lib/plugin/FuzzyPages.php

   1 <?php // -*-php-*-
   2 // $Id$
   3 /*
   4  * Copyright 1999, 2000, 2001, 2002 $ThePhpWikiProgrammingTeam
   5  * Copyright 2009 Marc-Etienne Vargenau, Alcatel-Lucent
   6  *
   7  * This file is part of PhpWiki.
   8  *
   9  * PhpWiki is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * PhpWiki is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License along
  20  * with PhpWiki; if not, write to the Free Software Foundation, Inc.,
  21  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  22  */
  23
  24 /**
  25  * FuzzyPages is plugin which searches for similar page titles.
  26  *
  27  * Pages are considered similar by averaging the similarity scores of
  28  * the spelling comparison and the metaphone comparison for each page
  29  * title in the database (php's metaphone() is an improved soundex
  30  * function).
  31  *
  32  * http://www.php.net/manual/en/function.similar-text.php
  33  * http://www.php.net/manual/en/function.metaphone.php
  34  */
  35 class WikiPlugin_FuzzyPages
  36 extends WikiPlugin
  37 {
  38     function getName() {
  39         return _("FuzzyPages");
  40     }
  41
  42     function getDescription() {
  43         return sprintf(_("Search for page titles similar to %s."),
  44                        '[pagename]');
  45     }
  46
  47     function getDefaultArguments() {
  48         return array('s'     => false,
  49                      'debug' => false);
  50     }
  51
  52     function spelling_similarity($subject) {
  53         $spelling_similarity_score = 0;
  54         similar_text($subject, $this->_searchterm,
  55                      $spelling_similarity_score);
  56         return $spelling_similarity_score;
  57     }
  58
  59     function sound_similarity($subject) {
  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         return ($this->spelling_similarity($subject)
  68                 + $this->sound_similarity($subject)) / 2;
  69     }
  70
  71     function collectSimilarPages(&$list, &$dbi) {
  72         if (! defined('MIN_SCORE_CUTOFF'))
  73             define('MIN_SCORE_CUTOFF', 33);
  74
  75         $this->_searchterm_metaphone = metaphone($this->_searchterm);
  76
  77         $allPages = $dbi->getAllPages();
  78
  79         while ($pagehandle = $allPages->next()) {
  80             $pagename = $pagehandle->getName();
  81             $similarity_score = $this->averageSimilarities($pagename);
  82             if ($similarity_score > MIN_SCORE_CUTOFF)
  83                 $list[$pagename] = $similarity_score;
  84         }
  85     }
  86
  87     function sortCollectedPages(&$list) {
  88         arsort($list, SORT_NUMERIC);
  89     }
  90
  91     function addTableCaption(&$table, &$dbi) {
  92         if ($dbi->isWikiPage($this->_searchterm))
  93             $link = WikiLink($this->_searchterm, 'auto');
  94         else
  95             $link = $this->_searchterm;
  96         $caption = fmt("These page titles match fuzzy with '%s'", $link);
  97         $table->pushContent(HTML::caption(array('align'=>'top'), $caption));
  98     }
  99
 100     function addTableHead(&$table) {
 101         $row = HTML::tr(HTML::th(_("Name")),
 102                         HTML::th(array('align' => 'right'), _("Score")));
 103
 104         if (defined('DEBUG') && DEBUG && $this->debug) {
 105             $this->_pushDebugHeadingTDinto($row);
 106         }
 107
 108         $table->pushContent(HTML::thead($row));
 109     }
 110
 111     function addTableBody(&$list, &$table) {
 112         if (! defined('HIGHLIGHT_ROWS_CUTOFF_SCORE'))
 113             define('HIGHLIGHT_ROWS_CUTOFF_SCORE', 60);
 114
 115         $tbody = HTML::tbody();
 116         foreach ($list as $found_pagename => $score) {
 117             $row = HTML::tr(array('class' =>
 118                                   $score > HIGHLIGHT_ROWS_CUTOFF_SCORE
 119                                   ? 'evenrow' : 'oddrow'),
 120                             HTML::td(WikiLink($found_pagename)),
 121                             HTML::td(array('align' => 'right'),
 122                                      round($score)));
 123
 124             if (defined('DEBUG') && DEBUG && $this->debug) {
 125                 $this->_pushDebugTDinto($row, $found_pagename);
 126             }
 127
 128             $tbody->pushContent($row);
 129         }
 130         $table->pushContent($tbody);
 131     }
 132
 133     function formatTable(&$list, &$dbi) {
 134
 135         if (empty($list)) {
 136            return HTML::p(fmt("No fuzzy matches with '%s'", $this->_searchterm));
 137         }
 138         $table = HTML::table(array('cellpadding' => 2,
 139                                    'cellspacing' => 1,
 140                                    'border'      => 0,
 141                                    'class' => 'pagelist'));
 142         $this->addTableCaption($table, $dbi);
 143         $this->addTableHead($table);
 144         $this->addTableBody($list, $table);
 145         return $table;
 146     }
 147
 148
 149     function run($dbi, $argstr, &$request, $basepage) {
 150         $args = $this->getArgs($argstr, $request);
 151         extract($args);
 152         if (empty($s)) {
 153             return HTML();
 154         }
 155
 156         if (defined('DEBUG') && DEBUG) {
 157             $this->debug = $debug;
 158         }
 159
 160         $this->_searchterm = $s;
 161         $this->_list = array();
 162
 163         $this->collectSimilarPages($this->_list, $dbi);
 164         $this->sortCollectedPages($this->_list);
 165         return $this->formatTable($this->_list, $dbi);
 166     }
 167
 168     function _pushDebugHeadingTDinto(&$row) {
 169         $row->pushContent(HTML::td(_("Spelling Score")),
 170                           HTML::td(_("Sound Score")),
 171                           HTML::td('Metaphones'));
 172     }
 173
 174     function _pushDebugTDinto(&$row, $pagename) {
 175         // This actually calculates everything a second time for each pagename
 176         // so the individual scores can be displayed separately for debugging.
 177         $debug_spelling = round($this->spelling_similarity($pagename), 1);
 178         $debug_sound = round($this->sound_similarity($pagename), 1);
 179         $debug_metaphone = sprintf("(%s, %s)", metaphone($pagename),
 180                                    $this->_searchterm_metaphone);
 181
 182         $row->pushcontent(HTML::td(array('align' => 'center'), $debug_spelling),
 183                           HTML::td(array('align' => 'center'), $debug_sound),
 184                           HTML::td($debug_metaphone));
 185     }
 186 };
 187
 188 // Local Variables:
 189 // mode: php
 190 // tab-width: 8
 191 // c-basic-offset: 4
 192 // c-hanging-comment-ender-p: nil
 193 // indent-tabs-mode: nil
 194 // End:
 195 ?>