lib/plugin/CreateToc.php

   1 <?php // -*-php-*-
   2 // rcs_id('$Id$');
   3 /*
   4  * Copyright 2004,2005 $ThePhpWikiProgrammingTeam
   5  * Copyright 2008-2010 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 getDefaultArguments() {
  56         return array('extracollapse' => 1,            // provide an entry +/- link to collapse
  57                      'firstlevelstyle' => 'number',   // 'number', 'letter' or 'roman'
  58                      'headers'   =>  "1,2,3,4,5",     // "!!!"=>h2, "!!"=>h3, "!"=>h4
  59                                                       // "1"=>h2, "2"=>h3, "3"=>h4, "4"=>h5, "5"=>h6
  60                      'indentstr' => '&nbsp;&nbsp;',
  61                      'jshide'    => 0,                // collapsed TOC as DHTML button
  62                      'liststyle' => 'dl',             // 'dl' or 'ul' or 'ol'
  63                      'noheader'  => 0,                // omit "Table of Contents" header
  64                      'notoc'     => 0,                // do not display TOC, only number headers
  65                      'pagename'  => '[pagename]',     // TOC of another page here?
  66                      'position'  => 'full',           // full, right or left
  67                      'width'     => '200px',
  68                      'with_counter' => 0,
  69                      'with_toclink' => 0,             // link back to TOC
  70                     );
  71     }
  72     // Initialisation of toc counter
  73     function _initTocCounter() {
  74         $counter = array(1=>0, 2=>0, 3=>0, 4=>0, 5=>0);
  75         return $counter;
  76     }
  77
  78     // Update toc counter with a new title
  79     function _tocCounter(&$counter, $level) {
  80         $counter[$level]++;
  81         for($i = $level+1; $i <= 5; $i++) {
  82             $counter[$i] = 0;
  83         }
  84     }
  85
  86     function _roman_counter($number) {
  87
  88         $n = intval($number);
  89         $result = '';
  90
  91         $lookup = array('C' => 100, 'XC' => 90, 'L' => 50, 'XL' => 40,
  92                         'X' => 10, 'IX' => 9, 'V' => 5, 'IV' => 4, 'I' => 1);
  93
  94         foreach ($lookup as $roman => $value) {
  95             $matches = intval($n / $value);
  96             $result .= str_repeat($roman, $matches);
  97             $n = $n % $value;
  98         }
  99         return $result;
 100     }
 101
 102     function _letter_counter($number) {
 103         if ($number <= 26) {
 104             return chr(ord("A") + $number - 1);
 105         } else {
 106             return chr(ord("A") + ($number/26) - 1) . chr(ord("A") + ($number%26));
 107         }
 108     }
 109
 110     // Get string corresponding to the current title
 111     function _getCounter(&$counter, $level, $firstlevelstyle) {
 112         if ($firstlevelstyle == 'roman') {
 113             $str= $this->_roman_counter($counter[1]);
 114         } else if ($firstlevelstyle == 'letter') {
 115             $str= $this->_letter_counter($counter[1]);
 116         } else {
 117             $str=$counter[1];
 118         }
 119         for($i = 2; $i <= 5; $i++) {
 120             if($counter[$i] != 0)
 121                 $str .= '.'.$counter[$i];
 122         }
 123         return $str;
 124     }
 125
 126     function preg_quote ($heading) {
 127         return str_replace(array("/",".","?","*"),
 128                                array('\/','\.','\?','\*'), $heading);
 129     }
 130
 131     // Get HTML header corresponding to current level (level is set of ! or =)
 132     function _getHeader($level) {
 133
 134         $count = substr_count($level,'!');
 135         switch ($count) {
 136             case 3: return "h2";
 137             case 2: return "h3";
 138             case 1: return "h4";
 139         }
 140         $count = substr_count($level,'=');
 141         switch ($count) {
 142             case 2: return "h2";
 143             case 3: return "h3";
 144             case 4: return "h4";
 145             case 5: return "h5";
 146             case 6: return "h6";
 147         }
 148         return "";
 149     }
 150
 151     function _quote($heading) {
 152         if (TOC_FULL_SYNTAX ) {
 153             $theading = TransformInline($heading);
 154             if ($theading)
 155                 return preg_quote($theading->asXML(), "/");
 156             else
 157                 return XmlContent::_quote(preg_quote($heading, "/"));
 158         } else {
 159             return XmlContent::_quote(preg_quote($heading, "/"));
 160         }
 161     }
 162
 163     /*
 164      * @param $hstart id (in $content) of heading start
 165      * @param $hend   id (in $content) of heading end
 166      */
 167     function searchHeader ($content, $start_index, $heading,
 168                            $level, &$hstart, &$hend, $basepage=false) {
 169         $hstart = 0;
 170         $hend = 0;
 171             $h = $this->_getHeader($level);
 172         $qheading = $this->_quote($heading);
 173             for ($j=$start_index; $j < count($content); $j++) {
 174             if (is_string($content[$j])) {
 175                     if (preg_match("/<$h>$qheading<\/$h>/",
 176                                    $content[$j]))
 177                         return $j;
 178             }
 179             elseif (isa($content[$j], 'cached_link'))
 180             {
 181                 if (method_exists($content[$j],'asXML')) {
 182                     $content[$j]->_basepage = $basepage;
 183                     $content[$j] = $content[$j]->asXML();
 184                 } else
 185                     $content[$j] = $content[$j]->asString();
 186                 // shortcut for single wikiword or link headers
 187                 if ($content[$j] == $heading
 188                     and substr($content[$j-1],-4,4) == "<$h>"
 189                     and substr($content[$j+1],0,5) == "</$h>")
 190                 {
 191                     $hstart = $j-1;
 192                     $hend = $j+1;
 193                     return $j; // single wikiword
 194                 }
 195                 elseif (TOC_FULL_SYNTAX) {
 196                     //DONE: To allow "!! WikiWord link" or !! http://anylink/
 197                     // Check against joined content (after cached_plugininvocation).
 198                     // The first link is the anchor then.
 199                     if (preg_match("/<$h>(?!.*<\/$h>)/", $content[$j-1])) {
 200                         $hstart = $j-1;
 201                         $joined = '';
 202                         for ($k=max($j-1,$start_index);$k<count($content);$k++) {
 203                             if (is_string($content[$k]))
 204                                 $joined .= $content[$k];
 205                             elseif (method_exists($content[$k],'asXML'))
 206                                 $joined .= $content[$k]->asXML();
 207                             else
 208                                 $joined .= $content[$k]->asString();
 209                             if (preg_match("/<$h>$qheading<\/$h>/",$joined)) {
 210                                 $hend=$k;
 211                                 return $k;
 212                             }
 213                         }
 214                     }
 215                 }
 216                 }
 217             }
 218             trigger_error("Heading <$h> $heading </$h> not found\n", E_USER_NOTICE);
 219             return 0;
 220     }
 221
 222     /** prevent from duplicate anchors,
 223      *  beautify spaces: " " => "_" and not "x20."
 224      */
 225     function _nextAnchor($s) {
 226         static $anchors = array();
 227
 228         $s = str_replace(' ','_',$s);
 229         $i = 1;
 230         $anchor = $s;
 231         while (!empty($anchors[$anchor])) {
 232             $anchor = sprintf("%s_%d",$s,$i++);
 233         }
 234         $anchors[$anchor] = $i;
 235         return $anchor;
 236     }
 237
 238     // We have to find headers in both:
 239     // - classic Phpwiki syntax (lines starting with "!", "!!" or "!!!")
 240     // - Wikicreole syntax (lines starting with "==", "===", etc.)
 241     // We must omit lines starting with "!" if inside a Mediawiki table
 242     // (they represent a table header)
 243     // Some constructs might incorrectly be recognized as a header
 244     // (e.g. lines starting with "!!!" or "==" inside <verbatim>)
 245     // Feature request: proper nesting; multiple levels (e.g. 1,3)
 246     function extractHeaders (&$content, &$markup, $backlink=0,
 247                              $counter=0, $levels=false, $firstlevelstyle='number', $basepage='')
 248     {
 249         if (!$levels) $levels = array(1,2);
 250         $tocCounter = $this->_initTocCounter();
 251         reset($levels);
 252         sort($levels);
 253         $headers = array();
 254         $j = 0;
 255         $insidetable = false;
 256         for ($i=0; $i<count($content); $i++) {
 257             if (preg_match('/^\s*{\|/', $content[$i])) {
 258                $insidetable = true;
 259                continue;
 260             }
 261             if (preg_match('/^\s*\|}/', $content[$i])) {
 262                $insidetable = false;
 263                continue;
 264             }
 265             if ($insidetable) {
 266                continue;
 267             }
 268             foreach ($levels as $level) {
 269                 if ($level < 1 or $level > 5) continue;
 270                 $phpwikiclassiclevel = 4 -$level;
 271                 $wikicreolelevel = $level + 1;
 272                 $trim = trim($content[$i]);
 273
 274                 if ((((strpos($trim, '=') === 0))
 275                       && (preg_match('/^\s*(={'.$wikicreolelevel.','.$wikicreolelevel.'})([^=].*)$/', $content[$i], $match)))
 276                    or (((strpos($trim, '!') === 0))
 277                       && ((preg_match('/^\s*(!{'.$phpwikiclassiclevel.','.$phpwikiclassiclevel.'})([^!].*)$/', $content[$i], $match))))) {
 278                     $this->_tocCounter($tocCounter, $level);
 279                     if (!strstr($content[$i],'#[')) {
 280                         $s = trim($match[2]);
 281                         // If it is Wikicreole syntax, remove '='s at the end
 282                         if (string_starts_with($match[1], "=")) {
 283                             $s = trim($s, "=");
 284                             $s = trim($s);
 285                         }
 286                         $anchor = $this->_nextAnchor($s);
 287                         $manchor = MangleXmlIdentifier($anchor);
 288                         $texts = $s;
 289                         if($counter) {
 290                             $texts = $this->_getCounter($tocCounter, $level, $firstlevelstyle).' '.$s;
 291                         }
 292                         $headers[] = array('text' => $texts,
 293                                            'anchor' => $anchor,
 294                                            'level' => $level);
 295                         // Change original wikitext, but that is useless art...
 296                         $content[$i] = $match[1]." #[|$manchor][$s|#TOC]";
 297                         // And now change the to be printed markup (XmlTree):
 298                         // Search <hn>$s</hn> line in markup
 299                         /* Url for backlink */
 300                         $url = WikiURL(new WikiPageName($basepage,false,"TOC"));
 301
 302                         $j = $this->searchHeader($markup->_content, $j, $s,
 303                                                  $match[1], $hstart, $hend,
 304                                                  $markup->_basepage);
 305                         if ($j and isset($markup->_content[$j])) {
 306                             $x = $markup->_content[$j];
 307                             $qheading = $this->_quote($s);
 308                             if ($counter)
 309                                  $counterString = $this->_getCounter($tocCounter, $level, $firstlevelstyle);
 310                             if (($hstart === 0) && is_string($markup->_content[$j])) {
 311                                 if ($backlink) {
 312                                     if ($counter)
 313                                         $anchorString = "<a href=\"$url\" id=\"$manchor\">$counterString</a> - \$2";
 314                                     else
 315                                         $anchorString = "<a href=\"$url\" id=\"$manchor\">\$2</a>";
 316                                 } else {
 317                                     $anchorString = "<a id=\"$manchor\"></a>";
 318                                     if ($counter)
 319                                         $anchorString .= "$counterString - ";
 320                                 }
 321                                 if ($x = preg_replace('/(<h\d>)('.$qheading.')(<\/h\d>)/',
 322                                                       "\$1$anchorString\$2\$3",$x,1)) {
 323                                     if ($backlink) {
 324                                         $x = preg_replace('/(<h\d>)('.$qheading.')(<\/h\d>)/',
 325                                                           "\$1$anchorString\$3",
 326                                                           $markup->_content[$j],1);
 327                                     }
 328                                     $markup->_content[$j] = $x;
 329                                 }
 330                             } else {
 331                                 $x = $markup->_content[$hstart];
 332                                 $h = $this->_getHeader($match[1]);
 333
 334                                 if ($backlink) {
 335                                     if ($counter) {
 336                                         $anchorString = "\$1<a href=\"$url\" id=\"$manchor\">$counterString</a> - ";
 337                                     } else {
 338                                         /* Not possible to make a backlink on a
 339                                          * title with a WikiWord */
 340                                         $anchorString = "\$1<a id=\"$manchor\"></a>";
 341                                     }
 342                                 }
 343                                 else {
 344                                     $anchorString = "\$1<a id=\"$manchor\"></a>";
 345                                     if ($counter)
 346                                         $anchorString .= "$counterString - ";
 347                                 }
 348                                 $x = preg_replace("/(<$h>)(?!.*<\/$h>)/",
 349                                                   $anchorString, $x, 1);
 350                                 if ($backlink) {
 351                                     $x =  preg_replace("/(<$h>)(?!.*<\/$h>)/",
 352                                                       $anchorString,
 353                                                       $markup->_content[$hstart],1);
 354                                 }
 355                                 $markup->_content[$hstart] = $x;
 356                             }
 357                         }
 358                     }
 359                 }
 360             }
 361         }
 362         return $headers;
 363     }
 364
 365     function run($dbi, $argstr, &$request, $basepage) {
 366         global $WikiTheme;
 367         extract($this->getArgs($argstr, $request));
 368         if ($pagename) {
 369             // Expand relative page names.
 370             $page = new WikiPageName($pagename, $basepage);
 371             $pagename = $page->name;
 372         }
 373         if (!$pagename) {
 374             return $this->error(_("no page specified"));
 375         }
 376         if (isBrowserIE() and browserDetect("Mac")) {
 377             $jshide = 0;
 378         }
 379         if (($notoc) or ($liststyle == 'ol')) {
 380             $with_counter = 1;
 381         }
 382
 383         // Check if user is allowed to get the Page.
 384         if (!mayAccessPage ('view', $pagename)) {
 385             return $this->error(sprintf(_("Illegal access to page %s: no read access"),
 386             $pagename));
 387         }
 388
 389         $page = $dbi->getPage($pagename);
 390         $current = $page->getCurrentRevision();
 391         //FIXME: I suspect this only to crash with Apache2
 392         if (!$current->get('markup') or $current->get('markup') < 2) {
 393             if (in_array(php_sapi_name(),array('apache2handler','apache2filter'))) {
 394                 trigger_error(_("CreateToc disabled for old markup"), E_USER_WARNING);
 395                 return '';
 396             }
 397         }
 398         $content = $current->getContent();
 399         $html = HTML::div(array('class' => 'toc', 'id'=> GenerateId("toc")));
 400         if ($notoc) {
 401             $html->setAttr('style','display:none;');
 402         }
 403         if (($position == "left") or ($position == "right")) {
 404             $html->setAttr('style','float:'.$position.'; width:'.$width.';');
 405         }
 406         $toclistid = GenerateId("toclist");
 407         $list = HTML::div(array('id'=>$toclistid, 'class'=>'toclist'));
 408         if (!strstr($headers,",")) {
 409             $headers = array($headers);
 410         } else {
 411             $headers = explode(",",$headers);
 412         }
 413         $levels = array();
 414         foreach ($headers as $h) {
 415             //replace !!! with level 1, ...
 416             if (strstr($h,"!")) {
 417                 $hcount = substr_count($h,'!');
 418                 $level = min(max(1, $hcount),3);
 419                 $levels[] = $level;
 420             } else {
 421                 $level = min(max(1, (int) $h), 5);
 422                 $levels[] = $level;
 423             }
 424         }
 425         if (TOC_FULL_SYNTAX)
 426             require_once("lib/InlineParser.php");
 427         if ($headers = $this->extractHeaders($content, $dbi->_markup,
 428                                              $with_toclink, $with_counter,
 429                                              $levels, $firstlevelstyle, $basepage))
 430         {
 431             foreach ($headers as $h) {
 432                 // proper heading indent
 433                 $level = $h['level'];
 434                 $indent = $level - 1;
 435                 $link = new WikiPageName($pagename,$page,$h['anchor']);
 436                 $li = WikiLink($link,'known',$h['text']);
 437                 // Hack to suppress pagename before #
 438                 // $li->_attr["href"] = strstr($li->_attr["href"], '#');
 439                 $list->pushContent(HTML::p(HTML::raw
 440                        (str_repeat($indentstr,$indent)),$li));
 441             }
 442         }
 443         $list->setAttr('style','display:'.($jshide?'none;':'block;'));
 444         $open = DATA_PATH.'/'.$WikiTheme->_findFile("images/folderArrowOpen.png");
 445         $close = DATA_PATH.'/'.$WikiTheme->_findFile("images/folderArrowClosed.png");
 446       if ($noheader) {
 447       } else {
 448         $toctoggleid = GenerateId("toctoggle");
 449         if ($extracollapse)
 450             $toclink = HTML(_("Table of Contents"),
 451                             " ",
 452                             HTML::a(array('id'=>'TOC')),
 453                             HTML::img(array(
 454                                             'id'=>$toctoggleid,
 455                                             'class'=>'wikiaction',
 456                                             'title'=>_("Click to display to TOC"),
 457                                             'onclick'=>"toggletoc(this, '".$open."', '".$close."', '".$toclistid."')",
 458                                             'alt' => 'toctoggle',
 459                                             'src' => $jshide ? $close : $open )));
 460         else
 461             $toclink = HTML::a(array('id'=>'TOC',
 462                                      'class'=>'wikiaction',
 463                                      'title'=>_("Click to display"),
 464                                      'onclick'=>"toggletoc(this, '".$open."', '".$close."', '".$toclistid."')"),
 465                                _("Table of Contents"),
 466                                HTML::span(array('style'=>'display:none',
 467                                                 'id'=>$toctoggleid)," "));
 468         $html->pushContent(HTML::p(array('class'=>'toctitle'), $toclink));
 469       }
 470       $html->pushContent($list);
 471       if (count($headers) == 0) {
 472           // Do not display an empty TOC
 473           $html->setAttr('style','display:none;');
 474       }
 475       return $html;
 476     }
 477 };
 478
 479 // Local Variables:
 480 // mode: php
 481 // tab-width: 8
 482 // c-basic-offset: 4
 483 // c-hanging-comment-ender-p: nil
 484 // indent-tabs-mode: nil
 485 // End:
 486 ?>