Merge pull request #59 from egulias/next-min-ver

Improve UTF8 detection, domain part validation, others

src/Egulias/EmailValidator/EmailLexer.php

@@ -75,8 +75,6 @@ class EmailLexer extends AbstractLexer
         '\0'   => self::C_NUL,
     );
 
-    protected $invalidASCII = array(226 => 1,);
-
     protected $hasInvalidTokens = false;
 
     protected $previous;
@@ -138,7 +136,8 @@ class EmailLexer extends AbstractLexer
     protected function getCatchablePatterns()
     {
         return array(
-            '[a-zA-Z_]+[46]?',
+            '[a-zA-Z_]+[46]?', //ASCII and domain literal
+            '[^\x00-\x7F]',  //UTF-8
             '[0-9]+',
             '\r\n',
             '::',
@@ -179,11 +178,6 @@ class EmailLexer extends AbstractLexer
             return self::INVALID;
         }
 
-        if ($this->isASCIIInvalid($value)) {
-            $this->hasInvalidTokens = true;
-            return self::INVALID;
-        }
-
         return  self::GENERIC;
     }
 
@@ -209,19 +203,6 @@ class EmailLexer extends AbstractLexer
         return false;
     }
 
-    /**
-     * @param $value
-     * @return bool
-     */
-    protected function isASCIIInvalid($value)
-    {
-        if (isset($this->invalidASCII[ord($value)])) {
-            return true;
-        }
-
-        return false;
-    }
-
     /**
      * @param $value
      * @return bool

src/Egulias/EmailValidator/Parser/DomainPart.php

@@ -104,9 +104,6 @@ class DomainPart extends Parser
     {
         $domain = '';
         do {
-            if ($this->lexer->token['type'] === EmailLexer::S_SEMICOLON) {
-                throw new \InvalidArgumentException('ERR_EXPECTING_ATEXT');
-            }
 
             $prev = $this->lexer->getPrevious();
 
@@ -221,9 +218,6 @@ class DomainPart extends Parser
         return $addressLiteral;
     }
 
-    /**
-     * @param string $addressLiteral
-     */
     protected function checkIPV4Tag($addressLiteral)
     {
         $matchesIP  = array();
@@ -249,6 +243,17 @@ class DomainPart extends Parser
 
     protected function checkDomainPartExceptions($prev)
     {
+        $invalidDomainTokens = array(
+            EmailLexer::S_DQUOTE => true,
+            EmailLexer::S_SEMICOLON => true,
+            EmailLexer::S_GREATERTHAN => true,
+            EmailLexer::S_LOWERTHAN => true,
+        );
+
+        if (isset($invalidDomainTokens[$this->lexer->token['type']])) {
+            throw new \InvalidArgumentException('ERR_EXPECTING_ATEXT');
+        }
+
         if ($this->lexer->token['type'] === EmailLexer::S_COMMA) {
             throw new \InvalidArgumentException('ERR_COMMA_IN_DOMAIN');
         }

tests/egulias/Tests/EmailValidator/EmailLexerTest.php

@@ -103,6 +103,17 @@ class EmailLexerTests extends \PHPUnit_Framework_TestCase
         $this->assertEquals(EmailLexer::S_HTAB, $lexer->token['type']);
     }
 
+    public function testLexerForUTF8()
+    {
+        $lexer = new EmailLexer();
+        $lexer->setInput("áÇ@bar.com");
+        $lexer->moveNext();
+        $lexer->moveNext();
+        $this->assertEquals(EmailLexer::GENERIC, $lexer->token['type']);
+        $lexer->moveNext();
+        $this->assertEquals(EmailLexer::GENERIC, $lexer->token['type']);
+    }
+
     public function testLexerSearchToken()
     {
         $lexer = new EmailLexer();
@@ -111,15 +122,6 @@ class EmailLexerTests extends \PHPUnit_Framework_TestCase
         $this->assertTrue($lexer->find(EmailLexer::S_HTAB));
     }
 
-    public function testLexerHasInvalidTokens()
-    {
-        $lexer = new EmailLexer();
-        $lexer->setInput(chr(226));
-        $lexer->moveNext();
-        $lexer->moveNext();
-        $this->assertTrue($lexer->hasInvalidTokens());
-    }
-
     public function getTokens()
     {
         return array(
@@ -152,7 +154,7 @@ class EmailLexerTests extends \PHPUnit_Framework_TestCase
             array('}', EmailLexer::S_CLOSEQBRACKET),
             array('',  EmailLexer::S_EMPTY),
             array(chr(31),  EmailLexer::INVALID),
-            array(chr(226),  EmailLexer::INVALID),
+            array(chr(226),  EmailLexer::GENERIC),
             array(chr(0),  EmailLexer::C_NUL)
         );
     }

tests/egulias/Tests/EmailValidator/EmailValidatorTest.php

@@ -37,6 +37,7 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
     public function getValidEmails()
     {
         return array(
+            array('â@iana.org'),
             array('fabien@symfony.com'),
             array('example@example.co.uk'),
             array('fabien_potencier@example.fr'),
@@ -56,6 +57,12 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
             array('""@iana.org'),
             array('"\""@iana.org'),
             array('müller@möller.de'),
+            array('test@email*'),
+            array('test@email!'),
+            array('test@email&'),
+            array('test@email^'),
+            array('test@email%'),
+            array('test@email$'),
         );
     }
 
@@ -112,6 +119,10 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
             array('test@foo;bar.com'),
             array('test;123@foobar.com'),
             array('test@example..com'),
+            array('email.email@email."'),
+            array('test@email>'),
+            array('test@email<'),
+            array('test@email{'),
         );
     }