CI_Utf8-related changes

 - Give priority to mb_convert_encoding() over iconv() in clean_string() (partially fixes #261)
 - Add more proper unit tests

2 files changed, 70 insertions, 2 deletions

diff --git a/tests/codeigniter/core/Input_test.php b/tests/codeigniter/core/Input_test.php
index 95833fc91..21ff6d81f 100644
--- a/tests/codeigniter/core/Input_test.php
+++ b/tests/codeigniter/core/Input_test.php
@@ -13,6 +13,8 @@ class Input_test extends CI_TestCase {
 		$this->ci_set_config('csrf_protection', FALSE);
 
 		$security = new Mock_Core_Security();
+
+		$this->ci_set_config('charset', 'UTF-8');
 		$utf8 = new Mock_Core_Utf8();
 
 		$this->input = new Mock_Core_Input($security, $utf8);
diff --git a/tests/codeigniter/core/Utf8_test.php b/tests/codeigniter/core/Utf8_test.php
index 2cf404841..7e6ffd930 100644
--- a/tests/codeigniter/core/Utf8_test.php
+++ b/tests/codeigniter/core/Utf8_test.php
@@ -4,22 +4,88 @@ class Utf8_test extends CI_TestCase {
 
 	public function set_up()
 	{
+		$this->ci_set_config('charset', 'UTF-8');
 		$this->utf8 = new Mock_Core_Utf8();
+		$this->ci_instance_var('utf8', $this->utf8);
 	}
 
 	// --------------------------------------------------------------------
 
-	public function test_convert_to_utf8()
+	/**
+	 * __construct() test
+	 *
+	 * @covers	CI_Utf8::__construct
+	 */
+	public function test___construct()
 	{
-		$this->assertEquals('ั‚ะตัั‚', $this->utf8->convert_to_utf8('๒ๅ๑๒', 'WINDOWS-1251'));
+		if (defined('PREG_BAD_UTF8_ERROR') && (ICONV_ENABLED === TRUE OR MB_ENABLED === TRUE) && strtoupper(config_item('charset')) === 'UTF-8')
+		{
+			$this->assertTrue(UTF8_ENABLED);
+		}
+		else
+		{
+			$this->assertFalse(UTF8_ENABLED);
+		}
 	}
 
 	// --------------------------------------------------------------------
 
+	/**
+	 * is_ascii() test
+	 *
+	 * Note: DO NOT move this below test_clean_string()
+	 */
 	public function test_is_ascii()
 	{
 		$this->assertTrue($this->utf8->is_ascii('foo bar'));
 		$this->assertFalse($this->utf8->is_ascii('ั‚ะตัั‚'));
 	}
 
+	// --------------------------------------------------------------------
+
+	/**
+	 * clean_string() test
+	 *
+	 * @depends	test_is_ascii
+	 * @covers	CI_Utf8::clean_string
+	 */
+	public function test_clean_string()
+	{
+		$this->assertEquals('foo bar', $this->utf8->clean_string('foo bar'));
+
+		$illegal_utf8 = "\xc0ั‚ะตัั‚";
+		if (MB_ENABLED)
+		{
+			$this->assertEquals('ั‚ะตัั‚', $this->utf8->clean_string($illegal_utf8));
+		}
+		elseif (ICONV_ENABLED)
+		{
+			// This is a known issue, iconv doesn't always work with //IGNORE
+			$this->assertTrue(in_array($this->utf8->clean_string($illegal_utf8), array('ั‚ะตัั‚', ''), TRUE));
+		}
+		else
+		{
+			$this->assertEquals($illegal_utf8, $this->utf8->clean_string($illegal_utf8));
+		}
+	}
+
+	// --------------------------------------------------------------------
+
+	/**
+	 * convert_to_utf8() test
+	 *
+	 * @covers	CI_Utf8::convert_to_utf8
+	 */
+	public function test_convert_to_utf8()
+	{
+		if (MB_ENABLED OR ICONV_ENABLED)
+		{
+			$this->assertEquals('ั‚ะตัั‚', $this->utf8->convert_to_utf8('๒ๅ๑๒', 'WINDOWS-1251'));
+		}
+		else
+		{
+			$this->assertFalse($this->utf8->convert_to_utf8('๒ๅ๑๒', 'WINDOWS-1251'));
+		}
+	}
+
 }
\ No newline at end of file