lib/plugin/CreateToc.php

   1 <?php // -*-php-*-
   2 // $Id$
   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 along
  20  * with PhpWiki; if not, write to the Free Software Foundation, Inc.,
  21  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 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  */
  38
  39 if (!defined('TOC_FULL_SYNTAX'))
  40     define('TOC_FULL_SYNTAX', 1);
  41
  42 class WikiPlugin_CreateToc
  43 extends WikiPlugin
  44 {
  45     function getName() {
  46         return _("CreateToc");
  47     }
  48
  49     function getDescription() {
  50         return _("Create a Table of Contents and automatically link to headers");
  51     }
  52
  53     function getDefaultArguments() {
  54         return array('extracollapse' => 1,            // provide an entry +/- link to collapse
  55                      'firstlevelstyle' => 'number',   // 'number', 'letter' or 'roman'
  56                      'headers'   =>  "1,2,3,4,5",     // "!!!"=>h2, "!!"=>h3, "!"=>h4
  57                                                       // "1"=>h2, "2"=>h3, "3"=>h4, "4"=>h5, "5"=>h6
  58                      'indentstr' => '&nbsp;&nbsp;',
  59                      'jshide'    => 0,                // collapsed TOC as DHTML button
  60                      'liststyle' => 'dl',             // 'dl' or 'ul' or 'ol'
  61                      'noheader'  => 0,                // omit "Table of Contents" header
  62                      'notoc'     => 0,                // do not display TOC, only number headers
  63                      'pagename'  => '[pagename]',     // TOC of another page here?
  64                      'position'  => 'full',           // full, right or left
  65                      'width'     => '200px',
  66                      'with_counter' => 0,
  67                      'with_toclink' => 0,             // link back to TOC
  68                      'version'   => false,
  69                     );
  70     }
  71     // Initialisation of toc counter
  72     function _initTocCounter() {
  73         $counter = array(1=>0, 2=>0, 3=>0, 4=>0, 5=>0);
  74         return $counter;
  75     }
  76
  77     // Update toc counter with a new title
  78     function _tocCounter(&$counter, $level) {
  79         $counter[$level]++;
  80         for($i = $level+1; $i <= 5; $i++) {
  81             $counter[$i] = 0;
  82         }
  83     }
  84
  85     function _roman_counter($number) {
  86
  87         $n = intval($number);
  88         $result = '';
  89
  90         $lookup = array('C' => 100, 'XC' => 90, 'L' => 50, 'XL' => 40,
  91                         'X' => 10, 'IX' => 9, 'V' => 5, 'IV' => 4, 'I' => 1);
  92
  93         foreach ($lookup as $roman => $value) {
  94             $matches = intval($n / $value);
  95             $result .= str_repeat($roman, $matches);
  96             $n = $n % $value;
  97         }
  98         return $result;
  99     }
 100
 101     function _letter_counter($number) {
 102         if ($number <= 26) {
 103             return chr(ord("A") + $number - 1);
 104         } else {
 105             return chr(ord("A") + ($number/26) - 1) . chr(ord("A") + ($number%26));
 106         }
 107     }
 108
 109     // Get string corresponding to the current title
 110     function _getCounter(&$counter, $level, $firstlevelstyle) {
 111         if ($firstlevelstyle == 'roman') {
 112             $str= $this->_roman_counter($counter[1]);
 113         } elseif ($firstlevelstyle == 'letter') {
 114             $str= $this->_letter_counter($counter[1]);
 115         } else {
 116             $str=$counter[1];
 117         }
 118         for($i = 2; $i <= 5; $i++) {
 119             if($counter[$i] != 0)
 120                 $str .= '.'.$counter[$i];
 121         }
 122         return $str;
 123     }
 124
 125     // Get HTML header corresponding to current level (level is set of ! or =)
 126     function _getHeader($level) {
 127
 128         $count = substr_count($level,'!');
 129         switch ($count) {
 130             case 3: return "h2";
 131             case 2: return "h3";
 132             case 1: return "h4";
 133         }
 134         $count = substr_count($level,'=');
 135         switch ($count) {
 136             case 2: return "h2";
 137             case 3: return "h3";
 138             case 4: return "h4";
 139             case 5: return "h5";
 140             case 6: return "h6";
 141         }
 142         return "";
 143     }
 144
 145     function _quote($heading) {
 146         if (TOC_FULL_SYNTAX ) {
 147             $theading = TransformInline($heading);
 148             if ($theading)
 149                 return preg_quote($theading->asXML(), "/");
 150             else
 151                 return XmlContent::_quote(preg_quote($heading, "/"));
 152         } else {
 153             return XmlContent::_quote(preg_quote($heading, "/"));
 154         }
 155     }
 156
 157     /*
 158      * @param $hstart id (in $content) of heading start
 159      * @param $hend   id (in $content) of heading end
 160      */
 161     function searchHeader ($content, $start_index, $heading,
 162                            $level, &$hstart, &$hend, $basepage=false) {
 163         $hstart = 0;
 164         $hend = 0;
 165         $h = $this->_getHeader($level);
 166         $qheading = $this->_quote($heading);
 167         for ($j=$start_index; $j < count($content); $j++) {
 168             if (is_string($content[$j])) {
 169                 if (preg_match("/<$h>$qheading<\/$h>/",
 170                                    $content[$j]))
 171                     return $j;
 172             }
 173             elseif (isa($content[$j], 'cached_link'))
 174             {
 175                 if (method_exists($content[$j],'asXML')) {
 176                     $content[$j]->_basepage = $basepage;
 177                     $content[$j] = $content[$j]->asXML();
 178                 } else
 179                     $content[$j] = $content[$j]->asString();
 180                 // shortcut for single wikiword or link headers
 181                 if ($content[$j] == $heading
 182                     and substr($content[$j-1],-4,4) == "<$h>"
 183                     and substr($content[$j+1],0,5) == "</$h>")
 184                 {
 185                     $hstart = $j-1;
 186                     $hend = $j+1;
 187                     return $j; // single wikiword
 188                 }
 189                 elseif (TOC_FULL_SYNTAX) {
 190                     //DONE: To allow "!! WikiWord link" or !! http://anylink/
 191                     // Check against joined content (after cached_plugininvocation).
 192                     // The first link is the anchor then.
 193                     if (preg_match("/<$h>(?!.*<\/$h>)/", $content[$j-1])) {
 194                         $hstart = $j-1;
 195                         $joined = '';
 196                         for ($k=max($j-1,$start_index);$k<count($content);$k++) {
 197                             if (is_string($content[$k]))
 198                                 $joined .= $content[$k];
 199                             elseif (method_exists($content[$k],'asXML'))
 200                                 $joined .= $content[$k]->asXML();
 201                             else
 202                                 $joined .= $content[$k]->asString();
 203                             if (preg_match("/<$h>$qheading<\/$h>/",$joined)) {
 204                                 $hend=$k;
 205                                 return $k;
 206                             }
 207                         }
 208                     }
 209                 }
 210             }
 211         }
 212         // Do not trigger error, it happens e.g. for "<<CreateToc pagename=AnotherPage>>"
 213         // trigger_error("Heading <$h> $heading </$h> not found\n", E_USER_NOTICE);
 214
 215         return 0;
 216     }
 217
 218     /** prevent from duplicate anchors,
 219      *  beautify spaces: " " => "_" and not "x20."
 220      */
 221     function _nextAnchor($s) {
 222         static $anchors = array();
 223
 224         $s = str_replace(' ','_',$s);
 225         $i = 1;
 226         $anchor = $s;
 227         while (!empty($anchors[$anchor])) {
 228             $anchor = sprintf("%s_%d",$s,$i++);
 229         }
 230         $anchors[$anchor] = $i;
 231         return $anchor;
 232     }
 233
 234     // We have to find headers in both:
 235     // - classic Phpwiki syntax (lines starting with "!", "!!" or "!!!")
 236     // - Wikicreole syntax (lines starting with "==", "===", etc.)
 237     // We must omit lines starting with "!" if inside a Mediawiki table
 238     // (they represent a table header)
 239     // Feature request: proper nesting; multiple levels (e.g. 1,3)
 240     function extractHeaders (&$content, &$markup, $backlink=0,
 241                              $counter=0, $levels=false, $firstlevelstyle='number', $basepage='')
 242     {
 243         if (!$levels) $levels = array(1,2);
 244         $tocCounter = $this->_initTocCounter();
 245         reset($levels);
 246         sort($levels);
 247         $headers = array();
 248         $j = 0;
 249         $insidetable = false;
 250         $insideverbatim = false;
 251         for ($i=0; $i<count($content); $i++) {
 252             if (preg_match('/^\s*{\|/', $content[$i])) {
 253                $insidetable = true;
 254                continue;
 255             } elseif (preg_match('/^\s*{{{/', $content[$i])
 256                     || preg_match('/^\s*<pre>/', $content[$i])
 257                     || preg_match('/^\s*<verbatim>/', $content[$i])) {
 258                $insideverbatim = true;
 259                continue;
 260             } elseif (preg_match('/^\s*\|}/', $content[$i])) {
 261                $insidetable = false;
 262                continue;
 263             } elseif (preg_match('/^\s*}}}/', $content[$i])
 264                     || preg_match('/^\s*<\/pre>/', $content[$i])
 265                     || preg_match('/^\s*<\/verbatim>/', $content[$i])) {
 266                $insideverbatim = false;
 267                continue;
 268             }
 269             if (($insidetable) || ($insideverbatim)) {
 270                continue;
 271             }
 272             foreach ($levels as $level) {
 273                 if ($level < 1 or $level > 5) continue;
 274                 $phpwikiclassiclevel = 4 -$level;
 275                 $wikicreolelevel = $level + 1;
 276                 $trim = trim($content[$i]);
 277
 278                 if ((((strpos($trim, '=') === 0))
 279                       && (preg_match('/^\s*(={'.$wikicreolelevel.','.$wikicreolelevel.'})([^=].*)$/', $content[$i], $match)))
 280                    or (((strpos($trim, '!') === 0))
 281                       && ((preg_match('/^\s*(!{'.$phpwikiclassiclevel.','.$phpwikiclassiclevel.'})([^!].*)$/', $content[$i], $match))))) {
 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\" id=\"$manchor\">$counterString</a> - \$2";
 318                                     else
 319                                         $anchorString = "<a href=\"$url\" id=\"$manchor\">\$2</a>";
 320                                 } else {
 321                                     $anchorString = "<a id=\"$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\" id=\"$manchor\">$counterString</a> - ";
 341                                     } else {
 342                                         /* Not possible to make a backlink on a
 343                                          * title with a WikiWord */
 344                                         $anchorString = "\$1<a id=\"$manchor\"></a>";
 345                                     }
 346                                 }
 347                                 else {
 348                                     $anchorString = "\$1<a id=\"$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(sprintf(_("A required argument '%s' is missing."), 'pagename'));
 379         }
 380         if (isBrowserIE() and browserDetect("Mac")) {
 381             $jshide = 0;
 382         }
 383         if (($notoc) or ($liststyle == 'ol')) {
 384             $with_counter = 1;
 385         }
 386         if ($firstlevelstyle and ($firstlevelstyle != 'number')
 387                              and ($firstlevelstyle != 'letter')
 388                              and ($firstlevelstyle != 'roman')) {
 389             return $this->error(_("Error: firstlevelstyle must be 'number', 'letter' or 'roman'"));
 390         }
 391
 392         // Check if page exists.
 393         if (!($dbi->isWikiPage($pagename))) {
 394             return $this->error(sprintf(_("Page '%s' does not exist."), $pagename));
 395         }
 396
 397         // Check if user is allowed to get the page.
 398         if (!mayAccessPage ('view', $pagename)) {
 399             return $this->error(sprintf(_("Illegal access to page %s: no read access"),
 400             $pagename));
 401         }
 402
 403         $page = $dbi->getPage($pagename);
 404
 405         if ($version) {
 406             if (!is_whole_number($version) or !($version>0)) {
 407                 return $this->error(_("Error: version must be a positive integer."));
 408             }
 409             $r = $page->getRevision($version);
 410             if ((!$r) || ($r->hasDefaultContents())) {
 411                 return $this->error(sprintf(_("%s: no such revision %d."),
 412                                             $pagename, $version));
 413             }
 414         } else {
 415             $r = $page->getCurrentRevision();
 416         }
 417
 418         $current = $page->getCurrentRevision();
 419         //FIXME: I suspect this only to crash with Apache2
 420         if (!$current->get('markup') or $current->get('markup') < 2) {
 421             if (in_array(php_sapi_name(),array('apache2handler','apache2filter'))) {
 422                 return $this->error(_("CreateToc disabled for old markup."));
 423             }
 424         }
 425
 426         $content = $r->getContent();
 427
 428         $html = HTML::div(array('class' => 'toc', 'id'=> GenerateId("toc")));
 429         if ($notoc) {
 430             $html->setAttr('style','display:none;');
 431         }
 432         if (($position == "left") or ($position == "right")) {
 433             $html->setAttr('style','float:'.$position.'; width:'.$width.';');
 434         }
 435         $toclistid = GenerateId("toclist");
 436         $list = HTML::div(array('id'=>$toclistid, 'class'=>'toclist'));
 437         if (!strstr($headers,",")) {
 438             $headers = array($headers);
 439         } else {
 440             $headers = explode(",",$headers);
 441         }
 442         $levels = array();
 443         foreach ($headers as $h) {
 444             //replace !!! with level 1, ...
 445             if (strstr($h,"!")) {
 446                 $hcount = substr_count($h,'!');
 447                 $level = min(max(1, $hcount),3);
 448                 $levels[] = $level;
 449             } else {
 450                 $level = min(max(1, (int) $h), 5);
 451                 $levels[] = $level;
 452             }
 453         }
 454         if (TOC_FULL_SYNTAX)
 455             require_once("lib/InlineParser.php");
 456         if ($headers = $this->extractHeaders($content, $dbi->_markup,
 457                                              $with_toclink, $with_counter,
 458                                              $levels, $firstlevelstyle, $basepage))
 459         {
 460             foreach ($headers as $h) {
 461                 // proper heading indent
 462                 $level = $h['level'];
 463                 $indent = $level - 1;
 464                 $link = new WikiPageName($pagename,$page,$h['anchor']);
 465                 $li = WikiLink($link,'known',$h['text']);
 466                 // Hack to suppress pagename before #
 467                 // $li->_attr["href"] = strstr($li->_attr["href"], '#');
 468                 $list->pushContent(HTML::p(HTML::raw
 469                        (str_repeat($indentstr,$indent)),$li));
 470             }
 471         }
 472         $list->setAttr('style','display:'.($jshide?'none;':'block;'));
 473         $open = DATA_PATH.'/'.$WikiTheme->_findFile("images/folderArrowOpen.png");
 474         $close = DATA_PATH.'/'.$WikiTheme->_findFile("images/folderArrowClosed.png");
 475       if ($noheader) {
 476       } else {
 477         $toctoggleid = GenerateId("toctoggle");
 478         if ($extracollapse)
 479             $toclink = HTML(_("Table of Contents"),
 480                             " ",
 481                             HTML::a(array('id'=>'TOC')),
 482                             HTML::img(array(
 483                                             'id'=>$toctoggleid,
 484                                             'class'=>'wikiaction',
 485                                             'title'=>_("Click to display to TOC"),
 486                                             'onclick'=>"toggletoc(this, '".$open."', '".$close."', '".$toclistid."')",
 487                                             'alt' => 'toctoggle',
 488                                             'src' => $jshide ? $close : $open )));
 489         else
 490             $toclink = HTML::a(array('id'=>'TOC',
 491                                      'class'=>'wikiaction',
 492                                      'title'=>_("Click to display"),
 493                                      'onclick'=>"toggletoc(this, '".$open."', '".$close."', '".$toclistid."')"),
 494                                _("Table of Contents"),
 495                                HTML::span(array('style'=>'display:none',
 496                                                 'id'=>$toctoggleid)," "));
 497         $html->pushContent(HTML::p(array('class'=>'toctitle'), $toclink));
 498       }
 499       $html->pushContent($list);
 500       if (count($headers) == 0) {
 501           // Do not display an empty TOC
 502           $html->setAttr('style','display:none;');
 503       }
 504       return $html;
 505     }
 506 };
 507
 508 // Local Variables:
 509 // mode: php
 510 // tab-width: 8
 511 // c-basic-offset: 4
 512 // c-hanging-comment-ender-p: nil
 513 // indent-tabs-mode: nil
 514 // End:
 515 ?>