lib/AkXhtmlValidator.php

   1 <?php
   2 /* vim: set expandtab tabstop=4 shiftwidth=4 softtabstop=4: */
   3 // +----------------------------------------------------------------------+
   4 // | Akelos Framework - http://www.akelos.org                             |
   5 // +----------------------------------------------------------------------+
   6 // | Copyright (c) 2002-2006, Akelos Media, S.L.  & Bermi Ferrer Martinez |
   7 // | Released under the GNU Lesser General Public License, see LICENSE.txt|
   8 // +----------------------------------------------------------------------+
   9
  10 /**
  11  * @package ActionView
  12  * @subpackage Utils
  13  * @author Bermi Ferrer <bermi a.t akelos c.om>
  14  * @copyright Copyright (c) 2002-2006, Akelos Media, S.L. http://www.akelos.org
  15  * @license GNU Lesser General Public License <http://www.gnu.org/copyleft/lesser.html>
  16  */
  17
  18 class AkXhtmlValidator
  19 {
  20     var $_attributes = array(
  21     'core' => array(
  22     'except' => array(
  23     'base',
  24     'head',
  25     'html',
  26     'meta',
  27     'param',
  28     'script',
  29     'style',
  30     'title'
  31     ) ,
  32     'attributes' => array(
  33     'class',
  34     'id',
  35     'style',
  36     'title',
  37     'accesskey',
  38     'tabindex'
  39     ) ,
  40     ) ,
  41     'language' => array(
  42     'except' => array(
  43     'base',
  44     'br',
  45     'hr',
  46     'iframe',
  47     'param',
  48     'script'
  49     ) ,
  50     'attributes' => array(
  51     'dir' => array(
  52     'ltr',
  53     'rtl'
  54     ) ,
  55     'lang',
  56     'xml:lang'
  57     ) ,
  58     ) ,
  59     'keyboard' => array(
  60     'attributes' => array(
  61     'accesskey' => '/^(\w){1}$/',
  62     'tabindex' => '/^(\d)+$/'
  63     ) ,
  64     ) ,
  65     );
  66     var $_events = array(
  67     'window' => array(
  68     'only' => array(
  69     'body'
  70     ) ,
  71     'attributes' => array(
  72     'onload',
  73     'onunload'
  74     ) ,
  75     ) ,
  76     'form' => array(
  77     'only' => array(
  78     'form',
  79     'input',
  80     'textarea',
  81     'select',
  82     'a',
  83     'label',
  84     'button'
  85     ) ,
  86     'attributes' => array(
  87     'onchange',
  88     'onsubmit',
  89     'onreset',
  90     'onselect',
  91     'onblur',
  92     'onfocus'
  93     ) ,
  94     ) ,
  95     'keyboard' => array(
  96     'except' => array(
  97     'base',
  98     'bdo',
  99     'br',
 100     'frame',
 101     'frameset',
 102     'head',
 103     'html',
 104     'iframe',
 105     'meta',
 106     'param',
 107     'script',
 108     'style',
 109     'title'
 110     ) ,
 111     'attributes' => array(
 112     'onkeydown',
 113     'onkeypress',
 114     'onkeyup'
 115     ) ,
 116     ) ,
 117     'mouse' => array(
 118     'except' => array(
 119     'base',
 120     'bdo',
 121     'br',
 122     'head',
 123     'html',
 124     'meta',
 125     'param',
 126     'script',
 127     'style',
 128     'title'
 129     ) ,
 130     'attributes' => array(
 131     'onclick',
 132     'ondblclick',
 133     'onmousedown',
 134     'onmousemove',
 135     'onmouseover',
 136     'onmouseout',
 137     'onmouseup'
 138     ) ,
 139     ) ,
 140     );
 141     var $_tags = array(
 142     'a' => array(
 143     'attributes' => array(
 144     'charset',
 145     'coords',
 146     'href',
 147     'hreflang',
 148     'name',
 149     'rel' => '/^(alternate|designates|stylesheet|start|next|prev|contents|index|glossary|copyright|chapter|section|subsection|appendix|help|bookmark| |shortcut|icon)+$/',
 150     'rev' => '/^(alternate|designates|stylesheet|start|next|prev|contents|index|glossary|copyright|chapter|section|subsection|appendix|help|bookmark| |shortcut|icon)+$/',
 151     'shape' => '/^(rect|rectangle|circ|circle|poly|polygon)$/',
 152     'type',
 153     ) ,
 154     ) ,
 155     'abbr',
 156     'acronym',
 157     'address',
 158     'area' => array(
 159     'attributes' => array(
 160     'alt',
 161     'coords',
 162     'href',
 163     'nohref' => '/^(true|false)$/',
 164     'shape' => '/^(rect|rectangle|circ|circle|poly|polygon)$/'
 165     ) ,
 166     'required' => array(
 167     'alt'
 168     ) ,
 169     ) ,
 170     'b',
 171     'base' => array(
 172     'attributes' => array(
 173     'href'
 174     ) ,
 175     'required' => array(
 176     'href'
 177     )
 178     ) ,
 179     'bdo' => array(
 180     'attributes' => array(
 181     'dir' => '/^(ltr|rtl)$/'
 182     ) ,
 183     'required' => array(
 184     'dir'
 185     )
 186     ) ,
 187     'big',
 188     'blockquote' => array(
 189     'attributes' => array(
 190     'cite'
 191     )
 192     ) ,
 193     'body',
 194     'br',
 195     'button' => array(
 196     'attributes' => array(
 197     'disabled' => '/^(disabled)$/',
 198     'type' => '/^(button|reset|submit)$/',
 199     'value'
 200     ) ,
 201     'inside' => 'form'
 202     ) ,
 203     'caption',
 204     'cite',
 205     'code',
 206     'col' => array(
 207     'attributes' => array(
 208     'align' => '/^(right|left|center|justify)$/',
 209     'char',
 210     'charoff',
 211     'span' => '/^(\d)+$/',
 212     'valign' => '/^(top|middle|bottom|baseline)$/',
 213     'width',
 214     ) ,
 215     'inside' => 'colgroup'
 216     ) ,
 217     'colgroup' => array(
 218     'attributes' => array(
 219     'align' => '/^(right|left|center|justify)$/',
 220     'char',
 221     'charoff',
 222     'span' => '/^(\d)+$/',
 223     'valign' => '/^(top|middle|bottom|baseline)$/',
 224     'width',
 225     )
 226     ) ,
 227     'dd',
 228     'del' => array(
 229     'attributes' => array(
 230     'cite',
 231     'datetime' => '/^([0-9]){8}/'
 232     )
 233     ) ,
 234     'div',
 235     'dfn',
 236     'dl',
 237     'dt',
 238     'em',
 239     'fieldset' => array(
 240     'inside' => 'form'
 241     ) ,
 242     'form' => array(
 243     'attributes' => array(
 244     'action',
 245     'accept',
 246     'accept-charset',
 247     'enctype',
 248     'method' => '/^(get|post)$/'
 249     ) ,
 250     'required' => array(
 251     'action'
 252     )
 253     ) ,
 254     'head' => array(
 255     'attributes' => array(
 256     'profile'
 257     )
 258     ) ,
 259     'h1',
 260     'h2',
 261     'h3',
 262     'h4',
 263     'h5',
 264     'h6',
 265     'hr',
 266     'html' => array(
 267     'attributes' => array(
 268     'xmlns'
 269     )
 270     ) ,
 271     'i',
 272     'img' => array(
 273     'attributes' => array(
 274     'alt',
 275     'src',
 276     'height',
 277     'ismap',
 278     'longdesc',
 279     'usemap',
 280     'width'
 281     ) ,
 282     'required' => array(
 283     'alt',
 284     'src'
 285     ) ,
 286     ) ,
 287     'input' => array(
 288     'attributes' => array(
 289     'accept',
 290     'alt',
 291     'checked' => '/^(checked)$/',
 292     'disabled' => '/^(disabled)$/',
 293     'maxlength' => '/^(\d)+$/',
 294     'name',
 295     'readonly' => '/^(readonly)$/',
 296     'size' => '/^(\d)+$/',
 297     'src',
 298     'type' => '/^(button|checkbox|file|hidden|image|password|radio|reset|submit|text)$/',
 299     'value'
 300     ) ,
 301     'inside' => 'form'
 302     ) ,
 303     'ins' => array(
 304     'attributes' => array(
 305     'cite',
 306     'datetime' => '/^([0-9]){8}/'
 307     )
 308     ) ,
 309     'kbd',
 310     'label' => array(
 311     'attributes' => array(
 312     'for'
 313     ) ,
 314     'inside' => 'form'
 315     ) ,
 316     'legend',
 317     'li',
 318     'link' => array(
 319     'attributes' => array(
 320     'charset',
 321     'href',
 322     'hreflang',
 323     'media' => '/^(all|braille|print|projection|screen|speech|,|;| )+$/i',
 324     'rel' => '/^(alternate|appendix|bookmark|chapter|contents|copyright|glossary|help|home|index|next|prev|section|start|stylesheet|subsection| |shortcut|icon)+$/i',
 325     'rev' => '/^(alternate|appendix|bookmark|chapter|contents|copyright|glossary|help|home|index|next|prev|section|start|stylesheet|subsection| |shortcut|icon)+$/i',
 326     'type'
 327     ) ,
 328     'inside' => 'head'
 329     ) ,
 330     'map' => array(
 331     'attributes' => array(
 332     'id',
 333     'name'
 334     ) ,
 335     'required' => array(
 336     'id'
 337     )
 338     ) ,
 339     'meta' => array(
 340     'attributes' => array(
 341     'content',
 342     'http-equiv' => '/^(content\-type|expires|refresh|set\-cookie)$/i',
 343     'name',
 344     'scheme'
 345     ) ,
 346     'required' => array(
 347     'content'
 348     )
 349     ) ,
 350     'noscript',
 351     'object' => array(
 352     'attributes' => array(
 353     'archive',
 354     'classid',
 355     'codebase',
 356     'codetype',
 357     'data',
 358     'declare',
 359     'height',
 360     'name',
 361     'standby',
 362     'type',
 363     'usemap',
 364     'width'
 365     )
 366     ) ,
 367     'ol',
 368     'optgroup' => array(
 369     'attributes' => array(
 370     'label',
 371     'disabled' => '/^(disabled)$/'
 372     ) ,
 373     'required' => array(
 374     'label'
 375     )
 376     ) ,
 377     'option' => array(
 378     'attributes' => array(
 379     'label',
 380     'disabled' => '/^(disabled)$/',
 381     'selected' => '/^(selected)$/',
 382     'value'
 383     ) ,
 384     'inside' => 'select',
 385     ) ,
 386     'p',
 387     'param' => array(
 388     'attributes' => array(
 389     'type',
 390     'valuetype' => '/^(data|ref|object)$/',
 391     'valuetype',
 392     'value'
 393     ) ,
 394     'required' => array(
 395     'name'
 396     ) ,
 397     ) ,
 398     'pre',
 399     'q' => array(
 400     'attributes' => array(
 401     'cite'
 402     )
 403     ) ,
 404     'samp',
 405     'script' => array(
 406     'attributes' => array(
 407     'type' => '/^(text\/ecmascript|text\/javascript|text\/jscript|text\/vbscript|text\/vbs|text\/xml)$/',
 408     'charset',
 409     'defer' => '/^(defer)$/',
 410     'src'
 411     ) ,
 412     'required' => array(
 413     'type'
 414     )
 415     ) ,
 416     'select' => array(
 417     'attributes' => array(
 418     'disabled' => '/^(disabled)$/',
 419     'multiple' => '/^(multiple)$/',
 420     'name',
 421     'size'
 422     ) ,
 423     'inside' => 'form'
 424     ) ,
 425     'small',
 426     'span',
 427     'strong',
 428     'style' => array(
 429     'attributes' => array(
 430     'type',
 431     'media' => '/^(screen|tty|tv|projection|handheld|print|braille|aural|all)$/'
 432     ) ,
 433     'required' => array(
 434     'type'
 435     )
 436     ) ,
 437     'sub',
 438     'sup',
 439     'table' => array(
 440     'attributes' => array(
 441     'border',
 442     'cellpadding',
 443     'cellspacing',
 444     'frame' => '/^(void|above|below|hsides|lhs|rhs|vsides|box|border)$/',
 445     'rules' => '/^(none|groups|rows|cols|all)$/',
 446     'summary',
 447     'width'
 448     )
 449     ) ,
 450     'tbody' => array(
 451     'attributes' => array(
 452     'align' => '/^(right|left|center|justify)$/',
 453     'char',
 454     'charoff',
 455     'valign' => '/^(top|middle|bottom|baseline)$/'
 456     )
 457     ) ,
 458     'td' => array(
 459     'attributes' => array(
 460     'abbr',
 461     'align' => '/^(left|right|center|justify|char)$/',
 462     'axis',
 463     'char',
 464     'charoff',
 465     'colspan' => '/^(\d)+$/',
 466     'headers',
 467     'rowspan' => '/^(\d)+$/',
 468     'scope' => '/^(col|colgroup|row|rowgroup)$/',
 469     'valign' => '/^(top|middle|bottom|baseline)$/'
 470     )
 471     ) ,
 472     'textarea' => array(
 473     'attributes' => array(
 474     'cols',
 475     'rows',
 476     'disabled',
 477     'name',
 478     'readonly'
 479     ) ,
 480     'required' => array(
 481     'cols',
 482     'rows'
 483     ) ,
 484     'inside' => 'form'
 485     ) ,
 486     'tfoot' => array(
 487     'attributes' => array(
 488     'align' => '/^(right|left|center|justify)$/',
 489     'char',
 490     'charoff',
 491     'valign' => '/^(top|middle|bottom)$/',
 492     'baseline'
 493     )
 494     ) ,
 495     'th' => array(
 496     'attributes' => array(
 497     'abbr',
 498     'align' => '/^(left|right|center|justify|char)$/',
 499     'axis',
 500     'char',
 501     'charoff',
 502     'colspan' => '/^(\d)+$/',
 503     'headers',
 504     'rowspan' => '/^(\d)+$/',
 505     'scope' => '/^(col|colgroup|row|rowgroup)$/',
 506     'valign' => '/^(top|middle|bottom|baseline)$/'
 507     )
 508     ) ,
 509     'thead' => array(
 510     'attributes' => array(
 511     'align' => '/^(right|left|center|justify)$/',
 512     'char',
 513     'charoff',
 514     'valign' => '/^(top|middle|bottom|baseline)$/'
 515     )
 516     ) ,
 517     'title',
 518     'tr' => array(
 519     'attributes' => array(
 520     'align' => '/^(right|left|center|justify|char)$/',
 521     'char',
 522     'charoff',
 523     'valign' => '/^(top|middle|bottom|baseline)$/'
 524     )
 525     ) ,
 526     'tt',
 527     'ul',
 528     'var',
 529     );
 530
 531     var $_entities = array(
 532     '&nbsp;' => '&#160;',
 533     '&iexcl;' => '&#161;',
 534     '&cent;' => '&#162;',
 535     '&pound;' => '&#163;',
 536     '&curren;' => '&#164;',
 537     '&yen;' => '&#165;',
 538     '&brvbar;' => '&#166;',
 539     '&sect;' => '&#167;',
 540     '&uml;' => '&#168;',
 541     '&copy;' => '&#169;',
 542     '&ordf;' => '&#170;',
 543     '&laquo;' => '&#171;',
 544     '&not;' => '&#172;',
 545     '&shy;' => '&#173;',
 546     '&reg;' => '&#174;',
 547     '&macr;' => '&#175;',
 548     '&deg;' => '&#176;',
 549     '&plusmn;' => '&#177;',
 550     '&sup2;' => '&#178;',
 551     '&sup3;' => '&#179;',
 552     '&acute;' => '&#180;',
 553     '&micro;' => '&#181;',
 554     '&para;' => '&#182;',
 555     '&middot;' => '&#183;',
 556     '&cedil;' => '&#184;',
 557     '&sup1;' => '&#185;',
 558     '&ordm;' => '&#186;',
 559     '&raquo;' => '&#187;',
 560     '&frac14;' => '&#188;',
 561     '&frac12;' => '&#189;',
 562     '&frac34;' => '&#190;',
 563     '&iquest;' => '&#191;',
 564     '&Agrave;' => '&#192;',
 565     '&Aacute;' => '&#193;',
 566     '&Acirc;' => '&#194;',
 567     '&Atilde;' => '&#195;',
 568     '&Auml;' => '&#196;',
 569     '&Aring;' => '&#197;',
 570     '&AElig;' => '&#198;',
 571     '&Ccedil;' => '&#199;',
 572     '&Egrave;' => '&#200;',
 573     '&Eacute;' => '&#201;',
 574     '&Ecirc;' => '&#202;',
 575     '&Euml;' => '&#203;',
 576     '&Igrave;' => '&#204;',
 577     '&Iacute;' => '&#205;',
 578     '&Icirc;' => '&#206;',
 579     '&Iuml;' => '&#207;',
 580     '&ETH;' => '&#208;',
 581     '&Ntilde;' => '&#209;',
 582     '&Ograve;' => '&#210;',
 583     '&Oacute;' => '&#211;',
 584     '&Ocirc;' => '&#212;',
 585     '&Otilde;' => '&#213;',
 586     '&Ouml;' => '&#214;',
 587     '&times;' => '&#215;',
 588     '&Oslash;' => '&#216;',
 589     '&Ugrave;' => '&#217;',
 590     '&Uacute;' => '&#218;',
 591     '&Ucirc;' => '&#219;',
 592     '&Uuml;' => '&#220;',
 593     '&Yacute;' => '&#221;',
 594     '&THORN;' => '&#222;',
 595     '&szlig;' => '&#223;',
 596     '&agrave;' => '&#224;',
 597     '&aacute;' => '&#225;',
 598     '&acirc;' => '&#226;',
 599     '&atilde;' => '&#227;',
 600     '&auml;' => '&#228;',
 601     '&aring;' => '&#229;',
 602     '&aelig;' => '&#230;',
 603     '&ccedil;' => '&#231;',
 604     '&egrave;' => '&#232;',
 605     '&eacute;' => '&#233;',
 606     '&ecirc;' => '&#234;',
 607     '&euml;' => '&#235;',
 608     '&igrave;' => '&#236;',
 609     '&iacute;' => '&#237;',
 610     '&icirc;' => '&#238;',
 611     '&iuml;' => '&#239;',
 612     '&eth;' => '&#240;',
 613     '&ntilde;' => '&#241;',
 614     '&ograve;' => '&#242;',
 615     '&oacute;' => '&#243;',
 616     '&ocirc;' => '&#244;',
 617     '&otilde;' => '&#245;',
 618     '&ouml;' => '&#246;',
 619     '&divide;' => '&#247;',
 620     '&oslash;' => '&#248;',
 621     '&ugrave;' => '&#249;',
 622     '&uacute;' => '&#250;',
 623     '&ucirc;' => '&#251;',
 624     '&uuml;' => '&#252;',
 625     '&yacute;' => '&#253;',
 626     '&thorn;' => '&#254;',
 627     '&yuml;' => '&#255;',
 628     '&fnof;' => '&#402;',
 629     '&Alpha;' => '&#913;',
 630     '&Beta;' => '&#914;',
 631     '&Gamma;' => '&#915;',
 632     '&Delta;' => '&#916;',
 633     '&Epsilon;' => '&#917;',
 634     '&Zeta;' => '&#918;',
 635     '&Eta;' => '&#919;',
 636     '&Theta;' => '&#920;',
 637     '&Iota;' => '&#921;',
 638     '&Kappa;' => '&#922;',
 639     '&Lambda;' => '&#923;',
 640     '&Mu;' => '&#924;',
 641     '&Nu;' => '&#925;',
 642     '&Xi;' => '&#926;',
 643     '&Omicron;' => '&#927;',
 644     '&Pi;' => '&#928;',
 645     '&Rho;' => '&#929;',
 646     '&Sigma;' => '&#931;',
 647     '&Tau;' => '&#932;',
 648     '&Upsilon;' => '&#933;',
 649     '&Phi;' => '&#934;',
 650     '&Chi;' => '&#935;',
 651     '&Psi;' => '&#936;',
 652     '&Omega;' => '&#937;',
 653     '&alpha;' => '&#945;',
 654     '&beta;' => '&#946;',
 655     '&gamma;' => '&#947;',
 656     '&delta;' => '&#948;',
 657     '&epsilon;' => '&#949;',
 658     '&zeta;' => '&#950;',
 659     '&eta;' => '&#951;',
 660     '&theta;' => '&#952;',
 661     '&iota;' => '&#953;',
 662     '&kappa;' => '&#954;',
 663     '&lambda;' => '&#955;',
 664     '&mu;' => '&#956;',
 665     '&nu;' => '&#957;',
 666     '&xi;' => '&#958;',
 667     '&omicron;' => '&#959;',
 668     '&pi;' => '&#960;',
 669     '&rho;' => '&#961;',
 670     '&sigmaf;' => '&#962;',
 671     '&sigma;' => '&#963;',
 672     '&tau;' => '&#964;',
 673     '&upsilon;' => '&#965;',
 674     '&phi;' => '&#966;',
 675     '&chi;' => '&#967;',
 676     '&psi;' => '&#968;',
 677     '&omega;' => '&#969;',
 678     '&thetasym;' => '&#977;',
 679     '&upsih;' => '&#978;',
 680     '&piv;' => '&#982;',
 681     '&bull;' => '&#8226;',
 682     '&hellip;' => '&#8230;',
 683     '&prime;' => '&#8242;',
 684     '&Prime;' => '&#8243;',
 685     '&oline;' => '&#8254;',
 686     '&frasl;' => '&#8260;',
 687     '&weierp;' => '&#8472;',
 688     '&image;' => '&#8465;',
 689     '&real;' => '&#8476;',
 690     '&trade;' => '&#8482;',
 691     '&alefsym;' => '&#8501;',
 692     '&larr;' => '&#8592;',
 693     '&uarr;' => '&#8593;',
 694     '&rarr;' => '&#8594;',
 695     '&darr;' => '&#8595;',
 696     '&harr;' => '&#8596;',
 697     '&crarr;' => '&#8629;',
 698     '&lArr;' => '&#8656;',
 699     '&uArr;' => '&#8657;',
 700     '&rArr;' => '&#8658;',
 701     '&dArr;' => '&#8659;',
 702     '&hArr;' => '&#8660;',
 703     '&forall;' => '&#8704;',
 704     '&part;' => '&#8706;',
 705     '&exist;' => '&#8707;',
 706     '&empty;' => '&#8709;',
 707     '&nabla;' => '&#8711;',
 708     '&isin;' => '&#8712;',
 709     '&notin;' => '&#8713;',
 710     '&ni;' => '&#8715;',
 711     '&prod;' => '&#8719;',
 712     '&sum;' => '&#8721;',
 713     '&minus;' => '&#8722;',
 714     '&lowast;' => '&#8727;',
 715     '&radic;' => '&#8730;',
 716     '&prop;' => '&#8733;',
 717     '&infin;' => '&#8734;',
 718     '&ang;' => '&#8736;',
 719     '&and;' => '&#8743;',
 720     '&or;' => '&#8744;',
 721     '&cap;' => '&#8745;',
 722     '&cup;' => '&#8746;',
 723     '&int;' => '&#8747;',
 724     '&there4;' => '&#8756;',
 725     '&sim;' => '&#8764;',
 726     '&cong;' => '&#8773;',
 727     '&asymp;' => '&#8776;',
 728     '&ne;' => '&#8800;',
 729     '&equiv;' => '&#8801;',
 730     '&le;' => '&#8804;',
 731     '&ge;' => '&#8805;',
 732     '&sub;' => '&#8834;',
 733     '&sup;' => '&#8835;',
 734     '&nsub;' => '&#8836;',
 735     '&sube;' => '&#8838;',
 736     '&supe;' => '&#8839;',
 737     '&oplus;' => '&#8853;',
 738     '&otimes;' => '&#8855;',
 739     '&perp;' => '&#8869;',
 740     '&sdot;' => '&#8901;',
 741     '&lceil;' => '&#8968;',
 742     '&rceil;' => '&#8969;',
 743     '&lfloor;' => '&#8970;',
 744     '&rfloor;' => '&#8971;',
 745     '&lang;' => '&#9001;',
 746     '&rang;' => '&#9002;',
 747     '&loz;' => '&#9674;',
 748     '&spades;' => '&#9824;',
 749     '&clubs;' => '&#9827;',
 750     '&hearts;' => '&#9829;',
 751     '&diams;' => '&#9830;',
 752     '&quot;' => '&#34;',
 753     '&amp;' => '&#38;',
 754     '&lt;' => '&#60;',
 755     '&gt;' => '&#62;',
 756     '&OElig;' => '&#338;',
 757     '&oelig;' => '&#339;',
 758     '&Scaron;' => '&#352;',
 759     '&scaron;' => '&#353;',
 760     '&Yuml;' => '&#376;',
 761     '&circ;' => '&#710;',
 762     '&tilde;' => '&#732;',
 763     '&ensp;' => '&#8194;',
 764     '&emsp;' => '&#8195;',
 765     '&thinsp;' => '&#8201;',
 766     '&zwnj;' => '&#8204;',
 767     '&zwj;' => '&#8205;',
 768     '&lrm;' => '&#8206;',
 769     '&rlm;' => '&#8207;',
 770     '&ndash;' => '&#8211;',
 771     '&mdash;' => '&#8212;',
 772     '&lsquo;' => '&#8216;',
 773     '&rsquo;' => '&#8217;',
 774     '&sbquo;' => '&#8218;',
 775     '&ldquo;' => '&#8220;',
 776     '&rdquo;' => '&#8221;',
 777     '&bdquo;' => '&#8222;',
 778     '&dagger;' => '&#8224;',
 779     '&Dagger;' => '&#8225;',
 780     '&permil;' => '&#8240;',
 781     '&lsaquo;' => '&#8249;',
 782     '&rsaquo;' => '&#8250;',
 783     '&euro;' => '&#8364;'
 784     );
 785
 786     var $_parser;
 787     var $_stack = array();
 788     var $_errors = array();
 789
 790     function AkXhtmlValidator()
 791     {
 792         $this->_parser = xml_parser_create('');
 793         xml_set_object($this->_parser, &$this);
 794         xml_set_element_handler($this->_parser, 'tagOpen', 'tagClose');
 795         xml_set_character_data_handler($this->_parser, 'cdata');
 796         xml_parser_set_option($this->_parser, XML_OPTION_CASE_FOLDING, false);
 797         xml_parser_set_option($this->_parser, XML_OPTION_TARGET_ENCODING, 'UTF-8');
 798     }
 799
 800     function validateTagAttributes($tag, $attributes = array())
 801     {
 802         $possible_attributes = $this->getPossibleTagAttributes($tag);
 803         foreach($attributes as $attribute => $value) {
 804             if (!in_array($attribute, $possible_attributes)) {
 805                 $this->addError(Ak::t("Attribute %attribute can't be used inside &lt;%tag> tags", array(
 806                 '%attribute' => $attribute,
 807                 '%tag' => $tag
 808                 )) , array(
 809                 array(
 810                 $attribute,
 811                 $tag
 812                 )
 813                 ));
 814             } elseif ($this->doesAttributeNeedsValidation($tag, $attribute)) {
 815                 $this->validateAttribute($tag, $attribute, $value);
 816             }
 817         }
 818     }
 819
 820     function doesAttributeNeedsValidation($tag, $attribute)
 821     {
 822         return isset($this->_tags[$tag]['attributes'][$attribute]) || isset($this->_tags[$tag]['required']) && in_array($attribute, $this->_tags[$tag]['required']);
 823     }
 824
 825     function validateAttribute($tag, $attribute, $value = null)
 826     {
 827         if (isset($this->_tags[$tag]['attributes'][$attribute]) && (strlen($value) > 0)) {
 828             if (!preg_match($this->_tags[$tag]['attributes'][$attribute], $value)) {
 829                 $this->addError(Ak::t("Invalid value on &lt;%tag %attribute=\"%value\"... Valid values must match the pattern \"%pattern\"", array(
 830                 '%tag' => $tag,
 831                 '%attribute' => $attribute,
 832                 '%value' => $value,
 833                 '%pattern' => htmlentities($this->_tags[$tag]['attributes'][$attribute])
 834                 )) , array(
 835                 array(
 836                 $attribute,
 837                 $value
 838                 )
 839                 ));
 840             }
 841         }
 842         if (isset($this->_tags[$tag]['required']) && in_array($attribute, $this->_tags[$tag]['required']) && (strlen($value) == 0)) {
 843             $this->addError(Ak::t("Missing required attribute %attribute on &lt;%tag&gt;", array(
 844             '%tag' => $tag,
 845             '%attribute' => $attribute
 846             )) , array(
 847             array(
 848             $tag,
 849             $attribute
 850             )
 851             ));
 852         }
 853     }
 854
 855     function addError($error, $highlight_text = array())
 856     {
 857         $this->_errors[] = $this->highlightError($error, $highlight_text) .' on line '.$this->getCurrentLine();
 858     }
 859
 860     function highlightError($error, $highlight_text = array())
 861     {
 862         if (empty($highlight_text)) {
 863             return $error;
 864         }
 865         require_once (AK_LIB_DIR.DS.'AkColor.php');
 866         require_once (AK_LIB_DIR.DS.'AkActionView'.DS.'helpers'.DS.'text_helper.php');
 867         $line = $this->getCurrentLine();
 868         $highlighted_error = '';
 869         foreach($highlight_text as $phrases) {
 870             $color = AkColor::getRandomHex();
 871             if (is_array($phrases)) {
 872                 $highlighted_error_line = $error;
 873                 foreach($phrases as $phrase) {
 874                     $this->_linesToHighlight[$line][$error] = array(
 875                     'color' => $color,
 876                     'phrase' => htmlentities($phrase)
 877                     );
 878                     $highlighted_error_line = TextHelper::highlight($highlighted_error_line, $phrase.' ', ' <strong style="border:2px solid #'.$color.'; background: #ffc;">\1</strong> ');
 879                 }
 880                 $highlighted_error.= $highlighted_error_line;
 881             } else {
 882                 $highlighted_error = TextHelper::highlight($error, $phrases.' ', ' <strong style="border:2px solid #'.$color.'; background: #ffc">\1</strong> ');
 883                 $this->_linesToHighlight[$line][$error] = array(
 884                 'color' => $color,
 885                 'phrase' => htmlentities($phrases)
 886                 );
 887             }
 888         }
 889         return $highlighted_error;
 890     }
 891
 892     function highlightErrors($xhtml)
 893     {
 894         $highlighted_xhtml = array();
 895         if (!empty($this->_linesToHighlight)) {
 896             $xhtml_arr = preg_split('/\n|\r/', $xhtml);
 897             foreach($xhtml_arr as $k => $xhtml_line) {
 898                 $pos = $k+$this->_startLine;
 899                 $highlighted_xhtml[$k] = $pos."&nbsp;&nbsp;&nbsp;&nbsp;";
 900                 $xhtml_line = htmlentities($xhtml_line);
 901                 if (isset($this->_linesToHighlight[$pos])) {
 902                     foreach($this->_linesToHighlight[$pos] as $highlight_details) {
 903                         $highlighted_xhtml[$k].= TextHelper::highlight($xhtml_line, $highlight_details['phrase'], '<strong style="border:2px solid #'.$highlight_details['color'].';padding:1px; margin:1px; background: #ffc;">\1</strong>');
 904                     }
 905                 } else {
 906                     $highlighted_xhtml[$k].= $xhtml_line;
 907                 }
 908                 $highlighted_xhtml[$k].= "<br />\n";
 909             }
 910         }
 911         return empty($highlighted_xhtml) ? $xhtml : join($highlighted_xhtml);
 912     }
 913
 914     function getCurrentLine()
 915     {
 916         return xml_get_current_line_number($this->_parser) +$this->_startLine;
 917     }
 918
 919     function hasErrors(&$xhtml)
 920     {
 921         $this->validateUniquenessOfIds();
 922         if (count($this->getErrors()) > 0) {
 923             $xhtml = $this->highlightErrors($xhtml);
 924             return true;
 925         } else {
 926             return false;
 927         }
 928     }
 929
 930     function getErrors()
 931     {
 932         return array_unique($this->_errors);
 933     }
 934
 935     function showErrors()
 936     {
 937         echo '<ul><li>'.join("</li>\n<li>", $this->getErrors()) .'</li></ul>';
 938     }
 939
 940     function getPossibleTagAttributes($tag)
 941     {
 942         static $cache;
 943         if (!isset($cache[$tag])) {
 944             $cache[$tag] = array_unique(array_merge($this->getUniqueAttributesAndEventsForTag($tag) , $this->getDefaultAttributesAndEventsForTag($tag)));
 945             sort($cache[$tag]);
 946         }
 947         return $cache[$tag];
 948     }
 949
 950     function validateRequiredAttributes($tag, $attributes)
 951     {
 952         $compulsory = $this->getCompulsoryTagAttributes($tag);
 953         $errors = array_diff($compulsory, array_keys($attributes));
 954         if (!empty($errors)) {
 955             $this->addError(Ak::t('Tag %tag requires %attributes to be defined', array(
 956             '%tag' => $tag,
 957             '%attributes' => (count($errors) == 1 ? 'attribute "' : 'attributes "') .join('", "', $errors) .'"'
 958             )) , array(
 959             $tag
 960             ));
 961         }
 962     }
 963
 964     function protectFromDuplicatedIds($tag, $attributes)
 965     {
 966         if (isset($attributes['id'])) {
 967             if (isset($this->_idTagXref[$attributes['id']])) {
 968                 $this->addError(Ak::t('Repeating id %id', array(
 969                 '%id' => $attributes['id']
 970                 )) , array(
 971                 $attributes['id']
 972                 ));
 973             }
 974             $this->_tagIdCounter[$attributes['id']] = isset($this->_tagIdCounter[$attributes['id']]) ? $this->_tagIdCounter[$attributes['id']]+1 : 1;
 975             $this->_idTagXref[$attributes['id']][] = $tag;
 976         }
 977     }
 978
 979     function validateUniquenessOfIds()
 980     {
 981         if (isset($this->_tagIdCounter) && max(array_values($this->_tagIdCounter)) > 1) {
 982             foreach($this->_tagIdCounter as $id => $count) {
 983                 if ($count > 1) {
 984                     $this->addError(Ak::t('You have repeated the id %id %count times on your xhtml code. Duplicated Ids found on %tags', array(
 985                     '%id' => "\"$id\"",
 986                     '%count' => $count,
 987                     '%tags' => (count($this->_idTagXref[$id]) == 1 ? 'tag "' : 'tag "') .join('", "', $this->_idTagXref[$id]) .'"'
 988                     )));
 989                 }
 990             }
 991         }
 992     }
 993
 994     function getCompulsoryTagAttributes($tag)
 995     {
 996         return !empty($this->_tags[$tag]['required']) ? (array)$this->_tags[$tag]['required'] : array();
 997     }
 998
 999     function getUniqueAttributesAndEventsForTag($tag)
1000     {
1001         $result = array();
1002         if (isset($this->_tags[$tag]['attributes']) && is_array($this->_tags[$tag]['attributes'])) {
1003             foreach($this->_tags[$tag]['attributes'] as $k => $candidate) {
1004                 $result[] = is_numeric($k) ? $candidate : $k;
1005             }
1006         }
1007         return $result;
1008     }
1009
1010     function getDefaultAttributesAndEventsForTag($tag)
1011     {
1012         $default = array();
1013         if (isset($this->_tags[$tag]) || in_array($tag, $this->_tags)) {
1014             foreach($this->getDefaultAttributesAndEventsForTags() as $defaults) {
1015                 if ((isset($defaults['except']) && in_array($tag, $defaults['except'])) || (isset($defaults['only']) && !in_array($tag, $defaults['only']))) {
1016                     continue;
1017                 }
1018                 foreach(isset($defaults['attributes']) ? $defaults['attributes'] : $defaults['events'] as $k => $candidate) {
1019                     $default[] = is_array($candidate) ? $k : $candidate;;
1020                 }
1021             }
1022         }
1023         return $default;
1024     }
1025
1026     function getDefaultAttributesAndEventsForTags()
1027     {
1028         if (!isset($this->default_values_for_tags)) {
1029             $this->default_values_for_tags = array_merge($this->_attributes, $this->_events);
1030         }
1031         return $this->default_values_for_tags;
1032     }
1033
1034     function getAvailableTags()
1035     {
1036         $tags = array();
1037         foreach(array_keys($this->_tags) as $k) {
1038             $tags[] = is_numeric($k) ? $this->_tags[$k] : $k;
1039         }
1040         sort($tags);
1041         return $tags;
1042     }
1043
1044     function validate(&$xhtml)
1045     {
1046         $this->_startLine = 1;
1047         $xhtml_copy = $this->removeDoctypeHeader($xhtml);
1048         $xhtml_copy = $this->removeCdata($xhtml_copy);
1049         $xhtml_copy = $this->convertLiteralEntitiesToNumericalEntities($xhtml_copy);
1050         $xhtml_copy = '<all>'.$xhtml_copy.'</all>';
1051         if (!xml_parse($this->_parser, $xhtml_copy)) {
1052             $this->addError(Ak::t('XHTML is not well-formed.') .' '.xml_error_string(xml_get_error_code($this->_parser)));
1053         }
1054         return !$this->hasErrors($xhtml);
1055     }
1056
1057     function removeDoctypeHeader($xhtml)
1058     {
1059         if (substr($xhtml, 0, 9) == '<!DOCTYPE') {
1060             $replacement = substr($xhtml, 0, strpos($xhtml, '>'));
1061             $this->_startLine = count(substr_count($replacement, "\n"));
1062         }
1063         return (isset($replacement)) ? substr($xhtml, strlen($replacement)) : $xhtml;
1064     }
1065
1066     function removeCdata($xhtml)
1067     {
1068         $xhtml = preg_replace('(<\!\[CDATA\[(.|\n)*\]\]>)', '', $xhtml);
1069         return str_replace(array('<![CDATA[',']]>') , '', $xhtml);
1070     }
1071
1072
1073     function convertLiteralEntitiesToNumericalEntities($xhtml)
1074     {
1075         return str_replace(array_keys($this->_entities), array_values($this->_entities), $xhtml);
1076     }
1077
1078     function tagOpen($parser, $tag, $attributes)
1079     {
1080         $this->_start_byte = xml_get_current_byte_index($parser);
1081         if ($tag == 'all') {
1082             $this->_stack[] = 'all';
1083             return;
1084         }
1085         $previous = $this->_stack[count($this->_stack) -1];
1086         $this->validateRequiredAttributes($tag, $attributes);
1087         $this->protectFromDuplicatedIds($tag, $attributes);
1088         if (!in_array($previous, $this->getAvailableTags())) {
1089             $this->validateTagAttributes($tag, $attributes);
1090             $this->_stack[] = $tag;
1091             return;
1092         }
1093         if (!in_array($tag, $this->getAvailableTags())) {
1094             $this->addError(Ak::t("Illegal tag: <code>%tag</code>", array(
1095             '%tag' => $tag
1096             )) , array(
1097             $tag
1098             ));
1099             $this->_stack[] = $tag;
1100             return;
1101         }
1102         // Is tag allowed in the current context?
1103         if (!$this->isTagAlowedOnCurrentContext($tag, $previous)) {
1104             if ($previous != 'all') {
1105                 //$this->addError(Ak::t("Tag <code>%tag</code> must occur inside another tag",array('%tag'=>$tag)));
1106                 //} else {
1107                 $this->addError(Ak::t("Tag %tag is not allowed within tag %previous", array(
1108                 '%tag' => $tag,
1109                 '%previous' => $previous
1110                 )) , array(
1111                 $tag
1112                 ));
1113             }
1114         }
1115         $this->validateTagAttributes($tag, $attributes);
1116         $this->_stack[] = $tag;
1117     }
1118
1119     function isTagAlowedOnCurrentContext($tag, $previous)
1120     {
1121         $rules = $this->getRules();
1122         $result = isset($rules[$previous]) ? in_array($tag, $rules[$previous]) : true;
1123         $inverse_rules = $this->getInverseRulesForTag($tag);
1124         $result = isset($inverse_rules[$tag]) ? in_array($previous, $inverse_rules[$tag]) : $result;
1125         return $result;
1126     }
1127
1128     function getRules()
1129     {
1130         static $rules;
1131         if (!isset($rules)) {
1132             //$inline = array ('abbr','cite','code','dfn','em','kbd','object','quote','q','samp','span','strong','var','a','sup','sub','acronym','img','#PCDATA');
1133             $inline = array(
1134             '#pcdata',
1135             'a',
1136             'abbr',
1137             'acronym',
1138             'applet',
1139             'b',
1140             'basefont',
1141             'bdo',
1142             'big',
1143             'br',
1144             'button',
1145             'cite',
1146             'code',
1147             'dfn',
1148             'em',
1149             'font',
1150             'i',
1151             'img',
1152             'input',
1153             'kbd',
1154             'label',
1155             'map',
1156             'object',
1157             'q',
1158             's',
1159             'samp',
1160             'select',
1161             'small',
1162             'span',
1163             'strike',
1164             'strong',
1165             'sub',
1166             'sup',
1167             'textarea',
1168             'tt',
1169             'u',
1170             'var'
1171             );
1172             //$block = array('dl','nl','ol','ul','address','blockcode','blockquote','div','p','pre','handler','section','separator','table');
1173             $block = array(
1174             'address',
1175             'blockcode',
1176             'blockquote',
1177             'center',
1178             'dir',
1179             'div',
1180             'dl',
1181             'fieldset',
1182             'form',
1183             'h1',
1184             'h2',
1185             'h3',
1186             'h4',
1187             'h5',
1188             'h6',
1189             'handler',
1190             'hr',
1191             'iframe',
1192             'isindex',
1193             'menu',
1194             'nl',
1195             'noframes',
1196             'script',
1197             'noscript',
1198             'ol',
1199             'p',
1200             'pre',
1201             'section',
1202             'separator',
1203             'table',
1204             'ul'
1205             );
1206             $flow = array_merge($block, $inline);
1207             $rules = array(
1208             'html' => array(
1209             'head',
1210             'body'
1211             ) ,
1212             'head' => array(
1213             'script',
1214             'style',
1215             'meta',
1216             'base',
1217             'link',
1218             'title'
1219             ) ,
1220             'body' => array_merge(array(
1221             'ins',
1222             'del'
1223             ) , $flow) ,
1224             'ul' => array(
1225             'li'
1226             ) ,
1227             'ol' => array(
1228             'li'
1229             ) ,
1230             //'p' => array_merge($inline, array('blockcode', 'blockquote', 'pre', 'table', 'dl', 'nl', 'ol', 'ul')),
1231             'blockquote' => $block,
1232             'dl' => array(
1233             'dt',
1234             'dd'
1235             ) ,
1236             'pre' => array_diff($inline, array(
1237             'img',
1238             'object',
1239             'big',
1240             'small',
1241             'sub',
1242             'sup'
1243             )) ,
1244             'form' => array_diff($flow, array(
1245             'form'
1246             )) ,
1247             // Tables
1248             'table' => array(
1249             'caption',
1250             'colgroup',
1251             'col',
1252             'thead',
1253             'tbody',
1254             'tr'
1255             ) ,
1256             'colgroup' => array(
1257             'col'
1258             ) ,
1259             'thead' => array(
1260             'tr'
1261             ) ,
1262             'tbody' => array(
1263             'tr'
1264             ) ,
1265             'tr' => array(
1266             'th',
1267             'td'
1268             ) ,
1269             'address' => array_merge($inline, array(
1270             'p'
1271             )) ,
1272             'fieldset' => array_merge($flow, array(
1273             'legend'
1274             )) ,
1275             'a' => array_diff($inline, array(
1276             'a'
1277             )) ,
1278             'object' => array_merge($flow, array(
1279             'param'
1280             )) ,
1281             'map' => array_merge($block, array(
1282             'area'
1283             )) ,
1284             'select' => array(
1285             'optgroup',
1286             'option'
1287             ) ,
1288             'optgroup' => array(
1289             'option'
1290             ) ,
1291             'label' => array_diff($inline, array(
1292             'label'
1293             )) ,
1294             'button' => array_diff($flow, array(
1295             'a',
1296             'input',
1297             'select',
1298             'textarea',
1299             'label',
1300             'button',
1301             'form',
1302             'fieldset',
1303             'iframe'
1304             )) ,
1305             );
1306             $flow_tags = array(
1307             'div',
1308             'center',
1309             'blockquote',
1310             'script',
1311             'noscript',
1312             'dd',
1313             'li',
1314             'th',
1315             'td'
1316             );
1317             foreach($flow_tags as $flow_tag) {
1318                 $rules[$flow_tag] = $flow;
1319             }
1320             $inline_tags = array(
1321             'p',
1322             'h1',
1323             'h2',
1324             'h3',
1325             'h4',
1326             'h5',
1327             'h6',
1328             'dt',
1329             'caption',
1330             'legend',
1331             'tt',
1332             'abbr',
1333             'acronym',
1334             'b',
1335             'bdo',
1336             'big',
1337             'cite',
1338             'code',
1339             'dfn',
1340             'em',
1341             'font',
1342             'i',
1343             'kbd',
1344             'q',
1345             's',
1346             'samp',
1347             'small',
1348             'span',
1349             'strike',
1350             'strong',
1351             'sub',
1352             'sup',
1353             'u',
1354             'var'
1355             );
1356             foreach($inline_tags as $inline_tag) {
1357                 $rules[$inline_tag] = $inline;
1358             }
1359         }
1360         return $rules;
1361     }
1362
1363     function getInverseRulesForTag($tag)
1364     {
1365         static $inverse_rules;
1366         if (!isset($inverse_rules[$tag])) {
1367             $inverse_rules[$tag] = array();
1368             $rules = $this->getRules();
1369             foreach($rules as $container_tag => $rule) {
1370                 if (in_array($tag, $rule)) {
1371                     $inverse_rules[$tag][] = $container_tag;
1372                 }
1373             }
1374         }
1375         return $inverse_rules[$tag];
1376     }
1377
1378     function cdata($parser, $cdata)
1379     {
1380         // Simply check that the 'previous' tag allows CDATA
1381         $previous = $this->_stack[count($this->_stack) -1];
1382         if ($cdata != '' && in_array($previous, array(
1383         'base',
1384         'area',
1385         'basefont',
1386         'br',
1387         'col',
1388         'hr',
1389         'img',
1390         'input',
1391         'link',
1392         'meta',
1393         'param'
1394         ))) {
1395             $this->addError(Ak::t("%previous tag is not a content tag. close it like this '&lt;%previous /&gt;'", array(
1396             '%previous' => $previous
1397             )) , array(
1398             $previous
1399             ));
1400         }
1401         // If previous tag is illegal, no point in running test
1402         if (!in_array($previous, $this->getAvailableTags())) {
1403             return;
1404         }
1405         if (trim($cdata) != '') {
1406             if (!$this->isTagAlowedOnCurrentContext('#pcdata', $previous)) {
1407                 $this->addError(Ak::t("Tag <code>%previous</code> may not contain raw character data", array(
1408                 '%previous' => $previous
1409                 )) , array(
1410                 $previous
1411                 ));
1412             }
1413         }
1414     }
1415
1416     function tagClose($parser, $tag)
1417     {
1418         if (in_array($tag, array(
1419         'base',
1420         'area',
1421         'basefont',
1422         'br',
1423         'col',
1424         'hr',
1425         'img',
1426         'input',
1427         'link',
1428         'meta',
1429         'param'
1430         ))) {
1431             $this->_end_byte = xml_get_current_byte_index($parser);
1432             if ($this->_end_byte-$this->_start_byte == 4) {
1433                 $this->addError(Ak::t("%tag tag is not a content tag. close it like this '&lt;%tag /&gt;'", array(
1434                 '%tag' => $tag
1435                 )) , array(
1436                 $tag
1437                 ));
1438             }
1439         }
1440         array_pop($this->_stack);
1441     }
1442
1443 }
1444
1445 ?>