From 538ebe4d15e96d3e468b6a219af371d0ddc8480b Mon Sep 17 00:00:00 2001
From: michalsn <michal@sniatala.pl>
Date: Fri, 26 Jun 2020 16:02:40 +0200
Subject: [PATCH 1/2] Support unicode for regular expressions in router. Fix
 #3169

---
 system/Router/RouteCollection.php  |  2 +-
 system/Router/Router.php           | 11 +++++++----
 tests/system/Router/RouterTest.php | 21 +++++++++++++++++++++
 3 files changed, 29 insertions(+), 5 deletions(-)

diff --git a/system/Router/RouteCollection.php b/system/Router/RouteCollection.php
index c89c0717a5..e952824327 100644
--- a/system/Router/RouteCollection.php
+++ b/system/Router/RouteCollection.php
@@ -1348,7 +1348,7 @@ class RouteCollection implements RouteCollectionInterface
 			// the expected param type.
 			$pos = strpos($from, $pattern);
 
-			if (preg_match('#' . $pattern . '#', $params[$index]))
+			if (preg_match('#^' . $pattern . '$#u', $params[$index]))
 			{
 				$from = substr_replace($from, $params[$index], $pos, strlen($pattern));
 			}
diff --git a/system/Router/Router.php b/system/Router/Router.php
index 888928a494..91c399edfa 100644
--- a/system/Router/Router.php
+++ b/system/Router/Router.php
@@ -171,6 +171,9 @@ class Router implements RouterInterface
 				: $this->controller;
 		}
 
+		// Decode URL-encoded string
+		$uri = urldecode($uri);
+
 		if ($this->checkRoutes($uri))
 		{
 			if ($this->collection->isFiltered($this->matchedRoute[0]))
@@ -420,12 +423,12 @@ class Router implements RouterInterface
 
 				// Replace it with a regex so it
 				// will actually match.
-                $key = str_replace('/', '\/', $key);
+				$key = str_replace('/', '\/', $key);
 				$key = str_replace('{locale}', '[^\/]+', $key);
 			}
 
 			// Does the RegEx match?
-			if (preg_match('#^' . $key . '$#', $uri, $matches))
+			if (preg_match('#^' . $key . '$#u', $uri, $matches))
 			{
 				// Is this route supposed to redirect to another?
 				if ($this->collection->isRedirect($key))
@@ -470,12 +473,12 @@ class Router implements RouterInterface
 				if (strpos($val, '$') !== false && strpos($key, '(') !== false && strpos($key, '/') !== false)
 				{
 					$replacekey = str_replace('/(.*)', '', $key);
-					$val        = preg_replace('#^' . $key . '$#', $val, $uri);
+					$val        = preg_replace('#^' . $key . '$#u', $val, $uri);
 					$val        = str_replace($replacekey, str_replace('/', '\\', $replacekey), $val);
 				}
 				elseif (strpos($val, '$') !== false && strpos($key, '(') !== false)
 				{
-					$val = preg_replace('#^' . $key . '$#', $val, $uri);
+					$val = preg_replace('#^' . $key . '$#u', $val, $uri);
 				}
 				elseif (strpos($val, '/') !== false)
 				{
diff --git a/tests/system/Router/RouterTest.php b/tests/system/Router/RouterTest.php
index 44459876c3..6786154d68 100644
--- a/tests/system/Router/RouterTest.php
+++ b/tests/system/Router/RouterTest.php
@@ -531,4 +531,25 @@ class RouterTest extends \CodeIgniter\Test\CIUnitTestCase
 		$this->assertEquals('Home', $router->controllerName());
 		$this->assertEquals('index', $router->methodName());
 	}
+
+	//--------------------------------------------------------------------
+
+	/**
+	 * @see https://github.com/codeigniter4/CodeIgniter4/issues/3169
+	 */
+	public function testRegularExpressionWithUnicode()
+	{
+		$this->collection->get('news/([a-z0-9\x{0980}-\x{09ff}-]+)', 'News::view/$1');
+
+		$router = new Router($this->collection, $this->request);
+
+		$router->handle('news/a0%E0%A6%80%E0%A7%BF-');
+		$this->assertEquals('\News', $router->controllerName());
+		$this->assertEquals('view', $router->methodName());
+
+		$expected = [
+			'a0ঀ৿-',
+		];
+		$this->assertEquals($expected, $router->params());
+	}
 }

From 6203e425c6d5d3c9926ed94ededae4ede5838d17 Mon Sep 17 00:00:00 2001
From: michalsn <michal@sniatala.pl>
Date: Sun, 28 Jun 2020 08:30:18 +0200
Subject: [PATCH 2/2] Test regex unicode with placeholder

---
 tests/system/Router/RouterTest.php | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/tests/system/Router/RouterTest.php b/tests/system/Router/RouterTest.php
index 6786154d68..9ccf2ae493 100644
--- a/tests/system/Router/RouterTest.php
+++ b/tests/system/Router/RouterTest.php
@@ -552,4 +552,24 @@ class RouterTest extends \CodeIgniter\Test\CIUnitTestCase
 		];
 		$this->assertEquals($expected, $router->params());
 	}
+
+	/**
+	 * @see https://github.com/codeigniter4/CodeIgniter4/issues/3169
+	 */
+	public function testRegularExpressionPlaceholderWithUnicode()
+	{
+		$this->collection->addPlaceholder('custom', '[a-z0-9\x{0980}-\x{09ff}-]+');
+		$this->collection->get('news/(:custom)', 'News::view/$1');
+
+		$router = new Router($this->collection, $this->request);
+
+		$router->handle('news/a0%E0%A6%80%E0%A7%BF-');
+		$this->assertEquals('\News', $router->controllerName());
+		$this->assertEquals('view', $router->methodName());
+
+		$expected = [
+			'a0ঀ৿-',
+		];
+		$this->assertEquals($expected, $router->params());
+	}
 }