lib/plugin/FuzzyPages.php

   1 <?php // -*-php-*-
   2 rcs_id('$Id: FuzzyPages.php,v 1.8 2002-08-20 18:26:12 rurban Exp $');
   3 /*
   4  Copyright 1999, 2000, 2001, 2002 $ThePhpWikiProgrammingTeam
   5
   6  This file is part of PhpWiki.
   7
   8  PhpWiki is free software; you can redistribute it and/or modify
   9  it under the terms of the GNU General Public License as published by
  10  the Free Software Foundation; either version 2 of the License, or
  11  (at your option) any later version.
  12
  13  PhpWiki is distributed in the hope that it will be useful,
  14  but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  GNU General Public License for more details.
  17
  18  You should have received a copy of the GNU General Public License
  19  along with PhpWiki; if not, write to the Free Software
  20  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  21  */
  22
  23
  24 //require_once('lib/PageList.php');
  25
  26 /**
  27  * FuzzyPages is plugin which searches for similar page titles.
  28  *
  29  * Pages are considered similar by averaging the similarity scores of
  30  * the spelling comparison and the metaphone comparison for each page
  31  * title in the database (php's metaphone() is an improved soundex
  32  * function).
  33  *
  34  * http://www.php.net/manual/en/function.similar-text.php
  35  * http://www.php.net/manual/en/function.metaphone.php
  36  */
  37 class WikiPlugin_FuzzyPages
  38 extends WikiPlugin
  39 {
  40     function getName() {
  41         return _("FuzzyPages");
  42     }
  43
  44     function getDescription() {
  45         return sprintf(_("List FuzzyPages for %s"), '[pagename]');
  46     }
  47
  48     function getDefaultArguments() {
  49         return array('s'     => false,
  50                      'debug' => false);
  51     }
  52
  53     function spelling_similarity($subject) {
  54         $spelling_similarity_score = 0;
  55         similar_text($subject, $this->_searchterm,
  56                      &$spelling_similarity_score);
  57         return $spelling_similarity_score;
  58     }
  59
  60     function sound_similarity($subject) {
  61         $sound_similarity_score = 0;
  62         similar_text(metaphone($subject), $this->_searchterm_metaphone,
  63                      &$sound_similarity_score);
  64         return $sound_similarity_score;
  65     }
  66
  67     function averageSimilarities($subject) {
  68         return ($this->spelling_similarity($subject)
  69                 + $this->sound_similarity($subject)) / 2;
  70     }
  71
  72     function collectSimilarPages(&$list, &$dbi) {
  73         if (! defined('MIN_SCORE_CUTOFF'))
  74             define('MIN_SCORE_CUTOFF', 33);
  75
  76         $this->_searchterm_metaphone = metaphone($this->_searchterm);
  77
  78         $allPages = $dbi->getAllPages();
  79
  80         while ($pagehandle = $allPages->next()) {
  81             $pagename = $pagehandle->getName();
  82             $similarity_score = $this->averageSimilarities($pagename);
  83             if ($similarity_score > MIN_SCORE_CUTOFF)
  84                 $list = array_merge($list,
  85                                     array($pagename => $similarity_score));
  86         }
  87     }
  88
  89     function sortCollectedPages(&$list) {
  90         array_multisort(&$list, SORT_NUMERIC, SORT_DESC);
  91     }
  92
  93
  94
  95     function addTableCaption(&$table, &$dbi) {
  96         if ($dbi->isWikiPage($this->_searchterm))
  97             $link = WikiLink($this->_searchterm, 'auto');
  98         else
  99             $link = $this->_searchterm;
 100         $caption = fmt("These page titles match fuzzy with '%s'", $link);
 101         $table->pushContent(HTML::caption(array('align'=>'top'), $caption));
 102     }
 103
 104     function addTableHead(&$table) {
 105         $row = HTML::tr(HTML::th(_("Name")),
 106                         HTML::th(array('align' => 'right'), _("Score")));
 107         if ($this->debug)
 108             $this->_pushDebugHeadingTDinto($row);
 109
 110         $table->pushContent(HTML::thead($row));
 111     }
 112
 113     function addTableBody(&$list, &$table) {
 114         if (! defined('HIGHLIGHT_ROWS_CUTOFF_SCORE'))
 115             define('HIGHLIGHT_ROWS_CUTOFF_SCORE', 60);
 116
 117         $tbody = HTML::tbody();
 118         foreach ($list as $found_pagename => $score) {
 119             $row = HTML::tr(array('class' =>
 120                                   $score > HIGHLIGHT_ROWS_CUTOFF_SCORE
 121                                   ? 'evenrow' : 'oddrow'),
 122                             HTML::td(WikiLink($found_pagename)),
 123                             HTML::td(array('align' => 'right'),
 124                                      round($score)));
 125
 126             if ($this->debug)
 127                 $this->_pushDebugTDinto($row, $found_pagename);
 128
 129             $tbody->pushContent($row);
 130         }
 131         $table->pushContent($tbody);
 132     }
 133
 134     function formatTable(&$list, &$dbi) {
 135
 136         $table = HTML::table(array('cellpadding' => 2,
 137                                    'cellspacing' => 1,
 138                                    'border'      => 0,
 139                                    'class' => 'pagelist'));
 140         $this->addTableCaption($table, &$dbi);
 141         $this->addTableHead($table);
 142         $this->addTableBody($list, $table);
 143         return $table;
 144     }
 145
 146
 147     function run($dbi, $argstr, $request) {
 148         $args = $this->getArgs($argstr, $request);
 149         extract($args);
 150         if (empty($s))
 151             return '';
 152         $this->debug = $debug;
 153
 154         $this->_searchterm = $s;
 155         $this->_list = array();
 156
 157         $this->collectSimilarPages($this->_list, &$dbi);
 158
 159         $this->sortCollectedPages($this->_list);
 160
 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 // Local Variables:
 187 // mode: php
 188 // tab-width: 8
 189 // c-basic-offset: 4
 190 // c-hanging-comment-ender-p: nil
 191 // indent-tabs-mode: nil
 192 // End:
 193 ?>