Adds a "fuzzy" search suggestion to the i18n_search plug-in for the GetSimple CMS

gistfile1.php

<?php

/**
 * Fuzzy Search Suggestion Snippet for i18n_search plug-in for the GetSimple CMS
 *
 * Sometimes a search query produces no matches. This is occassionaly due to a
 * miss-spelling in the query terms. This routine will attempt to offer a search
 * suggestion to the user when a search results in no matches.
 *
 * For example:
 *
 *    Sorry, I didn't find anything that matched your query: "typografy behavier"
 *    Did you mean: typography behaviour
 *
 * The script will also link the suggested search terms back to the search
 * engine. In most cases the suggestion will resolve to a positive query.
 * However, it's not (yet) fool-proof and will occasionally link a query that,
 * itself returns no results.
 *
 * I have only tested this with English words. As the script uses the metaphone
 * algorithm it's likely that it won't work too well for non-English languages.
 *
 * Still - it's better than nothing! :-)
 *
 * The script is running on my own website at http://darkblue.sdf.org/ - so you
 * can try it out there.
 *
 * @author  Jonathan M. Hollin <darkblue@sdf.lonestar.org>
 * @copyright Copyleft: http://www.gnu.org/copyleft/copyleft.html
 */

function stripColons(&$item, &$key) {
  $temp = explode(':', $item);
  $item = $temp[0];
}

// Search Suggestions
$candidates = array();
$individualWords = explode(' ', $words);
$suggestions = array();

// Process the word corpus
if ( !file_exists(GSDATAOTHERPATH . I18N_WORD_INDEX) ) create_i18n_search_index();
$corpus = array();
$f = fopen(GSDATAOTHERPATH . I18N_WORD_INDEX, 'r');
while ( ($line = fgets($f)) !== false ) {
  $items = explode(' ', $line);
  foreach ( $individualWords as $checkWord ) {
    if ($items[0] === $checkWord) {
      // Exact match!
      $items['hit'] = 1;
      $items['pattern'] = $checkWord;
      array_walk_recursive($items, 'stripColons');
      $candidates[] = $items;
      if ( ($key = array_search($items[0], $individualWords) ) !== false) unset($individualWords[$key]);
      continue;
    } else {
      // Looking for possible replacements
      $lev = levenshtein( metaphone($checkWord), metaphone($items[0]) );
      if ($lev < 1) {
        array_walk_recursive($items, 'stripColons');
        $items['pattern'] = $checkWord;
        $items['replace'] = $checkWord . '|' . $items[0];
        similar_text($items[0], $checkWord, $items['confidence']);
        $candidates[] = $items;
      }
    }
  }
}
fclose($f);

// Restore $individualWords
$individualWords = explode(' ', $words);

// Clean up $candidates
foreach ($candidates as $k => $v) {
  unset($key);
  if ( array_key_exists('hit', $v) ) $key = $v[0];
  if ($key) foreach ($candidates as $y => $z) if ( ($z['pattern'] === $key) && ($y !== $k) ) unset($candidates[$y]);
}
foreach ($candidates as $k => $v) {
  unset($candidates[$k]['pattern']);
  unset($candidates[$k]['hit']);
}
$candidates = array_values($candidates);

// Find match combinations that actually resolve to a page
$result = array();
if ( (count($candidates) >= 2) && (count($individualWords) !== 1) ) {
  // Start from the first array, and continue through all of them.
  for ($i = 0; $i < count($candidates); $i ++) {
    // Start with the current array of $i, +1. So that collisions never occur.
    for ($j = $i + 1; $j < count($candidates); $j ++) {
      // If there are common values (the array_intersect() function returns a non-empty array
      if (count(array_intersect($candidates[$i], $candidates[$j])) !== 0) {
        // Add the first array (if it's not there already)
        if (!in_array($candidates[$i], $result)) $result[] = $candidates[$i];
        // Add the second array (if it's not there already)
        if (!in_array($candidates[$j], $result)) $result[] = $candidates[$j];
      }
    }
  }
} else {
  if (count($individualWords) === 1) $result = $candidates;
}
unset($candidates);

// Compute the best suggestion
$condidence = 0;
$sentence = $replacements = array();
if (count($result)) {
  foreach ($result as $k => $v) {
    $replace = (isset($v['replace'])) ? explode('|', $v['replace']) : false;
    if (count($replace) === 2) {
      $key = array_search($replace[0], $individualWords);
      $confidence = ($v['confidence'] > $confidence) ? $v['confidence'] : $confidence;
      if ($replace && is_array($replace)) {
        if (isset($replacements[$replace[0]])) {
          if ($confidence > $replacements[$replace[0]]['confidence']) {
            $replacements[$replace[0]]['confidence'] = $v['confidence'];
            $replacements[$replace[0]]['with'] = $replace[1];
          }
        } else {
          $replacements[$replace[0]]['confidence'] = $v['confidence'];
          $replacements[$replace[0]]['with'] = $replace[1];
        }
      }

    }
  }
  unset($result);

  // Render the suggestion
  foreach ($individualWords as $word) {
    if (isset($replacements[$word])) {
      $sentence[] = '<span style="font-style: italic;">' . $replacements[$word]['with'] . '</span>';
    } else {
      $sentence[] = $word;
    }
  }
  unset($individualWords);
  unset($replacements);

  if (count($sentence)) echo '<p class="search-no-results">Did you mean: <a href="/search?words=' . strip_tags(implode('+', $sentence)) . '">' . implode(' ', $sentence) . '</a></p>';
}

?>

See: http://get-simple.info/forums/showthread.php?tid=1256&pid=29712#pid29712 for more info.