Chris@0: <?php
Chris@0: namespace Masterminds\HTML5\Parser;
Chris@0: 
Chris@0: /**
Chris@0:  * The scanner.
Chris@0:  *
Chris@0:  * This scans over an input stream.
Chris@0:  */
Chris@0: class Scanner
Chris@0: {
Chris@0: 
Chris@0:     const CHARS_HEX = 'abcdefABCDEF01234567890';
Chris@0: 
Chris@0:     const CHARS_ALNUM = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ01234567890';
Chris@0: 
Chris@0:     const CHARS_ALPHA = 'abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ';
Chris@0: 
Chris@0:     protected $is;
Chris@0: 
Chris@0:     // Flipping this to true will give minisculely more debugging info.
Chris@0:     public $debug = false;
Chris@0: 
Chris@0:     /**
Chris@0:      * Create a new Scanner.
Chris@0:      *
Chris@0:      * @param \Masterminds\HTML5\Parser\InputStream $input
Chris@0:      *            An InputStream to be scanned.
Chris@0:      */
Chris@0:     public function __construct($input)
Chris@0:     {
Chris@0:         $this->is = $input;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Get the current position.
Chris@0:      *
Chris@0:      * @return int The current intiger byte position.
Chris@0:      */
Chris@0:     public function position()
Chris@0:     {
Chris@0:         return $this->is->key();
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Take a peek at the next character in the data.
Chris@0:      *
Chris@0:      * @return string The next character.
Chris@0:      */
Chris@0:     public function peek()
Chris@0:     {
Chris@0:         return $this->is->peek();
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Get the next character.
Chris@0:      *
Chris@0:      * Note: This advances the pointer.
Chris@0:      *
Chris@0:      * @return string The next character.
Chris@0:      */
Chris@0:     public function next()
Chris@0:     {
Chris@0:         $this->is->next();
Chris@0:         if ($this->is->valid()) {
Chris@0:             if ($this->debug)
Chris@0:                 fprintf(STDOUT, "> %s\n", $this->is->current());
Chris@0:             return $this->is->current();
Chris@0:         }
Chris@0: 
Chris@0:         return false;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Get the current character.
Chris@0:      *
Chris@0:      * Note, this does not advance the pointer.
Chris@0:      *
Chris@0:      * @return string The current character.
Chris@0:      */
Chris@0:     public function current()
Chris@0:     {
Chris@0:         if ($this->is->valid()) {
Chris@0:             return $this->is->current();
Chris@0:         }
Chris@0: 
Chris@0:         return false;
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Silently consume N chars.
Chris@0:      */
Chris@0:     public function consume($count = 1)
Chris@0:     {
Chris@0:         for ($i = 0; $i < $count; ++ $i) {
Chris@0:             $this->next();
Chris@0:         }
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Unconsume some of the data.
Chris@0:      * This moves the data pointer backwards.
Chris@0:      *
Chris@0:      * @param int $howMany
Chris@0:      *            The number of characters to move the pointer back.
Chris@0:      */
Chris@0:     public function unconsume($howMany = 1)
Chris@0:     {
Chris@0:         $this->is->unconsume($howMany);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Get the next group of that contains hex characters.
Chris@0:      *
Chris@0:      * Note, along with getting the characters the pointer in the data will be
Chris@0:      * moved as well.
Chris@0:      *
Chris@0:      * @return string The next group that is hex characters.
Chris@0:      */
Chris@0:     public function getHex()
Chris@0:     {
Chris@0:         return $this->is->charsWhile(static::CHARS_HEX);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Get the next group of characters that are ASCII Alpha characters.
Chris@0:      *
Chris@0:      * Note, along with getting the characters the pointer in the data will be
Chris@0:      * moved as well.
Chris@0:      *
Chris@0:      * @return string The next group of ASCII alpha characters.
Chris@0:      */
Chris@0:     public function getAsciiAlpha()
Chris@0:     {
Chris@0:         return $this->is->charsWhile(static::CHARS_ALPHA);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Get the next group of characters that are ASCII Alpha characters and numbers.
Chris@0:      *
Chris@0:      * Note, along with getting the characters the pointer in the data will be
Chris@0:      * moved as well.
Chris@0:      *
Chris@0:      * @return string The next group of ASCII alpha characters and numbers.
Chris@0:      */
Chris@0:     public function getAsciiAlphaNum()
Chris@0:     {
Chris@0:         return $this->is->charsWhile(static::CHARS_ALNUM);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Get the next group of numbers.
Chris@0:      *
Chris@0:      * Note, along with getting the characters the pointer in the data will be
Chris@0:      * moved as well.
Chris@0:      *
Chris@0:      * @return string The next group of numbers.
Chris@0:      */
Chris@0:     public function getNumeric()
Chris@0:     {
Chris@0:         return $this->is->charsWhile('0123456789');
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Consume whitespace.
Chris@0:      *
Chris@0:      * Whitespace in HTML5 is: formfeed, tab, newline, space.
Chris@0:      */
Chris@0:     public function whitespace()
Chris@0:     {
Chris@0:         return $this->is->charsWhile("\n\t\f ");
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the current line that is being consumed.
Chris@0:      *
Chris@0:      * @return int The current line number.
Chris@0:      */
Chris@0:     public function currentLine()
Chris@0:     {
Chris@0:         return $this->is->currentLine();
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Read chars until something in the mask is encountered.
Chris@0:      */
Chris@0:     public function charsUntil($mask)
Chris@0:     {
Chris@0:         return $this->is->charsUntil($mask);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Read chars as long as the mask matches.
Chris@0:      */
Chris@0:     public function charsWhile($mask)
Chris@0:     {
Chris@0:         return $this->is->charsWhile($mask);
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Returns the current column of the current line that the tokenizer is at.
Chris@0:      *
Chris@0:      * Newlines are column 0. The first char after a newline is column 1.
Chris@0:      *
Chris@0:      * @return int The column number.
Chris@0:      */
Chris@0:     public function columnOffset()
Chris@0:     {
Chris@0:         return $this->is->columnOffset();
Chris@0:     }
Chris@0: 
Chris@0:     /**
Chris@0:      * Get all characters until EOF.
Chris@0:      *
Chris@0:      * This consumes characters until the EOF.
Chris@0:      *
Chris@0:      * @return int The number of characters remaining.
Chris@0:      */
Chris@0:     public function remainingChars()
Chris@0:     {
Chris@0:         return $this->is->remainingChars();
Chris@0:     }
Chris@0: }