lib/plugin/FuzzyPages.php

   1 <?php // -*-php-*-
   2 rcs_id('$Id: FuzzyPages.php,v 1.9 2003-01-18 21:41:02 carstenklapp Exp $');
   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: 1.9 $");
  51     }
  52
  53     function getDefaultArguments() {
  54         return array('s'     => false,
  55                      'debug' => false);
  56     }
  57
  58     function spelling_similarity($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                 $list = array_merge($list,
  90                                     array($pagename => $similarity_score));
  91         }
  92     }
  93
  94     function sortCollectedPages(&$list) {
  95         array_multisort(&$list, SORT_NUMERIC, SORT_DESC);
  96     }
  97
  98     function addTableCaption(&$table, &$dbi) {
  99         if ($dbi->isWikiPage($this->_searchterm))
 100             $link = WikiLink($this->_searchterm, 'auto');
 101         else
 102             $link = $this->_searchterm;
 103         $caption = fmt("These page titles match fuzzy with '%s'", $link);
 104         $table->pushContent(HTML::caption(array('align'=>'top'), $caption));
 105     }
 106
 107     function addTableHead(&$table) {
 108         $row = HTML::tr(HTML::th(_("Name")),
 109                         HTML::th(array('align' => 'right'), _("Score")));
 110         if ($this->debug)
 111             $this->_pushDebugHeadingTDinto($row);
 112
 113         $table->pushContent(HTML::thead($row));
 114     }
 115
 116     function addTableBody(&$list, &$table) {
 117         if (! defined('HIGHLIGHT_ROWS_CUTOFF_SCORE'))
 118             define('HIGHLIGHT_ROWS_CUTOFF_SCORE', 60);
 119
 120         $tbody = HTML::tbody();
 121         foreach ($list as $found_pagename => $score) {
 122             $row = HTML::tr(array('class' =>
 123                                   $score > HIGHLIGHT_ROWS_CUTOFF_SCORE
 124                                   ? 'evenrow' : 'oddrow'),
 125                             HTML::td(WikiLink($found_pagename)),
 126                             HTML::td(array('align' => 'right'),
 127                                      round($score)));
 128
 129             if ($this->debug)
 130                 $this->_pushDebugTDinto($row, $found_pagename);
 131
 132             $tbody->pushContent($row);
 133         }
 134         $table->pushContent($tbody);
 135     }
 136
 137     function formatTable(&$list, &$dbi) {
 138
 139         $table = HTML::table(array('cellpadding' => 2,
 140                                    'cellspacing' => 1,
 141                                    'border'      => 0,
 142                                    'class' => 'pagelist'));
 143         $this->addTableCaption($table, &$dbi);
 144         $this->addTableHead($table);
 145         $this->addTableBody($list, $table);
 146         return $table;
 147     }
 148
 149
 150     function run($dbi, $argstr, $request) {
 151         $args = $this->getArgs($argstr, $request);
 152         extract($args);
 153         if (empty($s))
 154             return '';
 155         $this->debug = $debug;
 156
 157         $this->_searchterm = $s;
 158         $this->_list = array();
 159
 160         $this->collectSimilarPages($this->_list, &$dbi);
 161
 162         $this->sortCollectedPages($this->_list);
 163
 164         return $this->formatTable($this->_list, &$dbi);
 165     }
 166
 167
 168
 169     function _pushDebugHeadingTDinto(&$row) {
 170         $row->pushContent(HTML::td(_("Spelling Score")),
 171                           HTML::td(_("Sound Score")),
 172                           HTML::td('Metaphones'));
 173     }
 174
 175     function _pushDebugTDinto(&$row, $pagename) {
 176         // This actually calculates everything a second time for each pagename
 177         // so the individual scores can be displayed separately for debugging.
 178         $debug_spelling = round($this->spelling_similarity($pagename), 1);
 179         $debug_sound = round($this->sound_similarity($pagename), 1);
 180         $debug_metaphone = sprintf("(%s, %s)", metaphone($pagename),
 181                                    $this->_searchterm_metaphone);
 182
 183         $row->pushcontent(HTML::td(array('align' => 'center'), $debug_spelling),
 184                           HTML::td(array('align' => 'center'), $debug_sound),
 185                           HTML::td($debug_metaphone));
 186     }
 187 };
 188
 189 // $Log: not supported by cvs2svn $
 190
 191 // Local Variables:
 192 // mode: php
 193 // tab-width: 8
 194 // c-basic-offset: 4
 195 // c-hanging-comment-ender-p: nil
 196 // indent-tabs-mode: nil
 197 // End:
 198 ?>