/**
* An attempt to measure word similarity in percent
*
* @param string $strA
* @param string $strB
*
* @return int distance from 0 to 100
*/
protected function _similarity($strA, $strB)
{
$lenA = $this->text->strlen($strA);
$lenB = $this->text->strlen($strB);
if ($lenA == 0 && $lenB == 0) {
return 100;
}
$distance = $this->_distance($strA, $strB);
$similarity = 100 - (int) round(200 * $distance / ($lenA + $lenB));
return $similarity >= 100 ? 100 : $similarity;
}
/**
* Convert $str to same register with $base
*
* @param string $str
* @param string $base
* @return string
*/
protected function toSameRegister($str, $base)
{
$minLen = min($this->text->strlen($base), $this->text->strlen($str));
for ($i = 0; $i < $minLen; $i++) {
$chr = $this->text->substr($base, $i, 1);
if ($chr != $this->text->strtolower($chr)) {
$chrN = $this->text->substr($str, $i, 1);
$chrN = strtoupper($chrN);
$str = substr_replace($str, $chrN, $i, 1);
}
}
return $str;
}
/**
* Split string to words
*
* @param string $string
* @param array &$results
* @param int $increment
* @return void
*/
protected function split($string, &$results, $increment = 1)
{
$string = $this->text->cleanString($string);
$words = $this->text->splitWords($string);
foreach ($words as $word) {
if ($this->text->strlen($word) >= $this->text->getGram() && !is_numeric($word) && $this->text->strlen($word) <= 10) {
$word = $this->text->strtolower($word);
if (!isset($results[$word])) {
$results[$word] = $increment;
} else {
$results[$word] += $increment;
}
}
}
}
