Fix labels length check (#278)

* Fix label length check

* Test label too long on single word hostname

.gitignore

@@ -1,3 +1,4 @@
+.phpunit.result.cache
 .idea
 .vscode
 composer.lock

.travis.yml

@@ -1,5 +1,4 @@
 language: php
-
 matrix:
   include:
     - php: 5.5.9
@@ -28,6 +27,29 @@ matrix:
       env:
         - psalm=yes
       dist: bionic
+#ppc64le support code
+    - php: 5.6
+      arch: ppc64le
+      env:
+      dist: xenial
+    - php: 7.0
+      arch: ppc64le
+      dist: xenial
+    - php: 7.1
+      arch: ppc64le
+      env:
+        - psalm=yes
+      dist: bionic
+    - php: 7.2
+      arch: ppc64le
+      env:
+        - psalm=yes
+      dist: bionic
+    - php: 7.4
+      arch: ppc64le
+      env:
+        - psalm=yes
+      dist: bionic
 
 install:
   - if [ "$deps" = "low" ]; then composer update --prefer-lowest; else composer install; fi

src/Parser/DomainPart.php

@@ -35,6 +35,7 @@ use Egulias\EmailValidator\Warning\TLD;
 class DomainPart extends Parser
 {
     const DOMAIN_MAX_LENGTH = 254;
+    const LABEL_MAX_LENGTH = 63;
 
     /**
      * @var string
@@ -160,6 +161,7 @@ class DomainPart extends Parser
     protected function doParseDomainPart()
     {
         $domain = '';
+        $label = '';
         $openedParenthesis = 0;
         do {
             $prev = $this->lexer->getPrevious();
@@ -190,7 +192,12 @@ class DomainPart extends Parser
                 $this->parseDomainLiteral();
             }
 
-            $this->checkLabelLength($prev);
+            if ($this->lexer->token['type'] === EmailLexer::S_DOT) {
+                $this->checkLabelLength($label);
+                $label = '';
+            } else {
+                $label .= $this->lexer->token['value'];
+            }
 
             if ($this->isFWS()) {
                 $this->parseFWS();
@@ -203,6 +210,8 @@ class DomainPart extends Parser
             }
         } while (null !== $this->lexer->token['type']);
 
+        $this->checkLabelLength($label);
+
         return $domain;
     }
 
@@ -386,16 +395,31 @@ class DomainPart extends Parser
         return true;
     }
 
-    protected function checkLabelLength(array $prev)
+    /**
+     * @param string $label
+     */
+    protected function checkLabelLength($label)
     {
-        if ($this->lexer->token['type'] === EmailLexer::S_DOT &&
+        if ($this->isLabelTooLong($label)) {
-            $prev['type'] === EmailLexer::GENERIC &&
-            strlen($prev['value']) > 63
-        ) {
             $this->warnings[LabelTooLong::CODE] = new LabelTooLong();
         }
     }
 
+    /**
+     * @param string $label
+     * @return bool
+     */
+    private function isLabelTooLong($label)
+    {
+        if (preg_match('/[^\x00-\x7F]/', $label)) {
+            idn_to_ascii($label, IDNA_DEFAULT, INTL_IDNA_VARIANT_UTS46, $idnaInfo);
+
+            return (bool) ($idnaInfo['errors'] & IDNA_ERROR_LABEL_TOO_LONG);
+        }
+
+        return strlen($label) > self::LABEL_MAX_LENGTH;
+    }
+
     protected function parseDomainComments()
     {
         $this->isUnclosedComment();

src/Warning/Warning.php

@@ -29,7 +29,7 @@ abstract class Warning
      */
     public function code()
     {
-        return self::CODE;
+        return static::CODE;
     }
 
     /**

tests/EmailValidator/Validation/NoRFCWarningsValidationTest.php

@@ -26,11 +26,22 @@ class NoRFCWarningsValidationTest extends TestCase
         $this->assertInstanceOf(RFCWarnings::class, $validation->getError());
     }
 
-    public function testEmailWithoutWarningsIsValid()
+    /**
+     * @dataProvider getValidEmailsWithoutWarnings
+     */
+    public function testEmailWithoutWarningsIsValid($email)
     {
         $validation = new NoRFCWarningsValidation();
 
-        $this->assertTrue($validation->isValid('example@example.com', new EmailLexer()));
+        $this->assertTrue($validation->isValid($email, new EmailLexer()));
         $this->assertNull($validation->getError());
     }
+
+    public function getValidEmailsWithoutWarnings()
+    {
+        return [
+            ['example@example.com',],
+            [sprintf('example@%s.com', str_repeat('ъ', 40)),],
+        ];
+    }
 }

tests/EmailValidator/Validation/RFCValidationTest.php

@@ -24,6 +24,7 @@ use Egulias\EmailValidator\Warning\CFWSWithFWS;
 use Egulias\EmailValidator\Warning\Comment;
 use Egulias\EmailValidator\Warning\DomainLiteral;
 use Egulias\EmailValidator\Warning\DomainTooLong;
+use Egulias\EmailValidator\Warning\EmailTooLong;
 use Egulias\EmailValidator\Warning\IPV6BadChar;
 use Egulias\EmailValidator\Warning\IPV6ColonEnd;
 use Egulias\EmailValidator\Warning\IPV6ColonStart;
@@ -226,12 +227,12 @@ class RFCValidationTest extends TestCase
         $this->assertTrue($this->validator->isValid($email, $this->lexer));
         $warnings = $this->validator->getWarnings();
         $this->assertCount(
-            count($warnings), $expectedWarnings,
+            count($expectedWarnings), $warnings,
             "Expected: " . implode(",", $expectedWarnings) . " and got " . implode(",", $warnings)
         );
 
         foreach ($warnings as $warning) {
-            $this->assertArrayHasKey($warning->code(), $expectedWarnings);
+            $this->assertContains($warning->code(), $expectedWarnings);
         }
     }
 
@@ -278,17 +279,29 @@ class RFCValidationTest extends TestCase
                 'example@toolonglocalparttoolonglocalparttoolonglocalparttoolonglocalpart.co.uk'
             ],
             [
-                [DomainTooLong::CODE, LabelTooLong::CODE,],
+                [DomainTooLong::CODE, LabelTooLong::CODE, EmailTooLong::CODE],
                 'example2@toolonglocalparttoolonglocalparttoolonglocalparttoolonglocalparttoolonglocalparttoolonglocal'.
                 'parttoolonglocalparttoolonglocalparttoolonglocalparttoolonglocalparttoolonglocalparttoolonglocalpart'.
                 'toolonglocalparttoolonglocalparttoolonglocalparttoolonglocalpart'
             ],
             [
-                [DomainTooLong::CODE, LabelTooLong::CODE,],
+                [DomainTooLong::CODE, LabelTooLong::CODE, EmailTooLong::CODE],
                 'example@toolonglocalparttoolonglocalparttoolonglocalparttoolonglocalparttoolonglocalparttoolonglocal'.
                 'parttoolonglocalparttoolonglocalparttoolonglocalparttoolonglocalparttoolonglocalparttoolonglocalpart'.
                 'toolonglocalparttoolonglocalparttoolonglocalparttoolonglocalpar'
             ],
+            [
+                [LabelTooLong::CODE,],
+                sprintf('example@%s.com', str_repeat('ъ', 60)),
+            ],
+            [
+                [LabelTooLong::CODE,],
+                sprintf('example@%s.com', str_repeat('a4t', 22)),
+            ],
+            [
+                [LabelTooLong::CODE,],
+                sprintf('example@%s', str_repeat('a4t', 22)),
+            ],
         ];
     }
 }