#19 and #10 - Improvement of QuotedPart parsing

Change DNS validation to default to true instead of false by @brianfreytag
COMMA is not allowed in domain part
2026-09-07 16:47:07 +00:00 · 2014-09-02 00:35:48 +02:00 · 2014-08-22 01:08:16 +02:00 · 2014-08-22 01:05:58 +02:00 · 2014-08-21 23:47:24 +02:00 · 2014-08-20 15:05:30 -04:00
7 changed files with 105 additions and 20 deletions
@@ -84,6 +84,7 @@ class EmailLexer extends AbstractLexer
        if (!$search->lookahead) {
            throw new \UnexpectedValueException($type . ' not found');
        }
+        return true;
    }

    /**
@@ -116,13 +117,13 @@ class EmailLexer extends AbstractLexer
    protected function getCatchablePatterns()
    {
        return array(
-            '[a-zA-Z_]+[4,6]?',
+            '[a-zA-Z_]+[46]?',
            '[0-9]+',
            '\r\n',
            '::',
            '\s+',
-            '[\x1-\x1F]+',
-            '.'
+            '[\x10-\x1F]+',
+            '.',
            );
    }

@@ -149,7 +150,7 @@ class EmailLexer extends AbstractLexer
            return $this->charValue[$value];
        }

-        if (preg_match('/[\x1-\x1F]+/', $value)) {
+        if (preg_match('/[\x10-\x1F]+/', $value)) {
            return self::INVALID;
        }

@@ -86,7 +86,7 @@ class EmailValidator
            return false;
        }

-        $dns = false;
+        $dns = true;
        if ($checkDNS) {
            $dns = $this->checkDNS();
        }
@@ -235,12 +235,18 @@ class DomainPart extends Parser

    protected function checkDomainPartExceptions($prev)
    {
+        if ($this->lexer->token['type'] === EmailLexer::S_COMMA) {
+            throw new \InvalidArgumentException('ERR_COMMA_IN_DOMAIN');
+        }
+
        if ($this->lexer->token['type'] === EmailLexer::S_AT) {
            throw new \InvalidArgumentException('ERR_CONSECUTIVEATS');
        }
+
        if ($this->lexer->token['type'] === EmailLexer::S_OPENQBRACKET && $prev['type'] !== EmailLexer::S_AT) {
            throw new \InvalidArgumentException('ERR_EXPECTING_ATEXT');
        }
+
        if ($this->lexer->token['type'] === EmailLexer::S_HYPHEN && $this->lexer->isNextToken(EmailLexer::S_DOT)) {
            throw new \InvalidArgumentException('ERR_DOMAINHYPHENEND');
        }
@@ -3,7 +3,6 @@
 namespace Egulias\EmailValidator\Parser;

 use Egulias\EmailValidator\EmailLexer;
-use Egulias\EmailValidator\Parser\Parser;
 use Egulias\EmailValidator\EmailValidator;


@@ -11,7 +10,9 @@ class LocalPart extends Parser
 {
    public function parse($localPart)
    {
+        $parseDQuote = true;
        $closingQuote = false;
+
        while ($this->lexer->token['type'] !== EmailLexer::S_AT && $this->lexer->token) {

            if ($this->lexer->token['type'] === EmailLexer::S_DOT && !$this->lexer->getPrevious()) {
@@ -19,6 +20,10 @@ class LocalPart extends Parser
            }

            $closingQuote = $this->checkDQUOTE($closingQuote);
+            if ($closingQuote && $parseDQuote) {
+                $this->parseDoubleQuote();
+                $parseDQuote = false;
+            }

            if ($this->lexer->token['type'] === EmailLexer::S_OPENPARENTHESIS) {
                $this->parseComments();
@@ -26,20 +31,15 @@ class LocalPart extends Parser

            $this->checkConsecutiveDots();

-            if ($this->lexer->token['type'] === EmailLexer::S_DOT && $this->lexer->isNextToken(EmailLexer::S_AT)) {
+            if (
+                $this->lexer->token['type'] === EmailLexer::S_DOT &&
+                $this->lexer->isNextToken(EmailLexer::S_AT)
+            ) {
                throw new \InvalidArgumentException('ERR_DOT_END');
            }

            $this->warnEscaping();
-
-            if ($this->lexer->isNextTokenAny(
-                array(
-                    EmailLexer::INVALID, EmailLexer::S_LOWERTHAN, EmailLexer::S_GREATERTHAN
-                )
-            )
-            ) {
-                throw new \InvalidArgumentException('ERR_EXPECTING_ATEXT');
-            }
+            $this->isInvalidToken($this->lexer->token, $closingQuote);

            if ($this->isFWS()) {
                $this->parseFWS();
@@ -53,4 +53,42 @@ class LocalPart extends Parser
            $this->warnings[] = EmailValidator::RFC5322_LOCAL_TOOLONG;
        }
    }
+
+    protected function parseDoubleQuote()
+    {
+        $special = array (
+            EmailLexer::S_CR => true,
+            EmailLexer::S_HTAB => true,
+            EmailLexer::S_LF => true
+        );
+        $setSpecialsWarning = true;
+
+        $this->lexer->moveNext();
+        while ($this->lexer->token['type'] !== EmailLexer::S_DQUOTE && $this->lexer->token) {
+            if (isset($special[$this->lexer->token['type']]) && $setSpecialsWarning) {
+                $this->warnings[] = EmailValidator::CFWS_FWS;
+                $setSpecialsWarning = false;
+            }
+            $this->lexer->moveNext();
+        }
+    }
+
+
+    protected function isInvalidToken($token, $closingQuote)
+    {
+        $forbidden = array(
+            EmailLexer::S_COMMA,
+            EmailLexer::S_CLOSEBRACKET,
+            EmailLexer::S_OPENBRACKET,
+            EmailLexer::S_GREATERTHAN,
+            EmailLexer::S_LOWERTHAN,
+            EmailLexer::S_COLON,
+            EmailLexer::S_SEMICOLON,
+            EmailLexer::INVALID
+        );
+
+        if (in_array($token['type'], $forbidden) && !$closingQuote) {
+            throw new \InvalidArgumentException('ERR_EXPECTING_ATEXT');
+        }
+    }
 }
@@ -102,6 +102,10 @@ abstract class Parser

    protected function isFWS()
    {
+        if ($this->escaped()) {
+            return false;
+        }
+
        if ($this->lexer->token['type'] === EmailLexer::S_SP ||
            $this->lexer->token['type'] === EmailLexer::S_HTAB ||
            $this->lexer->token['type'] === EmailLexer::S_CR ||
@@ -114,6 +118,21 @@ abstract class Parser
        return false;
    }

+    protected function escaped()
+    {
+        $previous = $this->lexer->getPrevious();
+
+        if ($previous['type'] === EmailLexer::S_BACKSLASH
+            &&
+            ($this->lexer->token['type'] === EmailLexer::S_SP ||
+            $this->lexer->token['type'] === EmailLexer::S_HTAB)
+        ) {
+            return true;
+        }
+
+        return false;
+    }
+
    protected function warnEscaping()
    {
        if ($this->lexer->token['type'] !== EmailLexer::S_BACKSLASH) {
@@ -36,6 +36,14 @@ class EmailLexerTests extends \PHPUnit_Framework_TestCase
        $this->assertEquals(EmailLexer::S_HTAB, $lexer->token['type']);
    }

+    public function testLexerSearchToken()
+    {
+        $lexer = new EmailLexer();
+        $lexer->setInput("foo\tbar");
+        $lexer->moveNext();
+        $this->assertTrue($lexer->find(EmailLexer::S_HTAB));
+    }
+
    public function getTokens()
    {
        return array(
@@ -34,18 +34,22 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
            array('fabien_potencier@example.fr'),
            array('example@localhost'),
            array('fab\'ien@symfony.com'),
+            array('fab\ ien@symfony.com'),
            array('example((example))@fakedfake.co.uk'),
            array('example@faked(fake).co.uk'),
            array('fabien+@symfony.com'),
            array('инфо@письмо.рф'),
-
+            array('"username"@example.com'),
+            array('"user,name"@example.com'),
+            array('"user name"@example.com'),
+            array('"user@name"@example.com'),
        );
    }

    /**
     * @dataProvider getInvalidEmails
     */
-    public function testAInvalidEmails($email)
+    public function testInvalidEmails($email)
    {
        $this->assertFalse($this->validator->isValid($email));
    }
@@ -65,6 +69,9 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
            array('username@ example . com'),
            array('example@(fake).com'),
            array('example@(fake.com'),
+            array('username@example,com'),
+            array('usern,ame@example.com'),
+            array('user[na]me@example.com'),
        );
    }

@@ -104,14 +111,13 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
            array(EmailValidator::ERR_CR_NO_LF, "example@exa\rmple.co.uk"),
            array(EmailValidator::ERR_CR_NO_LF, "example@[\r]"),
            array(EmailValidator::ERR_CR_NO_LF, "exam\rple@example.co.uk"),
-            array(EmailValidator::ERR_CR_NO_LF, "\"\r\"@localhost"),
        );
    }

    /**
     * @dataProvider getInvalidEmailsWithWarnings
     */
-    public function testInvalidEmailsWithWarningsCheck($warnings, $email)
+    public function testValidEmailsWithWarningsCheck($warnings, $email)
    {
        $this->assertTrue($this->validator->isValid($email, true));

@@ -138,6 +144,13 @@ class EmailValidatorTest extends \PHPUnit_Framework_TestCase
                ),
                "\"\t\"@example.co.uk"
            ),
+            array(
+                array(
+                    EmailValidator::RFC5321_QUOTEDSTRING,
+                    EmailValidator::CFWS_FWS,
+                ),
+                "\"\r\"@example.co.uk"
+            ),
            array(
                array(
                    EmailValidator::RFC5321_ADDRESSLITERAL,
Author	SHA1	Message	Date
Eduardo Gulias Davis	39b451bb2b	#19 and #10 - Improvement of QuotedPart parsing	2014-09-02 00:35:48 +02:00
Eduardo Gulias Davis	65afaf3f2b	Change DNS validation to default to true instead of false by @brianfreytag	2014-08-22 01:08:16 +02:00
Eduardo Gulias Davis	82f8514d89	COMMA is not allowed in domain part	2014-08-22 01:05:58 +02:00
Eduardo Gulias Davis	440bc61237	Merge pull request #17 from brianfreytag/remove_dns_requirement Change DNS validation to default to true instead of false	2014-08-21 23:47:24 +02:00
Brian Freytag	a96f0822e0	Change DNS validation to default to true instead of false There shouldn't be a dependency on the checkDNS() method in order to validate strict = true. We should only be concerned about the DNS if the library is specifically told to check DNS, so we default the DNS variable to true and then set it to false if checkDNS is set to true and checkDNS() returns false.	2014-08-20 15:05:30 -04:00