lib/plugin/FuzzyPages.php

   1 <?php
   2
   3 /*
   4  * Copyright 1999, 2000, 2001, 2002 $ThePhpWikiProgrammingTeam
   5  * Copyright 2009 Marc-Etienne Vargenau, Alcatel-Lucent
   6  *
   7  * This file is part of PhpWiki.
   8  *
   9  * PhpWiki is free software; you can redistribute it and/or modify
  10  * it under the terms of the GNU General Public License as published by
  11  * the Free Software Foundation; either version 2 of the License, or
  12  * (at your option) any later version.
  13  *
  14  * PhpWiki is distributed in the hope that it will be useful,
  15  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  16  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17  * GNU General Public License for more details.
  18  *
  19  * You should have received a copy of the GNU General Public License along
  20  * with PhpWiki; if not, write to the Free Software Foundation, Inc.,
  21  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  22  */
  23
  24 /**
  25  * FuzzyPages is plugin which searches for similar page titles.
  26  *
  27  * Pages are considered similar by averaging the similarity scores of
  28  * the spelling comparison and the metaphone comparison for each page
  29  * title in the database (php's metaphone() is an improved soundex
  30  * function).
  31  *
  32  * http://www.php.net/manual/en/function.similar-text.php
  33  * http://www.php.net/manual/en/function.metaphone.php
  34  */
  35 class WikiPlugin_FuzzyPages
  36     extends WikiPlugin
  37 {
  38     function getName()
  39     {
  40         return _("FuzzyPages");
  41     }
  42
  43     function getDescription()
  44     {
  45         return sprintf(_("Search for page titles similar to %s."),
  46             '[pagename]');
  47     }
  48
  49     function getDefaultArguments()
  50     {
  51         return array('s' => false,
  52             'debug' => false);
  53     }
  54
  55     function spelling_similarity($subject)
  56     {
  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     {
  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     {
  73         return ($this->spelling_similarity($subject)
  74             + $this->sound_similarity($subject)) / 2;
  75     }
  76
  77     function collectSimilarPages(&$list, &$dbi)
  78     {
  79         if (!defined('MIN_SCORE_CUTOFF'))
  80             define('MIN_SCORE_CUTOFF', 33);
  81
  82         $this->_searchterm_metaphone = metaphone($this->_searchterm);
  83
  84         $allPages = $dbi->getAllPages();
  85
  86         while ($pagehandle = $allPages->next()) {
  87             $pagename = $pagehandle->getName();
  88             $similarity_score = $this->averageSimilarities($pagename);
  89             if ($similarity_score > MIN_SCORE_CUTOFF)
  90                 $list[$pagename] = $similarity_score;
  91         }
  92     }
  93
  94     function sortCollectedPages(&$list)
  95     {
  96         arsort($list, SORT_NUMERIC);
  97     }
  98
  99     function addTableCaption(&$table, &$dbi)
 100     {
 101         if ($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         $table->pushContent(HTML::caption($caption));
 107     }
 108
 109     function addTableHead(&$table)
 110     {
 111         $row = HTML::tr(HTML::th(_("Name")), HTML::th(_("Score")));
 112
 113         if (defined('DEBUG') && DEBUG && $this->debug) {
 114             $this->_pushDebugHeadingTDinto($row);
 115         }
 116
 117         $table->pushContent(HTML::thead($row));
 118     }
 119
 120     function addTableBody(&$list, &$table)
 121     {
 122         if (!defined('HIGHLIGHT_ROWS_CUTOFF_SCORE'))
 123             define('HIGHLIGHT_ROWS_CUTOFF_SCORE', 60);
 124
 125         $tbody = HTML::tbody();
 126         foreach ($list as $found_pagename => $score) {
 127             $row = HTML::tr(array('class' =>
 128                 $score > HIGHLIGHT_ROWS_CUTOFF_SCORE
 129                     ? 'evenrow' : 'oddrow'),
 130                 HTML::td(WikiLink($found_pagename)),
 131                 HTML::td(array('class' => 'align-right'),
 132                     round($score)));
 133
 134             if (defined('DEBUG') && DEBUG && $this->debug) {
 135                 $this->_pushDebugTDinto($row, $found_pagename);
 136             }
 137
 138             $tbody->pushContent($row);
 139         }
 140         $table->pushContent($tbody);
 141     }
 142
 143     function formatTable(&$list, &$dbi)
 144     {
 145
 146         if (empty($list)) {
 147             return HTML::p(fmt("No fuzzy matches with “%s”", $this->_searchterm));
 148         }
 149         $table = HTML::table(array('class' => 'pagelist'));
 150         $this->addTableCaption($table, $dbi);
 151         $this->addTableHead($table);
 152         $this->addTableBody($list, $table);
 153         return $table;
 154     }
 155
 156     function run($dbi, $argstr, &$request, $basepage)
 157     {
 158         $args = $this->getArgs($argstr, $request);
 159         extract($args);
 160         if (empty($s)) {
 161             return HTML();
 162         }
 163
 164         if (defined('DEBUG') && DEBUG) {
 165             $this->debug = $debug;
 166         }
 167
 168         $this->_searchterm = $s;
 169         $this->_list = array();
 170
 171         $this->collectSimilarPages($this->_list, $dbi);
 172         $this->sortCollectedPages($this->_list);
 173         return $this->formatTable($this->_list, $dbi);
 174     }
 175
 176     private function _pushDebugHeadingTDinto(&$row)
 177     {
 178         $row->pushContent(HTML::td(_("Spelling Score")),
 179             HTML::td(_("Sound Score")),
 180             HTML::td('Metaphones'));
 181     }
 182
 183     private function _pushDebugTDinto(&$row, $pagename)
 184     {
 185         // This actually calculates everything a second time for each pagename
 186         // so the individual scores can be displayed separately for debugging.
 187         $debug_spelling = round($this->spelling_similarity($pagename), 1);
 188         $debug_sound = round($this->sound_similarity($pagename), 1);
 189         $debug_metaphone = sprintf("(%s, %s)", metaphone($pagename),
 190             $this->_searchterm_metaphone);
 191
 192         $row->pushcontent(HTML::td(array('class' => 'align-center'), $debug_spelling),
 193             HTML::td(array('class' => 'align-center'), $debug_sound),
 194             HTML::td($debug_metaphone));
 195     }
 196 }
 197
 198 // Local Variables:
 199 // mode: php
 200 // tab-width: 8
 201 // c-basic-offset: 4
 202 // c-hanging-comment-ender-p: nil
 203 // indent-tabs-mode: nil
 204 // End: