markusos/simple-search

View on GitHub
src/Search/Tokenizer/TokenizeTrait.php

Summary

Maintainability
A
0 mins
Test Coverage
<?php namespace Search\Tokenizer;

trait TokenizeTrait
{

    protected $tokens;
    protected $stemms;
    protected $stopWords;

    protected function toTokens($string)
    {
        return array_filter(preg_split("/[\\s\\.,?!;:()\\]\\[\\{\\}\\-\\_]+/", strtolower($string)));
    }

    protected function encode($tokens)
    {
        return array_map(function ($token) {
            return utf8_encode($token);
        }, $tokens);
    }

    public function setStopWords(array $stopWords)
    {
        $this->stopWords = $stopWords;
    }
}