Improve domain validation

src/EmailLexer.php

@@ -14,6 +14,9 @@ class EmailLexer extends AbstractLexer
     const S_DOT              = 46;
     const S_DQUOTE           = 34;
     const S_SQUOTE           = 39;
+    const S_BACKTICK         = 96;
+    const S_OPENCURLYBRACES  = 123;
+    const S_CLOSECURLYBRACES = 125;
     const S_OPENPARENTHESIS  = 49;
     const S_CLOSEPARENTHESIS = 261;
     const S_OPENBRACKET      = 262;
@@ -21,6 +24,11 @@ class EmailLexer extends AbstractLexer
     const S_HYPHEN           = 264;
     const S_COLON            = 265;
     const S_DOUBLECOLON      = 266;
+    const S_PIPE             = 124;
+    const S_TILDE            = 126;
+    const S_UNDERSCORE       = 95;
+    const S_PLUS             = 43;
+    const S_EQUAL            = 61;
     const S_SP               = 267;
     const S_HTAB             = 268;
     const S_CR               = 269;
@@ -52,6 +60,8 @@ class EmailLexer extends AbstractLexer
      * @var array
      */
     protected $charValue = array(
+        '{'    => self::S_OPENCURLYBRACES,
+        '}'    => self::S_CLOSECURLYBRACES,
         '('    => self::S_OPENPARENTHESIS,
         ')'    => self::S_CLOSEPARENTHESIS,
         '<'    => self::S_LOWERTHAN,
@@ -66,6 +76,7 @@ class EmailLexer extends AbstractLexer
         ','    => self::S_COMMA,
         '.'    => self::S_DOT,
         "'"    => self::S_SQUOTE,
+        "`"    => self::S_BACKTICK,
         '"'    => self::S_DQUOTE,
         '-'    => self::S_HYPHEN,
         '::'   => self::S_DOUBLECOLON,
@@ -75,8 +86,6 @@ class EmailLexer extends AbstractLexer
         "\n"   => self::S_LF,
         "\r\n" => self::CRLF,
         'IPv6' => self::S_IPV6TAG,
-        '{'    => self::S_OPENQBRACKET,
-        '}'    => self::S_CLOSEQBRACKET,
         ''     => self::S_EMPTY,
         '\0'   => self::C_NUL,
         '*'    => self::ASTERISK,
@@ -85,6 +94,11 @@ class EmailLexer extends AbstractLexer
         '^'    => self::CARET,
         '$'    => self::DOLLAR,
         '%'    => self::PERCENTAGE,
+        '~'    => self::S_TILDE,
+        '|'    => self::S_PIPE,
+        '_'    => self::S_UNDERSCORE,
+        '='    => self::S_EQUAL,
+        '+'    => self::S_PLUS,
     );
 
     /**
@@ -216,7 +230,7 @@ class EmailLexer extends AbstractLexer
     protected function getCatchablePatterns()
     {
         return array(
-            '[a-zA-Z_]+[46]?', //ASCII and domain literal
+            '[a-zA-Z]+[46]?', //ASCII and domain literal
             '[^\x00-\x7F]',  //UTF-8
             '[0-9]+',
             '\r\n',

src/Parser/DomainPart.php

@@ -245,14 +245,6 @@ class DomainPart extends Parser
      */
     protected function checkDomainPartExceptions(array $prev, bool $hasComments) : Result
     {
-        $invalidDomainTokens = array(
-            EmailLexer::S_DQUOTE => true,
-            EmailLexer::S_SQUOTE => true,
-            EmailLexer::S_SEMICOLON => true,
-            EmailLexer::S_GREATERTHAN => true,
-            EmailLexer::S_LOWERTHAN => true,
-        );
-
         $validDomainTokens = array(
             EmailLexer::GENERIC => true,
             EmailLexer::S_HYPHEN => true,

tests/EmailValidator/EmailLexerTest.php

@@ -138,6 +138,7 @@ class EmailLexerTests extends TestCase
             array(":", EmailLexer::S_COLON),
             array(".", EmailLexer::S_DOT),
             array("\"", EmailLexer::S_DQUOTE),
+            array("`", EmailLexer::S_BACKTICK),
             array("'", EmailLexer::S_SQUOTE),
             array("-", EmailLexer::S_HYPHEN),
             array("\\", EmailLexer::S_BACKSLASH),
@@ -152,8 +153,13 @@ class EmailLexerTests extends TestCase
             array(',', EmailLexer::S_COMMA),
             array('<', EmailLexer::S_LOWERTHAN),
             array('>', EmailLexer::S_GREATERTHAN),
-            array('{', EmailLexer::S_OPENQBRACKET),
-            array('}', EmailLexer::S_CLOSEQBRACKET),
+            array('{', EmailLexer::S_OPENCURLYBRACES),
+            array('}', EmailLexer::S_CLOSECURLYBRACES),
+            array('|', EmailLexer::S_PIPE),
+            array('~', EmailLexer::S_TILDE),
+            array('=', EmailLexer::S_EQUAL),
+            array('+', EmailLexer::S_PLUS),
+            array('_', EmailLexer::S_UNDERSCORE),
             array('',  EmailLexer::S_EMPTY),
             array(chr(31),  EmailLexer::INVALID),
             array(chr(226),  EmailLexer::GENERIC),

tests/EmailValidator/Validation/RFCValidationDomainPartTest.php

@@ -126,12 +126,20 @@ class RFCValidationDomainPartTest extends TestCase
             ['example@toolonglocalparttoolonglocalparttoolonglocalparttoolonglocalpart.co.uk'],
             ['example@toolonglocalparttoolonglocalparttoolonglocalparttoolonglocalpart.test.co.uk'],
             ['example@test.toolonglocalparttoolonglocalparttoolonglocalparttoolonglocalpart.co.uk'],
-            ['test@email*'],
-            ['test@email!'],
-            ['test@email&'],
-            ['test@email^'],
-            ['test@email%'],
-            ['test@email$'],
+            ['test@email*a.com'],
+            ['test@email!a.com'],
+            ['test@email&a.com'],
+            ['test@email^a.com'],
+            ['test@email%a.com'],
+            ['test@email$a.com'],
+            ['test@email`a.com'],
+            ['test@email|a.com'],
+            ['test@email~a.com'],
+            ['test@email{a.com'],
+            ['test@email}a.com'],
+            ['test@email=a.com'],
+            ['test@email+a.com'],
+            ['test@email_a.com'],
         ];
     }
 

tests/EmailValidator/Validation/RFCValidationTest.php

@@ -64,7 +64,8 @@ class RFCValidationTest extends TestCase
             ['fab\'ien@symfony.com'],
             ['fab\ ien@symfony.com'],
             ['example((example))@fakedfake.co.uk'],
-            ['fabien+@symfony.com'],
+            ['fabien+a@symfony.com'],
+            ['exampl=e@example.com'],
             ['инфо@письмо.рф'],
             ['"username"@example.com'],
             ['"user,name"@example.com'],