function re_categorize()
{
$pidfile = "/etc/artica-postfix/pids/" . basename(__FILE__) . "." . __FUNCTION__ . ".pid";
$oldpid = @file_get_contents($pidfile);
if ($oldpid < 100) {
$oldpid = null;
}
$unix = new unix();
if ($unix->process_exists($oldpid, basename(__FILE__))) {
if ($GLOBALS["VERBOSE"]) {
echo "Already executed pid {$oldpid}\n";
}
die;
}
if (systemMaxOverloaded()) {
writelogs_squid("Fatal: VERY Overloaded system, die();", __FUNCTION__, __FILE__, __LINE__, "stats");
return;
}
$mypid = getmypid();
@file_put_contents($pidfile, $mypid);
$sock = new sockets();
$RecategorizeSecondsToWaitOverload = $sock->GET_INFO("RecategorizeSecondsToWaitOverload");
$RecategorizeMaxExecutionTime = $sock->GET_INFO("RecategorizeSecondsToWaitOverload");
$RecategorizeProxyStats = $sock->GET_INFO("RecategorizeProxyStats");
if (!is_numeric($RecategorizeProxyStats)) {
$RecategorizeProxyStats = 1;
}
if (!is_numeric($RecategorizeSecondsToWaitOverload)) {
$RecategorizeSecondsToWaitOverload = 30;
}
if (!is_numeric($RecategorizeMaxExecutionTime)) {
$RecategorizeMaxExecutionTime = 210;
}
if ($RecategorizeProxyStats == 0) {
return;
}
$t = time();
$sql = "SELECT * FROM visited_sites";
$results = $GLOBALS["Q"]->QUERY_SQL($sql);
$num_rows = mysql_num_rows($results);
$c = 0;
while ($ligne = @mysql_fetch_array($results, MYSQL_ASSOC)) {
$website = trim($ligne["sitename"]);
if ($website == null) {
continue;
}
$category = trim($GLOBALS["Q"]->GET_CATEGORIES($website, true));
$GLOBALS["Q"]->QUERY_SQL("UPDATE visited_sites SET category='{$category}' WHERE sitename='{$website}'");
if (!$GLOBALS["Q"]->ok) {
writelogs_squid("Fatal: mysql error {$GLOBALS["Q"]->mysql_error}", __FUNCTION__, __FILE__, __LINE__, "stats");
return;
}
$c++;
if ($c > 5000) {
$distanceInSeconds = round(abs(time() - $t));
$distanceInMinutes = round($distanceInSeconds / 60);
if ($distanceInMinutes > $RecategorizeMaxExecutionTime) {
$took = $unix->distanceOfTimeInWords($t, time());
writelogs_squid("Re-categorized websites task aborted (Max execution time {$RecategorizeMaxExecutionTime}Mn) ({$took})", __FUNCTION__, __FILE__, __LINE__, "stats");
return;
}
$c = 0;
}
}
$took = $unix->distanceOfTimeInWords($t, time());
writelogs_squid("{$num_rows} re-categorized websites in main table ({$took})", __FUNCTION__, __FILE__, __LINE__, "stats");
__re_categorize_subtables($t);
}
function re_categorize($nopid = false)
{
$q = $GLOBALS["Q"];
if (!$nopid) {
$pidfile = "/etc/artica-postfix/pids/" . basename(__FILE__) . "." . __FUNCTION__ . ".pid";
$pid = @file_get_contents($pidfile);
if ($pid < 100) {
$pid = null;
}
$unix = new unix();
if ($unix->process_exists($pid, basename(__FILE__))) {
ufdbguard_admin_events("Already executed pid {$pid}", __FUNCTION__, __FILE__, __LINE__, "stats");
if ($GLOBALS["VERBOSE"]) {
echo "Already executed pid {$pid}\n";
}
return;
}
$mypid = getmypid();
@file_put_contents($pidfile, $mypid);
}
$sock = new sockets();
$RecategorizeSecondsToWaitOverload = $sock->GET_INFO("RecategorizeSecondsToWaitOverload");
$RecategorizeMaxExecutionTime = $sock->GET_INFO("RecategorizeSecondsToWaitOverload");
$RecategorizeProxyStats = $sock->GET_INFO("RecategorizeProxyStats");
if (!is_numeric($RecategorizeProxyStats)) {
$RecategorizeProxyStats = 1;
}
if (!is_numeric($RecategorizeSecondsToWaitOverload)) {
$RecategorizeSecondsToWaitOverload = 30;
}
if (!is_numeric($RecategorizeMaxExecutionTime)) {
$RecategorizeMaxExecutionTime = 210;
}
if ($RecategorizeProxyStats == 0) {
ufdbguard_admin_events("RecategorizeProxyStats=0, aborting...", __FUNCTION__, __FILE__, __LINE__, "stats");
return;
}
$t = time();
if (!$GLOBALS["Q"]->FIELD_EXISTS("visited_sites", "recatgorized")) {
$GLOBALS["Q"]->QUERY_SQL("ALTER TABLE `visited_sites` ADD `recatgorized` smallint(1) NOT NULL ,\n\t\t\t\tADD KEY `recatgorized` (`recatgorized`)");
}
$sql = "SELECT * FROM visited_sites WHERE recatgorized='0' LIMIT 0,300";
$results = $GLOBALS["Q"]->QUERY_SQL($sql);
$num_rows = mysql_num_rows($results);
stats_admin_events(2, "Trying to recategorize {$num_rows} visited websites", "", __FILE__, __LINE__);
$t = time();
$c = 0;
$L = 0;
while ($ligne = @mysql_fetch_array($results, MYSQL_ASSOC)) {
$website = trim($ligne["sitename"]);
$categorySrc = $ligne["category"];
if ($website == null) {
continue;
}
$category = trim($GLOBALS["Q"]->GET_CATEGORIES($website, true));
if ($category == $categorySrc) {
$GLOBALS["Q"]->QUERY_SQL("UPDATE visited_sites SET recatgorized=1 WHERE sitename='{$website}'");
continue;
}
if ($category == null) {
$GLOBALS["Q"]->QUERY_SQL("UPDATE visited_sites SET recatgorized=1 WHERE sitename='{$website}'");
continue;
}
$GLOBALS["Q"]->QUERY_SQL("UPDATE visited_sites SET category='{$category}',recatgorized=1 WHERE sitename='{$website}'");
if (!$GLOBALS["Q"]->ok) {
stats_admin_events(0, "Fatal: mysql error", "{$GLOBALS["Q"]->mysql_error}", __FILE__, __LINE__);
return;
}
$c++;
$L++;
$websites[$website] = $category;
if ($L > 400) {
$took = $unix->distanceOfTimeInWords($t, time());
if (SquidStatisticsTasksOverTime()) {
stats_admin_events(1, "Statistics overtime... Aborting", null, __FILE__, __LINE__);
return;
}
$L = 0;
}
}
$took = $unix->distanceOfTimeInWords($t, time());
stats_admin_events(2, "{$c} re-categorized websites in main table ({$took})", __FUNCTION__, __FILE__, __LINE__, "stats");
if (count($websites) > 0) {
__re_categorize_subtables($t, $websites);
}
}