/** URI pickup
* Return an array of URIs in the $string
* [OK] http://nasty.example.org#nasty_string
* [OK] http://nasty.example.org:80/foo/xxx#nasty_string/bar
* [OK] ftp://nasty.example.org:80/dfsdfs
* [OK] ftp://cnn.example.com&story=breaking_news@10.0.0.1/top_story.htm (from RFC3986)
* Not available for: IDN(ignored)
* @param string $string
* @return array
*/
static function uri_pickup($string = '')
{
if (!is_string($string)) {
return array();
}
$array = array();
preg_match_all('#(\\b[a-z][a-z0-9.+-]{1,8}):[/\\\\]+' . '(?:' . '([^\\s<>"\'\\[\\]/\\#?@]*)' . '@)?' . '(' . '\\[[0-9a-f:.]+\\]' . '|' . '(?:[0-9]{1,3}\\.){3}[0-9]{1,3}' . '|' . '[a-z0-9_-][a-z0-9_.-]+[a-z0-9_-]' . ')' . '(?::([0-9]*))?' . '((?:/+[^\\s<>"\'\\[\\]/\\#?]+)*/+)?' . '([^\\s<>"\'\\[\\]\\#?]+)?' . '(?:\\?([^\\s<>"\'\\[\\]\\#]+))?' . '(?:\\#([a-z0-9._~%!$&\'()*+,;=:@-]*))?' . '#i', $string, $array, PREG_SET_ORDER | PREG_OFFSET_CAPTURE);
// Reformat the $array
static $parts = array(1 => 'scheme', 2 => 'userinfo', 3 => 'host', 4 => 'port', 5 => 'path', 6 => 'file', 7 => 'query', 8 => 'fragment');
$default = array(0 => '', 1 => -1);
foreach (array_keys($array) as $uri) {
$_uri =& $array[$uri];
SpamUtility::array_rename_keys($_uri, $parts, TRUE, $default);
$offset = $_uri['scheme'][1];
// Scheme's offset = URI's offset
foreach (array_keys($_uri) as $part) {
$_uri[$part] = $_uri[$part][0];
// Remove offsets
}
}
foreach (array_keys($array) as $uri) {
$_uri =& $array[$uri];
if ($_uri['scheme'] === '') {
unset($array[$uri]);
// Considererd harmless
continue;
}
unset($_uri[0]);
// Matched string itself
$_uri['area']['offset'] = $offset;
// Area offset for area_measure()
}
return $array;
}
private static function summarize_detail_newtral($progress = array())
{
if (!isset($progress['hosts']) || !is_array($progress['hosts']) || empty($progress['hosts'])) {
return '';
}
// Generate a responsible $trie
$trie = array();
foreach ($progress['hosts'] as $value) {
// 'A.foo.bar.example.com'
$resp = SpamPickup::whois_responsibility($value);
// 'example.com'
if (empty($resp)) {
// One or more test, or do nothing here
$resp = strval($value);
$rest = '';
} else {
$rest = rtrim(substr($value, 0, -strlen($resp)), '.');
// 'A.foo.bar'
}
$trie = SpamUtility::array_merge_leaves($trie, array($resp => array($rest => NULL)), FALSE);
}
// Format: var_export_shrink() -like output
$result = array();
ksort_by_domain($trie);
foreach (array_keys($trie) as $key) {
ksort_by_domain($trie[$key]);
if (count($trie[$key]) == 1 && key($trie[$key]) == '') {
// Just one 'responsibility.example.com'
$result[] = ' \'' . $key . '\',';
} else {
// One subdomain-or-host, or several ones
$subs = array();
foreach (array_keys($trie[$key]) as $sub) {
if ($sub == '') {
$subs[] = $key;
// 'example.com'
} else {
$subs[] = $sub . '. ';
// 'A.foo.bar. '
}
}
$result[] = ' \'' . $key . '\' => \'' . implode(', ', $subs) . '\',';
}
unset($trie[$key]);
}
return 'array (' . "\n" . implode("\n", $result) . "\n" . ')';
}
