/mvc/components/filter

<?php
/**
 * User-entered data filter
 */
class filterComponent {
    /**
     * Generates a URL-safe SEO-friendly string from any text string; long strings are truncated to 100 chars
     *
     * @param string $str
     * @return string
     */
    public function urlString($str) {
        $str = preg_replace('/-{2,}/', '-', preg_replace('/[^a-z0-9-]/', '-', str_replace('_', '-',
 strtolower($str))));
        return trim(substr($str, 0, 100), '-');
    }

    /**
     * Returns a string containing only numeric digits
     *
     * @param string $data
     * @param bool $decimal
     * @return float
     */
    public function getNumbersOnly($data, $decimal = false) {
        return preg_replace(($decimal ? '/[^\d\.]/' : '/\D/'), '', $data);
    }

    /**
     * Replace deprecated font-sizes with accessibility-friendly sizes in CSS format
     *
     * @param mixed $size
     * @return string
     */
    protected function _fontsizereplace($size) {
        $sizes = array(-2 => 'xx-small', -1 => 'x-small', '+1' => 'small', '+2' => 'medium', '+3' => 'large',
                       '+4' => 'x-large', 0 => 'xx-small', '1' => 'x-small', '2' => 'small', '3' => 'medium',
                       '4' => 'large', '5' => 'x-large', '6' => 'x-large', '7' => 'x-large', '8' => 'x-large');
        if (isset($sizes[$size[2]])) {
            $size[2] = $sizes[$size[2]];
        }
        return '<span style="font-size:' . $size[2] . ';">';
    }

    /**
     * Cleans up messy broken HTML and returns pristine code that is XHTML-valid
     *
     * Requires the PHP Tidy extension to be enabled, although it will work without it but clean up will
     * be severly limited and it will not convert broken HTML into valid XHTML
     *
     * @param string $html
     * @return string
     */
    public function cleanHtml($html) {
        $html = preg_replace('/(<font .*?color="(.*?)".*?'.'>)/mi', '$1<span style="color:$2;">', $html);
        $html = preg_replace_callback('/(<font .*?size="(.*?)".*?'.'>)/mi', array($this, '_fontsizereplace'),
 $html);
        $deprecated = array('</font>' => '</span>',
                            '<center>' => '<div style="text-align:center;">', '</center>' => '</div>',
                            '<strike>' => '<span style="text-decoration:line-through;">', '</strike>' =>
 '</span>',
                            '<s>' => '<span style="text-decoration:line-through;">', '</s>' => '</span>',
                            '<u>' => '<span style="text-decoration:underline;">', '</u>' => '</span>');
        $html = str_ireplace(array_keys($deprecated), array_values($deprecated), $html);
        if (extension_loaded('tidy')) {
            $tidy = new tidy;
            $phpopen = '|TIDY_PHP_OPEN|';
            $phpclose = '|TIDY_PHP_CLOSE|';
            $html = str_replace(array('<?', '?>'), array($phpopen, $phpclose), $html);
            $tidy->parseString($html, array('show-body-only' => true, 'output-xhtml' => true, 'word-2000' =>
 true,
                                            'wrap' => 0, 'wrap-php' => false, 'hide-comments' => true,
                                            'drop-font-tags' => true, 'logical-emphasis' => true));
            $tidy->cleanRepair(); //the next line strips out placeholders and casts to string in the process
            $html = trim(str_replace(array($phpopen, $phpclose), array('<?', '?>'), $tidy));
        }
        $html = preg_replace('/(<img (.*?) align="(.*?)"(.*?)'.'>)/mi', '<img $2 style="float:$3;"$4>', $html);
        return $html;
    }
}