Message-id validator (#290)

* scafolding for MessageID validation * Improve domain valid tokens * Improved EmailParser to remove leaked logic. User of lexer recorder within parsers. * MessageIDParser passing tests. * change left for right, which is the right one * psaml errors * Better naming * comments are not allowed in IDLeft for message-id * Suppress psalm inheritance over tokens and dependencies * Update src/Parser.php Co-authored-by: Alexander M. Turek <me@derrabus.de> * Update src/Parser.php Co-authored-by: Alexander M. Turek <me@derrabus.de> * improve parser from comments Co-authored-by: Alexander M. Turek <me@derrabus.de>
egulias · Mar 6, 2021 · 8e526a5 · 8e526a5
1 parent 451b438
commit 8e526a5
Show file tree

Hide file tree

Showing 17 changed files with 557 additions and 240 deletions.
diff --git a/composer.lock b/composer.lock
diff --git a/src/EmailLexer.php b/src/EmailLexer.php
@@ -124,6 +124,7 @@ class EmailLexer extends AbstractLexer
      *
      * @var array
      *
+     * @psalm-suppress NonInvariantDocblockPropertyType
      * @psalm-var array{value:string, type:null|int, position:int}
      * @psalm-suppress NonInvariantDocblockPropertyType
      */

diff --git a/src/EmailParser.php b/src/EmailParser.php
@@ -9,18 +9,12 @@
 use Egulias\EmailValidator\Result\ValidEmail;
 use Egulias\EmailValidator\Result\InvalidEmail;
 use Egulias\EmailValidator\Warning\EmailTooLong;
-use Egulias\EmailValidator\Result\Reason\ExpectingATEXT;
 use Egulias\EmailValidator\Result\Reason\NoLocalPart;
 
-class EmailParser
+class EmailParser extends Parser
 {
     const EMAIL_MAX_LENGTH = 254;
 
-    /**
-     * @var array
-     */
-    protected $warnings = [];
-
     /**
      * @var string
      */
@@ -30,81 +24,54 @@ class EmailParser
      * @var string
      */
     protected $localPart = '';
-    /**
-     * @var EmailLexer
-     */
-    protected $lexer;
-
-    public function __construct(EmailLexer $lexer)
-    {
-        $this->lexer = $lexer;
-    }
 
     public function parse(string $str) : Result
     {
-        $this->lexer->setInput($str);
+        $result = parent::parse($str);
 
+        $this->addLongEmailWarning($this->localPart, $this->domainPart);
+
+        return $result;
+    }
+
+    protected function preLeftParsing(): Result
+    {
         if (!$this->hasAtToken()) {
             return new InvalidEmail(new NoLocalPart(), $this->lexer->token["value"]);
         }
+        return new ValidEmail();
+    }
 
-        $localPartResult = $this->processLocalPart();
-
-        if ($localPartResult->isInvalid()) {
-            return $localPartResult;
-        }
-
-        $domainPartResult = $this->processDomainPart();
-
-        if ($domainPartResult->isInvalid()) {
-            return $domainPartResult;
-        }
-
-        if ($this->lexer->hasInvalidTokens()) {
-            return new InvalidEmail(new ExpectingATEXT("Invalid tokens found"), $this->lexer->token["value"]);
-        }
-
-        $this->addLongEmailWarning($this->localPart, $this->domainPart);
+    protected function parseLeftFromAt(): Result
+    {
+        return $this->processLocalPart();
+    }
 
-        return new ValidEmail();
+    protected function parseRightFromAt(): Result
+    {
+        return $this->processDomainPart();
     }
 
     private function processLocalPart() : Result
     {
-        $this->lexer->startRecording();
         $localPartParser = new LocalPart($this->lexer);
         $localPartResult = $localPartParser->parse();
-        $this->lexer->stopRecording();
-        $this->localPart = rtrim($this->lexer->getAccumulatedValues(), '@');
+        $this->localPart = $localPartParser->localPart();
         $this->warnings = array_merge($localPartParser->getWarnings(), $this->warnings);
 
         return $localPartResult;
     }
 
     private function processDomainPart() : Result
     {
-        $this->lexer->clearRecorded();
-        $this->lexer->startRecording();
         $domainPartParser = new DomainPart($this->lexer);
         $domainPartResult = $domainPartParser->parse();
-        $this->lexer->stopRecording();
-        $this->domainPart = $this->lexer->getAccumulatedValues();
+        $this->domainPart = $domainPartParser->domainPart();
         $this->warnings = array_merge($domainPartParser->getWarnings(), $this->warnings);
 
         return $domainPartResult;
     }
 
-    /**
-     * @return Warning\Warning[]
-     */
-    public function getWarnings() : array
-    {
-        return $this->warnings;
-    }
-
-    /**
-     * @return string
-     */
     public function getDomainPart() : string
     {
         return $this->domainPart;
@@ -115,25 +82,7 @@ public function getLocalPart() : string
         return $this->localPart;
     }
 
-    /**
-     * @return bool
-     */
-    protected function hasAtToken() : bool
-    {
-        $this->lexer->moveNext();
-        $this->lexer->moveNext();
-        if ($this->lexer->token['type'] === EmailLexer::S_AT) {
-            return false;
-        }
-
-        return true;
-    }
-
-    /**
-     * @param string $localPart
-     * @param string $parsedDomainPart
-     */
-    protected function addLongEmailWarning($localPart, $parsedDomainPart) : void
+    private function addLongEmailWarning(string $localPart, string $parsedDomainPart) : void
     {
         if (strlen($localPart . '@' . $parsedDomainPart) > self::EMAIL_MAX_LENGTH) {
             $this->warnings[EmailTooLong::CODE] = new EmailTooLong();

diff --git a/src/MessageIDParser.php b/src/MessageIDParser.php
@@ -0,0 +1,93 @@
+<?php
+
+namespace Egulias\EmailValidator;
+
+use Egulias\EmailValidator\Parser;
+use Egulias\EmailValidator\EmailLexer;
+use Egulias\EmailValidator\Result\Result;
+use Egulias\EmailValidator\Parser\IDLeftPart;
+use Egulias\EmailValidator\Parser\IDRightPart;
+use Egulias\EmailValidator\Result\ValidEmail;
+use Egulias\EmailValidator\Result\InvalidEmail;
+use Egulias\EmailValidator\Warning\EmailTooLong;
+use Egulias\EmailValidator\Result\Reason\NoLocalPart;
+
+class MessageIDParser extends Parser
+{
+
+    const EMAILID_MAX_LENGTH = 254;
+
+    /**
+     * @var string
+     */
+    protected $idLeft = '';
+
+    /**
+     * @var string
+     */
+    protected $idRight = '';
+
+    public function parse(string $str) : Result
+    {
+        $result = parent::parse($str);
+
+        $this->addLongEmailWarning($this->idLeft, $this->idRight);
+
+        return $result;
+    }
+
+    protected function preLeftParsing(): Result
+    {
+        if (!$this->hasAtToken()) {
+            return new InvalidEmail(new NoLocalPart(), $this->lexer->token["value"]);
+        }
+        return new ValidEmail();
+    }
+
+    protected function parseLeftFromAt(): Result
+    {
+        return $this->processIDLeft();
+    }
+
+    protected function parseRightFromAt(): Result
+    {
+        return $this->processIDRight();
+    }
+
+    private function processIDLeft() : Result
+    {
+        $localPartParser = new IDLeftPart($this->lexer);
+        $localPartResult = $localPartParser->parse();
+        $this->idLeft = $localPartParser->localPart();
+        $this->warnings = array_merge($localPartParser->getWarnings(), $this->warnings);
+
+        return $localPartResult;
+    }
+
+    private function processIDRight() : Result
+    {
+        $domainPartParser = new IDRightPart($this->lexer);
+        $domainPartResult = $domainPartParser->parse();
+        $this->idRight = $domainPartParser->domainPart();
+        $this->warnings = array_merge($domainPartParser->getWarnings(), $this->warnings);
+
+        return $domainPartResult;
+    }
+
+    public function getLeftPart() : string
+    {
+        return $this->idLeft;
+    }
+
+    public function getRightPart() : string
+    {
+        return $this->idRight;
+    }
+
+    private function addLongEmailWarning(string $localPart, string $parsedDomainPart) : void
+    {
+        if (strlen($localPart . '@' . $parsedDomainPart) > self::EMAILID_MAX_LENGTH) {
+            $this->warnings[EmailTooLong::CODE] = new EmailTooLong();
+        }
+    }
+}
diff --git a/src/Parser.php b/src/Parser.php
@@ -0,0 +1,78 @@
+<?php
+
+namespace Egulias\EmailValidator;
+
+use Egulias\EmailValidator\Result\Result;
+use Egulias\EmailValidator\Result\ValidEmail;
+use Egulias\EmailValidator\Result\InvalidEmail;
+use Egulias\EmailValidator\Result\Reason\ExpectingATEXT;
+
+abstract class Parser
+{
+    /**
+     * @var Warning\Warning[]
+     */
+    protected $warnings = [];
+
+    /**
+     * @var EmailLexer
+     */
+    protected $lexer;
+
+    /**
+     * id-left "@" id-right
+     */
+    abstract protected function parseRightFromAt() : Result;
+    abstract protected function parseLeftFromAt() : Result;
+    abstract protected function preLeftParsing() : Result;
+
+
+    public function __construct(EmailLexer $lexer)
+    {
+        $this->lexer = $lexer;   
+    }
+
+    public function parse(string $str) : Result
+    {
+        $this->lexer->setInput($str);
+
+        if ($this->lexer->hasInvalidTokens()) {
+            return new InvalidEmail(new ExpectingATEXT("Invalid tokens found"), $this->lexer->token["value"]);
+        }
+
+        $preParsingResult = $this->preLeftParsing();
+        if ($preParsingResult->isInvalid()) {
+            return $preParsingResult;
+        }
+
+        $localPartResult = $this->parseLeftFromAt();
+
+        if ($localPartResult->isInvalid()) {
+            return $localPartResult;
+        }
+
+        $domainPartResult = $this->parseRightFromAt(); 
+
+        if ($domainPartResult->isInvalid()) {
+            return $domainPartResult;
+        }
+
+        return new ValidEmail();
+    }
+
+    /**
+     * @return Warning\Warning[]
+     */
+    public function getWarnings() : array
+    {
+        return $this->warnings;
+    }
+
+    protected function hasAtToken() : bool
+    {
+        $this->lexer->moveNext();
+        $this->lexer->moveNext();
+
+        return $this->lexer->token['type'] !== EmailLexer::S_AT;
+    }
+}
diff --git a/src/Parser/Comment.php b/src/Parser/Comment.php
@@ -11,7 +11,7 @@
 use Egulias\EmailValidator\Result\Reason\UnOpenedComment;
 use Egulias\EmailValidator\Warning\Comment as WarningComment;
 
-class Comment extends Parser
+class Comment extends PartParser
 {
     /**
      * @var int

diff --git a/src/Parser/DomainLiteral.php b/src/Parser/DomainLiteral.php
@@ -20,7 +20,7 @@
 use Egulias\EmailValidator\Result\Reason\UnusualElements;
 use Egulias\EmailValidator\Warning\DomainLiteral as WarningDomainLiteral;
 
-class DomainLiteral extends Parser
+class DomainLiteral extends PartParser
 {
     public function parse() : Result
     {