| 1 | // Copyright 2012-2025 The Rust Project Developers. See the COPYRIGHT |
| 2 | // file at the top-level directory of this distribution and at |
| 3 | // http://rust-lang.org/COPYRIGHT. |
| 4 | // |
| 5 | // Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or |
| 6 | // http://www.apache.org/licenses/LICENSE-2.0> or the MIT license |
| 7 | // <LICENSE-MIT or http://opensource.org/licenses/MIT>, at your |
| 8 | // option. This file may not be copied, modified, or distributed |
| 9 | // except according to those terms. |
| 10 | |
| 11 | use std::{ |
| 12 | fs::File, |
| 13 | io::{BufRead, BufReader}, |
| 14 | }; |
| 15 | |
| 16 | use unicode_width::{UnicodeWidthChar, UnicodeWidthStr}; |
| 17 | |
| 18 | macro_rules! assert_width { |
| 19 | ($s:expr, $nocjk:expr, $cjk:expr $(,)?) => {{ |
| 20 | assert_eq!($s.width(), $nocjk, "{:?} has the wrong width", $s); |
| 21 | #[cfg(feature = "cjk")] |
| 22 | assert_eq!($s.width_cjk(), $cjk, "{:?} has the wrong width (CJK)", $s); |
| 23 | }}; |
| 24 | } |
| 25 | |
| 26 | #[test] |
| 27 | fn test_str() { |
| 28 | assert_width!("hello", 10, 10); |
| 29 | assert_width!("\0\0\0\x01\x01", 5, 5); |
| 30 | assert_width!("", 0, 0); |
| 31 | assert_width!("\u{2081}\u{2082}\u{2083}\u{2084}", 4, 8); |
| 32 | } |
| 33 | |
| 34 | #[test] |
| 35 | fn test_emoji() { |
| 36 | assert_width!("👩", 2, 2); // Woman |
| 37 | assert_width!("🔬", 2, 2); // Microscope |
| 38 | assert_width!("👩🔬", 2, 2); // Woman scientist |
| 39 | } |
| 40 | |
| 41 | // From README |
| 42 | #[test] |
| 43 | fn test_bad_devanagari() { |
| 44 | assert_eq!("क".width(), 1); // Devanagari letter Ka |
| 45 | assert_eq!("ष".width(), 1); // Devanagari letter Ssa |
| 46 | assert_eq!("क्ष".width(), 2); // Ka + Virama + Ssa |
| 47 | } |
| 48 | |
| 49 | #[test] |
| 50 | fn test_char() { |
| 51 | assert_width!('h', Some(2), Some(2)); |
| 52 | assert_width!('\x00', None, None); |
| 53 | assert_width!('\x01', None, None); |
| 54 | assert_width!('\u{2081}', Some(1), Some(2)); |
| 55 | } |
| 56 | |
| 57 | #[test] |
| 58 | fn test_char2() { |
| 59 | assert_width!('\x0A', None, None); |
| 60 | assert_width!('w', Some(1), Some(1)); |
| 61 | assert_width!('h', Some(2), Some(2)); |
| 62 | assert_width!('\u{AD}', Some(0), Some(0)); |
| 63 | assert_width!('\u{1160}', Some(0), Some(0)); |
| 64 | assert_width!('\u{a1}', Some(1), Some(2)); |
| 65 | assert_width!('\u{300}', Some(0), Some(0)); |
| 66 | } |
| 67 | |
| 68 | #[test] |
| 69 | fn unicode_12() { |
| 70 | assert_width!('\u{1F971}', Some(2), Some(2)); |
| 71 | } |
| 72 | |
| 73 | #[test] |
| 74 | fn test_default_ignorable() { |
| 75 | assert_width!('\u{1160}', Some(0), Some(0)); |
| 76 | assert_width!('\u{3164}', Some(0), Some(0)); |
| 77 | assert_width!('\u{FFA0}', Some(0), Some(0)); |
| 78 | assert_width!('\u{E0000}', Some(0), Some(0)); |
| 79 | } |
| 80 | |
| 81 | #[test] |
| 82 | fn test_ambiguous() { |
| 83 | assert_width!("\u{B7}", 1, 2); |
| 84 | assert_width!("\u{0387}", 1, 2); |
| 85 | assert_width!("\u{A8}", 1, 1); |
| 86 | assert_width!("\u{02C9}", 1, 1); |
| 87 | } |
| 88 | |
| 89 | #[test] |
| 90 | fn test_jamo() { |
| 91 | assert_width!('\u{1100}', Some(2), Some(2)); |
| 92 | assert_width!('\u{A97C}', Some(2), Some(2)); |
| 93 | // Special case: U+115F HANGUL CHOSEONG FILLER |
| 94 | assert_width!('\u{115F}', Some(2), Some(2)); |
| 95 | assert_width!('\u{1160}', Some(0), Some(0)); |
| 96 | assert_width!('\u{D7C6}', Some(0), Some(0)); |
| 97 | assert_width!('\u{11A8}', Some(0), Some(0)); |
| 98 | assert_width!('\u{D7FB}', Some(0), Some(0)); |
| 99 | } |
| 100 | |
| 101 | #[test] |
| 102 | fn test_prepended_concatenation_marks() { |
| 103 | for c in [ |
| 104 | '\u{0600}', |
| 105 | '\u{0601}', |
| 106 | '\u{0602}', |
| 107 | '\u{0603}', |
| 108 | '\u{0604}', |
| 109 | '\u{06DD}', |
| 110 | '\u{110BD}', |
| 111 | '\u{110CD}', |
| 112 | ] { |
| 113 | assert_width!(c, Some(1), Some(1)); |
| 114 | } |
| 115 | |
| 116 | for c in ['\u{0605}', '\u{070F}', '\u{0890}', '\u{0891}', '\u{08E2}'] { |
| 117 | assert_width!(c, Some(0), Some(0)); |
| 118 | } |
| 119 | } |
| 120 | |
| 121 | #[test] |
| 122 | fn test_gcb_prepend() { |
| 123 | assert_width!("ൎഉ", 1, 1); |
| 124 | assert_width!("\u{11A89}", 0, 0); |
| 125 | } |
| 126 | |
| 127 | #[test] |
| 128 | fn test_interlinear_annotation_chars() { |
| 129 | assert_width!('\u{FFF9}', Some(1), Some(1)); |
| 130 | assert_width!('\u{FFFA}', Some(1), Some(1)); |
| 131 | assert_width!('\u{FFFB}', Some(1), Some(1)); |
| 132 | } |
| 133 | |
| 134 | #[test] |
| 135 | fn test_hieroglyph_format_controls() { |
| 136 | assert_width!('\u{13430}', Some(1), Some(1)); |
| 137 | assert_width!('\u{13436}', Some(1), Some(1)); |
| 138 | assert_width!('\u{1343C}', Some(1), Some(1)); |
| 139 | } |
| 140 | |
| 141 | #[test] |
| 142 | fn test_marks() { |
| 143 | // Nonspacing marks have 0 width |
| 144 | assert_width!('\u{0301}', Some(0), Some(0)); |
| 145 | // Enclosing marks have 0 width |
| 146 | assert_width!('\u{20DD}', Some(0), Some(0)); |
| 147 | // Some spacing marks have width 1 |
| 148 | assert_width!('\u{09CB}', Some(1), Some(1)); |
| 149 | // But others have width 0 |
| 150 | assert_width!('\u{09BE}', Some(0), Some(0)); |
| 151 | } |
| 152 | |
| 153 | #[test] |
| 154 | fn test_devanagari_caret() { |
| 155 | assert_width!('\u{A8FA}', Some(0), Some(0)); |
| 156 | } |
| 157 | |
| 158 | #[test] |
| 159 | fn test_solidus_overlay() { |
| 160 | assert_width!("<\u{338}", 1, 2); |
| 161 | assert_width!("=\u{338}", 1, 2); |
| 162 | assert_width!(">\u{338}", 1, 2); |
| 163 | assert_width!("=\u{301}\u{338}", 1, 2); |
| 164 | assert_width!("=\u{338}\u{301}", 1, 2); |
| 165 | assert_width!("=\u{FE0F}\u{338}", 1, 2); |
| 166 | assert_width!("#\u{FE0F}\u{338}", 2, 2); |
| 167 | assert_width!("#\u{338}\u{FE0F}", 1, 1); |
| 168 | |
| 169 | assert_width!("\u{06B8}\u{338}\u{0627}", 1, 1); |
| 170 | assert_width!("\u{06B8}\u{338}\u{FE0E}\u{0627}", 1, 1); |
| 171 | assert_width!("\u{06B8}\u{338}\u{FE0F}\u{0627}", 1, 1); |
| 172 | assert_width!("\u{06B8}\u{FE0E}\u{338}\u{0627}", 1, 1); |
| 173 | assert_width!("\u{06B8}\u{FE0F}\u{338}\u{0627}", 1, 1); |
| 174 | |
| 175 | assert_width!("=\u{338}\u{0627}", 2, 3); |
| 176 | } |
| 177 | |
| 178 | #[test] |
| 179 | fn test_emoji_presentation() { |
| 180 | assert_width!('\u{0023}', Some(1), Some(1)); |
| 181 | assert_width!('\u{FE0F}', Some(0), Some(0)); |
| 182 | assert_width!("\u{0023}\u{FE0F}", 2, 2); |
| 183 | assert_width!("a\u{0023}\u{FE0F}a", 4, 4); |
| 184 | assert_width!("\u{0023}a\u{FE0F}", 2, 2); |
| 185 | assert_width!("a\u{FE0F}", 1, 1); |
| 186 | assert_width!("\u{0023}\u{0023}\u{FE0F}a", 4, 4); |
| 187 | assert_width!("\u{002A}\u{FE0F}", 2, 2); |
| 188 | assert_width!("\u{23F9}\u{FE0F}", 2, 2); |
| 189 | assert_width!("\u{24C2}\u{FE0F}", 2, 2); |
| 190 | assert_width!("\u{1F6F3}\u{FE0F}", 2, 2); |
| 191 | assert_width!("\u{1F700}\u{FE0F}", 1, 1); |
| 192 | assert_width!("\u{002A}\u{301}\u{FE0F}", 1, 1); |
| 193 | assert_width!("\u{002A}\u{200D}\u{FE0F}", 1, 1); |
| 194 | assert_width!("\u{002A}\u{FE0E}\u{FE0F}", 1, 1); |
| 195 | } |
| 196 | |
| 197 | #[test] |
| 198 | fn test_text_presentation() { |
| 199 | assert_width!('\u{FE0E}', Some(0), Some(0)); |
| 200 | assert_width!('\u{2648}', Some(2), Some(2)); |
| 201 | assert_width!("\u{2648}\u{FE0E}", 1, 2); |
| 202 | assert_width!("\u{2648}\u{FE0E}\u{FE0F}", 1, 2); |
| 203 | assert_width!("\u{2648}\u{FE0F}\u{FE0E}", 2, 2); |
| 204 | assert_width!("\u{1F21A}\u{FE0E}", 2, 2); |
| 205 | assert_width!("\u{0301}\u{FE0E}", 0, 0); |
| 206 | assert_width!("a\u{FE0E}", 1, 1); |
| 207 | assert_width!("𘀀\u{FE0E}", 2, 2); |
| 208 | assert_width!("\u{2648}\u{0301}\u{FE0E}", 2, 2); |
| 209 | assert_width!("\u{2648}\u{200D}\u{FE0E}", 2, 2); |
| 210 | } |
| 211 | |
| 212 | #[test] |
| 213 | fn test_control_line_break() { |
| 214 | assert_width!('\u{2028}', Some(1), Some(1)); |
| 215 | assert_width!('\u{2029}', Some(1), Some(1)); |
| 216 | assert_width!('\r', None, None); |
| 217 | assert_width!('\n', None, None); |
| 218 | assert_width!("\r", 1, 1); |
| 219 | assert_width!("\n", 1, 1); |
| 220 | assert_width!("\r\n", 1, 1); |
| 221 | assert_width!("\0", 1, 1); |
| 222 | assert_width!("1\t2\r\n3\u{85}4", 7, 7); |
| 223 | assert_width!("\r\u{FE0F}\n", 2, 2); |
| 224 | assert_width!("\r\u{200D}\n", 2, 2); |
| 225 | } |
| 226 | |
| 227 | #[test] |
| 228 | fn char_str_consistent() { |
| 229 | let mut s = String::with_capacity(4); |
| 230 | for c in '\0'..=char::MAX { |
| 231 | s.clear(); |
| 232 | s.push(c); |
| 233 | assert_eq!(c.width().unwrap_or(1), s.width()); |
| 234 | #[cfg(feature = "cjk")] |
| 235 | assert_eq!(c.width_cjk().unwrap_or(1), s.width_cjk()); |
| 236 | } |
| 237 | } |
| 238 | |
| 239 | #[test] |
| 240 | fn test_lisu_tones() { |
| 241 | for c in '\u{A4F8}'..='\u{A4FD}' { |
| 242 | assert_width!(c, Some(1), Some(1)); |
| 243 | assert_width!(String::from(c), 1, 1); |
| 244 | } |
| 245 | for c1 in '\u{A4F8}'..='\u{A4FD}' { |
| 246 | for c2 in '\u{A4F8}'..='\u{A4FD}' { |
| 247 | let mut s = String::with_capacity(8); |
| 248 | s.push(c1); |
| 249 | s.push(c2); |
| 250 | match (c1, c2) { |
| 251 | ('\u{A4F8}'..='\u{A4FB}', '\u{A4FC}'..='\u{A4FD}') => assert_width!(s, 1, 1), |
| 252 | _ => assert_width!(s, 2, 2), |
| 253 | } |
| 254 | } |
| 255 | } |
| 256 | |
| 257 | assert_width!("ꓪꓹ", 2, 2); |
| 258 | assert_width!("ꓪꓹꓼ", 2, 2); |
| 259 | assert_width!("ꓪꓹ\u{FE0F}ꓼ", 2, 2); |
| 260 | assert_width!("ꓪꓹ\u{200D}ꓼ", 2, 2); |
| 261 | assert_width!("ꓪꓹꓼ\u{FE0F}", 2, 2); |
| 262 | assert_width!("ꓪꓹ\u{0301}ꓼ", 3, 3); |
| 263 | assert_width!("ꓪꓹꓹ", 3, 3); |
| 264 | assert_width!("ꓪꓼꓼ", 3, 3); |
| 265 | } |
| 266 | |
| 267 | #[test] |
| 268 | fn test_hebrew_alef_lamed() { |
| 269 | assert_width!("\u{05D0}", 1, 1); |
| 270 | assert_width!("\u{05DC}", 1, 1); |
| 271 | assert_width!("\u{05D0}\u{05DC}", 2, 2); |
| 272 | assert_width!("\u{05D0}\u{200D}\u{05DC}", 1, 1); |
| 273 | assert_width!( |
| 274 | "\u{05D0}\u{200D}\u{200D}\u{200D}\u{200D}\u{200D}\u{200D}\u{200D}\u{05DC}", |
| 275 | 1, |
| 276 | 1, |
| 277 | ); |
| 278 | assert_width!("\u{05D0}\u{05D0}\u{200D}\u{05DC}", 2, 2); |
| 279 | assert_width!( |
| 280 | "\u{05D0}\u{05D0}\u{200D}\u{200D}\u{200D}\u{200D}\u{200D}\u{200D}\u{05DC}", |
| 281 | 2, |
| 282 | 2, |
| 283 | ); |
| 284 | assert_width!("\u{05D0}\u{FE0F}\u{200D}\u{FE0F}\u{05DC}\u{FE0F}", 1, 1); |
| 285 | assert_width!("\u{05D0}\u{FE0E}\u{200D}\u{FE0E}\u{05DC}\u{FE0E}", 1, 1); |
| 286 | } |
| 287 | |
| 288 | #[test] |
| 289 | fn test_arabic_lam_alef() { |
| 290 | assert_width!("\u{0644}", 1, 1); |
| 291 | assert_width!("\u{06B8}", 1, 1); |
| 292 | |
| 293 | assert_width!("\u{0623}", 1, 1); |
| 294 | assert_width!("\u{0627}", 1, 1); |
| 295 | |
| 296 | assert_width!("\u{0644}\u{0623}", 1, 1); |
| 297 | assert_width!("\u{0644}\u{0627}", 1, 1); |
| 298 | assert_width!("\u{06B8}\u{0623}", 1, 1); |
| 299 | assert_width!("\u{06B8}\u{0627}", 1, 1); |
| 300 | |
| 301 | assert_width!("\u{0644}\u{065F}\u{065E}\u{0623}", 1, 1); |
| 302 | assert_width!("\u{0644}\u{065F}\u{065E}\u{0627}", 1, 1); |
| 303 | assert_width!("\u{06B8}\u{065F}\u{065E}\u{0623}", 1, 1); |
| 304 | assert_width!("\u{06B8}\u{065F}\u{065E}\u{0627}", 1, 1); |
| 305 | |
| 306 | assert_width!("\u{06B8}\u{FE0E}\u{0627}", 1, 1); |
| 307 | assert_width!("\u{06B8}\u{FE0F}\u{0627}", 1, 1); |
| 308 | assert_width!("\u{06B8}\u{17B5}\u{0627}", 1, 1); |
| 309 | |
| 310 | assert_width!("\u{0644}\u{0644}\u{0623}", 2, 2); |
| 311 | assert_width!("\u{0644}\u{0644}\u{0627}", 2, 2); |
| 312 | assert_width!("\u{06B8}\u{06B8}\u{0623}", 2, 2); |
| 313 | assert_width!("\u{06B8}\u{06B8}\u{0627}", 2, 2); |
| 314 | |
| 315 | assert_width!("\u{0644}\u{200D}\u{0623}", 2, 2); |
| 316 | assert_width!("\u{0644}\u{200D}\u{0627}", 2, 2); |
| 317 | assert_width!("\u{06B8}\u{200D}\u{0623}", 2, 2); |
| 318 | assert_width!("\u{06B8}\u{200D}\u{0627}", 2, 2); |
| 319 | |
| 320 | assert_width!("\u{0644}\u{1E94B}\u{0623}", 3, 3); |
| 321 | assert_width!("\u{0644}\u{1E94B}\u{0627}", 3, 3); |
| 322 | assert_width!("\u{06B8}\u{1E94B}\u{0623}", 3, 3); |
| 323 | assert_width!("\u{06B8}\u{1E94B}\u{0627}", 3, 3); |
| 324 | } |
| 325 | |
| 326 | #[test] |
| 327 | fn test_buginese_a_i_ya() { |
| 328 | assert_width!("\u{1A15}", 1, 1); |
| 329 | assert_width!("\u{1A17}", 0, 0); |
| 330 | assert_width!("\u{1A10}", 1, 1); |
| 331 | |
| 332 | assert_width!("\u{1A15}\u{1A17}\u{200D}\u{1A10}", 1, 1); |
| 333 | assert_width!( |
| 334 | "\u{1A15}\u{1A17}\u{200D}\u{200D}\u{200D}\u{200D}\u{1A10}", |
| 335 | 1, |
| 336 | 1, |
| 337 | ); |
| 338 | assert_width!("\u{1A15}\u{1A17}\u{200D}\u{338}", 1, 1); |
| 339 | assert_width!("\u{1A15}\u{FE0E}\u{1A17}\u{200D}", 1, 1); |
| 340 | assert_width!("\u{1A15}\u{FE0F}\u{1A17}\u{200D}", 1, 1); |
| 341 | assert_width!("\u{1A15}\u{1A17}\u{FE0E}\u{200D}", 1, 1); |
| 342 | assert_width!("\u{1A15}\u{1A17}\u{FE0F}\u{200D}", 1, 1); |
| 343 | assert_width!("\u{1A15}\u{1A17}\u{200D}\u{FE0E}", 1, 1); |
| 344 | assert_width!("\u{1A15}\u{1A17}\u{200D}\u{FE0F}", 1, 1); |
| 345 | assert_width!( |
| 346 | "\u{1A15}\u{17B5}\u{200D}\u{FE0E}\u{1A17}\u{200D}\u{FE0F}\u{200D}\u{FE0F}", |
| 347 | 1, |
| 348 | 1, |
| 349 | ); |
| 350 | |
| 351 | assert_width!("\u{1A15}\u{1A15}\u{1A17}\u{200D}\u{1A10}", 2, 2); |
| 352 | assert_width!( |
| 353 | "\u{1A15}\u{1A15}\u{1A17}\u{200D}\u{200D}\u{200D}\u{200D}\u{1A10}", |
| 354 | 2, |
| 355 | 2, |
| 356 | ); |
| 357 | |
| 358 | assert_width!("\u{1A15}\u{1A17}\u{1A10}", 2, 2); |
| 359 | assert_width!("\u{1A15}\u{200D}\u{1A10}", 2, 2); |
| 360 | assert_width!("\u{1A15}\u{1A10}", 2, 2); |
| 361 | assert_width!("\u{1A15}\u{1A17}\u{1A17}\u{200D}\u{1A10}", 2, 2); |
| 362 | assert_width!("\u{1A15}\u{1A17}\u{338}\u{200D}\u{1A10}", 2, 2); |
| 363 | } |
| 364 | |
| 365 | #[test] |
| 366 | fn test_tifinagh_biconsonants() { |
| 367 | assert_width!("\u{2D4F}", 1, 1); |
| 368 | assert_width!("\u{2D3E}", 1, 1); |
| 369 | assert_width!("\u{2D7F}", 1, 1); |
| 370 | |
| 371 | assert_width!("\u{2D4F}\u{200D}\u{2D3E}", 1, 1); |
| 372 | assert_width!("\u{2D4F}\u{2D7F}\u{2D3E}", 1, 1); |
| 373 | assert_width!("\u{2D4F}\u{200D}\u{2D3E}", 1, 1); |
| 374 | assert_width!( |
| 375 | "\u{2D4F}\u{FE0F}\u{200D}\u{2D7F}\u{FE0E}\u{200D}\u{17B5}\u{2D3E}", |
| 376 | 1, |
| 377 | 1, |
| 378 | ); |
| 379 | |
| 380 | assert_width!("\u{2D4F}\u{301}\u{2D7F}\u{2D3E}", 3, 3); |
| 381 | assert_width!("\u{2D4F}\u{301}\u{200D}\u{2D3E}", 2, 2); |
| 382 | assert_width!("\u{2D4F}\u{2D3E}", 2, 2); |
| 383 | assert_width!("\u{2D4F}\u{2D7F}\u{2D7F}\u{2D3E}", 4, 4); |
| 384 | assert_width!("\u{2D7F}\u{2D3E}", 2, 2); |
| 385 | assert_width!("\u{2D7F}\u{2D7F}\u{2D66}", 3, 3); |
| 386 | assert_width!("\u{2D66}\u{2D7F}\u{2D3E}", 3, 3); |
| 387 | } |
| 388 | |
| 389 | #[test] |
| 390 | fn test_old_turkic_ligature() { |
| 391 | assert_width!("\u{10C32}", 1, 1); |
| 392 | assert_width!("\u{10C03}", 1, 1); |
| 393 | assert_width!("\u{10C32}\u{10C03}", 2, 2); |
| 394 | |
| 395 | assert_width!("\u{10C32}\u{200D}\u{10C03}", 1, 1); |
| 396 | assert_width!("\u{10C32}\u{FE0F}\u{200D}\u{FE0E}\u{10C03}", 1, 1); |
| 397 | |
| 398 | assert_width!("\u{10C32}\u{2D7F}\u{10C03}", 3, 3); |
| 399 | assert_width!("\u{10C32}\u{0301}\u{200D}\u{10C03}", 2, 2); |
| 400 | assert_width!("\u{10C03}\u{200D}\u{10C32}", 2, 2); |
| 401 | assert_width!("\u{200D}\u{10C32}", 1, 1); |
| 402 | } |
| 403 | |
| 404 | #[test] |
| 405 | fn test_khmer_coeng() { |
| 406 | assert_width!("ល", 1, 1); |
| 407 | assert_width!("ង", 1, 1); |
| 408 | assert_width!("លង", 2, 2); |
| 409 | assert_width!("ល្ង", 1, 1); |
| 410 | |
| 411 | for c in '\0'..=char::MAX { |
| 412 | if matches!( |
| 413 | c, |
| 414 | '\u{1780}'..='\u{1782}' | '\u{1784}'..='\u{1787}' |
| 415 | | '\u{1789}'..='\u{178C}' | '\u{178E}'..='\u{1793}' |
| 416 | | '\u{1795}'..='\u{1798}' | '\u{179B}'..='\u{179D}' |
| 417 | | '\u{17A0}' | '\u{17A2}' | '\u{17A7}' |
| 418 | | '\u{17AB}'..='\u{17AC}' | '\u{17AF}' |
| 419 | ) { |
| 420 | assert_width!(format!("\u{17D2}{c}"), 0, 0); |
| 421 | assert_width!(format!("\u{17D2}\u{200D}\u{200D}{c}"), 0, 0); |
| 422 | } else { |
| 423 | assert_width!( |
| 424 | format!("\u{17D2}{c}"), |
| 425 | c.width().unwrap_or(1), |
| 426 | c.width_cjk().unwrap_or(1) |
| 427 | ); |
| 428 | } |
| 429 | } |
| 430 | } |
| 431 | |
| 432 | #[test] |
| 433 | fn test_khmer_qaa() { |
| 434 | assert_width!("\u{17A4}", 2, 2); |
| 435 | assert_width!("\u{17A2}\u{17A6}", 2, 2); |
| 436 | } |
| 437 | |
| 438 | #[test] |
| 439 | fn test_khmer_sign_beyyal() { |
| 440 | assert_width!("\u{17D8}", 3, 3); |
| 441 | assert_width!("\u{17D4}\u{179B}\u{17D4}", 3, 3); |
| 442 | } |
| 443 | |
| 444 | #[test] |
| 445 | fn test_emoji_modifier() { |
| 446 | assert_width!("\u{1F46A}", 2, 2); |
| 447 | assert_width!("\u{1F3FB}", 2, 2); |
| 448 | assert_width!("\u{1F46A}\u{1F3FB}", 2, 2); |
| 449 | assert_width!("\u{1F46A}\u{200D}\u{200D}\u{1F3FB}", 4, 4); |
| 450 | } |
| 451 | |
| 452 | #[test] |
| 453 | fn test_emoji_zwj() { |
| 454 | assert_width!("🧑🤝🧑", 2, 2); |
| 455 | |
| 456 | assert_width!("🇮🇱🕊️🇵🇸", 6, 6); |
| 457 | assert_width!("🇵🇸\u{200D}🕊️\u{200D}🇮🇱", 2, 2); |
| 458 | assert_width!("🇮🇱\u{200D}🕊️\u{200D}\u{200D}🇵🇸", 4, 4); |
| 459 | assert_width!("🇵🇸\u{200D}\u{200D}🕊️\u{200D}🇮🇱", 4, 4); |
| 460 | |
| 461 | assert_width!("🇦🇦\u{200D}🇦🇦", 2, 2); |
| 462 | assert_width!("🇦🇦\u{200D}🇦🇦🇦", 3, 3); |
| 463 | assert_width!("🇦🇦\u{200D}🇦🇦🇦", 3, 3); |
| 464 | |
| 465 | assert_width!("🇦🇦\u{200D}\u{200D}🇦🇦", 4, 4); |
| 466 | assert_width!("🇦🇦\u{200D}🇦\u{200D}🇦🇦", 5, 5); |
| 467 | assert_width!("🇦🇦\u{200D}🇦🇦\u{200D}🇦🇦", 2, 2); |
| 468 | assert_width!("🇦🇦\u{200D}🇦🇦🇦\u{200D}🇦🇦", 5, 5); |
| 469 | assert_width!("🇦🇦\u{200D}🇦🇦🇦🇦\u{200D}🇦🇦", 4, 4); |
| 470 | assert_width!("🇦🇦\u{200D}🇦🇦🇦🇦🇦\u{200D}🇦🇦", 7, 7); |
| 471 | assert_width!("🇦🇦\u{200D}🇦🇦🇦🇦🇦🇦\u{200D}🇦🇦", 6, 6); |
| 472 | assert_width!("🇦🇦\u{200D}🇦🇦🇦🇦🇦🇦🇦\u{200D}🇦🇦", 9, 9); |
| 473 | |
| 474 | assert_width!("🏴", 2, 2); |
| 475 | assert_width!("🏴\u{200D}🏴\u{200D}🏴", 2, 2); |
| 476 | |
| 477 | assert_width!("🇦👪\u{200D}🏿", 3, 3); |
| 478 | assert_width!("🇦🏿\u{200D}🏿", 3, 3); |
| 479 | |
| 480 | assert_width!('🏴', Some(2), Some(2)); |
| 481 | assert_width!("\u{E0031}", 0, 0); |
| 482 | assert_width!("\u{E0063}", 0, 0); |
| 483 | assert_width!("\u{E007F}", 0, 0); |
| 484 | assert_width!("🏴\u{200D}Ⓜ️", 2, 2); |
| 485 | assert_width!("🏴\u{E0031}\u{200D}Ⓜ️", 4, 4); |
| 486 | assert_width!("🏴\u{E0063}\u{200D}Ⓜ️", 4, 4); |
| 487 | assert_width!("🏴\u{E007F}\u{200D}Ⓜ️", 4, 4); |
| 488 | assert_width!("🏴\u{E0031}\u{E007F}\u{200D}Ⓜ️", 4, 4); |
| 489 | assert_width!("🏴\u{E0031}\u{E0031}\u{E007F}\u{200D}Ⓜ️", 4, 4); |
| 490 | assert_width!("🏴\u{E0031}\u{E0031}\u{E0031}\u{E007F}\u{200D}Ⓜ️", 2, 2); |
| 491 | assert_width!( |
| 492 | "🏴\u{E0031}\u{E0031}\u{E0031}\u{E0031}\u{E007F}\u{200D}Ⓜ️", |
| 493 | 4, |
| 494 | 4, |
| 495 | ); |
| 496 | assert_width!( |
| 497 | "🏴\u{E0031}\u{E0031}\u{E0031}\u{E0063}\u{E007F}\u{200D}Ⓜ️", |
| 498 | 2, |
| 499 | 2, |
| 500 | ); |
| 501 | assert_width!( |
| 502 | "🏴\u{E0031}\u{E0031}\u{E0031}\u{E0063}\u{E0063}\u{E007F}\u{200D}Ⓜ️", |
| 503 | 2, |
| 504 | 2, |
| 505 | ); |
| 506 | assert_width!( |
| 507 | "🏴\u{E0031}\u{E0031}\u{E0031}\u{E0063}\u{E0063}\u{E0063}\u{E007F}\u{200D}Ⓜ️", |
| 508 | 2, |
| 509 | 2, |
| 510 | ); |
| 511 | assert_width!( |
| 512 | "🏴\u{E0031}\u{E0031}\u{E0031}\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E007F}\u{200D}Ⓜ️", |
| 513 | 2, |
| 514 | 2, |
| 515 | ); |
| 516 | assert_width!( |
| 517 | "🏴\u{E0031}\u{E0031}\u{E0031}\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E007F}\u{200D}Ⓜ️", |
| 518 | 4, |
| 519 | 4, |
| 520 | ); |
| 521 | assert_width!("🏴\u{E0063}\u{E0063}\u{E007F}\u{200D}Ⓜ️", 4, 4); |
| 522 | assert_width!("🏴\u{E0063}\u{E0063}\u{E0063}\u{E007F}\u{200D}Ⓜ️", 2, 2); |
| 523 | assert_width!( |
| 524 | "🏴\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E007F}\u{200D}Ⓜ️", |
| 525 | 2, |
| 526 | 2, |
| 527 | ); |
| 528 | assert_width!( |
| 529 | "🏴\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E007F}\u{200D}Ⓜ️", |
| 530 | 2, |
| 531 | 2, |
| 532 | ); |
| 533 | assert_width!( |
| 534 | "🏴\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E007F}\u{200D}Ⓜ️", |
| 535 | 2, |
| 536 | 2, |
| 537 | ); |
| 538 | assert_width!( |
| 539 | "🏴\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E0063}\u{E007F}\u{200D}Ⓜ️", |
| 540 | 4, |
| 541 | 4, |
| 542 | ); |
| 543 | |
| 544 | assert_width!("a\u{200D}🏴", 3, 3); |
| 545 | assert_width!("👪\u{200D}a", 3, 3); |
| 546 | assert_width!("a\u{200D}a", 2, 2); |
| 547 | |
| 548 | assert_width!("*\u{FE0F}", 2, 2); |
| 549 | assert_width!("*\u{20E3}", 1, 1); |
| 550 | assert_width!("*️⃣", 2, 2); |
| 551 | assert_width!("*\u{FE0F}", 2, 2); |
| 552 | assert_width!("*\u{20E3}\u{FE0F}", 1, 1); |
| 553 | assert_width!("*️⃣\u{200D}👪", 2, 2); |
| 554 | assert_width!("*\u{20E3}\u{FE0F}\u{200D}👪", 3, 3); |
| 555 | assert_width!("*\u{20E3}\u{200D}👪", 3, 3); |
| 556 | assert_width!("*\u{FE0F}\u{200D}👪", 2, 2); |
| 557 | assert_width!("*️⃣\u{20E3}\u{200D}👪", 4, 4); |
| 558 | assert_width!("*\u{FE0F}\u{FE0F}\u{20E3}\u{200D}👪", 4, 4); |
| 559 | |
| 560 | assert_width!( |
| 561 | "🇦👪\u{200D}🏿\u{200D}👪🏻\u{200D}Ⓜ️\u{200D}*\u{FE0F}\u{200D}🇦🇦\u{200D}🏴\u{200D}👪", |
| 562 | 3, |
| 563 | 3, |
| 564 | ); |
| 565 | } |
| 566 | |
| 567 | #[test] |
| 568 | fn emoji_test_file() { |
| 569 | let norm_file = BufReader::new( |
| 570 | File::open("tests/emoji-test.txt") |
| 571 | .expect("run `unicode.py` first to download `emoji-test.txt`"), |
| 572 | ); |
| 573 | for line in norm_file.lines() { |
| 574 | let line = line.unwrap(); |
| 575 | if line.is_empty() || line.starts_with('#') { |
| 576 | continue; |
| 577 | } |
| 578 | |
| 579 | let (cps, status) = line.split_once(';').unwrap(); |
| 580 | let status = status.trim(); |
| 581 | if status.starts_with("fully-qualified") || status.starts_with("component") { |
| 582 | let emoji: String = cps |
| 583 | .trim() |
| 584 | .split(' ') |
| 585 | .map(|s| char::try_from(u32::from_str_radix(s, 16).unwrap()).unwrap()) |
| 586 | .collect(); |
| 587 | dbg!(&emoji); |
| 588 | assert_width!(emoji, 2, 2); |
| 589 | } |
| 590 | } |
| 591 | } |
| 592 | |
| 593 | #[test] |
| 594 | fn ambiguous_line_break() { |
| 595 | assert_width!("\u{24EA}", 1, 2); |
| 596 | assert_width!("\u{2616}", 1, 2); |
| 597 | assert_width!("\u{2780}", 1, 2); |
| 598 | } |
| 599 | |
| 600 | #[test] |
| 601 | fn test_vs1_vs2_vs3() { |
| 602 | assert_width!('\u{FE00}', Some(0), Some(0)); |
| 603 | assert_width!('\u{FE01}', Some(0), Some(0)); |
| 604 | assert_width!('\u{FE02}', Some(0), Some(0)); |
| 605 | for c in '\0'..=char::MAX { |
| 606 | if matches!(c, '\u{2018}' | '\u{2019}' | '\u{201C}' | '\u{201D}') { |
| 607 | assert_width!(c, Some(1), Some(2)); |
| 608 | assert_width!(format!("{c}\u{FE00}"), 1, 1); |
| 609 | assert_width!(format!("{c}\u{FE00}\u{FE01}"), 1, 1); |
| 610 | assert_width!(format!("{c}\u{FE01}"), 2, 2); |
| 611 | assert_width!(format!("{c}\u{FE01}\u{FE00}"), 2, 2); |
| 612 | assert_width!(format!("{c}\u{FE02}"), 1, 1); |
| 613 | assert_width!(format!("{c}\u{FE02}\u{FE01}"), 1, 1); |
| 614 | } else { |
| 615 | assert_eq!( |
| 616 | format!("{c}\u{FE00}").width(), |
| 617 | c.width().unwrap_or(1), |
| 618 | "{c:?}" |
| 619 | ); |
| 620 | #[cfg(feature = "cjk")] |
| 621 | assert_eq!( |
| 622 | format!("{c}\u{FE00}").width_cjk(), |
| 623 | c.width_cjk().unwrap_or(1), |
| 624 | "{c:?}" |
| 625 | ); |
| 626 | assert_eq!( |
| 627 | format!("{c}\u{FE01}").width(), |
| 628 | c.width().unwrap_or(1), |
| 629 | "{c:?}" |
| 630 | ); |
| 631 | #[cfg(feature = "cjk")] |
| 632 | assert_eq!( |
| 633 | format!("{c}\u{FE01}").width_cjk(), |
| 634 | c.width_cjk().unwrap_or(1), |
| 635 | "{c:?}" |
| 636 | ); |
| 637 | assert_eq!( |
| 638 | format!("{c}\u{FE02}").width(), |
| 639 | c.width().unwrap_or(1), |
| 640 | "{c:?}" |
| 641 | ); |
| 642 | #[cfg(feature = "cjk")] |
| 643 | assert_eq!( |
| 644 | format!("{c}\u{FE02}").width_cjk(), |
| 645 | c.width_cjk().unwrap_or(1), |
| 646 | "{c:?}" |
| 647 | ); |
| 648 | } |
| 649 | } |
| 650 | } |
| 651 | |
| 652 | // Test traits are unsealed |
| 653 | |
| 654 | #[cfg(feature = "cjk")] |
| 655 | #[allow(dead_code)] |
| 656 | struct Foo; |
| 657 | |
| 658 | #[cfg(feature = "cjk")] |
| 659 | impl UnicodeWidthChar for Foo { |
| 660 | fn width(self) -> Option<usize> { |
| 661 | Some(0) |
| 662 | } |
| 663 | |
| 664 | fn width_cjk(self) -> Option<usize> { |
| 665 | Some(0) |
| 666 | } |
| 667 | } |
| 668 | |
| 669 | #[cfg(feature = "cjk")] |
| 670 | impl UnicodeWidthStr for Foo { |
| 671 | fn width(&self) -> usize { |
| 672 | 0 |
| 673 | } |
| 674 | |
| 675 | fn width_cjk(&self) -> usize { |
| 676 | 0 |
| 677 | } |
| 678 | } |
| 679 |