lib/plugin/FuzzyPages.php

   1 <?php // -*-php-*-
   2 rcs_id('$Id$');
   3 /*
   4  Copyright 1999, 2000, 2001, 2002 $ThePhpWikiProgrammingTeam
   5
   6  This file is part of PhpWiki.
   7
   8  PhpWiki is free software; you can redistribute it and/or modify
   9  it under the terms of the GNU General Public License as published by
  10  the Free Software Foundation; either version 2 of the License, or
  11  (at your option) any later version.
  12
  13  PhpWiki is distributed in the hope that it will be useful,
  14  but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  GNU General Public License for more details.
  17
  18  You should have received a copy of the GNU General Public License
  19  along with PhpWiki; if not, write to the Free Software
  20  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  21  */
  22
  23 //require_once('lib/PageList.php');
  24
  25 /**
  26  * FuzzyPages is plugin which searches for similar page titles.
  27  *
  28  * Pages are considered similar by averaging the similarity scores of
  29  * the spelling comparison and the metaphone comparison for each page
  30  * title in the database (php's metaphone() is an improved soundex
  31  * function).
  32  *
  33  * http://www.php.net/manual/en/function.similar-text.php
  34  * http://www.php.net/manual/en/function.metaphone.php
  35  */
  36 class WikiPlugin_FuzzyPages
  37 extends WikiPlugin
  38 {
  39     function getName() {
  40         return _("FuzzyPages");
  41     }
  42
  43     function getDescription() {
  44         return sprintf(_("Search for page titles similar to %s."),
  45                        '[pagename]');
  46     }
  47
  48     function getVersion() {
  49         return preg_replace("/[Revision: $]/", '',
  50                             "\$Revision: 1.12 $");
  51     }
  52
  53     function getDefaultArguments() {
  54         return array('s'     => false,
  55                      'debug' => false);
  56     }
  57
  58     function spelling_similarity($subject) {
  59         $spelling_similarity_score = 0;
  60         similar_text($subject, $this->_searchterm,
  61                      $spelling_similarity_score);
  62         return $spelling_similarity_score;
  63     }
  64
  65     function sound_similarity($subject) {
  66         $sound_similarity_score = 0;
  67         similar_text(metaphone($subject), $this->_searchterm_metaphone,
  68                      $sound_similarity_score);
  69         return $sound_similarity_score;
  70     }
  71
  72     function averageSimilarities($subject) {
  73         return ($this->spelling_similarity($subject)
  74                 + $this->sound_similarity($subject)) / 2;
  75     }
  76
  77     function collectSimilarPages(&$list, &$dbi) {
  78         if (! defined('MIN_SCORE_CUTOFF'))
  79             define('MIN_SCORE_CUTOFF', 33);
  80
  81         $this->_searchterm_metaphone = metaphone($this->_searchterm);
  82
  83         $allPages = $dbi->getAllPages();
  84
  85         while ($pagehandle = $allPages->next()) {
  86             $pagename = $pagehandle->getName();
  87             $similarity_score = $this->averageSimilarities($pagename);
  88             if ($similarity_score > MIN_SCORE_CUTOFF)
  89                 $list[$pagename] = $similarity_score;
  90         }
  91     }
  92
  93     function sortCollectedPages(&$list) {
  94         arsort($list, SORT_NUMERIC);
  95     }
  96
  97     function addTableCaption(&$table, &$dbi) {
  98         if ($dbi->isWikiPage($this->_searchterm))
  99             $link = WikiLink($this->_searchterm, 'auto');
 100         else
 101             $link = $this->_searchterm;
 102         $caption = fmt("These page titles match fuzzy with '%s'", $link);
 103         $table->pushContent(HTML::caption(array('align'=>'top'), $caption));
 104     }
 105
 106     function addTableHead(&$table) {
 107         $row = HTML::tr(HTML::th(_("Name")),
 108                         HTML::th(array('align' => 'right'), _("Score")));
 109         if ($this->debug)
 110             $this->_pushDebugHeadingTDinto($row);
 111
 112         $table->pushContent(HTML::thead($row));
 113     }
 114
 115     function addTableBody(&$list, &$table) {
 116         if (! defined('HIGHLIGHT_ROWS_CUTOFF_SCORE'))
 117             define('HIGHLIGHT_ROWS_CUTOFF_SCORE', 60);
 118
 119         $tbody = HTML::tbody();
 120         foreach ($list as $found_pagename => $score) {
 121             $row = HTML::tr(array('class' =>
 122                                   $score > HIGHLIGHT_ROWS_CUTOFF_SCORE
 123                                   ? 'evenrow' : 'oddrow'),
 124                             HTML::td(WikiLink($found_pagename)),
 125                             HTML::td(array('align' => 'right'),
 126                                      round($score)));
 127
 128             if ($this->debug)
 129                 $this->_pushDebugTDinto($row, $found_pagename);
 130
 131             $tbody->pushContent($row);
 132         }
 133         $table->pushContent($tbody);
 134     }
 135
 136     function formatTable(&$list, &$dbi) {
 137
 138         $table = HTML::table(array('cellpadding' => 2,
 139                                    'cellspacing' => 1,
 140                                    'border'      => 0,
 141                                    'class' => 'pagelist'));
 142         $this->addTableCaption($table, $dbi);
 143         $this->addTableHead($table);
 144         $this->addTableBody($list, $table);
 145         return $table;
 146     }
 147
 148
 149     function run($dbi, $argstr, &$request, $basepage) {
 150         $args = $this->getArgs($argstr, $request);
 151         extract($args);
 152         if (empty($s))
 153             return '';
 154         $this->debug = $debug;
 155
 156         $this->_searchterm = $s;
 157         $this->_list = array();
 158
 159         $this->collectSimilarPages($this->_list, $dbi);
 160         $this->sortCollectedPages($this->_list);
 161         return $this->formatTable($this->_list, $dbi);
 162     }
 163
 164
 165
 166     function _pushDebugHeadingTDinto(&$row) {
 167         $row->pushContent(HTML::td(_("Spelling Score")),
 168                           HTML::td(_("Sound Score")),
 169                           HTML::td('Metaphones'));
 170     }
 171
 172     function _pushDebugTDinto(&$row, $pagename) {
 173         // This actually calculates everything a second time for each pagename
 174         // so the individual scores can be displayed separately for debugging.
 175         $debug_spelling = round($this->spelling_similarity($pagename), 1);
 176         $debug_sound = round($this->sound_similarity($pagename), 1);
 177         $debug_metaphone = sprintf("(%s, %s)", metaphone($pagename),
 178                                    $this->_searchterm_metaphone);
 179
 180         $row->pushcontent(HTML::td(array('align' => 'center'), $debug_spelling),
 181                           HTML::td(array('align' => 'center'), $debug_sound),
 182                           HTML::td($debug_metaphone));
 183     }
 184 };
 185
 186 // $Log: not supported by cvs2svn $
 187 // Revision 1.11  2004/02/17 12:11:36  rurban
 188 // added missing 4th basepage arg at plugin->run() to almost all plugins. This caused no harm so far, because it was silently dropped on normal usage. However on plugin internal ->run invocations it failed. (InterWikiSearch, IncludeSiteMap, ...)
 189 //
 190 // Revision 1.10  2003/02/22 20:49:55  dairiki
 191 // Fixes for "Call-time pass by reference has been deprecated" errors.
 192 //
 193 // Revision 1.9  2003/01/18 21:41:02  carstenklapp
 194 // Code cleanup:
 195 // Reformatting & tabs to spaces;
 196 // Added copyleft, getVersion, getDescription, rcs_id.
 197 //
 198
 199 // Local Variables:
 200 // mode: php
 201 // tab-width: 8
 202 // c-basic-offset: 4
 203 // c-hanging-comment-ender-p: nil
 204 // indent-tabs-mode: nil
 205 // End:
 206 ?>