public function testRootdomains()
{
foreach ($this->rootDomains as $root => $examples) {
foreach ($examples as $url) {
$parser = \Nabble\Domainparser::parseUrl($url);
$this->assertEquals($root, $parser['topleveldomain']);
}
}
}
<?php
/**
* Merge a list with the blocked domains
*/
require_once './../vendor/autoload.php';
$semaltBlockerSources = \Nabble\Semalt::getBlocklist();
echo "Old list: " . count($semaltBlockerSources) . " sources\n";
echo "Paste domains (one on each line), then press ENTER twice:\n\n";
$handle = fopen("php://stdin", "r");
$newDomains = [];
while ($line = fgets($handle)) {
if (trim($line) == '') {
echo "Done, got " . count($newDomains) . " sources, new domains is ";
$newList = array_merge($semaltBlockerSources, $newDomains);
foreach ($newList as &$source) {
$source = \Nabble\Domainparser::getToplevelDomain($source);
}
$newList = array_unique($newList);
sort($newList);
echo count($newList) . " sources:\n\n";
echo implode("\n", $newList) . "\n\n";
break;
} else {
$newDomains[] = trim($line);
}
}
// iterate all sources
foreach ($sources as $source => $regex) {
$raw = file_get_contents($source);
if (!empty($regex)) {
preg_match_all($regex, $raw, $list);
$list = array_filter($list[1], function ($v) {
return filter_var($v, FILTER_VALIDATE_URL) || filter_var('http://' . $v, FILTER_VALIDATE_URL);
});
} else {
$list = explode("\n", $raw);
}
$spammers = array_merge($spammers, $list);
}
// only top-level domains
foreach ($spammers as &$spammer) {
$spammer = \Nabble\Domainparser::getToplevelDomain($spammer);
}
// merge & cleanup spammers
$spammers = array_merge(\Nabble\Semalt::getBlocklist(), $spammers);
$spammers = array_map('strtolower', $spammers);
$spammers = array_map('trim', $spammers);
$spammers = array_unique($spammers);
$spammers = array_filter($spammers);
sort($spammers);
// echo some info
echo "New list: " . count($spammers) . " sources\n";
// write
if (count($spammers)) {
file_put_contents('../domains/blocked', implode("\n", $spammers));
}
echo "Done\n";
