StringUtils.php — Jetpack – WP Security, Backup, Speed, & Growth 12.6.4

jetpack / vendor / nojimage / twitter-text-php / lib / Twitter / Text / StringUtils.php

jetpack / vendor / nojimage / twitter-text-php / lib / Twitter / Text Last commit date

Autolink.php 5 years ago Configuration.php 5 years ago EmojiRegex.php 5 years ago Extractor.php 5 years ago HitHighlighter.php 5 years ago ParseResults.php 5 years ago Parser.php 5 years ago Regex.php 5 years ago StringUtils.php 5 years ago TldLists.php 5 years ago Validator.php 5 years ago

StringUtils.php

192 lines

1	<?php
2
3	/**
4	* @author Takashi Nojima
5	* @copyright Copyright 2014, Takashi Nojima
6	* @license http://www.apache.org/licenses/LICENSE-2.0 Apache License v2.0
7	* @package Twitter.Text
8	*/
9
10	namespace Twitter\Text;
11
12	/**
13	* String utility
14	*
15	* @author Takashi Nojima
16	* @copyright Copyright 2014, Takashi Nojima
17	* @license http://www.apache.org/licenses/LICENSE-2.0 Apache License v2.0
18	* @package Twitter
19	*/
20	class StringUtils
21	{
22
23	/**
24	* alias of mb_substr
25	*
26	* @param string $str
27	* @param integer $start (character)
28	* @param integer $length (character)
29	* @param string $encoding
30	* @return string
31	*/
32	public static function substr($str, $start, $length = null, $encoding = 'UTF-8')
33	{
34	if ($length === null) {
35	// for PHP <= 5.4.7
36	$length = mb_strlen($str, $encoding);
37	}
38	return mb_substr($str, $start, $length, $encoding);
39	}
40
41	/**
42	* alias of mb_strlen
43	*
44	* @param string $str
45	* @param string $encoding
46	* @return integer
47	*/
48	public static function strlen($str, $encoding = 'UTF-8')
49	{
50	return mb_strlen($str, $encoding);
51	}
52
53	/**
54	* alias of mb_strpos
55	*
56	* @param string $haystack
57	* @param string $needle
58	* @param integer $offset
59	* @param string $encoding
60	* @return integer
61	*/
62	public static function strpos($haystack, $needle, $offset = 0, $encoding = 'UTF-8')
63	{
64	return mb_strpos($haystack, $needle, $offset, $encoding);
65	}
66
67	/**
68	* A multibyte-aware substring replacement function.
69	*
70	* @param string $string The string to modify.
71	* @param string $replacement The replacement string.
72	* @param int $start The start of the replacement.
73	* @param int $length The number of characters to replace.
74	* @param string $encoding The encoding of the string.
75	*
76	* @return string The modified string.
77	*
78	* @see http://www.php.net/manual/en/function.substr-replace.php#90146
79	*/
80	public static function substrReplace($string, $replacement, $start, $length = null, $encoding = 'UTF-8')
81	{
82	$string_length = static::strlen($string, $encoding);
83	if ($start < 0) {
84	$start = max(0, $string_length + $start);
85	} elseif ($start > $string_length) {
86	$start = $string_length;
87	}
88	if ($length < 0) {
89	$length = max(0, $string_length - $start + $length);
90	} elseif (($length === null) \|\| ($length > $string_length)) {
91	$length = $string_length;
92	}
93	if (($start + $length) > $string_length) {
94	$length = $string_length - $start;
95	}
96
97	$suffixOffset = $start + $length;
98	$suffixLength = $string_length - $start - $length;
99
100	return static::substr($string, 0, $start, $encoding)
101	. $replacement
102	. static::substr($string, $suffixOffset, $suffixLength, $encoding);
103	}
104
105	/**
106	* idn_to_ascii wrapper
107	*
108	* @param string $domain as utf8
109	* @return string
110	*/
111	public static function idnToAscii($domain)
112	{
113	// INTL_IDNA_VARIANT_UTS46 defined PHP 5.4.0 or later
114	if (defined('INTL_IDNA_VARIANT_UTS46')) {
115	return idn_to_ascii($domain, IDNA_ALLOW_UNASSIGNED, INTL_IDNA_VARIANT_UTS46);
116	}
117
118	return idn_to_ascii($domain, IDNA_ALLOW_UNASSIGNED);
119	}
120
121	/**
122	* normalize text from NFC
123	*
124	* @param string $text
125	* @return string
126	*/
127	public static function normalizeFromNFC($text)
128	{
129	return normalizer_normalize($text);
130	}
131
132	/**
133	* get code point
134	*
135	* @param string $char
136	* @param string $encoding
137	* @return int
138	*/
139	public static function ord($char, $encoding = 'UTF-8')
140	{
141	if (mb_strlen($char, $encoding) > 1) {
142	$char = mb_substr($char, 0, 1, $encoding);
143	}
144
145	return current(unpack('N', mb_convert_encoding($char, 'UCS-4BE', $encoding)));
146	}
147
148	/**
149	* get code point at
150	*
151	* @param string $str
152	* @param int $offset
153	* @param string $encoding
154	* @return int
155	*/
156	public static function codePointAt($str, $offset, $encoding = 'UTF-8')
157	{
158	return static::ord(mb_substr($str, $offset, 1, $encoding), $encoding);
159	}
160
161	/**
162	* is surrogate pair char
163	*
164	* @param string $char
165	* @return bool
166	*/
167	public static function isSurrogatePair($char)
168	{
169	return preg_match('/[\\x{10000}-\\x{10FFFF}]/u', $char);
170	}
171
172	/**
173	* get the character code count
174	*
175	* @param $string
176	* @param string $encoding
177	* @return int
178	*/
179	public static function charCount($string, $encoding = 'UTF-8')
180	{
181	$count = 0;
182	$strlen = static::strlen($string);
183
184	for ($offset = 0; $offset < $strlen; $offset++) {
185	$char = static::substr($string, $offset, 1, $encoding);
186	$count += static::isSurrogatePair($char) ? 2 : 1;
187	}
188
189	return $count;
190	}
191	}
192