lib/plugin/FuzzyPages.php

   1 <?php // -*-php-*-
   2 rcs_id('$Id: FuzzyPages.php,v 1.6 2002-03-02 02:59:56 carstenklapp Exp $');
   3 /*
   4  Copyright 1999, 2000, 2001, 2002 $ThePhpWikiProgrammingTeam
   5
   6  This file is part of PhpWiki.
   7
   8  PhpWiki is free software; you can redistribute it and/or modify
   9  it under the terms of the GNU General Public License as published by
  10  the Free Software Foundation; either version 2 of the License, or
  11  (at your option) any later version.
  12
  13  PhpWiki is distributed in the hope that it will be useful,
  14  but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  GNU General Public License for more details.
  17
  18  You should have received a copy of the GNU General Public License
  19  along with PhpWiki; if not, write to the Free Software
  20  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  21  */
  22
  23
  24 //require_once('lib/PageList.php');
  25
  26 /**
  27  * FuzzyPages is plugin which searches for similar page titles.
  28  *
  29  * Pages are considered similar by averaging the similarity scores of
  30  * the spelling comparison and the metaphone comparison for each page
  31  * title in the database (php's metaphone() is an improved soundex
  32  * function).
  33  *
  34  * http://www.php.net/manual/en/function.similar-text.php
  35  * http://www.php.net/manual/en/function.metaphone.php
  36  */
  37 class WikiPlugin_FuzzyPages
  38 extends WikiPlugin
  39 {
  40     function getName() {
  41         return _("FuzzyPages");
  42     }
  43
  44     function getDescription() {
  45         return sprintf(_("List FuzzyPages for %s"), '[pagename]');
  46     }
  47
  48     function getDefaultArguments() {
  49         return array('page'  => '[pagename]',
  50                      's'     => false,
  51                      'debug' => false);
  52     }
  53
  54
  55
  56     function spelling_similarity($subject) {
  57         $spelling_similarity_score = 0;
  58         similar_text($subject, $this->_searchterm,
  59                      &$spelling_similarity_score);
  60         return $spelling_similarity_score;
  61     }
  62
  63     function sound_similarity($subject) {
  64         $sound_similarity_score = 0;
  65         // similar_text() automatically calculates a percentage.
  66         // http://www.php.net/manual/en/function.similar-text.php
  67         similar_text(metaphone($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(&$list, SORT_NUMERIC, SORT_DESC);
  96     }
  97
  98
  99
 100     function addTableCaption(&$table, &$dbi) {
 101         if ($dbi->isWikiPage($this->_searchterm))
 102             $link = WikiLink($this->_searchterm, 'auto');
 103         else
 104             $link = $this->_searchterm;
 105         $caption = fmt("These page titles match fuzzy with '%s'", $link);
 106         $table->pushContent(HTML::caption(array('align'=>'top'), $caption));
 107     }
 108
 109     function addTableHead(&$table) {
 110         $row = HTML::tr(HTML::td(_("Name")),
 111                         HTML::td(array('align' => 'right'), _("Score")));
 112         if ($this->debug)
 113             $this->_pushDebugHeadingTDinto($row);
 114
 115         $table->pushContent(HTML::thead($row));
 116     }
 117
 118     function addTableBody(&$list, &$table) {
 119         if (! defined('HIGHLIGHT_ROWS_CUTOFF_SCORE'))
 120             define('HIGHLIGHT_ROWS_CUTOFF_SCORE', 60);
 121
 122         $tbody = HTML::tbody();
 123         foreach ($list as $found_pagename => $score) {
 124             $row = HTML::tr(array('class' =>
 125                                   $score > HIGHLIGHT_ROWS_CUTOFF_SCORE
 126                                   ? 'oddrow' : 'evenrow'),
 127                             HTML::td(WikiLink($found_pagename)),
 128                             HTML::td(array('align' => 'right'),
 129                                      round($score)));
 130
 131             if ($this->debug)
 132                 $this->_pushDebugTDinto($row, $found_pagename);
 133
 134             $tbody->pushContent($row);
 135         }
 136         $table->pushContent($tbody);
 137     }
 138
 139     function formatTable(&$list, &$dbi) {
 140
 141         $table = HTML::table(array('cellpadding' => 2,
 142                                    'cellspacing' => 1,
 143                                    'border'      => 0,
 144                                    'class' => 'pagelist'));
 145         $this->addTableCaption($table, &$dbi);
 146         $this->addTableHead($table);
 147         $this->addTableBody($list, $table);
 148         return $table;
 149     }
 150
 151
 152
 153     function run($dbi, $argstr, $request) {
 154         $args = $this->getArgs($argstr, $request);
 155         extract($args);
 156         if (empty($page))
 157             return '';
 158         $this->debug = $debug;
 159
 160         $this->_searchterm = $s ? $s : $page;
 161         $this->_list = array();
 162
 163         $this->collectSimilarPages($this->_list, &$dbi);
 164
 165         $this->sortCollectedPages($this->_list);
 166
 167         return $this->formatTable($this->_list, &$dbi);
 168     }
 169
 170
 171
 172     function _pushDebugHeadingTDinto(&$row) {
 173             $row->pushContent(HTML::td(_("Spelling Score")),
 174                               HTML::td(_("Sound Score")),
 175                               HTML::td('Metaphones'));
 176     }
 177
 178     function _pushDebugTDinto(&$row, $pagename) {
 179         // This actually calculates everything a second time for each page
 180         // but the individual scores are retained separately.
 181         $debug_spelling = round($this->spelling_similarity($pagename), 1);
 182         $debug_sound = round($this->sound_similarity($pagename), 1);
 183         $debug_metaphone = sprintf("(%s, %s)", metaphone($pagename),
 184                                    $this->_searchterm_metaphone);
 185
 186         $row->pushcontent(HTML::td(array('align' => 'center'), $debug_spelling),
 187                           HTML::td(array('align' => 'center'), $debug_sound),
 188                           HTML::td($debug_metaphone));
 189     }
 190 };
 191
 192 // Local Variables:
 193 // mode: php
 194 // tab-width: 8
 195 // c-basic-offset: 4
 196 // c-hanging-comment-ender-p: nil
 197 // indent-tabs-mode: nil
 198 // End:
 199 ?>