note 36900 added to function.levenshtein

I wrote this function to have an "intelligent" comparison between data to be written in a DB and already existent data. Not ony calculating distances but also balancing distances for each field. <?php /* This function calculate a balanced percentage distance between an array of strings "$record" and a compared array "$compared", balanced through an array of weights "$weight". The three arrays must have the same indices. For an unbalanced distance, set all weights to 1. The used formula is: percentage distance = sum(field_levenshtein_distance * field_weight) / sum(record_field_length * field_weight) * 100 */ function search_similar($record, $weights, $compared, $precision=2) { $field_names = array_keys($record); # "Weighted length" of $record and "weighted distance". foreach ($field_names as $field_key) { $record_weight += strlen($record[$field_key]) * $weights[$field_key]; $weighted_distance += levenshtein($record[$field_key],$compared[$field_key]) * $weights[$field_key]; } # Building the result.. if ($record_weight) { return round(($weighted_distance / $record_weight * 100),$precision); } elseif ((strlen(implode("",$record)) == 0) && (strlen(implode("",$compared)) == 0)) { // empty records return round(0,$precision); } elseif (array_sum($weights) == 0) { // all weights == 0 return round(0,$precision); } else { return false; } /* Be very careful distinguising 0 result and false result. The function results 0 ('0.00' if $precision is 2 and so on) if: - $record and $compared are equals (even if $record and $compared are empty); - all weights are 0 (the meaning could be "no care about any field"). Conversely, the function results false if $record is empty, but the weights are not all 0 and $compared is not empty. That cause a "division by 0" error. I wrote this kind of check: if ($rel_dist = search_similar(...)) { print $rel_dist; } elseif ($rel_dist == "0.00") { // supposing that $precision is 2 print $rel_dist; } else { print "infinite"; } */ } ?> ---- Manual Page -- http://www.php.net/manual/en/function.levenshtein.php Edit -- http://master.php.net/manage/user-notes.php?action=edit+36900 Delete -- http://master.php.net/manage/user-notes.php?action=delete+36900&report=yes Reject -- http://master.php.net/manage/user-notes.php?action=reject+36900&report=yes Search -- http://master.php.net/manage/user-notes.php