Camspiers\StatisticalClassifier\DataSource\Directory::read PHP Method

    public function read()
    {
        $data = array();
        if (file_exists($this->options['directory'])) {
            $pattern = $this->options['mode'] == self::MODE_DIRECTORY_AS_CATEGORY ? '/*' : '';
            if (is_array($this->options['include']) && count($this->options['include']) !== 0) {
                $files = array();
                foreach ($this->options['include'] as $include) {
                    $files = array_merge($files, array_slice(glob("{$this->options['directory']}/{$include}{$pattern}", GLOB_NOSORT), 0, $this->options['limit']));
                }
            } else {
                $files = array_slice(glob("{$this->options['directory']}{$pattern}/*", GLOB_NOSORT), 0, $this->options['limit']);
            }
            foreach ($files as $filename) {
                if (is_file($filename)) {
                    if ($this->options['mode'] === self::MODE_DIRECTORY_AS_CATEGORY) {
                        $categoryPath = dirname($filename);
                    } else {
                        $categoryPath = $filename;
                    }
                    $data[] = array('category' => basename($categoryPath), 'document' => file_get_contents($filename));
                }
            }
        }
        return $data;
    }