socialde.pt
Laravel AT Protocol Client (alpha & unstable)
1<?php
2
3namespace SocialDept\AtpClient\RichText;
4
5class ByteCounter
6{
7 /**
8 * Get the byte length of a UTF-8 string
9 */
10 public static function length(string $text): int
11 {
12 return strlen($text);
13 }
14
15 /**
16 * Get byte position of character at given index
17 */
18 public static function bytePosition(string $text, int $charIndex): int
19 {
20 $chars = mb_str_split($text, 1, 'UTF-8');
21 $bytePos = 0;
22
23 for ($i = 0; $i < $charIndex && $i < count($chars); $i++) {
24 $bytePos += strlen($chars[$i]);
25 }
26
27 return $bytePos;
28 }
29
30 /**
31 * Get substring by byte positions
32 */
33 public static function substring(string $text, int $byteStart, int $byteEnd): string
34 {
35 return substr($text, $byteStart, $byteEnd - $byteStart);
36 }
37
38 /**
39 * Validate byte positions don't split multi-byte characters
40 */
41 public static function validateBytePositions(string $text, int $byteStart, int $byteEnd): bool
42 {
43 // Check if positions are within bounds
44 if ($byteStart < 0 || $byteEnd > strlen($text) || $byteStart > $byteEnd) {
45 return false;
46 }
47
48 // Ensure we're not splitting a multi-byte character
49 $before = substr($text, 0, $byteStart);
50 $middle = substr($text, $byteStart, $byteEnd - $byteStart);
51 $after = substr($text, $byteEnd);
52
53 // Check if reconstructed string is valid UTF-8
54 return mb_check_encoding($before, 'UTF-8')
55 && mb_check_encoding($middle, 'UTF-8')
56 && mb_check_encoding($after, 'UTF-8');
57 }
58}