opened parenthesis converted to class field

Remove duplicated tokens

.travis.yml

@@ -1,3 +1,5 @@
+sudo: false
+
 language: php
 
 php:
@@ -5,11 +7,25 @@ php:
   - 5.4
   - 5.5
   - 5.6
+  - 7.0
   - hhvm
 
-before_script:
+env:
- - wget http://getcomposer.org/composer.phar
+  global:
- - php composer.phar install --dev --no-interaction
+    - deps=no
+
+matrix:
+  fast_finish: true
+  include:
+    - php: 5.3
+      env: deps=low
+    - php: 5.6
+      env: deps=high
+
+install:
+  - if [ "$deps" = "no" ]; then composer install; fi
+  - if [ "$deps" = "low" ]; then composer update --prefer-lowest; fi
+  - if [ "$deps" = "high" ]; then composer update; fi
 
 script:
  - mkdir -p build/logs

README.md

@@ -1,12 +1,15 @@
-EmailValidator [![Build Status](https://travis-ci.org/egulias/EmailValidator.png?branch=master)](https://travis-ci.org/egulias/EmailValidator) [![Coverage Status](https://coveralls.io/repos/egulias/EmailValidator/badge.png?branch=master)](https://coveralls.io/r/egulias/EmailValidator?branch=master) [![SensioLabsInsight](https://insight.sensiolabs.com/projects/22ba6692-9c02-42e5-a65d-1c5696bfffc6/small.png)](https://insight.sensiolabs.com/projects/22ba6692-9c02-42e5-a65d-1c5696bfffc6)[![Scrutinizer Code Quality](https://scrutinizer-ci.com/g/egulias/EmailValidator/badges/quality-score.png?b=master)](https://scrutinizer-ci.com/g/egulias/EmailValidator/?branch=master)
+#EmailValidator
+[![Build Status](https://travis-ci.org/egulias/EmailValidator.png?branch=master)](https://travis-ci.org/egulias/EmailValidator) [![Coverage Status](https://coveralls.io/repos/egulias/EmailValidator/badge.png?branch=master)](https://coveralls.io/r/egulias/EmailValidator?branch=master) [![Scrutinizer Code Quality](https://scrutinizer-ci.com/g/egulias/EmailValidator/badges/quality-score.png?b=master)](https://scrutinizer-ci.com/g/egulias/EmailValidator/?branch=master) [![SensioLabsInsight](https://insight.sensiolabs.com/projects/22ba6692-9c02-42e5-a65d-1c5696bfffc6/small.png)](https://insight.sensiolabs.com/projects/22ba6692-9c02-42e5-a65d-1c5696bfffc6)
 =============================
+With the help of
 
+![Powered by PhpStorm](https://www.jetbrains.com/phpstorm/documentation/docs/logo_phpstorm.png)
 ##Installation##
 
 Run the command below to install via Composer
 
 ```shell
-composer require egulias/email-validator
+composer require egulias/email-validator "~1.2"
 ```
 
 ##Usage##

composer.json

@@ -3,22 +3,23 @@
   "description":  "A library for validating emails",
   "homepage":     "https://github.com/egulias/EmailValidator",
   "type":         "Library",
-  "keywords":     ["email", "validation", "validator"],
+  "keywords":     ["email", "validation", "validator", "emailvalidation", "emailvalidator"],
   "license":      "MIT",
   "authors": [
     {"name": "Eduardo Gulias Davis"}
   ],
   "extra": {
     "branch-alias": {
-      "dev-master": "1.3.x-dev"
+      "dev-master": "2.0.x-dev"
     }
   },
   "require":      {
     "php": ">= 5.3.3",
-    "doctrine/lexer": "dev-master"
+    "doctrine/lexer": "~1.0,>=1.0.1"
   },
   "require-dev" :   {
-    "satooshi/php-coveralls": "dev-master"
+    "satooshi/php-coveralls": "dev-master",
+    "phpunit/phpunit": "~4.4"
   },
   "autoload": {
     "psr-0": {

src/Egulias/EmailValidator/EmailLexer.php

@@ -67,16 +67,12 @@ class EmailLexer extends AbstractLexer
         "\n"   => self::S_LF,
         "\r\n" => self::CRLF,
         'IPv6' => self::S_IPV6TAG,
-        '<'    => self::S_LOWERTHAN,
-        '>'    => self::S_GREATERTHAN,
         '{'    => self::S_OPENQBRACKET,
         '}'    => self::S_CLOSEQBRACKET,
         ''     => self::S_EMPTY,
         '\0'   => self::C_NUL,
     );
 
-    protected $invalidASCII = array(226 => 1,);
-
     protected $hasInvalidTokens = false;
 
     protected $previous;
@@ -138,7 +134,8 @@ 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 +176,6 @@ class EmailLexer extends AbstractLexer
             return self::INVALID;
         }
 
-        if ($this->isASCIIInvalid($value)) {
-            $this->hasInvalidTokens = true;
-            return self::INVALID;
-        }
-
         return  self::GENERIC;
     }
 
@@ -209,19 +201,6 @@ 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/EmailValidator.php

@@ -33,6 +33,7 @@ class EmailValidator
     const ERR_FWS_CRLF_END       = 149;
     const ERR_CR_NO_LF           = 150;
     const ERR_DEPREC_REACHED     = 151;
+    const ERR_UNOPENEDCOMMENT    = 152;
     const RFC5321_TLD             = 9;
     const RFC5321_TLDNUMERIC      = 10;
     const RFC5321_QUOTEDSTRING    = 11;

src/Egulias/EmailValidator/Parser/DomainPart.php

@@ -4,7 +4,6 @@
 namespace Egulias\EmailValidator\Parser;
 
 use Egulias\EmailValidator\EmailLexer;
-use Egulias\EmailValidator\Parser\Parser;
 use Egulias\EmailValidator\EmailValidator;
 
 class DomainPart extends Parser
@@ -103,6 +102,7 @@ class DomainPart extends Parser
     protected function doParseDomainPart()
     {
         $domain = '';
+        $openedParenthesis = 0;
         do {
             $prev = $this->lexer->getPrevious();
 
@@ -112,7 +112,19 @@ class DomainPart extends Parser
 
             if ($this->lexer->token['type'] === EmailLexer::S_OPENPARENTHESIS) {
                 $this->parseComments();
+                $openedParenthesis += $this->getOpenedParenthesis();
                 $this->lexer->moveNext();
+                $tmpPrev = $this->lexer->getPrevious();
+                if ($tmpPrev['type'] === EmailLexer::S_CLOSEPARENTHESIS) {
+                    $openedParenthesis--;
+                }
+            }
+            if ($this->lexer->token['type'] === EmailLexer::S_CLOSEPARENTHESIS) {
+                if ($openedParenthesis === 0) {
+                    throw new \InvalidArgumentException('ERR_UNOPENEDCOMMENT');
+                } else {
+                    $openedParenthesis--;
+                }
             }
 
             $this->checkConsecutiveDots();
@@ -179,7 +191,7 @@ class DomainPart extends Parser
             }
 
             if ($this->lexer->isNextToken(EmailLexer::S_CR)) {
-                throw new \InvalidArgumentException("ERR_CR_NO_LF");
+                throw new \InvalidArgumentException('ERR_CR_NO_LF');
             }
             if ($this->lexer->token['type'] === EmailLexer::S_BACKSLASH) {
                 $this->warnings[] = EmailValidator::RFC5322_DOMLIT_OBSDTEXT;
@@ -217,9 +229,6 @@ class DomainPart extends Parser
         return $addressLiteral;
     }
 
-    /**
-     * @param string $addressLiteral
-     */
     protected function checkIPV4Tag($addressLiteral)
     {
         $matchesIP  = array();
@@ -245,6 +254,17 @@ 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');
         }

src/Egulias/EmailValidator/Parser/LocalPart.php

@@ -4,7 +4,6 @@ namespace Egulias\EmailValidator\Parser;
 
 use Egulias\EmailValidator\EmailLexer;
 use Egulias\EmailValidator\EmailValidator;
-use \InvalidArgumentException;
 
 class LocalPart extends Parser
 {
@@ -12,9 +11,9 @@ class LocalPart extends Parser
     {
         $parseDQuote = true;
         $closingQuote = false;
+        $openedParenthesis = 0;
 
         while ($this->lexer->token['type'] !== EmailLexer::S_AT && $this->lexer->token) {
-
             if ($this->lexer->token['type'] === EmailLexer::S_DOT && !$this->lexer->getPrevious()) {
                 throw new \InvalidArgumentException('ERR_DOT_START');
             }
@@ -26,12 +25,19 @@ class LocalPart extends Parser
 
             if ($this->lexer->token['type'] === EmailLexer::S_OPENPARENTHESIS) {
                 $this->parseComments();
+                $openedParenthesis += $this->getOpenedParenthesis();
+            }
+            if ($this->lexer->token['type'] === EmailLexer::S_CLOSEPARENTHESIS) {
+                if ($openedParenthesis === 0) {
+                    throw new \InvalidArgumentException('ERR_UNOPENEDCOMMENT');
+                } else {
+                    $openedParenthesis--;
+                }
             }
 
             $this->checkConsecutiveDots();
 
-            if (
+            if ($this->lexer->token['type'] === EmailLexer::S_DOT &&
-                $this->lexer->token['type'] === EmailLexer::S_DOT &&
                 $this->lexer->isNextToken(EmailLexer::S_AT)
             ) {
                 throw new \InvalidArgumentException('ERR_DOT_END');
@@ -82,7 +88,7 @@ class LocalPart extends Parser
             $this->lexer->moveNext();
 
             if (!$this->escaped() && isset($invalid[$this->lexer->token['type']])) {
-                throw new InvalidArgumentException("ERR_EXPECTED_ATEXT");
+                throw new \InvalidArgumentException('ERR_EXPECTED_ATEXT');
             }
         }
 
@@ -90,12 +96,12 @@ class LocalPart extends Parser
 
         if ($prev['type'] === EmailLexer::S_BACKSLASH) {
             if (!$this->checkDQUOTE(false)) {
-                throw new \InvalidArgumentException("ERR_UNCLOSED_DQUOTE");
+                throw new \InvalidArgumentException('ERR_UNCLOSED_DQUOTE');
             }
         }
 
         if (!$this->lexer->isNextToken(EmailLexer::S_AT) && $prev['type'] !== EmailLexer::S_BACKSLASH) {
-            throw new \InvalidArgumentException("ERR_EXPECED_AT");
+            throw new \InvalidArgumentException('ERR_EXPECED_AT');
         }
 
         return $parseAgain;

src/Egulias/EmailValidator/Parser/Parser.php

@@ -9,6 +9,7 @@ abstract class Parser
 {
     protected $warnings = array();
     protected $lexer;
+    protected $openedParenthesis = 0;
 
     public function __construct(EmailLexer $lexer)
     {
@@ -20,7 +21,13 @@ abstract class Parser
         return $this->warnings;
     }
 
-    abstract function parse($str);
+    abstract public function parse($str);
+
+    /** @return int */
+    public function getOpenedParenthesis()
+    {
+        return $this->openedParenthesis;
+    }
 
     /**
      * validateQuotedPair
@@ -35,15 +42,15 @@ abstract class Parser
         $this->warnings[] = EmailValidator::DEPREC_QP;
     }
 
-    /**
-     * @return string the the comment
-     * @throws \InvalidArgumentException
-     */
     protected function parseComments()
     {
+        $this->openedParenthesis = 1;
         $this->isUnclosedComment();
         $this->warnings[] = EmailValidator::CFWS_COMMENT;
         while (!$this->lexer->isNextToken(EmailLexer::S_CLOSEPARENTHESIS)) {
+            if ($this->lexer->isNextToken(EmailLexer::S_OPENPARENTHESIS)) {
+                $this->openedParenthesis++;
+            }
             $this->warnEscaping();
             $this->lexer->moveNext();
         }
@@ -75,11 +82,11 @@ abstract class Parser
         $this->checkCRLFInFWS();
 
         if ($this->lexer->token['type'] === EmailLexer::S_CR) {
-            throw new \InvalidArgumentException("ERR_CR_NO_LF");
+            throw new \InvalidArgumentException('ERR_CR_NO_LF');
         }
 
         if ($this->lexer->isNextToken(EmailLexer::GENERIC) && $previous['type']  !== EmailLexer::S_AT) {
-            throw new \InvalidArgumentException("ERR_ATEXT_AFTER_CFWS");
+            throw new \InvalidArgumentException('ERR_ATEXT_AFTER_CFWS');
         }
 
         if ($this->lexer->token['type'] === EmailLexer::S_LF || $this->lexer->token['type'] === EmailLexer::C_NUL) {
@@ -160,7 +167,7 @@ abstract class Parser
             return $hasClosingQuote;
         }
         $previous = $this->lexer->getPrevious();
-        if ($this->lexer->isNextToken(EmailLexer::GENERIC) && $previous['type'] === EmailLexer::GENERIC) {
+        if ($previous['type'] === EmailLexer::GENERIC && $this->lexer->isNextToken(EmailLexer::GENERIC)) {
             throw new \InvalidArgumentException('ERR_EXPECTING_ATEXT');
         }
 
@@ -181,10 +188,10 @@ abstract class Parser
             return;
         }
         if ($this->lexer->isNextToken(EmailLexer::CRLF)) {
-            throw new \InvalidArgumentException("ERR_FWS_CRLF_X2");
+            throw new \InvalidArgumentException('ERR_FWS_CRLF_X2');
         }
         if (!$this->lexer->isNextTokenAny(array(EmailLexer::S_SP, EmailLexer::S_HTAB))) {
-            throw new \InvalidArgumentException("ERR_FWS_CRLF_END");
+            throw new \InvalidArgumentException('ERR_FWS_CRLF_END');
         }
     }
 }

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'),
@@ -55,6 +56,13 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
             array('"test\ test"@iana.org'),
             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$'),
         );
     }
 
@@ -69,7 +77,7 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
     public function getInvalidEmails()
     {
         return array(
-
+            array('test@example.com test'),
             array('user  name@example.com'),
             array('user   name@example.com'),
             array('example.@example.co.uk'),
@@ -108,6 +116,13 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
             array('test@iana.org \r\n\r\n'),
             array('test@iana.org  \r\n\r\n '),
             array('test@iana/icann.org'),
+            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{'),
         );
     }
 
@@ -137,6 +152,11 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
             array(EmailValidator::ERR_DOT_END, 'example@localhost.'),
             array(EmailValidator::ERR_DOT_END, 'example.@example.co.uk'),
             array(EmailValidator::ERR_UNCLOSEDCOMMENT, '(example@localhost'),
+            array(EmailValidator::ERR_UNOPENEDCOMMENT, 'comment)example@localhost'),
+            array(EmailValidator::ERR_UNOPENEDCOMMENT, 'example(comment))@localhost'),
+            array(EmailValidator::ERR_UNOPENEDCOMMENT, 'example@comment)localhost'),
+            array(EmailValidator::ERR_UNOPENEDCOMMENT, 'example@localhost(comment))'),
+            array(EmailValidator::ERR_UNOPENEDCOMMENT, 'example@(comment))example.com'),
             array(EmailValidator::ERR_UNCLOSEDQUOTEDSTR, '"example@localhost'),
             array(EmailValidator::ERR_EXPECTING_ATEXT, 'exa"mple@localhost'),
             //This was the original. But atext is not allowed after \n
@@ -173,13 +193,32 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
     public function getInvalidEmailsWithWarnings()
     {
         return array(
-            array(array( EmailValidator::DEPREC_CFWS_NEAR_AT,), 'example @example.co.uk'),
+            array(
-            array(array( EmailValidator::DEPREC_CFWS_NEAR_AT,), 'example@ example.co.uk'),
+                array(
-            array(array( EmailValidator::CFWS_COMMENT,), 'example@example(examplecomment).co.uk'),
+                    EmailValidator::DEPREC_CFWS_NEAR_AT,
+                    EmailValidator::DNSWARN_NO_RECORD
+                ),
+                'example @example.co.uk'
+            ),
+            array(
+                array(
+                    EmailValidator::DEPREC_CFWS_NEAR_AT,
+                    EmailValidator::DNSWARN_NO_RECORD
+                ),
+                'example@ example.co.uk'
+            ),
+            array(
+                array(
+                    EmailValidator::CFWS_COMMENT,
+                    EmailValidator::DNSWARN_NO_RECORD
+                ),
+                'example@example(examplecomment).co.uk'
+            ),
             array(
                 array(
                     EmailValidator::CFWS_COMMENT,
                     EmailValidator::DEPREC_CFWS_NEAR_AT,
+                    EmailValidator::DNSWARN_NO_RECORD,
                 ),
                 'example(examplecomment)@example.co.uk'
             ),
@@ -187,6 +226,7 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
                 array(
                     EmailValidator::RFC5321_QUOTEDSTRING,
                     EmailValidator::CFWS_FWS,
+                    EmailValidator::DNSWARN_NO_RECORD,
                 ),
                 "\"\t\"@example.co.uk"
             ),
@@ -194,6 +234,7 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
                 array(
                     EmailValidator::RFC5321_QUOTEDSTRING,
                     EmailValidator::CFWS_FWS,
+                    EmailValidator::DNSWARN_NO_RECORD
                 ),
                 "\"\r\"@example.co.uk"
             ),
@@ -285,12 +326,14 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
             array(
                 array(
                     EmailValidator::RFC5321_QUOTEDSTRING,
+                    EmailValidator::DNSWARN_NO_RECORD
                 ),
                 '"example"@example.co.uk'
             ),
             array(
                 array(
                     EmailValidator::RFC5322_LOCAL_TOOLONG,
+                    EmailValidator::DNSWARN_NO_RECORD
                 ),
                 'too_long_localpart_too_long_localpart_too_long_localpart_too_long_localpart@example.co.uk'
             ),