Chris@0: <?php
Chris@0: 
Chris@0: namespace Egulias\EmailValidator;
Chris@0: 
Chris@0: use Doctrine\Common\Lexer\AbstractLexer;
Chris@0: 
Chris@0: class EmailLexer extends AbstractLexer
Chris@0: {
Chris@0:     //ASCII values
Chris@0:     const C_DEL              = 127;
Chris@0:     const C_NUL              = 0;
Chris@0:     const S_AT               = 64;
Chris@0:     const S_BACKSLASH        = 92;
Chris@0:     const S_DOT              = 46;
Chris@0:     const S_DQUOTE           = 34;
Chris@0:     const S_OPENPARENTHESIS  = 49;
Chris@0:     const S_CLOSEPARENTHESIS = 261;
Chris@0:     const S_OPENBRACKET      = 262;
Chris@0:     const S_CLOSEBRACKET     = 263;
Chris@0:     const S_HYPHEN           = 264;
Chris@0:     const S_COLON            = 265;
Chris@0:     const S_DOUBLECOLON      = 266;
Chris@0:     const S_SP               = 267;
Chris@0:     const S_HTAB             = 268;
Chris@0:     const S_CR               = 269;
Chris@0:     const S_LF               = 270;
Chris@0:     const S_IPV6TAG          = 271;
Chris@0:     const S_LOWERTHAN        = 272;
Chris@0:     const S_GREATERTHAN      = 273;
Chris@0:     const S_COMMA            = 274;
Chris@0:     const S_SEMICOLON        = 275;
Chris@0:     const S_OPENQBRACKET     = 276;
Chris@0:     const S_CLOSEQBRACKET    = 277;
Chris@0:     const S_SLASH            = 278;
Chris@0:     const S_EMPTY            = null;
Chris@0:     const GENERIC            = 300;
Chris@0:     const CRLF               = 301;
Chris@0:     const INVALID            = 302;
Chris@0:     const ASCII_INVALID_FROM = 127;
Chris@0:     const ASCII_INVALID_TO   = 199;
Chris@0: 
Chris@0:     /**
Chris@0:      * US-ASCII visible characters not valid for atext (@link http://tools.ietf.org/html/rfc5322#section-3.2.3)
Chris@0:      *
Chris@0:      * @var array
Chris@0:      */
Chris@0:     protected $charValue = array(
Chris@0:         '('    => self::S_OPENPARENTHESIS,
Chris@0:         ')'    => self::S_CLOSEPARENTHESIS,
Chris@0:         '<'    => self::S_LOWERTHAN,
Chris@0:         '>'    => self::S_GREATERTHAN,
Chris@0:         '['    => self::S_OPENBRACKET,
Chris@0:         ']'    => self::S_CLOSEBRACKET,
Chris@0:         ':'    => self::S_COLON,
Chris@0:         ';'    => self::S_SEMICOLON,
Chris@0:         '@'    => self::S_AT,
Chris@0:         '\\'   => self::S_BACKSLASH,
Chris@0:         '/'    => self::S_SLASH,
Chris@0:         ','    => self::S_COMMA,
Chris@0:         '.'    => self::S_DOT,
Chris@0:         '"'    => self::S_DQUOTE,
Chris@0:         '-'    => self::S_HYPHEN,
Chris@0:         '::'   => self::S_DOUBLECOLON,
Chris@0:         ' '    => self::S_SP,
Chris@0:         "\t"   => self::S_HTAB,
Chris@0:         "\r"   => self::S_CR,
Chris@0:         "\n"   => self::S_LF,
Chris@0:         "\r\n" => self::CRLF,
Chris@0:         'IPv6' => self::S_IPV6TAG,
Chris@0:         '{'    => self::S_OPENQBRACKET,
Chris@0:         '}'    => self::S_CLOSEQBRACKET,
Chris@0:         ''     => self::S_EMPTY,
Chris@0:         '\0'   => self::C_NUL,
Chris@0:     );
Chris@0: 
Chris@0:     protected $hasInvalidTokens = false;
Chris@0: 
Chris@0:     protected $previous;
Chris@0: 
Chris@0:     public function reset()
Chris@0:     {
Chris@0:         $this->hasInvalidTokens = false;
Chris@0:         parent::reset();
Chris@0:     }
Chris@0: 
Chris@0:     public function hasInvalidTokens()
Chris@0:     {
Chris@0:         return $this->hasInvalidTokens;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param $type
Chris@0:      * @throws \UnexpectedValueException
Chris@0:      * @return boolean
Chris@0:      */
Chris@0:     public function find($type)
Chris@0:     {
Chris@0:         $search = clone $this;
Chris@0:         $search->skipUntil($type);
Chris@0: 
Chris@0:         if (!$search->lookahead) {
Chris@0:             throw new \UnexpectedValueException($type . ' not found');
Chris@0:         }
Chris@0:         return true;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * getPrevious
Chris@0:      *
Chris@0:      * @return array token
Chris@0:      */
Chris@0:     public function getPrevious()
Chris@0:     {
Chris@0:         return $this->previous;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * moveNext
Chris@0:      *
Chris@0:      * @return boolean
Chris@0:      */
Chris@0:     public function moveNext()
Chris@0:     {
Chris@0:         $this->previous = $this->token;
Chris@0: 
Chris@0:         return parent::moveNext();
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Lexical catchable patterns.
Chris@0:      *
Chris@0:      * @return string[]
Chris@0:      */
Chris@0:     protected function getCatchablePatterns()
Chris@0:     {
Chris@0:         return array(
Chris@0:             '[a-zA-Z_]+[46]?', //ASCII and domain literal
Chris@0:             '[^\x00-\x7F]',  //UTF-8
Chris@0:             '[0-9]+',
Chris@0:             '\r\n',
Chris@0:             '::',
Chris@0:             '\s+?',
Chris@0:             '.',
Chris@0:             );
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Lexical non-catchable patterns.
Chris@0:      *
Chris@0:      * @return string[]
Chris@0:      */
Chris@0:     protected function getNonCatchablePatterns()
Chris@0:     {
Chris@0:         return array('[\xA0-\xff]+');
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Retrieve token type. Also processes the token value if necessary.
Chris@0:      *
Chris@0:      * @param string $value
Chris@0:      * @throws \InvalidArgumentException
Chris@0:      * @return integer
Chris@0:      */
Chris@0:     protected function getType(&$value)
Chris@0:     {
Chris@0:         if ($this->isNullType($value)) {
Chris@0:             return self::C_NUL;
Chris@0:         }
Chris@0: 
Chris@0:         if ($this->isValid($value)) {
Chris@0:             return $this->charValue[$value];
Chris@0:         }
Chris@0: 
Chris@0:         if ($this->isUTF8Invalid($value)) {
Chris@0:             $this->hasInvalidTokens = true;
Chris@0:             return self::INVALID;
Chris@0:         }
Chris@0: 
Chris@0:         return  self::GENERIC;
Chris@0:     }
Chris@0: 
Chris@0:     protected function isValid($value)
Chris@0:     {
Chris@0:         if (isset($this->charValue[$value])) {
Chris@0:             return true;
Chris@0:         }
Chris@0: 
Chris@0:         return false;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param $value
Chris@0:      * @return bool
Chris@0:      */
Chris@0:     protected function isNullType($value)
Chris@0:     {
Chris@0:         if ($value === "\0") {
Chris@0:             return true;
Chris@0:         }
Chris@0: 
Chris@0:         return false;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * @param $value
Chris@0:      * @return bool
Chris@0:      */
Chris@0:     protected function isUTF8Invalid($value)
Chris@0:     {
Chris@0:         if (preg_match('/\p{Cc}+/u', $value)) {
Chris@0:             return true;
Chris@0:         }
Chris@0: 
Chris@0:         return false;
Chris@0:     }
Chris@0: 
Chris@0:     protected function getModifiers()
Chris@0:     {
Chris@0:         return 'iu';
Chris@0:     }
Chris@0: }