library/Text_Highlighter/Text/Highlighter.php

   1 <?php
   2 /* vim: set expandtab tabstop=4 shiftwidth=4 softtabstop=4: */
   3 /**
   4  * Highlighter base class
   5  *
   6  * PHP versions 4 and 5
   7  *
   8  * LICENSE: This source file is subject to version 3.0 of the PHP license
   9  * that is available through the world-wide-web at the following URI:
  10  * http://www.php.net/license/3_0.txt.  If you did not receive a copy of
  11  * the PHP License and are unable to obtain it through the web, please
  12  * send a note to license@php.net so we can mail you a copy immediately.
  13  *
  14  * @category   Text
  15  * @package    Text_Highlighter
  16  * @author     Andrey Demenev <demenev@gmail.com>
  17  * @copyright  2004-2006 Andrey Demenev
  18  * @license    http://www.php.net/license/3_0.txt  PHP License
  19  * @version    CVS: $Id$
  20  * @link       http://pear.php.net/package/Text_Highlighter
  21  */
  22
  23 // require_once 'PEAR.php';
  24
  25 // {{{ BC constants
  26
  27 // BC trick : define constants related to default
  28 // renderer if needed
  29 if (!defined('HL_NUMBERS_LI')) {
  30     /**#@+
  31      * Constant for use with $options['numbers']
  32      * @see Text_Highlighter_Renderer_Html::_init()
  33      */
  34     /**
  35      * use numbered list
  36      */
  37     define ('HL_NUMBERS_LI'    ,    1);
  38     /**
  39      * Use 2-column table with line numbers in left column and code in  right column.
  40      * Forces $options['tag'] = HL_TAG_PRE
  41      */
  42     define ('HL_NUMBERS_TABLE'    , 2);
  43     /**#@-*/
  44 }
  45
  46 // }}}
  47 // {{{ constants
  48 /**
  49  * for our purpose, it is infinity
  50  */
  51 define ('HL_INFINITY',      1000000000);
  52
  53 // }}}
  54
  55 /**
  56  * Text highlighter base class
  57  *
  58  * @author     Andrey Demenev <demenev@gmail.com>
  59  * @copyright  2004-2006 Andrey Demenev
  60  * @license    http://www.php.net/license/3_0.txt  PHP License
  61  * @version    Release: @package_version@
  62  * @link       http://pear.php.net/package/Text_Highlighter
  63  */
  64
  65 // {{{ Text_Highlighter
  66
  67 /**
  68  * Text highlighter base class
  69  *
  70  * This class implements all functions necessary for highlighting,
  71  * but it does not contain highlighting rules. Actual highlighting is
  72  * done using a descendent of this class.
  73  *
  74  * One is not supposed to manually create descendent classes.
  75  * Instead, describe highlighting  rules in XML format and
  76  * use {@link Text_Highlighter_Generator} to create descendent class.
  77  * Alternatively, an instance of a descendent class can be created
  78  * directly.
  79  *
  80  * Use {@link Text_Highlighter::factory()} to create an
  81  * object for particular language highlighter
  82  *
  83  * Usage example
  84  * <code>
  85  *require_once 'Text/Highlighter.php';
  86  *$hlSQL = Text_Highlighter::factory('SQL',array('numbers'=>true));
  87  *echo $hlSQL->highlight('SELECT * FROM table a WHERE id = 12');
  88  * </code>
  89  *
  90  * @author Andrey Demenev <demenev@gmail.com>
  91  * @package Text_Highlighter
  92  * @access public
  93  */
  94
  95 class Text_Highlighter
  96 {
  97     // {{{ members
  98
  99     /**
 100      * Syntax highlighting rules.
 101      * Auto-generated classes set this var
 102      *
 103      * @access protected
 104      * @see _init
 105      * @var array
 106      */
 107     var $_syntax;
 108
 109     /**
 110      * Renderer object.
 111      *
 112      * @access private
 113      * @var array
 114      */
 115     var $_renderer;
 116
 117     /**
 118      * Options. Keeped for BC
 119      *
 120      * @access protected
 121      * @var array
 122      */
 123     var $_options = array();
 124
 125     /**
 126      * Conditionds
 127      *
 128      * @access protected
 129      * @var array
 130      */
 131     var $_conditions = array();
 132
 133     /**
 134      * Disabled keywords
 135      *
 136      * @access protected
 137      * @var array
 138      */
 139     var $_disabled = array();
 140
 141     /**
 142      * Language
 143      *
 144      * @access protected
 145      * @var string
 146      */
 147     var $_language = '';
 148
 149     // }}}
 150     // {{{ _checkDefines
 151
 152     /**
 153      * Called by subclssses' constructors to enable/disable
 154      * optional highlighter rules
 155      *
 156      * @param array $defines  Conditional defines
 157      *
 158      * @access protected
 159      */
 160     function _checkDefines()
 161     {
 162         if (isset($this->_options['defines'])) {
 163             $defines = $this->_options['defines'];
 164         } else {
 165             $defines = array();
 166         }
 167         foreach ($this->_conditions as $name => $actions) {
 168             foreach($actions as $action) {
 169                 $present = in_array($name, $defines);
 170                 if (!$action[1]) {
 171                     $present = !$present;
 172                 }
 173                 if ($present) {
 174                     unset($this->_disabled[$action[0]]);
 175                 } else {
 176                     $this->_disabled[$action[0]] = true;
 177                 }
 178             }
 179         }
 180     }
 181
 182     // }}}
 183     // {{{ factory
 184
 185     /**
 186      * Create a new Highlighter object for specified language
 187      *
 188      * @param string $lang    language, for example "SQL"
 189      * @param array  $options Rendering options. This
 190      * parameter is only keeped for BC reasons, use
 191      * {@link Text_Highlighter::setRenderer()} instead
 192      *
 193      * @return mixed a newly created Highlighter object, or
 194      * a PEAR error object on error
 195      *
 196      * @static
 197      * @access public
 198      */
 199     function &factory($lang, $options = array())
 200     {
 201         $lang = strtoupper($lang);
 202         @include_once 'Text/Highlighter/' . $lang . '.php';
 203
 204         $classname = 'Text_Highlighter_' . $lang;
 205
 206         if (!class_exists($classname)) {
 207             return PEAR::raiseError('Highlighter for ' . $lang . ' not found');
 208         }
 209
 210         $obj = new $classname($options);
 211
 212         return $obj;
 213     }
 214
 215     // }}}
 216     // {{{ setRenderer
 217
 218     /**
 219      * Set renderer object
 220      *
 221      * @param object $renderer  Text_Highlighter_Renderer
 222      *
 223      * @access public
 224      */
 225     function setRenderer(&$renderer)
 226     {
 227         $this->_renderer = $renderer;
 228     }
 229
 230     // }}}
 231
 232     /**
 233      * Helper function to find matching brackets
 234      *
 235      * @access private
 236      */
 237     function _matchingBrackets($str)
 238     {
 239         return strtr($str, '()<>[]{}', ')(><][}{');
 240     }
 241
 242
 243
 244
 245     function _getToken()
 246     {
 247         if (!empty($this->_tokenStack)) {
 248             return array_pop($this->_tokenStack);
 249         }
 250         if ($this->_pos >= $this->_len) {
 251             return NULL;
 252         }
 253
 254         if ($this->_state != -1 && preg_match($this->_endpattern, $this->_str, $m, PREG_OFFSET_CAPTURE, $this->_pos)) {
 255             $endpos = $m[0][1];
 256             $endmatch = $m[0][0];
 257         } else {
 258             $endpos = -1;
 259         }
 260         preg_match ($this->_regs[$this->_state], $this->_str, $m, PREG_OFFSET_CAPTURE, $this->_pos);
 261         $n = 1;
 262
 263
 264          foreach ($this->_counts[$this->_state] as $i=>$count) {
 265             if (!isset($m[$n])) {
 266                 break;
 267             }
 268             if ($m[$n][1]>-1 && ($endpos == -1 || $m[$n][1] < $endpos)) {
 269                 if ($this->_states[$this->_state][$i] != -1) {
 270                     $this->_tokenStack[] = array($this->_delim[$this->_state][$i], $m[$n][0]);
 271                 } else {
 272                     $inner = $this->_inner[$this->_state][$i];
 273                     if (isset($this->_parts[$this->_state][$i])) {
 274                         $parts = array();
 275                         $partpos = $m[$n][1];
 276                         for ($j=1; $j<=$count; $j++) {
 277                             if ($m[$j+$n][1] < 0) {
 278                                 continue;
 279                             }
 280                             if (isset($this->_parts[$this->_state][$i][$j])) {
 281                                 if ($m[$j+$n][1] > $partpos) {
 282                                     array_unshift($parts, array($inner, substr($this->_str, $partpos, $m[$j+$n][1]-$partpos)));
 283                                 }
 284                                 array_unshift($parts, array($this->_parts[$this->_state][$i][$j], $m[$j+$n][0]));
 285                             }
 286                             $partpos = $m[$j+$n][1] + strlen($m[$j+$n][0]);
 287                         }
 288                         if ($partpos < $m[$n][1] + strlen($m[$n][0])) {
 289                             array_unshift($parts, array($inner, substr($this->_str, $partpos, $m[$n][1] - $partpos + strlen($m[$n][0]))));
 290                         }
 291                         $this->_tokenStack = array_merge($this->_tokenStack, $parts);
 292                     } else {
 293                         foreach ($this->_keywords[$this->_state][$i] as $g => $re) {
 294                             if (isset($this->_disabled[$g])) {
 295                                 continue;
 296                             }
 297                             if (preg_match($re, $m[$n][0])) {
 298                                 $inner = $this->_kwmap[$g];
 299                                 break;
 300                             }
 301                         }
 302                         $this->_tokenStack[] = array($inner, $m[$n][0]);
 303                     }
 304                 }
 305                 if ($m[$n][1] > $this->_pos) {
 306                     $this->_tokenStack[] = array($this->_lastinner, substr($this->_str, $this->_pos, $m[$n][1]-$this->_pos));
 307                 }
 308                 $this->_pos = $m[$n][1] + strlen($m[$n][0]);
 309                 if ($this->_states[$this->_state][$i] != -1) {
 310                     $this->_stack[] = array($this->_state, $this->_lastdelim, $this->_lastinner, $this->_endpattern);
 311                     $this->_lastinner = $this->_inner[$this->_state][$i];
 312                     $this->_lastdelim = $this->_delim[$this->_state][$i];
 313                     $l = $this->_state;
 314                     $this->_state = $this->_states[$this->_state][$i];
 315                     $this->_endpattern = $this->_end[$this->_state];
 316                     if ($this->_subst[$l][$i]) {
 317                         for ($k=0; $k<=$this->_counts[$l][$i]; $k++) {
 318                             if (!isset($m[$i+$k])) {
 319                                 break;
 320                             }
 321                             $quoted = preg_quote($m[$n+$k][0], '/');
 322                             $this->_endpattern = str_replace('%'.$k.'%', $quoted, $this->_endpattern);
 323                             $this->_endpattern = str_replace('%b'.$k.'%', $this->_matchingBrackets($quoted), $this->_endpattern);
 324                         }
 325                     }
 326                 }
 327                 return array_pop($this->_tokenStack);
 328             }
 329             $n += $count + 1;
 330         }
 331
 332         if ($endpos > -1) {
 333             $this->_tokenStack[] = array($this->_lastdelim, $endmatch);
 334             if ($endpos > $this->_pos) {
 335                 $this->_tokenStack[] = array($this->_lastinner, substr($this->_str, $this->_pos, $endpos-$this->_pos));
 336             }
 337             list($this->_state, $this->_lastdelim, $this->_lastinner, $this->_endpattern) = array_pop($this->_stack);
 338             $this->_pos = $endpos + strlen($endmatch);
 339             return array_pop($this->_tokenStack);
 340         }
 341         $p = $this->_pos;
 342         $this->_pos = HL_INFINITY;
 343         return array($this->_lastinner, substr($this->_str, $p));
 344     }
 345
 346
 347
 348
 349     // {{{ highlight
 350
 351     /**
 352      * Highlights code
 353      *
 354      * @param  string $str      Code to highlight
 355      * @access public
 356      * @return string Highlighted text
 357      *
 358      */
 359
 360     function highlight($str)
 361     {
 362         if (!($this->_renderer)) {
 363             include_once('Text/Highlighter/Renderer/Html.php');
 364             $this->_renderer = new Text_Highlighter_Renderer_Html($this->_options);
 365         }
 366         $this->_state = -1;
 367         $this->_pos = 0;
 368         $this->_stack = array();
 369         $this->_tokenStack = array();
 370         $this->_lastinner = $this->_defClass;
 371         $this->_lastdelim = $this->_defClass;
 372         $this->_endpattern = '';
 373         $this->_renderer->reset();
 374         $this->_renderer->setCurrentLanguage($this->_language);
 375         $this->_str = $this->_renderer->preprocess($str);
 376         $this->_len = strlen($this->_str);
 377         while ($token = $this->_getToken()) {
 378             $this->_renderer->acceptToken($token[0], $token[1]);
 379         }
 380         $this->_renderer->finalize();
 381         return $this->_renderer->getOutput();
 382     }
 383
 384     // }}}
 385
 386 }
 387
 388 // }}}
 389
 390 /*
 391  * Local variables:
 392  * tab-width: 4
 393  * c-basic-offset: 4
 394  * c-hanging-comment-ender-p: nil
 395  * End:
 396  */
 397
 398 ?>