/**
* Return an appropriately formatted LIKE query and the clause
*/
static function mungeQuery($query, $prot)
{
$field = 'el_index';
$rv = LinkFilter::makeLike($query, $prot);
if ($rv === false) {
//makeLike doesn't handle wildcard in IP, so we'll have to munge here.
if (preg_match('/^(:?[0-9]{1,3}\\.)+\\*\\s*$|^(:?[0-9]{1,3}\\.){3}[0-9]{1,3}:[0-9]*\\*\\s*$/', $query)) {
$rv = $prot . rtrim($query, " \t*") . '%';
$field = 'el_to';
}
}
return array($rv, $field);
}
}
}
//------------------------------------------------------------------------------
$username = wfMsg('spambot_username');
$fname = $username;
$wgUser = User::newFromName($username);
// Create the user if necessary
if (!$wgUser->getID()) {
$wgUser->addToDatabase();
}
if (!isset($args[0])) {
print "Usage: php cleanupSpam.php <hostname>\n";
exit(1);
}
$spec = $args[0];
$like = LinkFilter::makeLike($spec);
if (!$like) {
print "Not a valid hostname specification: {$spec}\n";
exit(1);
}
$dbr =& wfGetDB(DB_SLAVE);
if ($options['all']) {
// Clean up spam on all wikis
$dbr =& wfGetDB(DB_SLAVE);
print "Finding spam on " . count($wgLocalDatabases) . " wikis\n";
$found = false;
foreach ($wgLocalDatabases as $db) {
$count = $dbr->selectField("`{$db}`.externallinks", 'COUNT(*)', array('el_index LIKE ' . $dbr->addQuotes($like)), $fname);
if ($count) {
$found = true;
passthru("php cleanupSpam.php {$db} {$spec} | sed s/^/{$db}: /");
private function run($resultPageSet = null)
{
$params = $this->extractRequestParams();
$protocol = $params['protocol'];
$query = $params['query'];
if (is_null($query)) {
$this->dieUsage('Missing required query parameter', 'params');
}
// Find the right prefix
global $wgUrlProtocols;
foreach ($wgUrlProtocols as $p) {
if (substr($p, 0, strlen($protocol)) === $protocol) {
$protocol = $p;
break;
}
}
$likeQuery = LinkFilter::makeLike($query, $protocol);
if (!$likeQuery) {
$this->dieUsage('Invalid query', 'bad_query');
}
$likeQuery = substr($likeQuery, 0, strpos($likeQuery, '%') + 1);
$this->addTables(array('page', 'externallinks'));
// must be in this order for 'USE INDEX'
$this->addOption('USE INDEX', 'el_index');
$db = $this->getDB();
$this->addWhere('page_id=el_from');
$this->addWhere('el_index LIKE ' . $db->addQuotes($likeQuery));
$this->addWhereFld('page_namespace', $params['namespace']);
$prop = array_flip($params['prop']);
$fld_ids = isset($prop['ids']);
$fld_title = isset($prop['title']);
$fld_url = isset($prop['url']);
if (is_null($resultPageSet)) {
$this->addFields(array('page_id', 'page_namespace', 'page_title'));
$this->addFieldsIf('el_to', $fld_url);
} else {
$this->addFields($resultPageSet->getPageTableFields());
}
$limit = $params['limit'];
$offset = $params['offset'];
$this->addOption('LIMIT', $limit + 1);
if (isset($offset)) {
$this->addOption('OFFSET', $offset);
}
$res = $this->select(__METHOD__);
$data = array();
$count = 0;
while ($row = $db->fetchObject($res)) {
if (++$count > $limit) {
// We've reached the one extra which shows that there are additional pages to be had. Stop here...
$this->setContinueEnumParameter('offset', $offset + $limit + 1);
break;
}
if (is_null($resultPageSet)) {
$vals = array();
if ($fld_ids) {
$vals['pageid'] = intval($row->page_id);
}
if ($fld_title) {
$title = Title::makeTitle($row->page_namespace, $row->page_title);
$vals['ns'] = intval($title->getNamespace());
$vals['title'] = $title->getPrefixedText();
}
if ($fld_url) {
$vals['url'] = $row->el_to;
}
$data[] = $vals;
} else {
$resultPageSet->processDbRow($row);
}
}
$db->freeResult($res);
if (is_null($resultPageSet)) {
$result = $this->getResult();
$result->setIndexedTagName($data, $this->getModulePrefix());
$result->addValue('query', $this->getModuleName(), $data);
}
}
function cleanUp($phrase, $database)
{
global $wgOut, $wgUser;
if (!isset($phrase) || "" == $phrase) {
$this->showForm(wfMsg('cleanupspam_error_empty'));
return;
}
/* do a check whether something actually _is_ inside $wgLocalDatabases */
if (!is_array($wgLocalDatabases) && 'local' == $this->mMode) {
$this->showForm(wfMsg('cleanupspam_no_local'));
return;
}
$like = LinkFilter::makeLike($phrase);
if (!$like) {
$this->showForm(wfMsg('cleanupspam_error_not_valid') . ": " . $phrase);
return;
}
$like = $phrase;
$dbr =& wfGetDB(DB_SLAVE);
switch ($this->mMode) {
case 'this':
/* Clean up spam just on this wiki */
$res = $dbr->select('externallinks', array('DISTINCT el_from'), array('el_to LIKE ' . $dbr->addQuotes("%//{$phrase}%")), $fname);
$count = $dbr->numRows($res);
if ($count) {
$wgOut->addWikiText("Found {$count} article(s) containing links to '''{$phrase}'''.\n");
while ($row = $dbr->fetchObject($res)) {
if ('revert' == $this->mDo) {
/* have eyes on this */
$this->cleanupArticle($row->el_from, $phrase);
} else {
/* just add more data and that should be fine */
$this->writeupArticle($row->el_from, $phrase, $row->el_to);
}
}
} else {
$wgOut->addWikiText(wfMsg('cleanupspam_count_zero', "'''" . $phrase . "'''."));
}
break;
case 'all':
/* todo check for no wikis in city_list */
$wikis = $this->fetchWikias();
if (!is_array($wikis)) {
return;
}
$wgOut->addWikiText("Finding spam on all (" . count($wikis) . ") wikis.\n");
foreach ($wikis as $db) {
$count = $dbr->selectField("`" . $db->city_dbname . "`.externallinks", 'COUNT(*)', array('el_to LIKE ' . $dbr->addQuotes("%//{$phrase}%")), $fname);
if ($count) {
$found = true;
$this->cleanUp($phrase, $db->city_dbname);
}
}
if ('revert' == $this->mDo) {
if ($found) {
$wgOut->addWikiText(wfMsg('cleanupspam_cleanup_finished'));
} else {
$wgOut->addWikiText(wfMsg('cleanupspam_none_found', $phrase));
}
}
break;
}
if ('revert' == $this->mDo) {
$wgOut->addWikiText(wfMsg('cleanupspam_cleanup_finished'));
}
$sk = $wgUser->getSkin();
$titleObj = Title::makeTitle(NS_SPECIAL, 'Cleanupspam');
$link_back = $sk->makeKnownLinkObj($titleObj, '<b>here</b>');
$wgOut->addHtml("<br/>" . wfMsg('cleanupspam_link_back') . " " . $link_back . ".");
}
/**
* Return an appropriately formatted LIKE query
*/
static function mungeQuery($query, $prot)
{
return LinkFilter::makeLike($query, $prot);
}
