1<?php
2
3namespace Sabre\VObject;
4
5/**
6 * Useful utilities for working with various strings.
7 *
8 * @copyright Copyright (C) fruux GmbH (https://fruux.com/)
9 * @author Evert Pot (http://evertpot.com/)
10 * @license http://sabre.io/license/ Modified BSD License
11 */
12class StringUtil {
13
14    /**
15     * Returns true or false depending on if a string is valid UTF-8
16     *
17     * @param string $str
18     * @return bool
19     */
20    static public function isUTF8($str) {
21
22        // Control characters
23        if (preg_match('%[\x00-\x08\x0B-\x0C\x0E\x0F]%', $str)) {
24            return false;
25        }
26
27        return (bool)preg_match('%%u', $str);
28
29    }
30
31    /**
32     * This method tries its best to convert the input string to UTF-8.
33     *
34     * Currently only ISO-5991-1 input and UTF-8 input is supported, but this
35     * may be expanded upon if we receive other examples.
36     *
37     * @param string $str
38     * @return string
39     */
40    static public function convertToUTF8($str) {
41
42        $encoding = mb_detect_encoding($str , array('UTF-8','ISO-8859-1', 'WINDOWS-1252'), true);
43
44        switch($encoding) {
45            case 'ISO-8859-1' :
46                $newStr = utf8_encode($str);
47                break;
48            /* Unreachable code. Not sure yet how we can improve this
49             * situation.
50            case 'WINDOWS-1252' :
51                $newStr = iconv('cp1252', 'UTF-8', $str);
52                break;
53             */
54            default :
55                 $newStr = $str;
56
57        }
58
59        // Removing any control characters
60        return (preg_replace('%(?:[\x00-\x08\x0B-\x0C\x0E-\x1F\x7F])%', '', $newStr));
61
62    }
63
64}
65
66