| 1 | // Copyright © SixtyFPS GmbH <info@slint.dev> |
| 2 | // SPDX-License-Identifier: GPL-3.0-only OR LicenseRef-Slint-Royalty-free-2.0 OR LicenseRef-Slint-Software-3.0 |
| 3 | |
| 4 | use core::ops::Range; |
| 5 | |
| 6 | use euclid::num::Zero; |
| 7 | |
| 8 | use super::glyphclusters::GlyphClusterIterator; |
| 9 | use super::{BreakOpportunity, LineBreakIterator, ShapeBuffer}; |
| 10 | |
| 11 | #[derive (Debug, PartialEq, Eq, Default)] |
| 12 | pub struct TextFragment<Length> { |
| 13 | pub byte_range: Range<usize>, |
| 14 | pub glyph_range: Range<usize>, |
| 15 | pub width: Length, |
| 16 | pub trailing_whitespace_width: Length, |
| 17 | pub trailing_whitespace_bytes: usize, |
| 18 | pub trailing_mandatory_break: bool, |
| 19 | } |
| 20 | |
| 21 | #[derive (Clone)] |
| 22 | pub struct TextFragmentIterator<'a, Length> { |
| 23 | line_breaks: LineBreakIterator<'a>, |
| 24 | glyph_clusters: GlyphClusterIterator<'a, Length>, |
| 25 | text_len: usize, |
| 26 | pub break_anywhere: bool, |
| 27 | } |
| 28 | |
| 29 | impl<'a, Length> TextFragmentIterator<'a, Length> { |
| 30 | pub fn new(text: &'a str, shape_buffer: &'a ShapeBuffer<Length>) -> Self { |
| 31 | Self { |
| 32 | line_breaks: LineBreakIterator::new(text), |
| 33 | glyph_clusters: GlyphClusterIterator::new(text, shaped_text:shape_buffer), |
| 34 | text_len: text.len(), |
| 35 | break_anywhere: false, |
| 36 | } |
| 37 | } |
| 38 | } |
| 39 | |
| 40 | impl<Length: Clone + Default + core::ops::AddAssign + Zero + Copy> Iterator |
| 41 | for TextFragmentIterator<'_, Length> |
| 42 | { |
| 43 | type Item = TextFragment<Length>; |
| 44 | |
| 45 | fn next(&mut self) -> Option<Self::Item> { |
| 46 | let first_glyph_cluster = self.glyph_clusters.next()?; |
| 47 | |
| 48 | let mut fragment = Self::Item::default(); |
| 49 | |
| 50 | let next_break_offset = if self.break_anywhere { |
| 51 | if first_glyph_cluster.is_line_or_paragraph_separator { |
| 52 | fragment.trailing_mandatory_break = true; |
| 53 | } |
| 54 | 0 |
| 55 | } else if let Some((next_break_offset, break_type)) = self.line_breaks.next() { |
| 56 | if matches!(break_type, BreakOpportunity::Mandatory) { |
| 57 | fragment.trailing_mandatory_break = true; |
| 58 | } |
| 59 | next_break_offset |
| 60 | } else { |
| 61 | self.text_len |
| 62 | }; |
| 63 | |
| 64 | if first_glyph_cluster.is_whitespace { |
| 65 | fragment.trailing_whitespace_width = first_glyph_cluster.width; |
| 66 | fragment.trailing_whitespace_bytes = first_glyph_cluster.byte_range.len(); |
| 67 | fragment.byte_range.start = first_glyph_cluster.byte_range.start; |
| 68 | fragment.byte_range.end = first_glyph_cluster.byte_range.start; |
| 69 | } else { |
| 70 | fragment.width = first_glyph_cluster.width; |
| 71 | fragment.byte_range = first_glyph_cluster.byte_range.clone(); |
| 72 | } |
| 73 | |
| 74 | let start = first_glyph_cluster.glyph_range.start; |
| 75 | let mut last_glyph_cluster = first_glyph_cluster; |
| 76 | |
| 77 | while last_glyph_cluster.byte_range.end < next_break_offset { |
| 78 | let next_glyph_cluster = match self.glyph_clusters.next() { |
| 79 | Some(cluster) => cluster, |
| 80 | None => break, |
| 81 | }; |
| 82 | |
| 83 | if next_glyph_cluster.is_line_or_paragraph_separator { |
| 84 | break; |
| 85 | } |
| 86 | |
| 87 | if next_glyph_cluster.is_whitespace { |
| 88 | fragment.trailing_whitespace_width += next_glyph_cluster.width; |
| 89 | fragment.trailing_whitespace_bytes += next_glyph_cluster.byte_range.len(); |
| 90 | } else { |
| 91 | // transition from whitespace to characters by treating previous trailing whitespace |
| 92 | // as regular characters |
| 93 | if last_glyph_cluster.is_whitespace { |
| 94 | fragment.width += core::mem::take(&mut fragment.trailing_whitespace_width); |
| 95 | fragment.width += next_glyph_cluster.width; |
| 96 | fragment.byte_range.end = next_glyph_cluster.byte_range.end; |
| 97 | fragment.trailing_whitespace_bytes = 0; |
| 98 | } else { |
| 99 | fragment.width += next_glyph_cluster.width; |
| 100 | fragment.byte_range.end = next_glyph_cluster.byte_range.end; |
| 101 | } |
| 102 | } |
| 103 | |
| 104 | last_glyph_cluster = next_glyph_cluster.clone(); |
| 105 | } |
| 106 | |
| 107 | fragment.glyph_range = Range { start, end: last_glyph_cluster.glyph_range.end }; |
| 108 | |
| 109 | // Make sure that adjacent fragments are advanced in their byte range: |
| 110 | // this assertion should hold: fragment.byte_range.end + fragment.trailing_whitespace_bytes == next_fragment.byte_range.start |
| 111 | // That means characters causing mandatory breaks need to be included. |
| 112 | if fragment.trailing_mandatory_break && !self.break_anywhere { |
| 113 | fragment.trailing_whitespace_bytes = next_break_offset - fragment.byte_range.end; |
| 114 | } |
| 115 | |
| 116 | Some(fragment) |
| 117 | } |
| 118 | } |
| 119 | |
| 120 | #[cfg (test)] |
| 121 | use super::{FixedTestFont, TextLayout}; |
| 122 | #[cfg (test)] |
| 123 | use std::{vec, vec::Vec}; |
| 124 | |
| 125 | #[test ] |
| 126 | fn fragment_iterator_simple() { |
| 127 | let font = FixedTestFont; |
| 128 | let text = "H WX" ; |
| 129 | let shape_buffer = ShapeBuffer::new(&TextLayout { font: &font, letter_spacing: None }, text); |
| 130 | let fragments = TextFragmentIterator::new(text, &shape_buffer).collect::<Vec<_>>(); |
| 131 | let expected = vec![ |
| 132 | TextFragment { |
| 133 | byte_range: Range { start: 0, end: 1 }, |
| 134 | glyph_range: Range { start: 0, end: 2 }, |
| 135 | width: 10., |
| 136 | trailing_whitespace_width: 10., |
| 137 | trailing_mandatory_break: false, |
| 138 | trailing_whitespace_bytes: 1, |
| 139 | }, |
| 140 | TextFragment { |
| 141 | byte_range: Range { start: 2, end: text.len() }, |
| 142 | glyph_range: Range { start: 2, end: text.len() }, |
| 143 | width: 20., |
| 144 | trailing_whitespace_width: 0., |
| 145 | trailing_mandatory_break: false, |
| 146 | trailing_whitespace_bytes: 0, |
| 147 | }, |
| 148 | ]; |
| 149 | assert_eq!(fragments, expected); |
| 150 | } |
| 151 | |
| 152 | #[test ] |
| 153 | fn fragment_iterator_simple_v2() { |
| 154 | let font = FixedTestFont; |
| 155 | let text = "Hello World" ; |
| 156 | let shape_buffer = ShapeBuffer::new(&TextLayout { font: &font, letter_spacing: None }, text); |
| 157 | let fragments = TextFragmentIterator::new(text, &shape_buffer).collect::<Vec<_>>(); |
| 158 | let expected = vec![ |
| 159 | TextFragment { |
| 160 | byte_range: Range { start: 0, end: 5 }, |
| 161 | glyph_range: Range { start: 0, end: 6 }, |
| 162 | width: 50., |
| 163 | trailing_whitespace_width: 10., |
| 164 | trailing_mandatory_break: false, |
| 165 | trailing_whitespace_bytes: 1, |
| 166 | }, |
| 167 | TextFragment { |
| 168 | byte_range: Range { start: 6, end: text.len() }, |
| 169 | glyph_range: Range { start: 6, end: text.len() }, |
| 170 | width: 10. * (text.len() - 6) as f32, |
| 171 | trailing_whitespace_width: 0., |
| 172 | trailing_whitespace_bytes: 0, |
| 173 | trailing_mandatory_break: false, |
| 174 | }, |
| 175 | ]; |
| 176 | assert_eq!(fragments, expected); |
| 177 | } |
| 178 | |
| 179 | #[test ] |
| 180 | fn fragment_iterator_forced_break() { |
| 181 | let font = FixedTestFont; |
| 182 | let text = "H \nW" ; |
| 183 | let shape_buffer = ShapeBuffer::new(&TextLayout { font: &font, letter_spacing: None }, text); |
| 184 | let fragments = TextFragmentIterator::new(text, &shape_buffer).collect::<Vec<_>>(); |
| 185 | assert_eq!( |
| 186 | fragments, |
| 187 | vec![ |
| 188 | TextFragment { |
| 189 | byte_range: Range { start: 0, end: 1 }, |
| 190 | glyph_range: Range { start: 0, end: 1 }, |
| 191 | width: 10., |
| 192 | trailing_whitespace_width: 0., |
| 193 | trailing_whitespace_bytes: 1, |
| 194 | trailing_mandatory_break: true, |
| 195 | }, |
| 196 | TextFragment { |
| 197 | byte_range: Range { start: 2, end: 3 }, |
| 198 | glyph_range: Range { start: 2, end: 3 }, |
| 199 | width: 10., |
| 200 | trailing_whitespace_width: 0., |
| 201 | trailing_whitespace_bytes: 0, |
| 202 | trailing_mandatory_break: false, |
| 203 | }, |
| 204 | ] |
| 205 | ); |
| 206 | } |
| 207 | |
| 208 | #[test ] |
| 209 | fn fragment_iterator_forced_break_multi() { |
| 210 | let font = FixedTestFont; |
| 211 | let text = "H \n\n\nW" ; |
| 212 | let shape_buffer = ShapeBuffer::new(&TextLayout { font: &font, letter_spacing: None }, text); |
| 213 | let fragments = TextFragmentIterator::new(text, &shape_buffer).collect::<Vec<_>>(); |
| 214 | assert_eq!( |
| 215 | fragments, |
| 216 | vec![ |
| 217 | TextFragment { |
| 218 | byte_range: Range { start: 0, end: 1 }, |
| 219 | glyph_range: Range { start: 0, end: 1 }, |
| 220 | width: 10., |
| 221 | trailing_whitespace_width: 0., |
| 222 | trailing_whitespace_bytes: 1, |
| 223 | trailing_mandatory_break: true, |
| 224 | }, |
| 225 | TextFragment { |
| 226 | byte_range: Range { start: 2, end: 2 }, |
| 227 | glyph_range: Range { start: 2, end: 3 }, |
| 228 | width: 0., |
| 229 | trailing_whitespace_width: 10., |
| 230 | trailing_whitespace_bytes: 1, |
| 231 | trailing_mandatory_break: true, |
| 232 | }, |
| 233 | TextFragment { |
| 234 | byte_range: Range { start: 3, end: 3 }, |
| 235 | glyph_range: Range { start: 3, end: 4 }, |
| 236 | width: 0., |
| 237 | trailing_whitespace_width: 10., |
| 238 | trailing_whitespace_bytes: 1, |
| 239 | trailing_mandatory_break: true, |
| 240 | }, |
| 241 | TextFragment { |
| 242 | byte_range: Range { start: 4, end: 5 }, |
| 243 | glyph_range: Range { start: 4, end: 5 }, |
| 244 | width: 10., |
| 245 | trailing_whitespace_width: 0., |
| 246 | trailing_whitespace_bytes: 0, |
| 247 | trailing_mandatory_break: false, |
| 248 | }, |
| 249 | ] |
| 250 | ); |
| 251 | } |
| 252 | |
| 253 | #[test ] |
| 254 | fn fragment_iterator_nbsp() { |
| 255 | let font = FixedTestFont; |
| 256 | let text = "X H \u{00a0}W" ; |
| 257 | let shape_buffer = ShapeBuffer::new(&TextLayout { font: &font, letter_spacing: None }, text); |
| 258 | let fragments = TextFragmentIterator::new(text, &shape_buffer).collect::<Vec<_>>(); |
| 259 | assert_eq!( |
| 260 | fragments, |
| 261 | vec![ |
| 262 | TextFragment { |
| 263 | byte_range: Range { start: 0, end: 1 }, |
| 264 | glyph_range: Range { start: 0, end: 2 }, |
| 265 | width: 10., |
| 266 | trailing_whitespace_width: 10., |
| 267 | trailing_whitespace_bytes: 1, |
| 268 | trailing_mandatory_break: false, |
| 269 | }, |
| 270 | TextFragment { |
| 271 | byte_range: Range { start: 2, end: 6 }, |
| 272 | glyph_range: Range { start: 2, end: 5 }, |
| 273 | width: 30., |
| 274 | trailing_whitespace_width: 0., |
| 275 | trailing_whitespace_bytes: 0, |
| 276 | trailing_mandatory_break: false, |
| 277 | } |
| 278 | ] |
| 279 | ); |
| 280 | } |
| 281 | |
| 282 | #[test ] |
| 283 | fn fragment_iterator_break_anywhere() { |
| 284 | let font = FixedTestFont; |
| 285 | let text = "AB \nCD \nEF" ; |
| 286 | let shape_buffer = ShapeBuffer::new(&TextLayout { font: &font, letter_spacing: None }, text); |
| 287 | let mut fragments = TextFragmentIterator::new(text, &shape_buffer); |
| 288 | assert_eq!( |
| 289 | fragments.next(), |
| 290 | Some(TextFragment { |
| 291 | byte_range: Range { start: 0, end: 2 }, |
| 292 | glyph_range: Range { start: 0, end: 2 }, |
| 293 | width: 20., |
| 294 | trailing_whitespace_width: 0., |
| 295 | trailing_whitespace_bytes: 1, |
| 296 | trailing_mandatory_break: true, |
| 297 | }) |
| 298 | ); |
| 299 | assert_eq!( |
| 300 | fragments.next(), |
| 301 | Some(TextFragment { |
| 302 | byte_range: Range { start: 3, end: 5 }, |
| 303 | glyph_range: Range { start: 3, end: 5 }, |
| 304 | width: 20., |
| 305 | trailing_whitespace_width: 0., |
| 306 | trailing_whitespace_bytes: 1, |
| 307 | trailing_mandatory_break: true, |
| 308 | },) |
| 309 | ); |
| 310 | fragments.break_anywhere = true; |
| 311 | let last_two = fragments.by_ref().take(2).collect::<Vec<_>>(); |
| 312 | assert_eq!( |
| 313 | last_two, |
| 314 | vec![ |
| 315 | TextFragment { |
| 316 | byte_range: Range { start: 6, end: 7 }, |
| 317 | glyph_range: Range { start: 6, end: 7 }, |
| 318 | width: 10., |
| 319 | trailing_whitespace_width: 0., |
| 320 | trailing_whitespace_bytes: 0, |
| 321 | trailing_mandatory_break: false, |
| 322 | }, |
| 323 | TextFragment { |
| 324 | byte_range: Range { start: 7, end: 8 }, |
| 325 | glyph_range: Range { start: 7, end: 8 }, |
| 326 | width: 10., |
| 327 | trailing_whitespace_width: 0., |
| 328 | trailing_whitespace_bytes: 0, |
| 329 | trailing_mandatory_break: false, |
| 330 | }, |
| 331 | ] |
| 332 | ); |
| 333 | } |
| 334 | |
| 335 | #[test ] |
| 336 | fn fragment_iterator_leading_nbsp() { |
| 337 | let font = FixedTestFont; |
| 338 | let text = "A \n\u{00a0}\u{00a0}AB" ; |
| 339 | let shape_buffer = ShapeBuffer::new(&TextLayout { font: &font, letter_spacing: None }, text); |
| 340 | let fragments = TextFragmentIterator::new(text, &shape_buffer).collect::<Vec<_>>(); |
| 341 | assert_eq!( |
| 342 | fragments, |
| 343 | vec![ |
| 344 | TextFragment { |
| 345 | byte_range: Range { start: 0, end: 1 }, |
| 346 | glyph_range: Range { start: 0, end: 1 }, |
| 347 | width: 10., |
| 348 | trailing_whitespace_width: 0., |
| 349 | trailing_whitespace_bytes: 1, |
| 350 | trailing_mandatory_break: true, |
| 351 | }, |
| 352 | TextFragment { |
| 353 | byte_range: Range { start: 2, end: 8 }, |
| 354 | glyph_range: Range { start: 2, end: 6 }, |
| 355 | width: 40., |
| 356 | trailing_whitespace_width: 0., |
| 357 | trailing_whitespace_bytes: 0, |
| 358 | trailing_mandatory_break: false, |
| 359 | } |
| 360 | ] |
| 361 | ); |
| 362 | } |
| 363 | |