reifydb_value/util/
unicode.rs

1// SPDX-License-Identifier: MIT
2// Copyright (c) 2026 ReifyDB
3
4pub trait UnicodeWidthStr {
5	fn width(&self) -> usize;
6}
7
8impl UnicodeWidthStr for str {
9	fn width(&self) -> usize {
10		self.chars().map(char_width).sum()
11	}
12}
13
14impl UnicodeWidthStr for &str {
15	fn width(&self) -> usize {
16		self.chars().map(char_width).sum()
17	}
18}
19
20fn char_width(ch: char) -> usize {
21	match ch {
22		'\x00'..='\x1F' | '\x7F'..='\u{9F}' => 0,
23
24		'\u{1100}'..='\u{115F}'
25		| '\u{2E80}'..='\u{2EFF}'
26		| '\u{2F00}'..='\u{2FDF}'
27		| '\u{3000}'..='\u{303F}'
28		| '\u{3040}'..='\u{309F}'
29		| '\u{30A0}'..='\u{30FF}'
30		| '\u{3100}'..='\u{312F}'
31		| '\u{3130}'..='\u{318F}'
32		| '\u{31A0}'..='\u{31BF}'
33		| '\u{31F0}'..='\u{31FF}'
34		| '\u{3200}'..='\u{32FF}'
35		| '\u{3300}'..='\u{33FF}'
36		| '\u{3400}'..='\u{4DBF}'
37		| '\u{4E00}'..='\u{9FFF}'
38		| '\u{A000}'..='\u{A48F}'
39		| '\u{A490}'..='\u{A4CF}'
40		| '\u{AC00}'..='\u{D7AF}'
41		| '\u{F900}'..='\u{FAFF}'
42		| '\u{FE30}'..='\u{FE4F}'
43		| '\u{FF00}'..='\u{FF60}'
44		| '\u{FFE0}'..='\u{FFE6}'
45		| '\u{20000}'..='\u{2FFFD}' => 2,
46
47		'\u{1F300}'..='\u{1F6FF}'
48		| '\u{1F700}'..='\u{1F77F}'
49		| '\u{1F780}'..='\u{1F7FF}'
50		| '\u{1F800}'..='\u{1F8FF}'
51		| '\u{1F900}'..='\u{1F9FF}'
52		| '\u{1FA00}'..='\u{1FA6F}'
53		| '\u{1FA70}'..='\u{1FAFF}' => 2,
54
55		'\u{200B}'..='\u{200F}' | '\u{2028}'..='\u{202E}' | '\u{2060}'..='\u{206F}' => 0,
56
57		'\u{0300}'..='\u{036F}'
58		| '\u{1AB0}'..='\u{1AFF}'
59		| '\u{1DC0}'..='\u{1DFF}'
60		| '\u{FE20}'..='\u{FE2F}' => 0,
61
62		'\u{FE00}'..='\u{FE0F}' => 0,
63		'\u{E0100}'..='\u{E01EF}' => 0,
64
65		'\u{2600}'..='\u{27BF}' => 2,
66
67		_ => 1,
68	}
69}
70
71#[cfg(test)]
72pub mod tests {
73	use super::*;
74
75	#[test]
76	fn test_ascii() {
77		assert_eq!("Hello".width(), 5);
78		assert_eq!("Hello, World!".width(), 13);
79		assert_eq!("".width(), 0);
80	}
81
82	#[test]
83	fn test_cjk() {
84		assert_eq!("你好".width(), 4); // Two Chinese characters
85		assert_eq!("こんにちは".width(), 10); // Five Japanese characters
86		assert_eq!("안녕하세요".width(), 10); // Five Korean characters
87	}
88
89	#[test]
90	fn test_mixed() {
91		assert_eq!("Hello 世界".width(), 10); // 6 for "Hello " + 4 for two CJK chars
92	}
93
94	#[test]
95	fn test_control_chars() {
96		assert_eq!("\x00\x01\x02".width(), 0);
97		assert_eq!("Hello\nWorld".width(), 10); // newline has 0 width
98		assert_eq!("Hello\tWorld".width(), 10); // tab has 0 width
99	}
100
101	#[test]
102	fn test_combining_marks() {
103		// e with combining acute accent
104		assert_eq!("e\u{0301}".width(), 1);
105		// a with combining tilde
106		assert_eq!("a\u{0303}".width(), 1);
107	}
108
109	#[test]
110	fn test_emoji() {
111		assert_eq!("🚀".width(), 2); // Rocket emoji
112		assert_eq!("😀".width(), 2); // Smiley face
113		assert_eq!("🎉".width(), 2); // Party popper
114		assert_eq!("Unicode: 🚀 ñ é ü".width(), 17); // "Unicode: " (9) + 🚀 (2) + " ñ é ü" (6) = 17
115	}
116}
reifydb_value/util/unicode.rs

reifydb_value/util/
unicode.rs
