www/plugins-dist/textwheel/engine/textwheel.php

   1 <?php
   2
   3 /*
   4  * TextWheel 0.1
   5  *
   6  * let's reinvent the wheel one last time
   7  *
   8  * This library of code is meant to be a fast and universal replacement
   9  * for any and all text-processing systems written in PHP
  10  *
  11  * It is dual-licensed for any use under the GNU/GPL2 and MIT licenses,
  12  * as suits you best
  13  *
  14  * (c) 2009 Fil - fil@rezo.net
  15  * Documentation & http://zzz.rezo.net/-TextWheel-
  16  *
  17  * Usage: $wheel = new TextWheel(); echo $wheel->text($text);
  18  *
  19  */
  20
  21 if (!defined('_ECRIRE_INC_VERSION')) return;
  22
  23 require_once dirname(__FILE__)."/textwheelruleset.php";
  24
  25 class TextWheel {
  26         protected $ruleset;
  27         protected static $subwheel = array();
  28
  29         protected $compiled = array();
  30
  31         /**
  32          * Constructor
  33          * @param TextWheelRuleSet $ruleset
  34          */
  35         public function TextWheel($ruleset = null) {
  36                 $this->setRuleSet($ruleset);
  37         }
  38
  39         /**
  40          * Set RuleSet
  41          * @param TextWheelRuleSet $ruleset
  42          */
  43         public function setRuleSet($ruleset){
  44                 if (!is_object($ruleset))
  45                         $ruleset = new TextWheelRuleSet ($ruleset);
  46                 $this->ruleset = $ruleset;
  47         }
  48
  49         /**
  50          * Apply all rules of RuleSet to a text
  51          *
  52          * @param string $t
  53          * @return string
  54          */
  55         public function text($t) {
  56                 $rules = & $this->ruleset->getRules();
  57                 ## apply each in order
  58                 foreach ($rules as $name => $rule) #php4+php5
  59                 {
  60                         $this->apply($rules[$name], $t);
  61                 }
  62                 #foreach ($this->rules as &$rule) #smarter &reference, but php5 only
  63                 #       $this->apply($rule, $t);
  64                 return $t;
  65         }
  66
  67         private function export($x) {
  68                 return addcslashes(var_export($x, true), "\n\r\t");
  69         }
  70
  71         public function compile($b = null) {
  72                 $rules = & $this->ruleset->getRules();
  73
  74                 ## apply each in order
  75                 $pre = array();
  76                 $comp = array();
  77
  78                 foreach ($rules as $name => $rule)
  79                 {
  80                         $rule->name = $name;
  81                         $this->initRule($rule);
  82                         if (is_string($rule->replace)
  83                         AND isset($this->compiled[$rule->replace])
  84                         AND $fun = $this->compiled[$rule->replace]) {
  85                                 $pre[] = "\n###\n## $name\n###\n" . $fun;
  86                                 preg_match(',function (\w+),', $fun, $r);
  87                                 $rule->compilereplace = $r[1]; # ne pas modifier ->replace sinon on casse l'execution...
  88                         }
  89
  90                         $r = "\t/* $name */\n";
  91
  92                         if ($rule->require)
  93                                 $r .= "\t".'require_once '.TextWheel::export($rule->require).';'."\n";
  94                         if ($rule->if_str)
  95                                 $r .= "\t".'if (strpos($t, '.TextWheel::export($rule->if_str).') === false)'."\n";
  96                         if ($rule->if_stri)
  97                                 $r .= "\t".'if (stripos($t, '.TextWheel::export($rule->if_stri).') === false)'."\n";
  98                         if ($rule->if_match)
  99                                 $r .= "\t".'if (preg_match('.TextWheel::export($rule->if_match).', $t))'."\n";
 100
 101                         if ($rule->func_replace !== 'replace_identity') {
 102                                 $fun = 'TextWheel::'.$rule->func_replace;
 103                                 switch($fun) {
 104                                         case 'TextWheel::replace_all_cb':
 105                                                 $fun = $rule->replace; # trim()...
 106                                                 break;
 107                                         case 'TextWheel::replace_preg':
 108                                                 $fun = 'preg_replace';
 109                                                 break;
 110                                         case 'TextWheel::replace_str':
 111                                                 $fun = 'str_replace';
 112                                                 break;
 113                                         case 'TextWheel::replace_preg_cb':
 114                                                 $fun = 'preg_replace_callback';
 115                                                 break;
 116                                         default:
 117                                                 break;
 118                                 }
 119                                 $r .= "\t".'$t = '.$fun.'('.TextWheel::export($rule->match).', '.TextWheel::export($rule->replace).', $t);'."\n";
 120                         }
 121
 122                         $comp[] = $r;
 123                 }
 124                 $code = join ("\n", $comp);
 125                 $code = 'function '.$b.'($t) {' . "\n". $code . "\n\treturn \$t;\n}\n\n";
 126                 $code = join ("\n", $pre) . $code;
 127
 128                 return $code;
 129         }
 130
 131
 132         /**
 133          * Get an internal global subwheel
 134          * read acces for annymous function only
 135          *
 136          * @param int $n
 137          * @return TextWheel
 138          */
 139         public static function &getSubWheel($n){
 140                 return TextWheel::$subwheel[$n];
 141         }
 142
 143         /**
 144          * Create SubWheel (can be overriden in debug class)
 145          * @param TextWheelRuleset $rules
 146          * @return TextWheel
 147          */
 148         protected function &createSubWheel(&$rules){
 149                 $tw = new TextWheel($rules);
 150                 return $tw;
 151         }
 152
 153         /**
 154          * Initializing a rule a first call
 155          * including file, creating function or wheel
 156          * optimizing tests
 157          *
 158          * @param TextWheelRule $rule
 159          */
 160         protected function initRule(&$rule){
 161                 # language specific
 162                 if ($rule->require){
 163                         require_once $rule->require;
 164                 }
 165
 166                 # optimization: strpos or stripos?
 167                 if (isset($rule->strpos)) {
 168                         if (strtolower($rule->if_str) !== strtoupper($rule->if_str)) {
 169                                 $rule->if_stri = $rule->if_str;
 170                                 unset($rule->if_str);
 171                         }
 172                 }
 173
 174                 if ($rule->create_replace){
 175                         $compile = $rule->replace.'($t)';
 176                         $rule->replace = create_function('$m', $rule->replace);
 177                         $this->compiled[$rule->replace] = $compile;
 178                         $rule->create_replace = false;
 179                         $rule->is_callback = true;
 180                 }
 181                 elseif ($rule->is_wheel){
 182                         $n = count(TextWheel::$subwheel);
 183                         TextWheel::$subwheel[] = $this->createSubWheel($rule->replace);
 184                         $var = '$m['.intval($rule->pick_match).']';
 185                         if ($rule->type=='all' OR $rule->type=='str' OR $rule->type=='split' OR !isset($rule->match))
 186                                 $var = '$m';
 187                         $code = 'return TextWheel::getSubWheel('.$n.')->text('.$var.');';
 188                         $rule->replace = create_function('$m', $code);
 189                         $cname = 'compiled_'.str_replace('-','_', $rule->name);
 190                         $compile = TextWheel::getSubWheel($n)->compile($cname);
 191                         $this->compiled[$rule->replace] = $compile;
 192                         $rule->is_wheel = false;
 193                         $rule->is_callback = true;
 194                 }
 195
 196                 # optimization
 197                 $rule->func_replace = '';
 198                 if (isset($rule->replace)) {
 199                         switch($rule->type) {
 200                                 case 'all':
 201                                         $rule->func_replace = 'replace_all';
 202                                         break;
 203                                 case 'str':
 204                                         $rule->func_replace = 'replace_str';
 205                                         // test if quicker strtr usable
 206                                         if (!$rule->is_callback
 207                                                 AND is_array($rule->match) AND is_array($rule->replace)
 208                                                 AND $c = array_map('strlen',$rule->match)
 209                                                 AND $c = array_unique($c)
 210                                                 AND count($c)==1
 211                                                 AND reset($c)==1
 212                                                 AND $c = array_map('strlen',$rule->replace)
 213                                                 AND $c = array_unique($c)
 214                                                 AND count($c)==1
 215                                                 AND reset($c)==1
 216                                                 ){
 217                                                 $rule->match = implode('',$rule->match);
 218                                                 $rule->replace = implode('',$rule->replace);
 219                                                 $rule->func_replace = 'replace_strtr';
 220                                         }
 221                                         break;
 222                                 case 'split':
 223                                         $rule->func_replace = 'replace_split';
 224                                         $rule->match = array($rule->match,  is_null($rule->glue)?$rule->match:$rule->glue);
 225                                         break;
 226                                 case 'preg':
 227                                 default:
 228                                         $rule->func_replace = 'replace_preg';
 229                                         break;
 230                         }
 231                         if ($rule->is_callback)
 232                                 $rule->func_replace .= '_cb';
 233                 }
 234                 if (!method_exists("TextWheel", $rule->func_replace)){
 235                         $rule->disabled = true;
 236                         $rule->func_replace = 'replace_identity';
 237                 }
 238                 # /end
 239         }
 240
 241         /**
 242          * Apply a rule to a text
 243          *
 244          * @param TextWheelRule $rule
 245          * @param string $t
 246          * @param int $count
 247          */
 248         protected function apply(&$rule, &$t, &$count=null) {
 249
 250                 if ($rule->disabled)
 251                         return;
 252
 253                 if (isset($rule->if_chars) AND (strpbrk($t, $rule->if_chars) === false))
 254                         return;
 255
 256                 if (isset($rule->if_str) AND strpos($t, $rule->if_str) === false)
 257                         return;
 258
 259                 if (isset($rule->if_stri) AND stripos($t, $rule->if_str) === false)
 260                         return;
 261
 262                 if (isset($rule->if_match) AND !preg_match($rule->if_match, $t))
 263                         return;
 264
 265                 if (!isset($rule->func_replace))
 266                         $this->initRule($rule);
 267
 268                 $func = $rule->func_replace;
 269                 TextWheel::$func($rule->match,$rule->replace,$t,$count);
 270         }
 271
 272         /**
 273          * No Replacement function
 274          * fall back in case of unknown method for replacing
 275          * should be called max once per rule
 276          *
 277          * @param mixed $match
 278          * @param mixed $replace
 279          * @param string $t
 280          * @param int $count
 281          */
 282         protected static function replace_identity(&$match,&$replace,&$t,&$count){
 283         }
 284
 285         /**
 286          * Static replacement of All text
 287          * @param mixed $match
 288          * @param mixed $replace
 289          * @param string $t
 290          * @param int $count
 291          */
 292         protected static function replace_all(&$match,&$replace,&$t,&$count){
 293                 # special case: replace $0 with $t
 294                 #   replace: "A$0B" will surround the string with A..B
 295                 #   replace: "$0$0" will repeat the string
 296                 if (strpos($replace, '$0')!==FALSE)
 297                         $t = str_replace('$0', $t, $replace);
 298                 else
 299                         $t = $replace;
 300         }
 301
 302         /**
 303          * Call back replacement of All text
 304          * @param mixed $match
 305          * @param mixed $replace
 306          * @param string $t
 307          * @param int $count
 308          */
 309         protected static function replace_all_cb(&$match,&$replace,&$t,&$count){
 310                 $t = $replace($t);
 311         }
 312
 313         /**
 314          * Static string replacement
 315          *
 316          * @param mixed $match
 317          * @param mixed $replace
 318          * @param string $t
 319          * @param int $count
 320          */
 321         protected static function replace_str(&$match,&$replace,&$t,&$count){
 322                 if (!is_string($match) OR strpos($t,$match)!==FALSE)
 323                         $t = str_replace($match, $replace, $t, $count);
 324         }
 325
 326         /**
 327          * Fast Static string replacement one char to one char
 328          *
 329          * @param mixed $match
 330          * @param mixed $replace
 331          * @param string $t
 332          * @param int $count
 333          */
 334         protected static function replace_strtr(&$match,&$replace,&$t,&$count){
 335                 $t = strtr( $t, $match, $replace);
 336         }
 337
 338         /**
 339          * Callback string replacement
 340          *
 341          * @param mixed $match
 342          * @param mixed $replace
 343          * @param string $t
 344          * @param int $count
 345          */
 346         protected static function replace_str_cb(&$match,&$replace,&$t,&$count){
 347                 if (strpos($t,$match)!==FALSE)
 348                         if (count($b = explode($match, $t)) > 1)
 349                                 $t = join($replace($match), $b);
 350         }
 351
 352         /**
 353          * Static Preg replacement
 354          *
 355          * @param mixed $match
 356          * @param mixed $replace
 357          * @param string $t
 358          * @param int $count
 359          */
 360         protected static function replace_preg(&$match,&$replace,&$t,&$count){
 361                 $t = preg_replace($match, $replace, $t, -1, $count);
 362         }
 363
 364         /**
 365          * Callback Preg replacement
 366          * @param mixed $match
 367          * @param mixed $replace
 368          * @param string $t
 369          * @param int $count
 370          */
 371         protected static function replace_preg_cb(&$match,&$replace,&$t,&$count){
 372                 $t = preg_replace_callback($match, $replace, $t, -1, $count);
 373         }
 374
 375
 376         /**
 377          * Static split replacement : invalid
 378          * @param mixed $match
 379          * @param mixed $replace
 380          * @param string $t
 381          * @param int $count
 382          */
 383         protected static function replace_split(&$match,&$replace,&$t,&$count){
 384                 throw new InvalidArgumentException('split rule always needs a callback function as replace');
 385         }
 386
 387         /**
 388          * Callback split replacement
 389          * @param array $match
 390          * @param mixed $replace
 391          * @param string $t
 392          * @param int $count
 393          */
 394         protected static function replace_split_cb(&$match,&$replace,&$t,&$count){
 395                 $a = explode($match[0], $t);
 396                 $t = join($match[1], array_map($replace,$a));
 397         }
 398 }
 399
 400 class TextWheelDebug extends TextWheel {
 401         static protected $t; #tableaux des temps
 402         static protected $tu; #tableaux des temps (rules utilises)
 403         static protected $tnu; #tableaux des temps (rules non utilises)
 404         static protected $u; #compteur des rules utiles
 405         static protected $w; #compteur des rules appliques
 406         static $total;
 407
 408         /**
 409          * Timer for profiling
 410          *
 411          * @staticvar int $time
 412          * @param string $t
 413          * @param bool $raw
 414          * @return int/strinf
 415          */
 416         protected function timer($t='rien', $raw = false) {
 417                 static $time;
 418                 $a=time(); $b=microtime();
 419                 // microtime peut contenir les microsecondes et le temps
 420                 $b=explode(' ',$b);
 421                 if (count($b)==2) $a = end($b); // plus precis !
 422                 $b = reset($b);
 423                 if (!isset($time[$t])) {
 424                         $time[$t] = $a + $b;
 425                 } else {
 426                         $p = ($a + $b - $time[$t]) * 1000;
 427                         unset($time[$t]);
 428                         if ($raw) return $p;
 429                         if ($p < 1000)
 430                                 $s = '';
 431                         else {
 432                                 $s = sprintf("%d ", $x = floor($p/1000));
 433                                 $p -= ($x*1000);
 434                         }
 435                         return $s . sprintf("%.3f ms", $p);
 436                 }
 437         }
 438
 439         /**
 440          * Apply all rules of RuleSet to a text
 441          *
 442          * @param string $t
 443          * @return string
 444          */
 445         public function text($t) {
 446                 $rules = & $this->ruleset->getRules();
 447                 ## apply each in order
 448                 foreach ($rules as $name => $rule) #php4+php5
 449                 {
 450                         if (is_int($name))
 451                                 $name .= ' '.$rule->match;
 452                         $this->timer($name);
 453                         $b = $t;
 454                         $this->apply($rule, $t);
 455                         TextWheelDebug::$w[$name] ++; # nombre de fois appliquee
 456                         $v = $this->timer($name, true); # timer
 457                         TextWheelDebug::$t[$name] += $v;
 458                         if ($t !== $b) {
 459                                 TextWheelDebug::$u[$name] ++; # nombre de fois utile
 460                                 TextWheelDebug::$tu[$name] += $v;
 461                         } else {
 462                                 TextWheelDebug::$tnu[$name] += $v;
 463                         }
 464
 465                 }
 466                 #foreach ($this->rules as &$rule) #smarter &reference, but php5 only
 467                 #       $this->apply($rule, $t);
 468                 return $t;
 469         }
 470
 471         /**
 472          * Ouputs data stored for profiling/debuging purposes
 473          */
 474         public static function outputDebug(){
 475                 if (isset(TextWheelDebug::$t)) {
 476                         $time = array_flip(array_map('strval', TextWheelDebug::$t));
 477                         krsort($time);
 478                         echo "
 479                         <div class='textwheeldebug'>
 480                         <style type='text/css'>
 481                                 .textwheeldebug table { margin:1em 0; }
 482                                 .textwheeldebug th,.textwheeldebug td { padding-left: 15px }
 483                                 .textwheeldebug .prof-0 .number { padding-right: 60px }
 484                                 .textwheeldebug .prof-1 .number { padding-right: 30px }
 485                                 .textwheeldebug .prof-1 .name { padding-left: 30px }
 486                                 .textwheeldebug .prof-2 .name { padding-left: 60px }
 487                                 .textwheeldebug .zero { color:orange; }
 488                                 .textwheeldebug .number { text-align:right; }
 489                                 .textwheeldebug .strong { font-weight:bold; }
 490                         </style>
 491                         <table class='sortable'>
 492                         <caption>Temps par rule</caption>
 493                         <thead><tr><th>temps&nbsp;(ms)</th><th>rule</th><th>application</th><th>t/u&nbsp;(ms)</th><th>t/n-u&nbsp;(ms)</th></tr></thead>\n";
 494                         $total = 0;
 495                         foreach($time as $t => $r) {
 496                                 $applications = intval(TextWheelDebug::$u[$r]);
 497                                 $total += $t;
 498                                 if(intval($t*10))
 499                                         echo "<tr>
 500                                         <td class='number strong'>".number_format(round($t*10)/10,1)."</td><td> ".spip_htmlspecialchars($r)."</td>
 501                                         <td"
 502                                         . (!$applications ? " class='zero'" : "")
 503                                         .">".$applications."/".intval(TextWheelDebug::$w[$r])."</td>
 504                                         <td class='number'>".($applications?number_format(round(TextWheelDebug::$tu[$r]/$applications*100)/100,2):"") ."</td>
 505                                         <td class='number'>".(($nu = intval(TextWheelDebug::$w[$r])-$applications)?number_format(round(TextWheelDebug::$tnu[$r]/$nu*100)/100,2):"") ."</td>
 506                                         </tr>";
 507                         }
 508                         echo "</table>\n";
 509
 510                         echo "
 511                         <table>
 512                         <caption>Temps total par rule</caption>
 513                         <thead><tr><th>temps</th><th>rule</th></tr></thead>\n";
 514                         ksort($GLOBALS['totaux']);
 515                         TextWheelDebug::outputTotal($GLOBALS['totaux']);
 516                         echo "</table>";
 517                         # somme des temps des rules, ne tient pas compte des subwheels
 518                         echo "<p>temps total rules: ".round($total)."&nbsp;ms</p>\n";
 519                         echo "</div>\n";
 520                 }
 521         }
 522
 523         public static function outputTotal($liste, $profondeur=0) {
 524                 ksort($liste);
 525                 foreach ($liste as $cause => $duree) {
 526                         if (is_array($duree)) {
 527                                 TextWheelDebug::outputTotal($duree, $profondeur+1);
 528                         } else {
 529                                 echo "<tr class='prof-$profondeur'>
 530                                         <td class='number'><b>".intval($duree)."</b>&nbsp;ms</td>
 531                                         <td class='name'>".spip_htmlspecialchars($cause)."</td>
 532                                         </tr>\n";
 533                         }
 534                 }
 535         }
 536
 537         /**
 538          * Create SubWheel (can be overriden in debug class)
 539          * @param TextWheelRuleset $rules
 540          * @return TextWheel
 541          */
 542         protected function &createSubWheel(&$rules){
 543                 return new TextWheelDebug($rules);
 544         }
 545
 546 }
 547
 548
 549
 550 /**
 551  * stripos for php4
 552  */
 553 if (!function_exists('stripos')) {
 554         function stripos($haystack, $needle) {
 555                 return strpos($haystack, stristr( $haystack, $needle ));
 556         }
 557 }
 558
 559 /**
 560  * approximation of strpbrk for php4
 561  * return false if no char of $char_list is in $haystack
 562  */
 563 if (!function_exists('strpbrk')) {
 564         function strpbrk($haystack, $char_list) {
 565     $result = strcspn($haystack, $char_list);
 566     if ($result != strlen($haystack)) {
 567         return $result;
 568     }
 569     return false;
 570         }
 571 }