Url: optimized unescape() performance

Author: JanTvrdik

src/Http/Url.php

@@ -451,7 +451,7 @@ public function __toString()
 
 
 	/**
-	 * Similar to rawurldecode, but preserve reserved chars encoded.
+	 * Similar to rawurldecode, but preserves reserved chars encoded.
 	 * @param  string to decode
 	 * @param  string reserved characters
 	 * @return string
@@ -461,14 +461,15 @@ public static function unescape($s, $reserved = '%;/?:@&=+$,')
 		// reserved (@see RFC 2396) = ";" | "/" | "?" | ":" | "@" | "&" | "=" | "+" | "$" | ","
 		// within a path segment, the characters "/", ";", "=", "?" are reserved
 		// within a query component, the characters ";", "/", "?", ":", "@", "&", "=", "+", ",", "$" are reserved.
-		preg_match_all('#(?<=%)[a-f0-9][a-f0-9]#i', $s, $matches, PREG_OFFSET_CAPTURE | PREG_SET_ORDER);
-		foreach (array_reverse($matches) as $match) {
-			$ch = chr(hexdec($match[0][0]));
-			if (strpos($reserved, $ch) === FALSE) {
-				$s = substr_replace($s, $ch, $match[0][1] - 1, 3);
-			}
+		if ($reserved === '') {
+			return rawurldecode($s);
+		}
+		$pattern = '#((?:%(?:' . implode('|', str_split(bin2hex($reserved), 2)) . '))++)#i';
+		$parts = preg_split($pattern, $s, -1, PREG_SPLIT_DELIM_CAPTURE);
+		for ($i = 0; $i < count($parts); $i += 2) {
+			$parts[$i] = rawurldecode($parts[$i]);
 		}
-		return $s;
+		return implode('', $parts);
 	}
 
 }

tests/Http/Url.unescape.phpt

@@ -0,0 +1,21 @@
+<?php
+
+/**
+ * Test: Nette\Http\Url unescape.
+ */
+
+use Nette\Http\Url,
+	Tester\Assert;
+
+
+require __DIR__ . '/../bootstrap.php';
+
+
+Assert::same( 'foo + bar', Url::unescape('foo + bar') );
+Assert::same( 'foo + bar', Url::unescape('foo + bar', '') );
+Assert::same( 'foo', Url::unescape('%66%6F%6F', '') );
+Assert::same( 'f%6F%6F', Url::unescape('%66%6F%6F', 'o') );
+Assert::same( '%66oo', Url::unescape('%66%6F%6F', 'f') );
+Assert::same( '%66%6F%6F', Url::unescape('%66%6F%6F', 'fo') );
+Assert::same( '%66%6f%6f', Url::unescape('%66%6f%6f', 'fo') );
+Assert::same( "%00\x01%02", Url::unescape('%00%01%02', "\x00\x02") );