lib/plugin/FuzzyPages.php

   1 <?php // -*-php-*-
   2 rcs_id('$Id$');
   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
  20  along with PhpWiki; if not, write to the Free Software
  21  Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  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 getName() {
  39         return _("FuzzyPages");
  40     }
  41
  42     function getDescription() {
  43         return sprintf(_("Search for page titles similar to %s."),
  44                        '[pagename]');
  45     }
  46
  47     function getVersion() {
  48         return preg_replace("/[Revision: $]/", '',
  49                             "\$Revision$");
  50     }
  51
  52     function getDefaultArguments() {
  53         return array('s'     => false,
  54                      'debug' => false);
  55     }
  56
  57     function spelling_similarity($subject) {
  58         $spelling_similarity_score = 0;
  59         similar_text($subject, $this->_searchterm,
  60                      $spelling_similarity_score);
  61         return $spelling_similarity_score;
  62     }
  63
  64     function sound_similarity($subject) {
  65         $sound_similarity_score = 0;
  66         similar_text(metaphone($subject), $this->_searchterm_metaphone,
  67                      $sound_similarity_score);
  68         return $sound_similarity_score;
  69     }
  70
  71     function averageSimilarities($subject) {
  72         return ($this->spelling_similarity($subject)
  73                 + $this->sound_similarity($subject)) / 2;
  74     }
  75
  76     function collectSimilarPages(&$list, &$dbi) {
  77         if (! defined('MIN_SCORE_CUTOFF'))
  78             define('MIN_SCORE_CUTOFF', 33);
  79
  80         $this->_searchterm_metaphone = metaphone($this->_searchterm);
  81
  82         $allPages = $dbi->getAllPages();
  83
  84         while ($pagehandle = $allPages->next()) {
  85             $pagename = $pagehandle->getName();
  86             $similarity_score = $this->averageSimilarities($pagename);
  87             if ($similarity_score > MIN_SCORE_CUTOFF)
  88                 $list[$pagename] = $similarity_score;
  89         }
  90     }
  91
  92     function sortCollectedPages(&$list) {
  93         arsort($list, SORT_NUMERIC);
  94     }
  95
  96     function addTableCaption(&$table, &$dbi) {
  97         if ($dbi->isWikiPage($this->_searchterm))
  98             $link = WikiLink($this->_searchterm, 'auto');
  99         else
 100             $link = $this->_searchterm;
 101         $caption = fmt("These page titles match fuzzy with '%s'", $link);
 102         $table->pushContent(HTML::caption(array('align'=>'top'), $caption));
 103     }
 104
 105     function addTableHead(&$table) {
 106         $row = HTML::tr(HTML::th(_("Name")),
 107                         HTML::th(array('align' => 'right'), _("Score")));
 108         if ($this->debug)
 109             $this->_pushDebugHeadingTDinto($row);
 110
 111         $table->pushContent(HTML::thead($row));
 112     }
 113
 114     function addTableBody(&$list, &$table) {
 115         if (! defined('HIGHLIGHT_ROWS_CUTOFF_SCORE'))
 116             define('HIGHLIGHT_ROWS_CUTOFF_SCORE', 60);
 117
 118         $tbody = HTML::tbody();
 119         foreach ($list as $found_pagename => $score) {
 120             $row = HTML::tr(array('class' =>
 121                                   $score > HIGHLIGHT_ROWS_CUTOFF_SCORE
 122                                   ? 'evenrow' : 'oddrow'),
 123                             HTML::td(WikiLink($found_pagename)),
 124                             HTML::td(array('align' => 'right'),
 125                                      round($score)));
 126
 127             if ($this->debug)
 128                 $this->_pushDebugTDinto($row, $found_pagename);
 129
 130             $tbody->pushContent($row);
 131         }
 132         $table->pushContent($tbody);
 133     }
 134
 135     function formatTable(&$list, &$dbi) {
 136
 137         $table = HTML::table(array('cellpadding' => 2,
 138                                    'cellspacing' => 1,
 139                                    'border'      => 0,
 140                                    'class' => 'pagelist'));
 141         $this->addTableCaption($table, $dbi);
 142         $this->addTableHead($table);
 143         $this->addTableBody($list, $table);
 144         return $table;
 145     }
 146
 147
 148     function run($dbi, $argstr, &$request, $basepage) {
 149         $args = $this->getArgs($argstr, $request);
 150         extract($args);
 151         if (empty($s)) {
 152             return HTML::div(array('class' => "error"), "Please provide 's' argument to the plugin.");
 153         }
 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 // 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 ?>