exists['spider_id'];
$upddate = $test_exists['upddate'];
$last_modif_old = $test_exists['last_modified'];
} else {
$exists_spider_id = 0;
}
$url_indexing = $full_url . $temp_path . $temp_file;
$url_indexing = ereg_replace("[?]\$", "", $url_indexing);
$url_print = $url . $temp_path . $temp_file;
$url_print = ereg_replace("[?]\$", "", $url_print);
//verify if 'revisit-after' date is expired or if page doesn't exists, or force is on.
if ($exists_spider_id == 0 || $upddate < $date || $force_first_reindex == 1 && ($level == 0 || $already_indexed == 0)) {
//test content-type of this page if not excluded
$result_test_http = '';
if (!phpdigReadRobots($exclude, $temp_path . $temp_file) && !eregi(FORBIDDEN_EXTENSIONS, $temp_path . $temp_file)) {
$result_test_http = phpdigTestUrl($url_indexing, 'date', $cookies);
}
if (is_array($result_test_http) && !in_array($result_test_http['status'], array('NOHOST', 'NOFILE', 'LOOP', 'NEWHOST'))) {
$tested_url = phpdigRewriteUrl($result_test_http['path']);
$cookies = $result_test_http['cookies'];
// update URI if redirect in same host...
if ($tested_url['path'] != $temp_path || $tested_url['file'] != $temp_file) {
$temp_path = $tested_url['path'];
$temp_file = $tested_url['file'];
$query = "UPDATE " . PHPDIG_DB_PREFIX . "tempspider SET path='{$temp_path}', file='{$temp_file}', WHERE id={$tempspider_id}";
mysql_query($query, $id_connect);
$url_indexing = $full_url . $temp_path . $temp_file;
$url_indexing = ereg_replace("[?]\$", "", $url_indexing);
$url_print = $url . $temp_path . $temp_file;
$url_print = ereg_replace("[?]\$", "", $url_print);
function phpdigDetectDir($link, $exclude = '', $cookies = array(), $site_id = '', $id_connect = '')
{
$test = parse_url($link['path'] . $link['file']);
//test the exclude with robots.txt
if (phpdigReadRobots($exclude, $link['path'] . $link['file']) == 1 || isset($exclude['@ALL@'])) {
$link['ok'] = 0;
} elseif (!isset($test['query']) && !eregi('[.][a-z0-9]{1,4}$', $link['path'] . $link['file']) && ($status = phpdigTestUrl($link['url'] . $link['path'] . $link['file'] . '/', 'date', $cookies)) && isset($status['status']) && $status['status'] == "HTML") {
$link['path'] = ereg_replace('/+$', '/', $link['path'] . $link['file'] . '/');
if ($link['path'] == '/') {
$link['path'] = '';
}
$link['file'] = "";
$link['ok'] = 1;
} else {
$status = phpdigTestUrl($link['url'] . $link['path'] . $link['file'], 'date', $cookies);
if (!in_array($status['status'], array('NOHOST', 'NOFILE', 'LOOP', 'NEWHOST'))) {
$link['ok'] = 1;
} else {
$link['ok'] = 0;
}
}
if (is_numeric($site_id) && LIMIT_TO_DIRECTORY) {
$query = "SELECT DISTINCT in_id, in_path FROM " . PHPDIG_DB_PREFIX . "includes WHERE in_site_id='{$site_id}'";
if (is_array($list_include = phpdigMySelect($id_connect, $query))) {
foreach ($list_include as $add_include) {
if ($link['path'] != $add_include['in_path']) {
$link['ok'] = 0;
}
}
}
}
if (!$link['ok'] && isset($status)) {
$link['status'] = $status['status'];
$link['host'] = $status['host'];
$link['path'] = $status['path'];
$link['cookies'] = $status['cookies'];
}
return $link;
}
