Charset Validator

README.md

@@ -204,6 +204,7 @@ Reference
   * v::alnum()
   * v::alpha()
   * v::between()
+  * v::charset()
   * v::consonants()
   * v::contains()
   * v::digits()
@@ -485,6 +486,17 @@ See also:
 
   * v::call() - A more elaborated building block validator
 
+#### v::charset()
+
+Validates if a string is in a specific charset.
+
+
+    v::charset('ASCII', 'açúcar'); //false
+    v::charset('ASCII', 'sugar');  //true
+    v::charset(array('ISO-8859-1', 'EUC-JP'), '日本国'); // true
+
+The array format is a logic OR, not AND.
+
 #### v::countryCode
 
 Validates an ISO country code like US or BR.

library/Respect/Validation/Exceptions/CharsetException.php

@@ -0,0 +1,18 @@
+<?php
+
+namespace Respect\Validation\Exceptions;
+
+class CharsetException extends ValidationException
+{
+
+    public static $defaultTemplates = array(
+        self::MODE_DEFAULT => array(
+            self::STANDARD => '{{name}} must be in the {{charset}} charset'
+        ),
+        self::MODE_NEGATIVE => array(
+            self::STANDARD => '{{name}} must not be in the {{charset}} charset'
+        )
+    );
+
+}
+

library/Respect/Validation/Rules/Charset.php

@@ -0,0 +1,36 @@
+<?php
+
+namespace Respect\Validation\Rules;
+
+use Respect\Validation\Exceptions\ComponentException;
+
+class Charset extends AbstractRule
+{
+
+    public $charset = null;
+
+    public function __construct($charset)
+    {
+        $available = mb_list_encodings();
+        $charset = is_array($charset) ? $charset : array($charset);
+        $charset = array_filter($charset, function ($c) use ($available) {
+            return in_array($c, $available, true);
+        });
+
+        if (!$charset) {
+            throw new ComponentException(
+                'Invalid charset'
+            );
+        }
+        $this->charset = $charset;
+    }
+
+    public function validate($input)
+    {
+        $detectedEncoding = mb_detect_encoding($input, $this->charset, true);
+
+        return in_array($detectedEncoding, $this->charset, true);
+    }
+
+}
+

tests/library/Respect/Validation/Rules/CharsetTest.php

@@ -0,0 +1,73 @@
+<?php
+
+namespace Respect\Validation\Rules;
+
+class CharsetTest extends \PHPUnit_Framework_TestCase
+{
+
+    /**
+     * @dataProvider providerForValidCharset
+     */
+    public function test_valid_data_with_charset_should_return_true($charset, $input)
+    {
+        $validator = new Charset($charset);
+        $this->assertTrue($validator->validate($input));
+    }
+
+    /**
+     * @dataProvider providerForInvalidCharset
+     * @expectedException Respect\Validation\Exceptions\CharsetException
+     */
+    public function test_invalid_charset_should_fail_and_throw_CharsetException($charset, $input)
+    {
+        $validator = new Charset($charset);
+        $this->assertFalse($validator->validate($input));
+        $this->assertFalse($validator->assert($input));
+    }
+
+    /**
+     * @dataProvider providerForInvalidParams
+     * @expectedException Respect\Validation\Exceptions\ComponentException
+     */
+    public function test_invalid_constructor_params_should_throw_ComponentException_upon_instantiation($charset)
+    {
+        $validator = new Charset($charset);
+    }
+
+    public function providerForInvalidParams()
+    {
+        return array(
+            array(new \stdClass),
+            array(array()),
+            array(null),
+            array('16'),
+            array('aeiou'),
+            array('a'),
+            array('Foo'),
+            array('basic'),
+            array(10)
+        );
+    }
+
+    public function providerForValidCharset()
+    {
+        return array(
+            array('ISO-8859-1', mb_convert_encoding('açaí', 'ISO-8859-1')),
+            array(array('UTF-8', 'ASCII'), 'strawberry'),
+            array('ASCII', mb_convert_encoding('strawberry', 'ASCII')),
+            array('UTF-8', '日本国'),
+            array(array('ISO-8859-1', 'EUC-JP'), '日本国'),
+            array('UTF-8', 'açaí'),
+            array('ISO-8859-1', 'açaí'),
+        );
+    }
+
+    public function providerForInvalidCharset()
+    {
+        return array(
+            array('ASCII', '日本国'),
+            array('ASCII', 'açaí')
+        );
+    }
+
+}