lib/plugin/CreateToc.php

   1 <?php // -*-php-*-
   2 rcs_id('$Id$');
   3 /*
   4  * Copyright 2004,2005 $ThePhpWikiProgrammingTeam
   5  * Copyright 2008-2009 Marc-Etienne Vargenau, Alcatel-Lucent
   6  *
   7  * This file is part of PhpWiki.
   8  *
   9  * PhpWiki is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * PhpWiki is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License
  20  * along with PhpWiki; if not, write to the Free Software
  21  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  22  */
  23
  24 /**
  25  * CreateToc:  Create a Table of Contents and automatically link to headers
  26  *
  27  * Usage:
  28  *  <<CreateToc arguments>>
  29  * @author:  Reini Urban, Marc-Etienne Vargenau
  30  *
  31  * Known problems:
  32  * - MacIE will not work with jshide.
  33  * - it will crash with old markup and Apache2 (?)
  34  * - Certain corner-edges will not work with TOC_FULL_SYNTAX.
  35  *   I believe I fixed all of them now, but who knows?
  36  * - bug #969495 "existing labels not honored" seems to be fixed.
  37  * - some constructs might incorrectly be recognized as a header
  38  *   (e.g. lines starting with "!!!" or "==" inside <verbatim>)
  39  */
  40
  41 if (!defined('TOC_FULL_SYNTAX'))
  42     define('TOC_FULL_SYNTAX', 1);
  43
  44 class WikiPlugin_CreateToc
  45 extends WikiPlugin
  46 {
  47     function getName() {
  48         return _("CreateToc");
  49     }
  50
  51     function getDescription() {
  52         return _("Create a Table of Contents and automatically link to headers");
  53     }
  54
  55     function getVersion() {
  56         return preg_replace("/[Revision: $]/", '',
  57                             "\$Revision$");
  58     }
  59
  60     function getDefaultArguments() {
  61         return array('extracollapse' => 1,            // provide an entry +/- link to collapse
  62                      'firstlevelstyle' => 'number',   // 'number', 'letter' or 'roman'
  63                      'headers'   =>  "1,2,3,4,5",     // "!!!"=>h2, "!!"=>h3, "!"=>h4
  64                                                       // "1"=>h2, "2"=>h3, "3"=>h4, "4"=>h5, "5"=>h6
  65                      'indentstr' => '&nbsp;&nbsp;',
  66                      'jshide'    => 0,                // collapsed TOC as DHTML button
  67                      'liststyle' => 'dl',             // 'dl' or 'ul' or 'ol'
  68                      'noheader'  => 0,                // omit "Table of Contents" header
  69                      'notoc'     => 0,                // do not display TOC, only number headers
  70                      'pagename'  => '[pagename]',     // TOC of another page here?
  71                      'position'  => 'full',           // full, right or left
  72                      'width'     => '200px',
  73                      'with_counter' => 0,
  74                      'with_toclink' => 0,             // link back to TOC
  75                     );
  76     }
  77     // Initialisation of toc counter
  78     function _initTocCounter() {
  79         $counter = array(1=>0, 2=>0, 3=>0, 4=>0, 5=>0);
  80         return $counter;
  81     }
  82
  83     // Update toc counter with a new title
  84     function _tocCounter(&$counter, $level) {
  85         $counter[$level]++;
  86         for($i = $level+1; $i <= 5; $i++) {
  87             $counter[$i] = 0;
  88         }
  89     }
  90
  91     function _roman_counter($number) {
  92
  93         $n = intval($number);
  94         $result = '';
  95
  96         $lookup = array('C' => 100, 'XC' => 90, 'L' => 50, 'XL' => 40,
  97                         'X' => 10, 'IX' => 9, 'V' => 5, 'IV' => 4, 'I' => 1);
  98
  99         foreach ($lookup as $roman => $value) {
 100             $matches = intval($n / $value);
 101             $result .= str_repeat($roman, $matches);
 102             $n = $n % $value;
 103         }
 104         return $result;
 105     }
 106
 107     function _letter_counter($number) {
 108         if ($number <= 26) {
 109             return chr(ord("A") + $number - 1);
 110         } else {
 111             return chr(ord("A") + ($number/26) - 1) . chr(ord("A") + ($number%26));
 112         }
 113     }
 114
 115     // Get string corresponding to the current title
 116     function _getCounter(&$counter, $level, $firstlevelstyle) {
 117         if ($firstlevelstyle == 'roman') {
 118             $str= $this->_roman_counter($counter[1]);
 119         } else if ($firstlevelstyle == 'letter') {
 120             $str= $this->_letter_counter($counter[1]);
 121         } else {
 122             $str=$counter[1];
 123         }
 124         for($i = 2; $i <= 5; $i++) {
 125             if($counter[$i] != 0)
 126                 $str .= '.'.$counter[$i];
 127         }
 128         return $str;
 129     }
 130
 131     function preg_quote ($heading) {
 132         return str_replace(array("/",".","?","*"),
 133                                array('\/','\.','\?','\*'), $heading);
 134     }
 135
 136     // Get HTML header corresponding to current level (level is set of ! or =)
 137     function _getHeader($level) {
 138
 139         $count = substr_count($level,'!');
 140         switch ($count) {
 141             case 3: return "h2";
 142             case 2: return "h3";
 143             case 1: return "h4";
 144         }
 145         $count = substr_count($level,'=');
 146         switch ($count) {
 147             case 2: return "h2";
 148             case 3: return "h3";
 149             case 4: return "h4";
 150             case 5: return "h5";
 151             case 6: return "h6";
 152         }
 153         return "";
 154     }
 155
 156     function _quote($heading) {
 157         if (TOC_FULL_SYNTAX ) {
 158             $theading = TransformInline($heading);
 159             if ($theading)
 160                 return preg_quote($theading->asXML(), "/");
 161             else
 162                 return XmlContent::_quote(preg_quote($heading, "/"));
 163         } else {
 164             return XmlContent::_quote(preg_quote($heading, "/"));
 165         }
 166     }
 167
 168     /*
 169      * @param $hstart id (in $content) of heading start
 170      * @param $hend   id (in $content) of heading end
 171      */
 172     function searchHeader ($content, $start_index, $heading,
 173                            $level, &$hstart, &$hend, $basepage=false) {
 174         $hstart = 0;
 175         $hend = 0;
 176             $h = $this->_getHeader($level);
 177         $qheading = $this->_quote($heading);
 178             for ($j=$start_index; $j < count($content); $j++) {
 179             if (is_string($content[$j])) {
 180                     if (preg_match("/<$h>$qheading<\/$h>/",
 181                                    $content[$j]))
 182                         return $j;
 183             }
 184             elseif (isa($content[$j], 'cached_link'))
 185             {
 186                 if (method_exists($content[$j],'asXML')) {
 187                     $content[$j]->_basepage = $basepage;
 188                     $content[$j] = $content[$j]->asXML();
 189                 } else
 190                     $content[$j] = $content[$j]->asString();
 191                 // shortcut for single wikiword or link headers
 192                 if ($content[$j] == $heading
 193                     and substr($content[$j-1],-4,4) == "<$h>"
 194                     and substr($content[$j+1],0,5) == "</$h>")
 195                 {
 196                     $hstart = $j-1;
 197                     $hend = $j+1;
 198                     return $j; // single wikiword
 199                 }
 200                 elseif (TOC_FULL_SYNTAX) {
 201                     //DONE: To allow "!! WikiWord link" or !! http://anylink/
 202                     // Check against joined content (after cached_plugininvocation).
 203                     // The first link is the anchor then.
 204                     if (preg_match("/<$h>(?!.*<\/$h>)/", $content[$j-1])) {
 205                         $hstart = $j-1;
 206                         $joined = '';
 207                         for ($k=max($j-1,$start_index);$k<count($content);$k++) {
 208                             if (is_string($content[$k]))
 209                                 $joined .= $content[$k];
 210                             elseif (method_exists($content[$k],'asXML'))
 211                                 $joined .= $content[$k]->asXML();
 212                             else
 213                                 $joined .= $content[$k]->asString();
 214                             if (preg_match("/<$h>$qheading<\/$h>/",$joined)) {
 215                                 $hend=$k;
 216                                 return $k;
 217                             }
 218                         }
 219                     }
 220                 }
 221                 }
 222             }
 223             trigger_error("Heading <$h> $heading </$h> not found\n", E_USER_NOTICE);
 224             return 0;
 225     }
 226
 227     /** prevent from duplicate anchors,
 228      *  beautify spaces: " " => "_" and not "x20."
 229      */
 230     function _nextAnchor($s) {
 231         static $anchors = array();
 232
 233         $s = str_replace(' ','_',$s);
 234         $i = 1;
 235         $anchor = $s;
 236         while (!empty($anchors[$anchor])) {
 237             $anchor = sprintf("%s_%d",$s,$i++);
 238         }
 239         $anchors[$anchor] = $i;
 240         return $anchor;
 241     }
 242
 243     // We have to find headers in both:
 244     // - classic Phpwiki syntax (lines starting with "!", "!!" or "!!!")
 245     // - Wikicreole syntax (lines starting with "==", "===", etc.)
 246     // We must omit lines starting with "!" if inside a Mediawiki table
 247     // (they represent a table header)
 248     // Some constructs might incorrectly be recognized as a header
 249     // (e.g. lines starting with "!!!" or "==" inside <verbatim>)
 250     // Feature request: proper nesting; multiple levels (e.g. 1,3)
 251     function extractHeaders (&$content, &$markup, $backlink=0,
 252                              $counter=0, $levels=false, $firstlevelstyle='number', $basepage='')
 253     {
 254
 255         if (!$levels) $levels = array(1,2);
 256         $tocCounter = $this->_initTocCounter();
 257         reset($levels);
 258         sort($levels);
 259         $headers = array();
 260         $j = 0;
 261         $insidetable = false;
 262         for ($i=0; $i<count($content); $i++) {
 263             if (preg_match('/^\s*{\|/', $content[$i])) {
 264                $insidetable = true;
 265                continue;
 266             }
 267             if (preg_match('/^\s*\|}/', $content[$i])) {
 268                $insidetable = false;
 269                continue;
 270             }
 271             if ($insidetable) {
 272                continue;
 273             }
 274             foreach ($levels as $level) {
 275                 if ($level < 1 or $level > 5) continue;
 276                 $phpwikiclassiclevel = 4 -$level;
 277                 $wikicreolelevel = $level + 1;
 278                 if ($phpwikiclassiclevel < 1 or $phpwikiclassiclevel > 3) continue;
 279                 if ((preg_match('/^\s*(!{'.$phpwikiclassiclevel.','.$phpwikiclassiclevel.'})([^!].*)$/', $content[$i], $match))
 280                  or (preg_match('/^\s*(={'.$wikicreolelevel.','.$wikicreolelevel.'})([^=].*)$/', $content[$i], $match)) )
 281                 {
 282                     $this->_tocCounter($tocCounter, $level);
 283                     if (!strstr($content[$i],'#[')) {
 284                         $s = trim($match[2]);
 285                         // If it is Wikicreole syntax, remove '='s at the end
 286                         if (string_starts_with($match[1], "=")) {
 287                             $s = trim($s, "=");
 288                             $s = trim($s);
 289                         }
 290                         $anchor = $this->_nextAnchor($s);
 291                         $manchor = MangleXmlIdentifier($anchor);
 292                         $texts = $s;
 293                         if($counter) {
 294                             $texts = $this->_getCounter($tocCounter, $level, $firstlevelstyle).' '.$s;
 295                         }
 296                         $headers[] = array('text' => $texts,
 297                                            'anchor' => $anchor,
 298                                            'level' => $level);
 299                         // Change original wikitext, but that is useless art...
 300                         $content[$i] = $match[1]." #[|$manchor][$s|#TOC]";
 301                         // And now change the to be printed markup (XmlTree):
 302                         // Search <hn>$s</hn> line in markup
 303                         /* Url for backlink */
 304                         $url = WikiURL(new WikiPageName($basepage,false,"TOC"));
 305
 306                         $j = $this->searchHeader($markup->_content, $j, $s,
 307                                                  $match[1], $hstart, $hend,
 308                                                  $markup->_basepage);
 309                         if ($j and isset($markup->_content[$j])) {
 310                             $x = $markup->_content[$j];
 311                             $qheading = $this->_quote($s);
 312                             if ($counter)
 313                                  $counterString = $this->_getCounter($tocCounter, $level, $firstlevelstyle);
 314                             if (($hstart === 0) && is_string($markup->_content[$j])) {
 315                                 if ($backlink) {
 316                                     if ($counter)
 317                                         $anchorString = "<a href=\"$url\" name=\"$manchor\">$counterString</a> - \$2";
 318                                     else
 319                                         $anchorString = "<a href=\"$url\" name=\"$manchor\">\$2</a>";
 320                                 } else {
 321                                     $anchorString = "<a name=\"$manchor\"></a>";
 322                                     if ($counter)
 323                                         $anchorString .= "$counterString - ";
 324                                 }
 325                                 if ($x = preg_replace('/(<h\d>)('.$qheading.')(<\/h\d>)/',
 326                                                       "\$1$anchorString\$2\$3",$x,1)) {
 327                                     if ($backlink) {
 328                                         $x = preg_replace('/(<h\d>)('.$qheading.')(<\/h\d>)/',
 329                                                           "\$1$anchorString\$3",
 330                                                           $markup->_content[$j],1);
 331                                     }
 332                                     $markup->_content[$j] = $x;
 333                                 }
 334                             } else {
 335                                 $x = $markup->_content[$hstart];
 336                                 $h = $this->_getHeader($match[1]);
 337
 338                                 if ($backlink) {
 339                                     if ($counter) {
 340                                         $anchorString = "\$1<a href=\"$url\" name=\"$manchor\">$counterString</a> - ";
 341                                     } else {
 342                                         /* Not possible to make a backlink on a
 343                                          * title with a WikiWord */
 344                                         $anchorString = "\$1<a name=\"$manchor\"></a>";
 345                                     }
 346                                 }
 347                                 else {
 348                                     $anchorString = "\$1<a name=\"$manchor\"></a>";
 349                                     if ($counter)
 350                                         $anchorString .= "$counterString - ";
 351                                 }
 352                                 $x = preg_replace("/(<$h>)(?!.*<\/$h>)/",
 353                                                   $anchorString, $x, 1);
 354                                 if ($backlink) {
 355                                     $x =  preg_replace("/(<$h>)(?!.*<\/$h>)/",
 356                                                       $anchorString,
 357                                                       $markup->_content[$hstart],1);
 358                                 }
 359                                 $markup->_content[$hstart] = $x;
 360                             }
 361                         }
 362                     }
 363                 }
 364             }
 365         }
 366         return $headers;
 367     }
 368
 369     function run($dbi, $argstr, &$request, $basepage) {
 370         global $WikiTheme;
 371         extract($this->getArgs($argstr, $request));
 372         if ($pagename) {
 373             // Expand relative page names.
 374             $page = new WikiPageName($pagename, $basepage);
 375             $pagename = $page->name;
 376         }
 377         if (!$pagename) {
 378             return $this->error(_("no page specified"));
 379         }
 380         if ($jshide and isBrowserIE() and browserDetect("Mac")) {
 381             //trigger_error(_("jshide set to 0 on Mac IE"), E_USER_NOTICE);
 382             $jshide = 0;
 383         }
 384         if (($notoc) or ($liststyle == 'ol')) {
 385             $with_counter = 1;
 386         }
 387
 388         // Check if user is allowed to get the Page.
 389         if (!mayAccessPage ('view', $pagename)) {
 390                 return $this->error(sprintf(_("Illegal access to page %s: no read access"),
 391                 $pagename));
 392         }
 393
 394         $page = $dbi->getPage($pagename);
 395         $current = $page->getCurrentRevision();
 396         //FIXME: I suspect this only to crash with Apache2
 397         if (!$current->get('markup') or $current->get('markup') < 2) {
 398             if (in_array(php_sapi_name(),array('apache2handler','apache2filter'))) {
 399                 trigger_error(_("CreateToc disabled for old markup"), E_USER_WARNING);
 400                 return '';
 401             }
 402         }
 403         $content = $current->getContent();
 404         $html = HTML::div(array('class' => 'toc', 'id'=> GenerateId("toc")));
 405         if ($notoc) {
 406             $html->setAttr('style','display:none;');
 407         }
 408         if (($position == "left") or ($position == "right")) {
 409             $html->setAttr('style','float:'.$position.'; width:'.$width.';');
 410         }
 411         $toclistid = GenerateId("toclist");
 412         $list = HTML::div(array('id'=>$toclistid, 'class'=>'toclist'));
 413         if (!strstr($headers,",")) {
 414             $headers = array($headers);
 415         } else {
 416             $headers = explode(",",$headers);
 417         }
 418         $levels = array();
 419         foreach ($headers as $h) {
 420             //replace !!! with level 1, ...
 421             if (strstr($h,"!")) {
 422                 $hcount = substr_count($h,'!');
 423                 $level = min(max(1, $hcount),3);
 424                 $levels[] = $level;
 425             } else {
 426                 $level = min(max(1, (int) $h), 5);
 427                 $levels[] = $level;
 428             }
 429         }
 430         if (TOC_FULL_SYNTAX)
 431             require_once("lib/InlineParser.php");
 432         if ($headers = $this->extractHeaders($content, $dbi->_markup,
 433                                              $with_toclink, $with_counter,
 434                                              $levels, $firstlevelstyle, $basepage))
 435         {
 436             foreach ($headers as $h) {
 437                 // proper heading indent
 438                 $level = $h['level'];
 439                 $indent = $level - 1;
 440                 $link = new WikiPageName($pagename,$page,$h['anchor']);
 441                 $li = WikiLink($link,'known',$h['text']);
 442                 // Hack to suppress pagename before #
 443                 // $li->_attr["href"] = strstr($li->_attr["href"], '#');
 444                 $list->pushContent(HTML::p(HTML::raw
 445                        (str_repeat($indentstr,$indent)),$li));
 446             }
 447         }
 448         $list->setAttr('style','display:'.($jshide?'none;':'block;'));
 449         $open = DATA_PATH.'/'.$WikiTheme->_findFile("images/folderArrowOpen.png");
 450         $close = DATA_PATH.'/'.$WikiTheme->_findFile("images/folderArrowClosed.png");
 451       if ($noheader) {
 452       } else {
 453         $toctoggleid = GenerateId("toctoggle");
 454         if ($extracollapse)
 455             $toclink = HTML(_("Table of Contents"),
 456                             " ",
 457                             HTML::a(array('name'=>'TOC')),
 458                             HTML::img(array(
 459                                             'id'=>$toctoggleid,
 460                                             'class'=>'wikiaction',
 461                                             'title'=>_("Click to display to TOC"),
 462                                             'onclick'=>"toggletoc(this, '".$open."', '".$close."', '".$toclistid."')",
 463                                             'border' => 0,
 464                                             'alt' => 'toctoggle',
 465                                             'src' => $jshide ? $close : $open )));
 466         else
 467             $toclink = HTML::a(array('name'=>'TOC',
 468                                      'class'=>'wikiaction',
 469                                      'title'=>_("Click to display"),
 470                                      'onclick'=>"toggletoc(this, '".$open."', '".$close."', '".$toclistid."')"),
 471                                _("Table of Contents"),
 472                                HTML::span(array('style'=>'display:none',
 473                                                 'id'=>$toctoggleid)," "));
 474         $html->pushContent(HTML::p(array('class'=>'toctitle'), $toclink));
 475       }
 476       $html->pushContent($list);
 477       if (count($headers) == 0) {
 478           // Do not display an empty TOC
 479           $html->setAttr('style','display:none;');
 480       }
 481       return $html;
 482     }
 483 };
 484
 485 // For emacs users
 486 // Local Variables:
 487 // mode: php
 488 // tab-width: 8
 489 // c-basic-offset: 4
 490 // c-hanging-comment-ender-p: nil
 491 // indent-tabs-mode: nil
 492 // End:
 493 ?>