kazhnuz/koblog
1
0
Fork 0
Code Issues 30 Pull requests Projects 3 Releases Packages Wiki Activity Actions

Update fuzz.php

Browse source 
Set the strings in lower case to avoid missing results because of the case.
This commit is contained in:
znvcx 2022-11-23 20:19:36 +00:00 • committed by GitHub
parent 55d3b162bf
commit 64612b85f9
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 41 additions and 6 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

47
bl-plugins/search/vendors/fuzz.php vendored
View file

@ -34,6 +34,7 @@ class Fuzz
*/ */
public function __construct($source, $maxResults, $searchMode, $useLCS) public function __construct($source, $maxResults, $searchMode, $useLCS)
{ {
$this->_source = $source; $this->_source = $source;
$this->_sourceLen = count($source); $this->_sourceLen = count($source);
$this->_maxResults = max($maxResults, 1); $this->_maxResults = max($maxResults, 1);
@ -60,6 +61,9 @@ class Fuzz
{ {
$results = []; $results = [];
$scores = []; $scores = [];
// put the text in lower case to avoid missing a result because of the case
$search = strtolower($search);
// Nullify these parameters if they are irrelevant to searchMode // Nullify these parameters if they are irrelevant to searchMode
if (!$this->_useLCS) $minLCS = null; if (!$this->_useLCS) $minLCS = null;
@ -67,7 +71,7 @@ class Fuzz
// Cycle through result pool // Cycle through result pool
//for ($i = 0; $i < $this->_sourceLen; $i++) { //for ($i = 0; $i < $this->_sourceLen; $i++) {
foreach ($this->_source as $pageKey => $data) { foreach ($this->_source as $pageKey => $data) {
$allLev = []; $allLev = [];
$allJaros = []; $allJaros = [];
$allLCSs = []; $allLCSs = [];
@ -125,6 +129,11 @@ class Fuzz
{ {
$suffix = []; $suffix = [];
$result = 0; $result = 0;
// put the text in lower case to avoid missing a result because of the case
$source = strtolower($source);
$target = strtolower($target);
$n = mb_strlen($source, CHARSET); $n = mb_strlen($source, CHARSET);
$m = mb_strlen($target, CHARSET); $m = mb_strlen($target, CHARSET);
@ -157,6 +166,10 @@ class Fuzz
$matrix = []; $matrix = [];
$n = mb_strlen($source, CHARSET); $n = mb_strlen($source, CHARSET);
$m = mb_strlen($target, CHARSET); $m = mb_strlen($target, CHARSET);
// put the text in lower case to avoid missing a result because of the case
$source = strtolower($source);
$target = strtolower($target);
if ($n === 0) { if ($n === 0) {
return $m; return $m;
@ -205,8 +218,12 @@ class Fuzz
*/ */
public function getJaroWinkler($first, $second) public function getJaroWinkler($first, $second)
{ {
$shorter = ''; $shorter;
$longer = ''; $longer;
// put the text in lower case to avoid missing a result because of the case
$first = strtolower($first);
$second = strtolower($second);
if (mb_strlen($first, CHARSET) > mb_strlen($second, CHARSET)) { if (mb_strlen($first, CHARSET) > mb_strlen($second, CHARSET)) {
$longer = mb_strtolower($first, CHARSET); $longer = mb_strtolower($first, CHARSET);
@ -254,7 +271,12 @@ class Fuzz
private function _getCharMatch($first, $second, $limit) private function _getCharMatch($first, $second, $limit)
{ {
$common = ''; $common = '';
$copy = $second; $copy = $second;
// put the text in lower case to avoid missing a result because of the case
$first = strtolower($first);
$second = strtolower($second);
$firstLen = mb_strlen($first, CHARSET); $firstLen = mb_strlen($first, CHARSET);
$secondLen = mb_strlen($second, CHARSET); $secondLen = mb_strlen($second, CHARSET);
@ -285,7 +307,12 @@ class Fuzz
private function _getTranspositions($first, $second) private function _getTranspositions($first, $second)
{ {
$trans = 0; $trans = 0;
$firstLen = mb_strlen($first, CHARSET);
// put the text in lower case to avoid missing a result because of the case
$first = strtolower($first);
$second = strtolower($second);
$firstLen = mb_strlen($first, CHARSET);
for ($i = 0; $i < $firstLen; $i++) { for ($i = 0; $i < $firstLen; $i++) {
if ($first[$i] != $second[$i]) { if ($first[$i] != $second[$i]) {
@ -307,6 +334,10 @@ class Fuzz
*/ */
private function _getPrefix($first, $second) private function _getPrefix($first, $second)
{ {
// put the text in lower case to avoid missing a result because of the case
$first = strtolower($first);
$second = strtolower($second);
if (mb_strlen($first, CHARSET) == 0 || mb_strlen($second, CHARSET) == 0) { if (mb_strlen($first, CHARSET) == 0 || mb_strlen($second, CHARSET) == 0) {
return ''; return '';
} }
@ -330,7 +361,11 @@ class Fuzz
* @return Return index of first difference * @return Return index of first difference
*/ */
private function _getDiffIndex($first, $second) private function _getDiffIndex($first, $second)
{ {
// put the text in lower case to avoid missing a result because of the case
$first = strtolower($first);
$second = strtolower($second);
if ($first == $second) { if ($first == $second) {
return -1; return -1;
} }