/**
* The basic URL parser.
* @link https://url.spec.whatwg.org/#concept-basic-url-parser URL Standard
* @param string $input A UTF-8 string.
* @param URL|null $base A base URL.
* @param string|null $encodingOverride A valid name of an encoding.
* @param (URL|string)[]|null $urlAndStateOverride An URL ("url" key) and a state override ("state override" key).
* @throws \DomainException If $urlAndStateOverride['state override'] is invalid.
* @return URL|false|null
*/
public static function parseBasicURL($input, self $base = null, $encodingOverride = null, array $urlAndStateOverride = null)
{
$input = str_replace(["\t", "\n", "\r"], '', $input);
if ($urlAndStateOverride) {
$url = $urlAndStateOverride['url'];
$stateOverride = (string) $urlAndStateOverride['state override'];
$string = (string) $input;
$state = $stateOverride;
} else {
$url = new self();
$stateOverride = null;
$string = trim($input, ".. ");
$state = 'scheme start state';
}
$encoding = $encodingOverride ? URLencoding::getOutputEncoding((string) $encodingOverride) : 'UTF-8';
$buffer = '';
$atFlag = false;
$bracketFlag = false;
$codePoints = preg_split('//u', $string, -1, PREG_SPLIT_NO_EMPTY);
for ($pointer = 0; true; $pointer++) {
$c = isset($codePoints[$pointer]) ? $codePoints[$pointer] : '';
switch ($state) {
case 'scheme start state':
if (stripos('abcdefghijklmnopqrstuvwxyz', $c) !== false) {
$buffer .= strtolower($c);
$state = 'scheme state';
} elseif (!$stateOverride) {
$state = 'no scheme state';
$pointer--;
} else {
return;
}
break;
case 'scheme state':
if (stripos('0123456789abcdefghijklmnopqrstuvwxyz+-.', $c) !== false) {
$buffer .= strtolower($c);
} elseif ($c === ':') {
if ($stateOverride && array_key_exists($url->scheme, self::$specialSchemes) !== array_key_exists($buffer, self::$specialSchemes)) {
return;
}
$url->scheme = $buffer;
$buffer = '';
if ($stateOverride) {
return;
}
if ($url->scheme === 'file') {
$state = 'file state';
} elseif ($url->isSpecial() && $base && $base->scheme === $url->scheme) {
$state = 'special relative or authority state';
} elseif ($url->isSpecial()) {
$state = 'special authority slashes state';
} elseif (isset($codePoints[$pointer + 1]) && $codePoints[$pointer + 1] === '/') {
$state = 'path or authority state';
$pointer++;
} else {
$url->nonRelativeFlag = $url->cannotBeABaseURLFlag = true;
$url->path[] = '';
$state = 'non-relative path state';
}
} elseif (!$stateOverride) {
$buffer = '';
$state = 'no scheme state';
$pointer = -1;
} else {
return;
}
break;
case 'no scheme state':
if (!$base || $base->cannotBeABaseURLFlag && $c !== '#') {
return false;
} elseif ($base->cannotBeABaseURLFlag && $c === '#') {
$url->scheme = $base->scheme;
$url->path = $base->path;
$url->query = $base->query;
$url->fragment = '';
$url->nonRelativeFlag = $url->cannotBeABaseURLFlag = true;
$state = 'fragment state';
} elseif ($base->scheme !== 'file') {
$state = 'relative state';
$pointer--;
} else {
$state = 'file state';
$pointer--;
}
break;
case 'special relative or authority state':
if ($c === '/' && isset($codePoints[$pointer + 1]) && $codePoints[$pointer + 1] === '/') {
$state = 'special authority ignore slashes state';
$pointer++;
} else {
$state = 'relative state';
$pointer--;
}
break;
case 'path or authority state':
if ($c === '/') {
$state = 'authority state';
} else {
$state = 'path state';
$pointer--;
}
break;
case 'relative state':
$url->scheme = $base->scheme;
switch ($c) {
case '':
$url->username = $base->username;
$url->password = $base->password;
$url->host = $base->host;
$url->port = $base->port;
$url->path = $base->path;
$url->query = $base->query;
break;
case '/':
$state = 'relative slash state';
break;
case '?':
$url->username = $base->username;
$url->password = $base->password;
$url->host = $base->host;
$url->port = $base->port;
$url->path = $base->path;
$url->query = '';
$state = 'query state';
break;
case '#':
$url->username = $base->username;
$url->password = $base->password;
$url->host = $base->host;
$url->port = $base->port;
$url->path = $base->path;
$url->query = $base->query;
$url->fragment = '';
$state = 'fragment state';
break;
default:
if ($c === '\\' && $url->isSpecial()) {
$state = 'relative slash state';
} else {
$url->username = $base->username;
$url->password = $base->password;
$url->host = $base->host;
$url->port = $base->port;
$url->path = $base->path;
array_pop($url->path);
$state = 'path state';
$pointer--;
}
}
break;
case 'relative slash state':
if ($c === '/' || $c === '\\' && $url->isSpecial()) {
$state = 'special authority ignore slashes state';
} else {
$url->username = $base->username;
$url->password = $base->password;
$url->host = $base->host;
$url->port = $base->port;
$state = 'path state';
$pointer--;
}
break;
case 'special authority slashes state':
if ($c === '/' && isset($codePoints[$pointer + 1]) && $codePoints[$pointer + 1] === '/') {
$state = 'special authority ignore slashes state';
$pointer++;
} else {
$state = 'special authority ignore slashes state';
$pointer--;
}
break;
case 'special authority ignore slashes state':
if (!in_array($c, ['/', '\\'])) {
$state = 'authority state';
$pointer--;
}
break;
case 'authority state':
if ($c === '@') {
if ($atFlag) {
$buffer = '%40' . $buffer;
}
$atFlag = true;
$usernameAndPassword = explode(':', $buffer, 2);
$url->username .= self::percentEncodeCodePoints(Infrastructure::USERINFO_ENCODE_SET, $usernameAndPassword[0]);
if (isset($usernameAndPassword[1])) {
$url->password .= self::percentEncodeCodePoints(Infrastructure::USERINFO_ENCODE_SET, $usernameAndPassword[1]);
}
$buffer = '';
} elseif (in_array($c, ['', '/', '?', '#']) || $c === '\\' && $url->isSpecial()) {
$pointer -= mb_strlen($buffer, 'UTF-8') + 1;
$buffer = '';
$state = 'host state';
} else {
$buffer .= $c;
}
break;
case 'host state':
case 'hostname state':
if ($c === ':' && !$bracketFlag) {
if ($buffer === '' && $url->isSpecial()) {
return false;
}
$host = HostProcessing::parseHost($buffer);
if ($host === false) {
return false;
}
$url->host = $host;
$buffer = '';
$state = 'port state';
if ($stateOverride === 'hostname state') {
return;
}
} elseif (in_array($c, ['', '/', '?', '#']) || $c === '\\' && $url->isSpecial()) {
$pointer--;
if ($buffer === '' && $url->isSpecial()) {
return false;
}
$host = HostProcessing::parseHost($buffer);
if ($host === false) {
return false;
}
$url->host = $host;
$buffer = '';
$state = 'path start state';
if ($stateOverride) {
return;
}
} else {
if ($c === '[') {
$bracketFlag = true;
}
if ($c === ']') {
$bracketFlag = false;
}
$buffer .= $c;
}
break;
case 'port state':
if (ctype_digit($c)) {
$buffer .= $c;
} elseif (in_array($c, ['', '/', '?', '#']) || $c === '\\' && $url->isSpecial() || $stateOverride) {
if ($buffer !== '') {
$port = (int) $buffer;
if ($port > pow(2, 16) - 1) {
return false;
}
$url->port = isset(self::$specialSchemes[$url->scheme]) && self::$specialSchemes[$url->scheme] === $port ? null : $port;
$buffer = '';
}
if ($stateOverride) {
return;
}
$state = 'path start state';
$pointer--;
} else {
return false;
}
break;
case 'file state':
$url->scheme = 'file';
switch ($c) {
case '':
if ($base && $base->scheme === 'file') {
$url->host = $base->host;
$url->path = $base->path;
$url->query = $base->query;
}
break;
case '\\':
case '/':
$state = 'file slash state';
break;
case '?':
if ($base && $base->scheme === 'file') {
$url->host = $base->host;
$url->path = $base->path;
$url->query = '';
$state = 'query state';
}
break;
case '#':
if ($base && $base->scheme === 'file') {
$url->host = $base->host;
$url->path = $base->path;
$url->query = $base->query;
$url->fragment = '';
$state = 'fragment state';
}
break;
default:
$remaining = array_slice($codePoints, $pointer + 1);
if ($base && $base->scheme === 'file' && isset($remaining[0]) && preg_match(Infrastructure::WINDOWS_DRIVE_LETTER, $c . $remaining[0]) === 0 && (count($remaining) === 1 || isset($remaining[1]) && strpos('/\\?#', $remaining[1]) === false)) {
$url->host = $base->host;
$url->path = $base->path;
$url->popPath();
}
$state = 'path state';
$pointer--;
}
break;
case 'file slash state':
if ($c === '/' || $c === '\\') {
$state = 'file host state';
} else {
if ($base && $base->scheme === 'file' && isset($base->path[0]) && preg_match(Infrastructure::NORMALIZED_WINDOWS_DRIVE_LETTER, $base->path[0]) === 1) {
$url->path[] = $base->path[0];
}
$state = 'path state';
$pointer--;
}
break;
case 'file host state':
if (in_array($c, ['', '/', '\\', '?', '#'])) {
$pointer--;
if (preg_match(Infrastructure::WINDOWS_DRIVE_LETTER, $buffer) === 1) {
$state = 'path state';
} elseif ($buffer === '') {
$state = 'path start state';
} else {
$host = HostProcessing::parseHost($buffer);
if ($host === false) {
return false;
}
if ($host !== 'localhost') {
$url->host = $host;
}
$buffer = '';
$state = 'path start state';
}
} else {
$buffer .= $c;
}
break;
case 'path start state':
$state = 'path state';
if (!($c === '/' || $c === '\\' && $url->isSpecial())) {
$pointer--;
}
break;
case 'path state':
if (in_array($c, ['', '/']) || $c === '\\' && $url->isSpecial() || !$stateOverride && in_array($c, ['?', '#'])) {
if (preg_match(self::DOUBLE_DOT_PATH_SEGMENT, $buffer) === 1) {
$url->popPath();
if (!($c === '/' || $c === '\\' && $url->isSpecial())) {
$url->path[] = '';
}
} elseif (preg_match(self::SINGLE_DOT_PATH_SEGMENT, $buffer) === 1 && !($c === '/' || $c === '\\' && $url->isSpecial())) {
$url->path[] = '';
} elseif (preg_match(self::SINGLE_DOT_PATH_SEGMENT, $buffer) !== 1) {
if ($url->scheme === 'file' && !$url->path && preg_match(Infrastructure::WINDOWS_DRIVE_LETTER, $buffer) === 1) {
$url->host = null;
$buffer[1] = ':';
}
$url->path[] = $buffer;
}
$buffer = '';
if ($c === '?') {
$url->query = '';
$state = 'query state';
} elseif ($c === '#') {
$url->fragment = '';
$state = 'fragment state';
}
} else {
if (stripos(implode('', array_slice($codePoints, $pointer)), '%2e') === 0) {
$buffer .= '.';
$pointer += 2;
} else {
$buffer .= Infrastructure::utf8PercentEncode(Infrastructure::DEFAULT_ENCODE_SET, $c);
}
}
break;
case 'non-relative path state':
if ($c === '?') {
$url->query = '';
$state = 'query state';
} elseif ($c === '#') {
$url->fragment = '';
$state = 'fragment state';
} else {
if ($c !== '') {
$url->path[0] .= Infrastructure::utf8PercentEncode(Infrastructure::SIMPLE_ENCODE_SET, $c);
}
}
break;
case 'query state':
if ($c === '' || !$stateOverride && $c === '#') {
if (!$url->isSpecial() || $url->scheme === 'ws' || $url->scheme === 'wss') {
$encoding = 'UTF-8';
}
$buffer = URLencoding::encode($buffer, $encoding);
$url->query = self::percentEncodeCodePoints('/[^!$-;=?-~]/', $buffer);
$buffer = '';
if ($c === '#') {
$url->fragment = '';
$state = 'fragment state';
}
} else {
$buffer .= $c;
}
break;
case 'fragment state':
if ($c !== '') {
$url->fragment .= str_replace("", '', implode('', array_slice($codePoints, $pointer)));
}
break 2;
default:
throw new \DomainException(sprintf('"%s" is an unknown state', $state));
}
if ($pointer >= 0 && !isset($codePoints[$pointer])) {
break;
}
}
return $url;
}
/**
* @param string $input
* @param boolean|null $unicodeFlag
* @param string|integer[]|false $domain
* @param string|null $message
* @dataProvider hostProvider
*/
public function testParseHost($input, $unicodeFlag, $domain, $message = null)
{
$this->assertSame($domain, $unicodeFlag === null ? HostProcessing::parseHost($input) : HostProcessing::parseHost($input, $unicodeFlag), $message);
}