From fd02fa968c895209576baebed2aad543687dea25 Mon Sep 17 00:00:00 2001
From: Kijin Sung <kijin@kijinsung.com>
Date: Sat, 11 Feb 2017 17:48:52 +0900
Subject: [PATCH] Fix #702 Punycode error on attempting to handle long URLs

---
 common/framework/url.php         | 50 +++++++++++++++++++++++++++-----
 tests/unit/framework/URLTest.php |  6 ++++
 2 files changed, 48 insertions(+), 8 deletions(-)

diff --git a/common/framework/url.php b/common/framework/url.php
index f810c0082..f41c5dc09 100644
--- a/common/framework/url.php
+++ b/common/framework/url.php
@@ -183,38 +183,72 @@ class URL
 	/**
 	 * Encode UTF-8 domain into IDNA (punycode)
 	 * 
-	 * @param string $domain
+	 * @param string $url
 	 * @return string
 	 */
-	public static function encodeIdna($domain)
+	public static function encodeIdna($url)
 	{
+		if (preg_match('@[:/#]@', $url))
+		{
+			$domain = parse_url($url, \PHP_URL_HOST);
+			$position = strpos($url, $domain);
+			if ($position === false)
+			{
+				return $url;
+			}
+		}
+		else
+		{
+			$domain = $url;
+			$position = 0;
+		}
+		
 		if (function_exists('idn_to_ascii'))
 		{
-			return idn_to_ascii($domain);
+			$new_domain = idn_to_ascii($domain);
 		}
 		else
 		{
 			$encoder = new \TrueBV\Punycode();
-			return $encoder->encode($domain);
+			$new_domain = $encoder->encode($domain);
 		}
+		
+		return substr_replace($url, $new_domain, $position, strlen($domain));
 	}
 
 	/**
 	 * Convert IDNA (punycode) domain into UTF-8
 	 * 
-	 * @param string $domain
+	 * @param string $url
 	 * @return string
 	 */
-	public static function decodeIdna($domain)
+	public static function decodeIdna($url)
 	{
+		if (preg_match('@[:/#]@', $url))
+		{
+			$domain = parse_url($url, \PHP_URL_HOST);
+			$position = strpos($url, $domain);
+			if ($position === false)
+			{
+				return $url;
+			}
+		}
+		else
+		{
+			$domain = $url;
+			$position = 0;
+		}
+		
 		if (function_exists('idn_to_utf8'))
 		{
-			return idn_to_utf8($domain);
+			$new_domain = idn_to_utf8($domain);
 		}
 		else
 		{
 			$decoder = new \TrueBV\Punycode();
-			return $decoder->decode($domain);
+			$new_domain = $decoder->decode($domain);
 		}
+		
+		return substr_replace($url, $new_domain, $position, strlen($domain));
 	}
 }
diff --git a/tests/unit/framework/URLTest.php b/tests/unit/framework/URLTest.php
index ed60020ee..22c8c1a38 100644
--- a/tests/unit/framework/URLTest.php
+++ b/tests/unit/framework/URLTest.php
@@ -105,10 +105,16 @@ class URLTest extends \Codeception\TestCase\Test
 	public function testEncodeIdna()
 	{
 		$this->assertEquals('xn--9i1bl3b186bf9e.xn--3e0b707e', Rhymix\Framework\URL::encodeIdna('퓨니코드.한국'));
+		$this->assertEquals('http://xn--9i1bl3b186bf9e.xn--3e0b707e', Rhymix\Framework\URL::encodeIdna('http://퓨니코드.한국'));
+		$this->assertEquals('//xn--9i1bl3b186bf9e.xn--3e0b707e#한글부분', Rhymix\Framework\URL::encodeIdna('//퓨니코드.한국#한글부분'));
+		$this->assertEquals('https://xn--9i1bl3b186bf9e.xn--3e0b707e/hello/world/라이믹스.php?i=4', Rhymix\Framework\URL::encodeIdna('https://퓨니코드.한국/hello/world/라이믹스.php?i=4'));
 	}
 	
 	public function testDecodeIdna()
 	{
 		$this->assertEquals('퓨니코드.한국', Rhymix\Framework\URL::decodeIdna('xn--9i1bl3b186bf9e.xn--3e0b707e'));
+		$this->assertEquals('http://퓨니코드.한국', Rhymix\Framework\URL::decodeIdna('http://xn--9i1bl3b186bf9e.xn--3e0b707e'));
+		$this->assertEquals('//퓨니코드.한국#한글부분', Rhymix\Framework\URL::decodeIdna('//xn--9i1bl3b186bf9e.xn--3e0b707e#한글부분'));
+		$this->assertEquals('https://퓨니코드.한국/hello/world/라이믹스.php?i=4', Rhymix\Framework\URL::decodeIdna('https://xn--9i1bl3b186bf9e.xn--3e0b707e/hello/world/라이믹스.php?i=4'));
 	}
 }