Chris@18: <?php
Chris@18: 
Chris@18: namespace Egulias\EmailValidator\Parser;
Chris@18: 
Chris@18: use Egulias\EmailValidator\EmailLexer;
Chris@18: use Egulias\EmailValidator\Exception\CharNotAllowed;
Chris@18: use Egulias\EmailValidator\Exception\CommaInDomain;
Chris@18: use Egulias\EmailValidator\Exception\ConsecutiveAt;
Chris@18: use Egulias\EmailValidator\Exception\CRLFAtTheEnd;
Chris@18: use Egulias\EmailValidator\Exception\CRNoLF;
Chris@18: use Egulias\EmailValidator\Exception\DomainHyphened;
Chris@18: use Egulias\EmailValidator\Exception\DotAtEnd;
Chris@18: use Egulias\EmailValidator\Exception\DotAtStart;
Chris@18: use Egulias\EmailValidator\Exception\ExpectingATEXT;
Chris@18: use Egulias\EmailValidator\Exception\ExpectingDomainLiteralClose;
Chris@18: use Egulias\EmailValidator\Exception\ExpectingDTEXT;
Chris@18: use Egulias\EmailValidator\Exception\NoDomainPart;
Chris@18: use Egulias\EmailValidator\Exception\UnopenedComment;
Chris@18: use Egulias\EmailValidator\Warning\AddressLiteral;
Chris@18: use Egulias\EmailValidator\Warning\CFWSWithFWS;
Chris@18: use Egulias\EmailValidator\Warning\DeprecatedComment;
Chris@18: use Egulias\EmailValidator\Warning\DomainLiteral;
Chris@18: use Egulias\EmailValidator\Warning\DomainTooLong;
Chris@18: use Egulias\EmailValidator\Warning\IPV6BadChar;
Chris@18: use Egulias\EmailValidator\Warning\IPV6ColonEnd;
Chris@18: use Egulias\EmailValidator\Warning\IPV6ColonStart;
Chris@18: use Egulias\EmailValidator\Warning\IPV6Deprecated;
Chris@18: use Egulias\EmailValidator\Warning\IPV6DoubleColon;
Chris@18: use Egulias\EmailValidator\Warning\IPV6GroupCount;
Chris@18: use Egulias\EmailValidator\Warning\IPV6MaxGroups;
Chris@18: use Egulias\EmailValidator\Warning\LabelTooLong;
Chris@18: use Egulias\EmailValidator\Warning\ObsoleteDTEXT;
Chris@18: use Egulias\EmailValidator\Warning\TLD;
Chris@18: 
Chris@18: class DomainPart extends Parser
Chris@18: {
Chris@18:     const DOMAIN_MAX_LENGTH = 254;
Chris@18:     protected $domainPart = '';
Chris@18: 
Chris@18:     public function parse($domainPart)
Chris@18:     {
Chris@18:         $this->lexer->moveNext();
Chris@18: 
Chris@18:         if ($this->lexer->token['type'] === EmailLexer::S_DOT) {
Chris@18:             throw new DotAtStart();
Chris@18:         }
Chris@18: 
Chris@18:         if ($this->lexer->token['type'] === EmailLexer::S_EMPTY) {
Chris@18:             throw new NoDomainPart();
Chris@18:         }
Chris@18:         if ($this->lexer->token['type'] === EmailLexer::S_HYPHEN) {
Chris@18:             throw new DomainHyphened();
Chris@18:         }
Chris@18: 
Chris@18:         if ($this->lexer->token['type'] === EmailLexer::S_OPENPARENTHESIS) {
Chris@18:             $this->warnings[DeprecatedComment::CODE] = new DeprecatedComment();
Chris@18:             $this->parseDomainComments();
Chris@18:         }
Chris@18: 
Chris@18:         $domain = $this->doParseDomainPart();
Chris@18: 
Chris@18:         $prev = $this->lexer->getPrevious();
Chris@18:         $length = strlen($domain);
Chris@18: 
Chris@18:         if ($prev['type'] === EmailLexer::S_DOT) {
Chris@18:             throw new DotAtEnd();
Chris@18:         }
Chris@18:         if ($prev['type'] === EmailLexer::S_HYPHEN) {
Chris@18:             throw new DomainHyphened();
Chris@18:         }
Chris@18:         if ($length > self::DOMAIN_MAX_LENGTH) {
Chris@18:             $this->warnings[DomainTooLong::CODE] = new DomainTooLong();
Chris@18:         }
Chris@18:         if ($prev['type'] === EmailLexer::S_CR) {
Chris@18:             throw new CRLFAtTheEnd();
Chris@18:         }
Chris@18:         $this->domainPart = $domain;
Chris@18:     }
Chris@18: 
Chris@18:     public function getDomainPart()
Chris@18:     {
Chris@18:         return $this->domainPart;
Chris@18:     }
Chris@18: 
Chris@18:     public function checkIPV6Tag($addressLiteral, $maxGroups = 8)
Chris@18:     {
Chris@18:         $prev = $this->lexer->getPrevious();
Chris@18:         if ($prev['type'] === EmailLexer::S_COLON) {
Chris@18:             $this->warnings[IPV6ColonEnd::CODE] = new IPV6ColonEnd();
Chris@18:         }
Chris@18: 
Chris@18:         $IPv6       = substr($addressLiteral, 5);
Chris@18:         //Daniel Marschall's new IPv6 testing strategy
Chris@18:         $matchesIP  = explode(':', $IPv6);
Chris@18:         $groupCount = count($matchesIP);
Chris@18:         $colons     = strpos($IPv6, '::');
Chris@18: 
Chris@18:         if (count(preg_grep('/^[0-9A-Fa-f]{0,4}$/', $matchesIP, PREG_GREP_INVERT)) !== 0) {
Chris@18:             $this->warnings[IPV6BadChar::CODE] = new IPV6BadChar();
Chris@18:         }
Chris@18: 
Chris@18:         if ($colons === false) {
Chris@18:             // We need exactly the right number of groups
Chris@18:             if ($groupCount !== $maxGroups) {
Chris@18:                 $this->warnings[IPV6GroupCount::CODE] = new IPV6GroupCount();
Chris@18:             }
Chris@18:             return;
Chris@18:         }
Chris@18: 
Chris@18:         if ($colons !== strrpos($IPv6, '::')) {
Chris@18:             $this->warnings[IPV6DoubleColon::CODE] = new IPV6DoubleColon();
Chris@18:             return;
Chris@18:         }
Chris@18: 
Chris@18:         if ($colons === 0 || $colons === (strlen($IPv6) - 2)) {
Chris@18:             // RFC 4291 allows :: at the start or end of an address
Chris@18:             //with 7 other groups in addition
Chris@18:             ++$maxGroups;
Chris@18:         }
Chris@18: 
Chris@18:         if ($groupCount > $maxGroups) {
Chris@18:             $this->warnings[IPV6MaxGroups::CODE] = new IPV6MaxGroups();
Chris@18:         } elseif ($groupCount === $maxGroups) {
Chris@18:             $this->warnings[IPV6Deprecated::CODE] = new IPV6Deprecated();
Chris@18:         }
Chris@18:     }
Chris@18: 
Chris@18:     protected function doParseDomainPart()
Chris@18:     {
Chris@18:         $domain = '';
Chris@18:         $openedParenthesis = 0;
Chris@18:         do {
Chris@18:             $prev = $this->lexer->getPrevious();
Chris@18: 
Chris@18:             $this->checkNotAllowedChars($this->lexer->token);
Chris@18: 
Chris@18:             if ($this->lexer->token['type'] === EmailLexer::S_OPENPARENTHESIS) {
Chris@18:                 $this->parseComments();
Chris@18:                 $openedParenthesis += $this->getOpenedParenthesis();
Chris@18:                 $this->lexer->moveNext();
Chris@18:                 $tmpPrev = $this->lexer->getPrevious();
Chris@18:                 if ($tmpPrev['type'] === EmailLexer::S_CLOSEPARENTHESIS) {
Chris@18:                     $openedParenthesis--;
Chris@18:                 }
Chris@18:             }
Chris@18:             if ($this->lexer->token['type'] === EmailLexer::S_CLOSEPARENTHESIS) {
Chris@18:                 if ($openedParenthesis === 0) {
Chris@18:                     throw new UnopenedComment();
Chris@18:                 } else {
Chris@18:                     $openedParenthesis--;
Chris@18:                 }
Chris@18:             }
Chris@18: 
Chris@18:             $this->checkConsecutiveDots();
Chris@18:             $this->checkDomainPartExceptions($prev);
Chris@18: 
Chris@18:             if ($this->hasBrackets()) {
Chris@18:                 $this->parseDomainLiteral();
Chris@18:             }
Chris@18: 
Chris@18:             $this->checkLabelLength($prev);
Chris@18: 
Chris@18:             if ($this->isFWS()) {
Chris@18:                 $this->parseFWS();
Chris@18:             }
Chris@18: 
Chris@18:             $domain .= $this->lexer->token['value'];
Chris@18:             $this->lexer->moveNext();
Chris@18:         } while ($this->lexer->token);
Chris@18: 
Chris@18:         return $domain;
Chris@18:     }
Chris@18: 
Chris@18:     private function checkNotAllowedChars($token)
Chris@18:     {
Chris@18:         $notAllowed = [EmailLexer::S_BACKSLASH => true, EmailLexer::S_SLASH=> true];
Chris@18:         if (isset($notAllowed[$token['type']])) {
Chris@18:             throw new CharNotAllowed();
Chris@18:         }
Chris@18:     }
Chris@18: 
Chris@18:     protected function parseDomainLiteral()
Chris@18:     {
Chris@18:         if ($this->lexer->isNextToken(EmailLexer::S_COLON)) {
Chris@18:             $this->warnings[IPV6ColonStart::CODE] = new IPV6ColonStart();
Chris@18:         }
Chris@18:         if ($this->lexer->isNextToken(EmailLexer::S_IPV6TAG)) {
Chris@18:             $lexer = clone $this->lexer;
Chris@18:             $lexer->moveNext();
Chris@18:             if ($lexer->isNextToken(EmailLexer::S_DOUBLECOLON)) {
Chris@18:                 $this->warnings[IPV6ColonStart::CODE] = new IPV6ColonStart();
Chris@18:             }
Chris@18:         }
Chris@18: 
Chris@18:         return $this->doParseDomainLiteral();
Chris@18:     }
Chris@18: 
Chris@18:     protected function doParseDomainLiteral()
Chris@18:     {
Chris@18:         $IPv6TAG = false;
Chris@18:         $addressLiteral = '';
Chris@18:         do {
Chris@18:             if ($this->lexer->token['type'] === EmailLexer::C_NUL) {
Chris@18:                 throw new ExpectingDTEXT();
Chris@18:             }
Chris@18: 
Chris@18:             if ($this->lexer->token['type'] === EmailLexer::INVALID ||
Chris@18:                 $this->lexer->token['type'] === EmailLexer::C_DEL   ||
Chris@18:                 $this->lexer->token['type'] === EmailLexer::S_LF
Chris@18:             ) {
Chris@18:                 $this->warnings[ObsoleteDTEXT::CODE] = new ObsoleteDTEXT();
Chris@18:             }
Chris@18: 
Chris@18:             if ($this->lexer->isNextTokenAny(array(EmailLexer::S_OPENQBRACKET, EmailLexer::S_OPENBRACKET))) {
Chris@18:                 throw new ExpectingDTEXT();
Chris@18:             }
Chris@18: 
Chris@18:             if ($this->lexer->isNextTokenAny(
Chris@18:                 array(EmailLexer::S_HTAB, EmailLexer::S_SP, $this->lexer->token['type'] === EmailLexer::CRLF)
Chris@18:             )) {
Chris@18:                 $this->warnings[CFWSWithFWS::CODE] = new CFWSWithFWS();
Chris@18:                 $this->parseFWS();
Chris@18:             }
Chris@18: 
Chris@18:             if ($this->lexer->isNextToken(EmailLexer::S_CR)) {
Chris@18:                 throw new CRNoLF();
Chris@18:             }
Chris@18: 
Chris@18:             if ($this->lexer->token['type'] === EmailLexer::S_BACKSLASH) {
Chris@18:                 $this->warnings[ObsoleteDTEXT::CODE] = new ObsoleteDTEXT();
Chris@18:                 $addressLiteral .= $this->lexer->token['value'];
Chris@18:                 $this->lexer->moveNext();
Chris@18:                 $this->validateQuotedPair();
Chris@18:             }
Chris@18:             if ($this->lexer->token['type'] === EmailLexer::S_IPV6TAG) {
Chris@18:                 $IPv6TAG = true;
Chris@18:             }
Chris@18:             if ($this->lexer->token['type'] === EmailLexer::S_CLOSEQBRACKET) {
Chris@18:                 break;
Chris@18:             }
Chris@18: 
Chris@18:             $addressLiteral .= $this->lexer->token['value'];
Chris@18: 
Chris@18:         } while ($this->lexer->moveNext());
Chris@18: 
Chris@18:         $addressLiteral = str_replace('[', '', $addressLiteral);
Chris@18:         $addressLiteral = $this->checkIPV4Tag($addressLiteral);
Chris@18: 
Chris@18:         if (false === $addressLiteral) {
Chris@18:             return $addressLiteral;
Chris@18:         }
Chris@18: 
Chris@18:         if (!$IPv6TAG) {
Chris@18:             $this->warnings[DomainLiteral::CODE] = new DomainLiteral();
Chris@18:             return $addressLiteral;
Chris@18:         }
Chris@18: 
Chris@18:         $this->warnings[AddressLiteral::CODE] = new AddressLiteral();
Chris@18: 
Chris@18:         $this->checkIPV6Tag($addressLiteral);
Chris@18: 
Chris@18:         return $addressLiteral;
Chris@18:     }
Chris@18: 
Chris@18:     protected function checkIPV4Tag($addressLiteral)
Chris@18:     {
Chris@18:         $matchesIP  = array();
Chris@18: 
Chris@18:         // Extract IPv4 part from the end of the address-literal (if there is one)
Chris@18:         if (preg_match(
Chris@18:             '/\\b(?:(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\\.){3}(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)$/',
Chris@18:             $addressLiteral,
Chris@18:             $matchesIP
Chris@18:         ) > 0
Chris@18:         ) {
Chris@18:             $index = strrpos($addressLiteral, $matchesIP[0]);
Chris@18:             if ($index === 0) {
Chris@18:                 $this->warnings[AddressLiteral::CODE] = new AddressLiteral();
Chris@18:                 return false;
Chris@18:             }
Chris@18:             // Convert IPv4 part to IPv6 format for further testing
Chris@18:             $addressLiteral = substr($addressLiteral, 0, $index) . '0:0';
Chris@18:         }
Chris@18: 
Chris@18:         return $addressLiteral;
Chris@18:     }
Chris@18: 
Chris@18:     protected function checkDomainPartExceptions($prev)
Chris@18:     {
Chris@18:         $invalidDomainTokens = array(
Chris@18:             EmailLexer::S_DQUOTE => true,
Chris@18:             EmailLexer::S_SEMICOLON => true,
Chris@18:             EmailLexer::S_GREATERTHAN => true,
Chris@18:             EmailLexer::S_LOWERTHAN => true,
Chris@18:         );
Chris@18: 
Chris@18:         if (isset($invalidDomainTokens[$this->lexer->token['type']])) {
Chris@18:             throw new ExpectingATEXT();
Chris@18:         }
Chris@18: 
Chris@18:         if ($this->lexer->token['type'] === EmailLexer::S_COMMA) {
Chris@18:             throw new CommaInDomain();
Chris@18:         }
Chris@18: 
Chris@18:         if ($this->lexer->token['type'] === EmailLexer::S_AT) {
Chris@18:             throw new ConsecutiveAt();
Chris@18:         }
Chris@18: 
Chris@18:         if ($this->lexer->token['type'] === EmailLexer::S_OPENQBRACKET && $prev['type'] !== EmailLexer::S_AT) {
Chris@18:             throw new ExpectingATEXT();
Chris@18:         }
Chris@18: 
Chris@18:         if ($this->lexer->token['type'] === EmailLexer::S_HYPHEN && $this->lexer->isNextToken(EmailLexer::S_DOT)) {
Chris@18:             throw new DomainHyphened();
Chris@18:         }
Chris@18: 
Chris@18:         if ($this->lexer->token['type'] === EmailLexer::S_BACKSLASH
Chris@18:             && $this->lexer->isNextToken(EmailLexer::GENERIC)) {
Chris@18:             throw new ExpectingATEXT();
Chris@18:         }
Chris@18:     }
Chris@18: 
Chris@18:     protected function hasBrackets()
Chris@18:     {
Chris@18:         if ($this->lexer->token['type'] !== EmailLexer::S_OPENBRACKET) {
Chris@18:             return false;
Chris@18:         }
Chris@18: 
Chris@18:         try {
Chris@18:             $this->lexer->find(EmailLexer::S_CLOSEBRACKET);
Chris@18:         } catch (\RuntimeException $e) {
Chris@18:             throw new ExpectingDomainLiteralClose();
Chris@18:         }
Chris@18: 
Chris@18:         return true;
Chris@18:     }
Chris@18: 
Chris@18:     protected function checkLabelLength($prev)
Chris@18:     {
Chris@18:         if ($this->lexer->token['type'] === EmailLexer::S_DOT &&
Chris@18:             $prev['type'] === EmailLexer::GENERIC &&
Chris@18:             strlen($prev['value']) > 63
Chris@18:         ) {
Chris@18:             $this->warnings[LabelTooLong::CODE] = new LabelTooLong();
Chris@18:         }
Chris@18:     }
Chris@18: 
Chris@18:     protected function parseDomainComments()
Chris@18:     {
Chris@18:         $this->isUnclosedComment();
Chris@18:         while (!$this->lexer->isNextToken(EmailLexer::S_CLOSEPARENTHESIS)) {
Chris@18:             $this->warnEscaping();
Chris@18:             $this->lexer->moveNext();
Chris@18:         }
Chris@18: 
Chris@18:         $this->lexer->moveNext();
Chris@18:         if ($this->lexer->isNextToken(EmailLexer::S_DOT)) {
Chris@18:             throw new ExpectingATEXT();
Chris@18:         }
Chris@18:     }
Chris@18: 
Chris@18:     protected function addTLDWarnings()
Chris@18:     {
Chris@18:         if ($this->warnings[DomainLiteral::CODE]) {
Chris@18:             $this->warnings[TLD::CODE] = new TLD();
Chris@18:         }
Chris@18:     }
Chris@18: }