Uri: unicode domain name awareness

Author: codemasher

src/Psr7/Uri.php

@@ -13,7 +13,7 @@
 use InvalidArgumentException;
 use Psr\Http\Message\UriInterface;
 
-use function call_user_func_array, explode, filter_var, is_array, is_string, ltrim, parse_url,
+use function call_user_func_array, explode, filter_var, is_array, is_string, ltrim, mb_strtolower,
 	preg_replace_callback, rawurlencode, strpos, strtolower, ucfirst, var_export;
 
 use const FILTER_FLAG_IPV6, FILTER_VALIDATE_IP;
@@ -48,7 +48,7 @@ public function __construct($uri = null){
 		if($uri !== null){
 
 			if(is_string($uri)){
-				$uri = parse_url($uri);
+				$uri = parseUrl($uri);
 			}
 
 			if(!is_array($uri)){
@@ -238,7 +238,7 @@ protected function filterHost($host):string{
 			$host = '['.$host.']';
 		}
 
-		return strtolower($host);
+		return mb_strtolower($host);
 	}
 
 	/**

src/Psr7/message_helpers.php

@@ -13,8 +13,8 @@
 
 use function array_filter, array_keys, array_map, array_values, count, explode,
 	gzdecode, gzinflate, gzuncompress, implode, is_array, is_numeric, is_scalar, is_string,
-	json_decode, json_encode, rawurldecode, rawurlencode, simplexml_load_string, strtolower, trim,
-	ucfirst;
+	json_decode, json_encode, parse_url, preg_match, preg_replace_callback, rawurldecode, rawurlencode,
+	simplexml_load_string, strtolower, trim, ucfirst, urlencode;
 
 const PSR7_INCLUDES = true;
 
@@ -511,3 +511,37 @@ function uriWithQueryValue(UriInterface $uri, string $key, string $value = null)
 	return $uri->withQuery(implode('&', $result));
 }
 
+/**
+ * UTF-8 aware \parse_url() replacement.
+ *
+ * The internal function produces broken output for non ASCII domain names
+ * (IDN) when used with locales other than "C".
+ *
+ * On the other hand, cURL understands IDN correctly only when UTF-8 locale
+ * is configured ("C.UTF-8", "en_US.UTF-8", etc.).
+ *
+ * @see https://bugs.php.net/bug.php?id=52923
+ * @see https://www.php.net/manual/en/function.parse-url.php#114817
+ * @see https://curl.haxx.se/libcurl/c/CURLOPT_URL.html#ENCODING
+ *
+ * @link https://github.com/guzzle/psr7/blob/c0dcda9f54d145bd4d062a6d15f54931a67732f9/src/Uri.php#L89-L130
+ */
+function parseUrl(string $url):?array{
+	// If IPv6
+	$prefix = '';
+	/** @noinspection RegExpRedundantEscape */
+	if(preg_match('%^(.*://\[[0-9:a-f]+\])(.*?)$%', $url, $matches)){
+		/** @var array{0:string, 1:string, 2:string} $matches */
+		$prefix = $matches[1];
+		$url    = $matches[2];
+	}
+
+	$encodedUrl = preg_replace_callback('%[^:/@?&=#]+%usD', fn($matches) => urlencode($matches[0]), $url);
+	$result     = parse_url($prefix.$encodedUrl);
+
+	if($result === false){
+		return null;
+	}
+
+	return array_map('urldecode', $result);
+}

tests/Psr7/UriTest.php

@@ -2,7 +2,7 @@
 /**
  * Class UriTest
  *
- * @link https://github.com/guzzle/psr7/blob/4b981cdeb8c13d22a6c193554f8c686f53d5c958/tests/UriTest.php
+ * @link https://github.com/guzzle/psr7/blob/c0dcda9f54d145bd4d062a6d15f54931a67732f9/tests/UriTest.php
  * @link https://github.com/bakame-php/psr7-uri-interface-tests/blob/5a556fdfe668a6c6a14772efeba6134c0b7dae34/tests/AbstractUriTestCase.php
  *
  * @created      10.08.2018
@@ -662,4 +662,22 @@ public function testWithPartSamePart():void{
 		$this::assertSame($expected, (string)$uri);
 	}
 
+	public function testInternationalizedDomainName():void{
+		$uri = new Uri('https://яндекс.рф');
+		self::assertSame('яндекс.рф', $uri->getHost());
+
+		$uri = new Uri('https://яндекAс.рф');
+		self::assertSame('яндекaс.рф', $uri->getHost());
+	}
+
+	public function testIPv6Host():void{
+		$uri = new Uri('https://[2a00:f48:1008::212:183:10]');
+		self::assertSame('[2a00:f48:1008::212:183:10]', $uri->getHost());
+
+		$uri = new Uri('http://[2a00:f48:1008::212:183:10]:56?foo=bar');
+		self::assertSame('[2a00:f48:1008::212:183:10]', $uri->getHost());
+		self::assertSame(56, $uri->getPort());
+		self::assertSame('foo=bar', $uri->getQuery());
+	}
+
 }