maintenance/wiki-mangleme.php

   1 <?php
   2 /**
   3
   4 Author : Nick Jenkins, http://nickj.org/
   5 Date   : 18 May 2006.
   6 License: GPL v 2.
   7
   8 Desc:
   9   Performs fuzz-style testing of MediaWiki's parser.
  10   The script feeds the parser some randomized malformed wiki-text, and stores
  11   the HTML output.
  12
  13   Checks the HTML output for:
  14     - unclosed tags
  15     - errors in Tidy
  16   both can indicate potential security issues.
  17
  18     Can optionally W3C validate of the HTML output (indicates malformed HTML
  19   output).
  20
  21 Background:
  22     Contains a PHP port, of a "shameless" Python PORT, OF LCAMTUF'S MANGELME
  23           http://www.securiteam.com/tools/6Z00N1PBFK.html
  24
  25 Requirements:
  26     You need PHP4 or PHP5, with PHP-curl enabled, and Tidy installed.
  27
  28 Usage:
  29     Update the "Configuration" section, especially the "WIKI_URL" to point
  30   to a local wiki you can test stuff on. You can optionally set
  31   "VALIDATE_ON_WEB" to true, although at the moment very few generated pages
  32   will validate. Then run "php wiki-mangleme.php".
  33
  34     This will print a list of HTML output that had unclosed tags, and/or that
  35   caused tidy errors. It will keep running until you press Ctrl-C. All output
  36   files are stored in the "mangleme" subdirectory.
  37 */
  38
  39 # This is a command line script, load mediawiki env:
  40 include('commandLine.inc');
  41
  42 // Configuration:
  43
  44 # The directory name where we store the output
  45 # for windows: "c:\\temp\\mangleme"
  46 define("DIRECTORY",      "/tmp/mangleme");
  47
  48 # URL to some wiki on which we can run our tests:
  49 define("WIKI_URL", $wgServer . $wgScriptPath . '/index.php?title=WIKIMANGLE' );
  50
  51 # Should our test output include binary strings?
  52 define("INCLUDE_BINARY",  false);
  53
  54 # Whether we want to send output on the web for validation:
  55 define("VALIDATE_ON_WEB", false);
  56 # URL to use to validate our output:
  57 define("VALIDATOR_URL",  "http://validator.w3.org/check");
  58
  59
  60 // If it goes wrong, we want to know about it.
  61 error_reporting(E_ALL);
  62
  63 /////////////////////  DEFINE THE DATA THAT WILL BE USED //////////////////////
  64 /* Note: Only some HTML tags are understood by MediaWiki, the rest is ignored.
  65          The tags that are ignored have been commented out below. */
  66
  67 $data = array();
  68 // $data["A"] = array("NAME", "HREF", "REF", "REV", "TITLE", "TARGET", "SHAPE", "onLoad", "STYLE");
  69 // $data["APPLET"] = array("CODEBASE", "CODE", "NAME", "ALIGN", "ALT", "HEIGHT", "WIDTH", "HSPACE", "VSPACE", "DOWNLOAD", "HEIGHT", "NAME", "TITLE", "onLoad", "STYLE");
  70 // $data["AREA"] = array("SHAPE", "ALT", "CO-ORDS", "HREF", "onLoad", "STYLE");
  71 $data["B"] = array("onLoad", "STYLE");
  72 // $data["BANNER"] = array("onLoad", "STYLE");
  73 // $data["BASE"] = array("HREF", "TARGET", "onLoad", "STYLE");
  74 // $data["BASEFONT"] = array("SIZE", "onLoad", "STYLE");
  75 // $data["BGSOUND"] = array("SRC", "LOOP", "onLoad", "STYLE");
  76 // $data["BQ"] = array("CLEAR", "NOWRAP", "onLoad", "STYLE");
  77 // $data["BODY"] = array("BACKGROUND", "BGCOLOR", "TEXT", "LINK", "ALINK", "VLINK", "LEFTMARGIN", "TOPMARGIN", "BGPROPERTIES", "onLoad", "STYLE");
  78 $data["CAPTION"] = array("ALIGN", "VALIGN", "onLoad", "STYLE");
  79 $data["CENTER"] = array("onLoad", "STYLE");
  80 // $data["COL"] = array("ALIGN", "SPAN", "onLoad", "STYLE");
  81 // $data["COLGROUP"] = array("ALIGN", "VALIGN", "HALIGN", "WIDTH", "SPAN", "onLoad", "STYLE");
  82 $data["DIV"] = array("ALIGN", "CLASS", "LANG", "onLoad", "STYLE");
  83 // $data["EMBED"] = array("SRC", "HEIGHT", "WIDTH", "UNITS", "NAME", "PALETTE", "onLoad", "STYLE");
  84 // $data["FIG"] = array("SRC", "ALIGN", "HEIGHT", "WIDTH", "UNITS", "IMAGEMAP", "onLoad", "STYLE");
  85 // $data["FN"] = array("ID", "onLoad", "STYLE");
  86 $data["FONT"] = array("SIZE", "COLOR", "FACE", "onLoad", "STYLE");
  87 // $data["FORM"] = array("ACTION", "METHOD", "ENCTYPE", "TARGET", "SCRIPT", "onLoad", "STYLE");
  88 // $data["FRAME"] = array("SRC", "NAME", "MARGINWIDTH", "MARGINHEIGHT", "SCROLLING", "FRAMESPACING", "onLoad", "STYLE");
  89 // $data["FRAMESET"] = array("ROWS", "COLS", "onLoad", "STYLE");
  90 $data["H1"] = array("SRC", "DINGBAT", "onLoad", "STYLE");
  91 // $data["HEAD"] = array("onLoad", "STYLE");
  92 $data["HR"] = array("SRC", "SIZE", "WIDTH", "ALIGN", "COLOR", "onLoad", "STYLE");
  93 // $data["HTML"] = array("onLoad", "STYLE");
  94 // $data["IFRAME"] = array("ALIGN", "FRAMEBORDER", "HEIGHT", "MARGINHEIGHT", "MARGINWIDTH", "NAME", "SCROLLING", "SRC", "ADDRESS", "WIDTH", "onLoad", "STYLE");
  95 // $data["IMG"] = array("ALIGN", "ALT", "SRC", "BORDER", "DYNSRC", "HEIGHT", "HSPACE", "ISMAP", "LOOP", "LOWSRC", "START", "UNITS", "USEMAP", "WIDTH", "VSPACE", "onLoad", "STYLE");
  96 // $data["INPUT"] = array("TYPE", "NAME", "VALUE", "onLoad", "STYLE");
  97 // $data["ISINDEX"] = array("HREF", "PROMPT", "onLoad", "STYLE");
  98 $data["LI"] = array("SRC", "DINGBAT", "SKIP", "TYPE", "VALUE", "onLoad", "STYLE");
  99 // $data["LINK"] = array("REL", "REV", "HREF", "TITLE", "onLoad", "STYLE");
 100 // $data["MAP"] = array("NAME", "onLoad", "STYLE");
 101 // $data["MARQUEE"] = array("ALIGN", "BEHAVIOR", "BGCOLOR", "DIRECTION", "HEIGHT", "HSPACE", "LOOP", "SCROLLAMOUNT", "SCROLLDELAY", "WIDTH", "VSPACE", "onLoad", "STYLE");
 102 // $data["MENU"] = array("onLoad", "STYLE");
 103 // $data["META"] = array("HTTP-EQUIV", "CONTENT", "NAME", "onLoad", "STYLE");
 104 // $data["MULTICOL"] = array("COLS", "GUTTER", "WIDTH", "onLoad", "STYLE");
 105 // $data["NOFRAMES"] = array("onLoad", "STYLE");
 106 // $data["NOTE"] = array("CLASS", "SRC", "onLoad", "STYLE");
 107 // $data["OVERLAY"] = array("SRC", "X", "Y", "HEIGHT", "WIDTH", "UNITS", "IMAGEMAP", "onLoad", "STYLE");
 108 // $data["PARAM"] = array("NAME", "VALUE", "onLoad", "STYLE");
 109 // $data["RANGE"] = array("FROM", "UNTIL", "onLoad", "STYLE");
 110 // $data["SCRIPT"] = array("LANGUAGE", "onLoad", "STYLE");
 111 // $data["SELECT"] = array("NAME", "SIZE", "MULTIPLE", "WIDTH", "HEIGHT", "UNITS", "onLoad", "STYLE");
 112 // $data["OPTION"] = array("VALUE", "SHAPE", "onLoad", "STYLE");
 113 // $data["SPACER"] = array("TYPE", "SIZE", "WIDTH", "HEIGHT", "ALIGN", "onLoad", "STYLE");
 114 // $data["SPOT"] = array("ID", "onLoad", "STYLE");
 115 // $data["TAB"] = array("INDENT", "TO", "ALIGN", "DP", "onLoad", "STYLE");
 116 $data["TABLE"] = array("ALIGN", "WIDTH", "BORDER", "CELLPADDING", "CELLSPACING", "BGCOLOR", "VALIGN", "COLSPEC", "UNITS", "DP", "onLoad", "STYLE");
 117 // $data["TBODY"] = array("CLASS", "ID", "onLoad", "STYLE");
 118 $data["TD"] = array("COLSPAN", "ROWSPAN", "ALIGN", "VALIGN", "BGCOLOR", "onLoad", "STYLE");
 119 // $data["TEXTAREA"] = array("NAME", "COLS", "ROWS", "onLoad", "STYLE");
 120 // $data["TEXTFLOW"] = array("CLASS", "ID", "onLoad", "STYLE");
 121 // $data["TFOOT"] = array("COLSPAN", "ROWSPAN", "ALIGN", "VALIGN", "BGCOLOR", "onLoad", "STYLE");
 122 $data["TH"] = array("ALIGN", "CLASS", "ID", "onLoad", "STYLE");
 123 // $data["TITLE"] = array("onLoad", "STYLE");
 124 $data["TR"] = array("ALIGN", "VALIGN", "BGCOLOR", "CLASS", "onLoad", "STYLE");
 125 $data["UL"] = array("SRC", "DINGBAT", "SKIP", "TYPE", "VALUE", "onLoad", "STYLE");
 126
 127 // Now add in a few that were not in the original, but which MediaWiki understands, even with
 128 // extraneous attributes:
 129 $data["gallery"] = array("CLASS", "ID", "onLoad", "STYLE");
 130 $data["pre"]     = array("CLASS", "ID", "onLoad", "STYLE");
 131 $data["nowiki"]  = array("CLASS", "ID", "onLoad", "STYLE");
 132 $data["blockquote"] = array("CLASS", "ID", "onLoad", "STYLE");
 133 $data["span"]    = array("CLASS", "ID", "onLoad", "STYLE");
 134 $data["code"]    = array("CLASS", "ID", "onLoad", "STYLE");
 135 $data["tt"]      = array("CLASS", "ID", "onLoad", "STYLE");
 136 $data["small"]   = array("CLASS", "ID", "onLoad", "STYLE");
 137 $data["big"]     = array("CLASS", "ID", "onLoad", "STYLE");
 138 $data["s"]       = array("CLASS", "ID", "onLoad", "STYLE");
 139 $data["u"]       = array("CLASS", "ID", "onLoad", "STYLE");
 140 $data["del"]     = array("CLASS", "ID", "onLoad", "STYLE");
 141 $data["ins"]     = array("CLASS", "ID", "onLoad", "STYLE");
 142 $data["sub"]     = array("CLASS", "ID", "onLoad", "STYLE");
 143 $data["ol"]      = array("CLASS", "ID", "onLoad", "STYLE");
 144
 145
 146 // The types of the HTML that we will be testing were defined above
 147 $types = array_keys($data);
 148
 149 // Some attribute values.
 150 $other = array("&","=",":","?","\"","\n","%n%n%n%n%n%n%n%n%n%n%n%n","\\");
 151 $ints = array("0","-1","127","7897","89000","808080","90928345","74326794236234","0xfffffff","ffff");
 152
 153 ///////////////////////////////// WIKI-SYNTAX ///////////////////////////
 154 /* Note: Defines various wiki-related bits of syntax, that can potentially cause
 155          MediaWiki to do something other than just print that literal text */
 156 $ext = array(
 157 "[[", "]]", "\n{|", "|}", "{{", "}}", "|", "[[image:", "[", "]",
 158 "=", "==", "===", "====", "=====", "======", "\n*", "*", "\n:", ":",
 159 "{{{", "}}}",
 160 "\n", "\n#", "#", "\n;", ";", "\n ",
 161 "----", "\n----",
 162 "|]]", "~~~", "#REDIRECT [[", "'''", "''",
 163 "ISBN 2", "\n|-", "| ", "\n| ",
 164 "<!--", "-->",
 165 "\"", "'",
 166 ">",
 167 "http://","https://","url://","ftp://","file://","irc://","javascript:",
 168 "!",
 169 "\n! ",
 170 "!!",
 171 "||",
 172 ".gif",
 173 ".png",
 174 ".jpg",
 175 ".jpeg",
 176 "<!--()()",
 177 '%08X',
 178 '/',
 179 ":x{|",
 180 "\n|-",
 181 "\n|+",
 182 "<noinclude>",
 183 "</noinclude>",
 184 "\n-----",
 185 "UNIQ25f46b0524f13e67NOPARSE",
 186 " \302\273",
 187 " :",
 188 " !",
 189 " ;",
 190 "\302\253",
 191 "RFC 000",
 192 "PMID 000",
 193 "?=",
 194 "(",
 195 ")".
 196 "]]]",
 197 "../",
 198 "{{{{",
 199 "}}}}",
 200 "{{subst:",
 201 '__NOTOC__',
 202 '__FORCETOC__',
 203 '__NOEDITSECTION__',
 204 '__START__',
 205 '{{PAGENAME}}',
 206 '{{PAGENAMEE}}',
 207 '{{NAMESPACE}}',
 208 '{{MSG:',
 209 '{{MSGNW:',
 210 '__END__',
 211 '{{INT:',
 212 '{{SITENAME}}',
 213 '{{NS:',
 214 '{{LOCALURL:',
 215 '{{LOCALURLE:',
 216 '{{SCRIPTPATH}}',
 217 '{{GRAMMAR:',
 218 '__NOTITLECONVERT__',
 219 '__NOCONTENTCONVERT__',
 220 "<!--MWTEMPLATESECTION=",
 221 "<!--LINK 987-->",
 222 "<!--IWLINK 987-->",
 223 "Image:",
 224 "[[category:",
 225 "{{REVISIONID}}",
 226 "{{SUBPAGENAME}}",
 227 "{{SUBPAGENAMEE}}",
 228 "{{ns:0}}",
 229 "[[:Image",
 230 "[[Special:",
 231 "{{fullurl:}}",
 232 '__TOC__',
 233 "<includeonly>",
 234 "</includeonly>",
 235 "<math>",
 236 "</math>"
 237 );
 238
 239
 240 /////////////////////  A CLASS THAT GENERATES RANDOM STRINGS OF DATA //////////////////////
 241
 242 class htmler {
 243         var $maxparams = 4;
 244         var $maxtypes = 40;
 245
 246         function randnum($finish,$start=0) {
 247                 return mt_rand($start,$finish);
 248         }
 249
 250         function randstring() {
 251                 global $ext;
 252                 $thestring = "";
 253
 254                 for ($i=0; $i<40; $i++) {
 255                         $what = $this->randnum(1);
 256
 257                         if ($what == 0) { // include some random wiki syntax
 258                                 $which = $this->randnum(count($ext) - 1);
 259                                 $thestring .= $ext[$which];
 260                         }
 261                         else { // include some random text
 262                                 $char = chr(INCLUDE_BINARY ? $this->randnum(255) : $this->randnum(126,32));
 263                                 if ($char == "<") $char = ""; // we don't want the '<' character, it stuffs us up.
 264                                 $length = $this->randnum(8);
 265                                 $thestring .= str_repeat ($char, $length);
 266                         }
 267                 }
 268                 return $thestring;
 269         }
 270
 271         function makestring() {
 272                 global $ints, $other;
 273                 $what = $this->randnum(2);
 274                 if ($what == 0) {
 275                         return $this->randstring();
 276                 }
 277                 elseif ($what == 1) {
 278                         return $ints[$this->randnum(count($ints) - 1)];
 279                 }
 280                 else {
 281                         return $other[$this->randnum(count($other) - 1)];
 282                 }
 283         }
 284
 285     function loop() {
 286                 global $types, $data;
 287                 $string = "";
 288                 $i = $this->randnum(count($types) - 1);
 289                 $t = $types[$i];
 290                 $arr = $data[$t];
 291                 $string .= "<" . $types[$i] . " ";
 292                 for ($z=0; $z<$this->maxparams; $z++) {
 293                         $badparam = $arr[$this->randnum(count($arr) - 1)];
 294                         $badstring = $this->makestring();
 295                         $string .= $badparam . "=" . $badstring . " ";
 296                 }
 297                 $string .= ">\n";
 298                 return $string;
 299                 }
 300
 301     function main() {
 302                 $page = "";
 303                 for ($k=0; $k<$this->maxtypes; $k++) {
 304                         $page .= $this->loop();
 305                 }
 306                 return $page;
 307                 }
 308 }
 309
 310
 311 ////////////////////  SAVING OUTPUT  /////////////////////////
 312
 313
 314 /**
 315 ** @desc: Utility function for saving a file. Currently has no error checking.
 316 */
 317 function saveFile($string, $name) {
 318         $fp = fopen ( DIRECTORY . "/" . $name, "w");
 319         fwrite($fp, $string);
 320         fclose ($fp);
 321 }
 322
 323
 324 //////////////////// MEDIAWIKI PREVIEW /////////////////////////
 325
 326 /*
 327 ** @desc: Asks MediaWiki for a preview of a string. Returns the HTML.
 328 */
 329 function wikiPreview($text) {
 330
 331         $params = array (
 332                 "action"      => "submit",
 333                 "wpMinoredit" => "1",
 334                 "wpPreview"   => "Show preview",
 335                 "wpSection"   => "new",
 336                 "wpEdittime"  => "",
 337                 "wpSummary"   => "This is a test",
 338                 "wpTextbox1"  => $text
 339         );
 340
 341         if( function_exists('curl_init') ) {
 342                 $ch = curl_init();
 343         } else {
 344                 die("Could not found 'curl_init' function. Is curl extension enabled ?\n");
 345         }
 346
 347         curl_setopt($ch, CURLOPT_POST, 1);                    // save form using a POST
 348         curl_setopt($ch, CURLOPT_POSTFIELDS, $params);        // load the POST variables
 349         curl_setopt($ch, CURLOPT_URL, WIKI_URL);              // set url to post to
 350         curl_setopt($ch, CURLOPT_RETURNTRANSFER,1);           // return into a variable
 351
 352         $result=curl_exec ($ch);
 353
 354         // if we encountered an error, then log it, and exit.
 355         if (curl_error($ch)) {
 356                 trigger_error("Curl error #: " . curl_errno($ch) . " - " . curl_error ($ch) );
 357                 print "Curl error #: " . curl_errno($ch) . " - " . curl_error ($ch) . " - exiting.\n";
 358                 exit();
 359         }
 360
 361         curl_close ($ch);
 362
 363         return $result;
 364 }
 365
 366
 367 //////////////////// HTML VALIDATION /////////////////////////
 368
 369 /*
 370 ** @desc: Asks the validator whether this is valid HTML, or not.
 371 */
 372 function validateHTML($text) {
 373
 374         $params = array ("fragment"   => $text);
 375
 376         $ch = curl_init();
 377
 378         curl_setopt($ch, CURLOPT_POST, 1);                    // save form using a POST
 379         curl_setopt($ch, CURLOPT_POSTFIELDS, $params);        // load the POST variables
 380         curl_setopt($ch, CURLOPT_URL, VALIDATOR_URL);         // set url to post to
 381         curl_setopt($ch, CURLOPT_RETURNTRANSFER,1);           // return into a variable
 382
 383         $result=curl_exec ($ch);
 384
 385         // if we encountered an error, then log it, and exit.
 386         if (curl_error($ch)) {
 387                 trigger_error("Curl error #: " . curl_errno($ch) . " - " . curl_error ($ch) );
 388                 print "Curl error #: " . curl_errno($ch) . " - " . curl_error ($ch) . " - exiting.\n";
 389                 exit();
 390         }
 391
 392         curl_close ($ch);
 393
 394         $valid = (strpos($result, "Failed validation") === false ? true : false);
 395
 396         return array($valid, $result);
 397 }
 398
 399
 400
 401 /**
 402 ** @desc: checks the string to see if tags are balanced.
 403 */
 404 function checkOpenCloseTags($string, $filename) {
 405         $valid = true;
 406
 407         $lines = explode("\n", $string);
 408
 409         $num_lines = count($lines);
 410         // print "Num lines: " . $num_lines . "\n";
 411
 412         foreach ($lines as $line_num => $line) {
 413
 414                 // skip mediawiki's own unbalanced lines.
 415                 if ($line_num == 15) continue;
 416                 if ($line == "\t\t<style type=\"text/css\">/*<![CDATA[*/") continue;
 417                 if ($line == "<textarea tabindex='1' accesskey=\",\" name=\"wpTextbox1\" id=\"wpTextbox1\" rows='25'") continue;
 418
 419                 if ($line == "/*<![CDATA[*/") continue;
 420                 if ($line == "/*]]>*/") continue;
 421                 if (ereg("^<form id=\"editform\" name=\"editform\" method=\"post\" action=\"", $line)) continue;
 422                 if (ereg("^enctype=\"multipart/form-data\"><input type=\"hidden\" name=\"wikidb_session\" value=\"", $line)) continue; // line num and content changes.
 423                 if ($line == "<textarea tabindex='1' accesskey=\",\" name=\"wpTextbox1\" rows='25'") continue;
 424                 if (ereg("^cols='80'>", $line)) continue; //  line num and content changes.
 425
 426                 if ($num_lines - $line_num == 246) continue;
 427                 if ($num_lines - $line_num == 65) continue;
 428                 if ($num_lines - $line_num == 62) continue;
 429                 if ($num_lines - $line_num == 52) continue;
 430                 if ($num_lines - $line_num == 50) continue;
 431                 if ($num_lines - $line_num == 29) continue;
 432                 if ($num_lines - $line_num == 28) continue;
 433                 if ($num_lines - $line_num == 27) continue;
 434                 if ($num_lines - $line_num == 23) continue;
 435
 436                 if (substr_count($line, "<") > substr_count($line, ">")) {
 437                         print "\nUnclosed tag in " . DIRECTORY . "/" . $filename . " on line: " . ($line_num + 1) . " \n$line\n";
 438                         $valid = false;
 439                 }
 440         }
 441         return $valid;
 442 }
 443
 444
 445 /**
 446 ** @desc: Get tidy to check for no HTML errors in the output file (e.g. unescaped strings).
 447 */
 448 function tidyCheckFile($name) {
 449         $file = DIRECTORY . "/" . $name;
 450         $x = `tidy -errors -quiet --show-warnings false $file 2>&1`;
 451         if (trim($x) != "") {
 452                 print "Tidy errors found in $file:\n$x";
 453                 return false;
 454         } else {
 455                 return true;
 456         }
 457 }
 458
 459
 460 ////////////////////// TESTING FUNCTION ////////////////////////
 461 /**
 462 ** @desc: takes a wiki markup string, and tests it for security or validation problems.
 463 */
 464 function testWikiMarkup($raw_markup, $testname) {
 465
 466            // don't overwrite a previous test of the same name.
 467            while (file_exists(DIRECTORY . "/" . $testname . ".raw_markup.txt")) {
 468                            $testname .= "-" . mt_rand(0,9);
 469            }
 470
 471                 // upload to MediaWiki install.
 472                 $wiki_preview = wikiPreview($raw_markup);
 473
 474                 // save output files
 475                 saveFile($raw_markup,  $testname . ".raw_markup.txt");
 476                 saveFile($wiki_preview,  $testname . ".wiki_preview.html");
 477
 478                 // validate result
 479                 $valid = true;
 480                 if (VALIDATE_ON_WEB) list ($valid, $validator_output) = validateHTML($wiki_preview);
 481                 $valid = $valid && checkOpenCloseTags ($wiki_preview, $testname . ".wiki_preview.html");
 482                 $valid = $valid && tidyCheckFile( $testname . ".wiki_preview.html" );
 483
 484
 485                 if( $valid ) {
 486                                 // Remove valid tests:
 487                                 unlink( DIRECTORY . "/" . $testname . ".raw_markup.txt" );
 488                                 unlink( DIRECTORY . "/" . $testname . ".wiki_preview.html");
 489                 } elseif( VALIDATE_ON_WEB ) {
 490                                 saveFile($validator_output,  $testname . ".validator_output.html");
 491                 }
 492 }
 493
 494
 495 //////////////////////  MAIN LOOP  ////////////////////////
 496
 497 // Make directory if doesn't exist
 498 if (!is_dir(DIRECTORY)) {
 499         mkdir (DIRECTORY, 0700 );
 500 }
 501 // otherwise, retest the things that we have found in previous runs
 502 else {
 503            print "Retesting previously found problems.\n";
 504
 505            // create a handler for the directory
 506            $handler = opendir(DIRECTORY);
 507
 508            // keep going until all files in directory have been read
 509            while ($file = readdir($handler)) {
 510
 511                            // if file is not raw markup, or is a retest, then skip it.
 512                            if (!ereg("\.raw_markup.txt$", $file)) continue;
 513                                 if ( ereg("^retest-", $file)) continue;
 514
 515                                 print "Retesting " . DIRECTORY . "/" . $file . "\n";
 516
 517                            // get file contents
 518                            $markup = file_get_contents(DIRECTORY . "/" . $file);
 519
 520                            // run retest
 521                            testWikiMarkup($markup, "retest-" . $file);
 522            }
 523
 524            // tidy up: close the handler
 525            closedir($handler);
 526
 527            print "Done retesting.\n";
 528 }
 529
 530 // seed the random number generator
 531 mt_srand(crc32(microtime()));
 532
 533 // main loop.
 534 $h = new htmler();
 535
 536 print "Beginning main loop. Results are stored in the ".DIRECTORY." directory.\n";
 537 print "Press CTRL+C to stop testing.\n";
 538 for ($count=0; true /*$count<10000 */ ; $count++) { // while (true)
 539         switch( $count % 4 ) {
 540                 case '0': print "\r/"; break;
 541                 case '1': print "\r-"; break;
 542                 case '2': print "\r\\"; break;
 543                 case '3': print "\r|"; break;
 544         }
 545         print " $count";
 546
 547         // generate and save text to test.
 548         $raw_markup = $h->main();
 549
 550         // test this wiki markup
 551         testWikiMarkup($raw_markup, $count);
 552 }
 553 ?>