reifydb_type/util/
unicode.rs

1// Copyright (c) reifydb.com 2025
2// This file is licensed under the AGPL-3.0-or-later, see license.md file
3
4//! Simple unicode width calculation implementation
5
6/// Trait for calculating display width of strings
7pub trait UnicodeWidthStr {
8	/// Returns the display width of the string
9	fn width(&self) -> usize;
10}
11
12impl UnicodeWidthStr for str {
13	fn width(&self) -> usize {
14		self.chars().map(char_width).sum()
15	}
16}
17
18impl UnicodeWidthStr for &str {
19	fn width(&self) -> usize {
20		self.chars().map(char_width).sum()
21	}
22}
23
24/// Calculate the display width of a single character
25fn char_width(ch: char) -> usize {
26	match ch {
27        // Control characters have no width (includes \n, \r, \t)
28        '\x00'..='\x1F' | '\x7F'..='\u{9F}' => 0,
29
30        // Most CJK characters are double-width
31        '\u{1100}'..='\u{115F}' |  // Hangul Jamo
32        '\u{2E80}'..='\u{2EFF}' |  // CJK Radicals Supplement
33        '\u{2F00}'..='\u{2FDF}' |  // Kangxi Radicals
34        '\u{3000}'..='\u{303F}' |  // CJK Symbols and Punctuation
35        '\u{3040}'..='\u{309F}' |  // Hiragana
36        '\u{30A0}'..='\u{30FF}' |  // Katakana
37        '\u{3100}'..='\u{312F}' |  // Bopomofo
38        '\u{3130}'..='\u{318F}' |  // Hangul Compatibility Jamo
39        '\u{31A0}'..='\u{31BF}' |  // Bopomofo Extended
40        '\u{31F0}'..='\u{31FF}' |  // Katakana Phonetic Extensions
41        '\u{3200}'..='\u{32FF}' |  // Enclosed CJK Letters and Months
42        '\u{3300}'..='\u{33FF}' |  // CJK Compatibility
43        '\u{3400}'..='\u{4DBF}' |  // CJK Unified Ideographs Extension A
44        '\u{4E00}'..='\u{9FFF}' |  // CJK Unified Ideographs
45        '\u{A000}'..='\u{A48F}' |  // Yi Syllables
46        '\u{A490}'..='\u{A4CF}' |  // Yi Radicals
47        '\u{AC00}'..='\u{D7AF}' |  // Hangul Syllables
48        '\u{F900}'..='\u{FAFF}' |  // CJK Compatibility Ideographs
49        '\u{FE30}'..='\u{FE4F}' |  // CJK Compatibility Forms
50        '\u{FF00}'..='\u{FF60}' |  // Fullwidth Forms (part)
51        '\u{FFE0}'..='\u{FFE6}' |  // Fullwidth Forms (part)
52        '\u{20000}'..='\u{2FFFD}' => 2,
53
54        // Emoji and symbols (generally double-width)
55        '\u{1F300}'..='\u{1F6FF}' |  // Emoji & Pictographs (includes Regional Indicators U+1F1E6-1F1FF)
56        '\u{1F700}'..='\u{1F77F}' |  // Alchemical Symbols
57        '\u{1F780}'..='\u{1F7FF}' |  // Geometric Shapes Extended
58        '\u{1F800}'..='\u{1F8FF}' |  // Supplemental Arrows-C
59        '\u{1F900}'..='\u{1F9FF}' |  // Supplemental Symbols and Pictographs
60        '\u{1FA00}'..='\u{1FA6F}' |  // Chess Symbols
61        '\u{1FA70}'..='\u{1FAFF}' => 2, // Symbols and Pictographs Extended-A
62
63        // Zero-width characters
64        '\u{200B}'..='\u{200F}' |  // Zero width space, joiners, etc.
65        '\u{2028}'..='\u{202E}' |  // Line/paragraph separators, directional formatting
66        '\u{2060}'..='\u{206F}' => 0,
67
68        // Combining marks (zero width)
69        '\u{0300}'..='\u{036F}' |  // Combining Diacritical Marks
70        '\u{1AB0}'..='\u{1AFF}' |  // Combining Diacritical Marks Extended
71        '\u{1DC0}'..='\u{1DFF}' |  // Combining Diacritical Marks for Symbols
72        '\u{FE20}'..='\u{FE2F}' => 0,
73
74        // Variation selectors (zero width)
75        '\u{FE00}'..='\u{FE0F}' => 0,
76        '\u{E0100}'..='\u{E01EF}' => 0,
77
78        // Some specific double-width symbols
79        '\u{2600}'..='\u{27BF}' => 2, // Miscellaneous Symbols, Dingbats
80        // Default: single width
81        _ => 1}
82}
83
84#[cfg(test)]
85mod tests {
86	use super::*;
87
88	#[test]
89	fn test_ascii() {
90		assert_eq!("Hello".width(), 5);
91		assert_eq!("Hello, World!".width(), 13);
92		assert_eq!("".width(), 0);
93	}
94
95	#[test]
96	fn test_cjk() {
97		assert_eq!("你好".width(), 4); // Two Chinese characters
98		assert_eq!("こんにちは".width(), 10); // Five Japanese characters
99		assert_eq!("안녕하세요".width(), 10); // Five Korean characters
100	}
101
102	#[test]
103	fn test_mixed() {
104		assert_eq!("Hello 世界".width(), 10); // 6 for "Hello " + 4 for two CJK chars
105	}
106
107	#[test]
108	fn test_control_chars() {
109		assert_eq!("\x00\x01\x02".width(), 0);
110		assert_eq!("Hello\nWorld".width(), 10); // newline has 0 width
111		assert_eq!("Hello\tWorld".width(), 10); // tab has 0 width
112	}
113
114	#[test]
115	fn test_combining_marks() {
116		// e with combining acute accent
117		assert_eq!("e\u{0301}".width(), 1);
118		// a with combining tilde
119		assert_eq!("a\u{0303}".width(), 1);
120	}
121
122	#[test]
123	fn test_emoji() {
124		assert_eq!("🚀".width(), 2); // Rocket emoji
125		assert_eq!("😀".width(), 2); // Smiley face
126		assert_eq!("🎉".width(), 2); // Party popper
127		assert_eq!("Unicode: 🚀 ñ é ü".width(), 17); // "Unicode: " (9) + 🚀 (2) + " ñ é ü" (6) = 17
128	}
129}
reifydb_type/util/unicode.rs

reifydb_type/util/
unicode.rs
