lib/plugin/FuzzyPages.php

   1 <?php
   2
   3 /*
   4  * Copyright 1999, 2000, 2001, 2002 $ThePhpWikiProgrammingTeam
   5  * Copyright 2009 Marc-Etienne Vargenau, Alcatel-Lucent
   6  *
   7  * This file is part of PhpWiki.
   8  *
   9  * PhpWiki is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * PhpWiki is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License along
  20  * with PhpWiki; if not, write to the Free Software Foundation, Inc.,
  21  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  22  */
  23
  24 /**
  25  * FuzzyPages is plugin which searches for similar page titles.
  26  *
  27  * Pages are considered similar by averaging the similarity scores of
  28  * the spelling comparison and the metaphone comparison for each page
  29  * title in the database (php's metaphone() is an improved soundex
  30  * function).
  31  *
  32  * http://www.php.net/manual/en/function.similar-text.php
  33  * http://www.php.net/manual/en/function.metaphone.php
  34  */
  35 class WikiPlugin_FuzzyPages
  36     extends WikiPlugin
  37 {
  38     function getDescription()
  39     {
  40         return sprintf(_("Search for page titles similar to %s."),
  41             '[pagename]');
  42     }
  43
  44     function getDefaultArguments()
  45     {
  46         return array('s' => false,
  47             'debug' => false);
  48     }
  49
  50     function spelling_similarity($subject)
  51     {
  52         $spelling_similarity_score = 0;
  53         similar_text($subject, $this->_searchterm,
  54             $spelling_similarity_score);
  55         return $spelling_similarity_score;
  56     }
  57
  58     function sound_similarity($subject)
  59     {
  60         $sound_similarity_score = 0;
  61         similar_text(metaphone($subject), $this->_searchterm_metaphone,
  62             $sound_similarity_score);
  63         return $sound_similarity_score;
  64     }
  65
  66     function averageSimilarities($subject)
  67     {
  68         return ($this->spelling_similarity($subject)
  69             + $this->sound_similarity($subject)) / 2;
  70     }
  71
  72     function collectSimilarPages(&$list, &$dbi)
  73     {
  74         if (!defined('MIN_SCORE_CUTOFF'))
  75             define('MIN_SCORE_CUTOFF', 33);
  76
  77         $this->_searchterm_metaphone = metaphone($this->_searchterm);
  78
  79         $allPages = $dbi->getAllPages();
  80
  81         while ($pagehandle = $allPages->next()) {
  82             $pagename = $pagehandle->getName();
  83             $similarity_score = $this->averageSimilarities($pagename);
  84             if ($similarity_score > MIN_SCORE_CUTOFF)
  85                 $list[$pagename] = $similarity_score;
  86         }
  87     }
  88
  89     function sortCollectedPages(&$list)
  90     {
  91         arsort($list, SORT_NUMERIC);
  92     }
  93
  94     function addTableCaption(&$table, &$dbi)
  95     {
  96         if ($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         $table->pushContent(HTML::caption($caption));
 102     }
 103
 104     function addTableHead(&$table)
 105     {
 106         $row = HTML::tr(HTML::th(_("Name")), HTML::th(_("Score")));
 107
 108         if (defined('DEBUG') && DEBUG && $this->debug) {
 109             $this->pushDebugHeadingTDinto($row);
 110         }
 111
 112         $table->pushContent(HTML::thead($row));
 113     }
 114
 115     function addTableBody(&$list, &$table)
 116     {
 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('class' => 'align-right'),
 127                     round($score)));
 128
 129             if (defined('DEBUG') && DEBUG && $this->debug) {
 130                 $this->pushDebugTDinto($row, $found_pagename);
 131             }
 132
 133             $tbody->pushContent($row);
 134         }
 135         $table->pushContent($tbody);
 136     }
 137
 138     function formatTable(&$list, &$dbi)
 139     {
 140
 141         if (empty($list)) {
 142             return HTML::p(fmt("No fuzzy matches with “%s”", $this->_searchterm));
 143         }
 144         $table = HTML::table(array('class' => 'pagelist'));
 145         $this->addTableCaption($table, $dbi);
 146         $this->addTableHead($table);
 147         $this->addTableBody($list, $table);
 148         return $table;
 149     }
 150
 151     function run($dbi, $argstr, &$request, $basepage)
 152     {
 153         $args = $this->getArgs($argstr, $request);
 154         extract($args);
 155         if (empty($s)) {
 156             return HTML();
 157         }
 158
 159         if (defined('DEBUG') && DEBUG) {
 160             $this->debug = $debug;
 161         }
 162
 163         $this->_searchterm = $s;
 164         $this->_list = array();
 165
 166         $this->collectSimilarPages($this->_list, $dbi);
 167         $this->sortCollectedPages($this->_list);
 168         return $this->formatTable($this->_list, $dbi);
 169     }
 170
 171     private function pushDebugHeadingTDinto(&$row)
 172     {
 173         $row->pushContent(HTML::td(_("Spelling Score")),
 174             HTML::td(_("Sound Score")),
 175             HTML::td('Metaphones'));
 176     }
 177
 178     private function pushDebugTDinto(&$row, $pagename)
 179     {
 180         // This actually calculates everything a second time for each pagename
 181         // so the individual scores can be displayed separately for debugging.
 182         $debug_spelling = round($this->spelling_similarity($pagename), 1);
 183         $debug_sound = round($this->sound_similarity($pagename), 1);
 184         $debug_metaphone = sprintf("(%s, %s)", metaphone($pagename),
 185             $this->_searchterm_metaphone);
 186
 187         $row->pushcontent(HTML::td(array('class' => 'align-center'), $debug_spelling),
 188             HTML::td(array('class' => 'align-center'), $debug_sound),
 189             HTML::td($debug_metaphone));
 190     }
 191 }
 192
 193 // Local Variables:
 194 // mode: php
 195 // tab-width: 8
 196 // c-basic-offset: 4
 197 // c-hanging-comment-ender-p: nil
 198 // indent-tabs-mode: nil
 199 // End: