1mod offset_utf16;
2mod point;
3mod point_utf16;
4mod unclipped;
5
6use arrayvec::ArrayString;
7use bromberg_sl2::HashMatrix;
8use smallvec::SmallVec;
9use std::{
10 cmp, fmt, io, mem,
11 ops::{AddAssign, Range},
12 str,
13};
14use sum_tree::{Bias, Dimension, SumTree};
15use util::debug_panic;
16
17pub use offset_utf16::OffsetUtf16;
18pub use point::Point;
19pub use point_utf16::PointUtf16;
20pub use unclipped::Unclipped;
21
22#[cfg(test)]
23const CHUNK_BASE: usize = 6;
24
25#[cfg(not(test))]
26const CHUNK_BASE: usize = 16;
27
28/// Type alias to [HashMatrix], an implementation of a homomorphic hash function. Two [Rope] instances
29/// containing the same text will produce the same fingerprint. This hash function is special in that
30/// it allows us to hash individual chunks and aggregate them up the [Rope]'s tree, with the resulting
31/// hash being equivalent to hashing all the text contained in the [Rope] at once.
32pub type RopeFingerprint = HashMatrix;
33
34#[derive(Clone, Default, Debug)]
35pub struct Rope {
36 chunks: SumTree<Chunk>,
37}
38
39impl Rope {
40 pub fn new() -> Self {
41 Self::default()
42 }
43
44 pub fn append(&mut self, rope: Rope) {
45 let mut chunks = rope.chunks.cursor::<()>();
46 chunks.next(&());
47 if let Some(chunk) = chunks.item() {
48 if self.chunks.last().map_or(false, |c| c.0.len() < CHUNK_BASE)
49 || chunk.0.len() < CHUNK_BASE
50 {
51 self.push(&chunk.0);
52 chunks.next(&());
53 }
54 }
55
56 self.chunks.push_tree(chunks.suffix(&()), &());
57 self.check_invariants();
58 }
59
60 pub fn replace(&mut self, range: Range<usize>, text: &str) {
61 let mut new_rope = Rope::new();
62 let mut cursor = self.cursor(0);
63 new_rope.append(cursor.slice(range.start));
64 cursor.seek_forward(range.end);
65 new_rope.push(text);
66 new_rope.append(cursor.suffix());
67 *self = new_rope;
68 }
69
70 pub fn slice(&self, range: Range<usize>) -> Rope {
71 let mut cursor = self.cursor(0);
72 cursor.seek_forward(range.start);
73 cursor.slice(range.end)
74 }
75
76 pub fn slice_rows(&self, range: Range<u32>) -> Rope {
77 //This would be more efficient with a forward advance after the first, but it's fine
78 let start = self.point_to_offset(Point::new(range.start, 0));
79 let end = self.point_to_offset(Point::new(range.end, 0));
80 self.slice(start..end)
81 }
82
83 pub fn push(&mut self, text: &str) {
84 let mut new_chunks = SmallVec::<[_; 16]>::new();
85 let mut new_chunk = ArrayString::new();
86 for ch in text.chars() {
87 if new_chunk.len() + ch.len_utf8() > 2 * CHUNK_BASE {
88 new_chunks.push(Chunk(new_chunk));
89 new_chunk = ArrayString::new();
90 }
91
92 new_chunk.push(ch);
93 }
94 if !new_chunk.is_empty() {
95 new_chunks.push(Chunk(new_chunk));
96 }
97
98 let mut new_chunks = new_chunks.into_iter();
99 let mut first_new_chunk = new_chunks.next();
100 self.chunks.update_last(
101 |last_chunk| {
102 if let Some(first_new_chunk_ref) = first_new_chunk.as_mut() {
103 if last_chunk.0.len() + first_new_chunk_ref.0.len() <= 2 * CHUNK_BASE {
104 last_chunk.0.push_str(&first_new_chunk.take().unwrap().0);
105 } else {
106 let mut text = ArrayString::<{ 4 * CHUNK_BASE }>::new();
107 text.push_str(&last_chunk.0);
108 text.push_str(&first_new_chunk_ref.0);
109 let (left, right) = text.split_at(find_split_ix(&text));
110 last_chunk.0.clear();
111 last_chunk.0.push_str(left);
112 first_new_chunk_ref.0.clear();
113 first_new_chunk_ref.0.push_str(right);
114 }
115 }
116 },
117 &(),
118 );
119
120 self.chunks
121 .extend(first_new_chunk.into_iter().chain(new_chunks), &());
122 self.check_invariants();
123 }
124
125 pub fn push_front(&mut self, text: &str) {
126 let suffix = mem::replace(self, Rope::from(text));
127 self.append(suffix);
128 }
129
130 fn check_invariants(&self) {
131 #[cfg(test)]
132 {
133 // Ensure all chunks except maybe the last one are not underflowing.
134 // Allow some wiggle room for multibyte characters at chunk boundaries.
135 let mut chunks = self.chunks.cursor::<()>().peekable();
136 while let Some(chunk) = chunks.next() {
137 if chunks.peek().is_some() {
138 assert!(chunk.0.len() + 3 >= CHUNK_BASE);
139 }
140 }
141 }
142 }
143
144 pub fn summary(&self) -> TextSummary {
145 self.chunks.summary().text.clone()
146 }
147
148 pub fn len(&self) -> usize {
149 self.chunks.extent(&())
150 }
151
152 pub fn is_empty(&self) -> bool {
153 self.len() == 0
154 }
155
156 pub fn max_point(&self) -> Point {
157 self.chunks.extent(&())
158 }
159
160 pub fn max_point_utf16(&self) -> PointUtf16 {
161 self.chunks.extent(&())
162 }
163
164 pub fn cursor(&self, offset: usize) -> Cursor {
165 Cursor::new(self, offset)
166 }
167
168 pub fn chars(&self) -> impl Iterator<Item = char> + '_ {
169 self.chars_at(0)
170 }
171
172 pub fn chars_at(&self, start: usize) -> impl Iterator<Item = char> + '_ {
173 self.chunks_in_range(start..self.len()).flat_map(str::chars)
174 }
175
176 pub fn reversed_chars_at(&self, start: usize) -> impl Iterator<Item = char> + '_ {
177 self.reversed_chunks_in_range(0..start)
178 .flat_map(|chunk| chunk.chars().rev())
179 }
180
181 pub fn bytes_in_range(&self, range: Range<usize>) -> Bytes {
182 Bytes::new(self, range)
183 }
184
185 pub fn chunks(&self) -> Chunks {
186 self.chunks_in_range(0..self.len())
187 }
188
189 pub fn chunks_in_range(&self, range: Range<usize>) -> Chunks {
190 Chunks::new(self, range, false)
191 }
192
193 pub fn reversed_chunks_in_range(&self, range: Range<usize>) -> Chunks {
194 Chunks::new(self, range, true)
195 }
196
197 pub fn offset_to_offset_utf16(&self, offset: usize) -> OffsetUtf16 {
198 if offset >= self.summary().len {
199 return self.summary().len_utf16;
200 }
201 let mut cursor = self.chunks.cursor::<(usize, OffsetUtf16)>();
202 cursor.seek(&offset, Bias::Left, &());
203 let overshoot = offset - cursor.start().0;
204 cursor.start().1
205 + cursor.item().map_or(Default::default(), |chunk| {
206 chunk.offset_to_offset_utf16(overshoot)
207 })
208 }
209
210 pub fn offset_utf16_to_offset(&self, offset: OffsetUtf16) -> usize {
211 if offset >= self.summary().len_utf16 {
212 return self.summary().len;
213 }
214 let mut cursor = self.chunks.cursor::<(OffsetUtf16, usize)>();
215 cursor.seek(&offset, Bias::Left, &());
216 let overshoot = offset - cursor.start().0;
217 cursor.start().1
218 + cursor.item().map_or(Default::default(), |chunk| {
219 chunk.offset_utf16_to_offset(overshoot)
220 })
221 }
222
223 pub fn offset_to_point(&self, offset: usize) -> Point {
224 if offset >= self.summary().len {
225 return self.summary().lines;
226 }
227 let mut cursor = self.chunks.cursor::<(usize, Point)>();
228 cursor.seek(&offset, Bias::Left, &());
229 let overshoot = offset - cursor.start().0;
230 cursor.start().1
231 + cursor
232 .item()
233 .map_or(Point::zero(), |chunk| chunk.offset_to_point(overshoot))
234 }
235
236 pub fn offset_to_point_utf16(&self, offset: usize) -> PointUtf16 {
237 if offset >= self.summary().len {
238 return self.summary().lines_utf16();
239 }
240 let mut cursor = self.chunks.cursor::<(usize, PointUtf16)>();
241 cursor.seek(&offset, Bias::Left, &());
242 let overshoot = offset - cursor.start().0;
243 cursor.start().1
244 + cursor.item().map_or(PointUtf16::zero(), |chunk| {
245 chunk.offset_to_point_utf16(overshoot)
246 })
247 }
248
249 pub fn point_to_point_utf16(&self, point: Point) -> PointUtf16 {
250 if point >= self.summary().lines {
251 return self.summary().lines_utf16();
252 }
253 let mut cursor = self.chunks.cursor::<(Point, PointUtf16)>();
254 cursor.seek(&point, Bias::Left, &());
255 let overshoot = point - cursor.start().0;
256 cursor.start().1
257 + cursor.item().map_or(PointUtf16::zero(), |chunk| {
258 chunk.point_to_point_utf16(overshoot)
259 })
260 }
261
262 pub fn point_to_offset(&self, point: Point) -> usize {
263 if point >= self.summary().lines {
264 return self.summary().len;
265 }
266 let mut cursor = self.chunks.cursor::<(Point, usize)>();
267 cursor.seek(&point, Bias::Left, &());
268 let overshoot = point - cursor.start().0;
269 cursor.start().1
270 + cursor
271 .item()
272 .map_or(0, |chunk| chunk.point_to_offset(overshoot))
273 }
274
275 pub fn point_utf16_to_offset(&self, point: PointUtf16) -> usize {
276 self.point_utf16_to_offset_impl(point, false)
277 }
278
279 pub fn unclipped_point_utf16_to_offset(&self, point: Unclipped<PointUtf16>) -> usize {
280 self.point_utf16_to_offset_impl(point.0, true)
281 }
282
283 fn point_utf16_to_offset_impl(&self, point: PointUtf16, clip: bool) -> usize {
284 if point >= self.summary().lines_utf16() {
285 return self.summary().len;
286 }
287 let mut cursor = self.chunks.cursor::<(PointUtf16, usize)>();
288 cursor.seek(&point, Bias::Left, &());
289 let overshoot = point - cursor.start().0;
290 cursor.start().1
291 + cursor
292 .item()
293 .map_or(0, |chunk| chunk.point_utf16_to_offset(overshoot, clip))
294 }
295
296 pub fn unclipped_point_utf16_to_point(&self, point: Unclipped<PointUtf16>) -> Point {
297 if point.0 >= self.summary().lines_utf16() {
298 return self.summary().lines;
299 }
300 let mut cursor = self.chunks.cursor::<(PointUtf16, Point)>();
301 cursor.seek(&point.0, Bias::Left, &());
302 let overshoot = Unclipped(point.0 - cursor.start().0);
303 cursor.start().1
304 + cursor.item().map_or(Point::zero(), |chunk| {
305 chunk.unclipped_point_utf16_to_point(overshoot)
306 })
307 }
308
309 pub fn clip_offset(&self, mut offset: usize, bias: Bias) -> usize {
310 let mut cursor = self.chunks.cursor::<usize>();
311 cursor.seek(&offset, Bias::Left, &());
312 if let Some(chunk) = cursor.item() {
313 let mut ix = offset - cursor.start();
314 while !chunk.0.is_char_boundary(ix) {
315 match bias {
316 Bias::Left => {
317 ix -= 1;
318 offset -= 1;
319 }
320 Bias::Right => {
321 ix += 1;
322 offset += 1;
323 }
324 }
325 }
326 offset
327 } else {
328 self.summary().len
329 }
330 }
331
332 pub fn clip_offset_utf16(&self, offset: OffsetUtf16, bias: Bias) -> OffsetUtf16 {
333 let mut cursor = self.chunks.cursor::<OffsetUtf16>();
334 cursor.seek(&offset, Bias::Right, &());
335 if let Some(chunk) = cursor.item() {
336 let overshoot = offset - cursor.start();
337 *cursor.start() + chunk.clip_offset_utf16(overshoot, bias)
338 } else {
339 self.summary().len_utf16
340 }
341 }
342
343 pub fn clip_point(&self, point: Point, bias: Bias) -> Point {
344 let mut cursor = self.chunks.cursor::<Point>();
345 cursor.seek(&point, Bias::Right, &());
346 if let Some(chunk) = cursor.item() {
347 let overshoot = point - cursor.start();
348 *cursor.start() + chunk.clip_point(overshoot, bias)
349 } else {
350 self.summary().lines
351 }
352 }
353
354 pub fn clip_point_utf16(&self, point: Unclipped<PointUtf16>, bias: Bias) -> PointUtf16 {
355 let mut cursor = self.chunks.cursor::<PointUtf16>();
356 cursor.seek(&point.0, Bias::Right, &());
357 if let Some(chunk) = cursor.item() {
358 let overshoot = Unclipped(point.0 - cursor.start());
359 *cursor.start() + chunk.clip_point_utf16(overshoot, bias)
360 } else {
361 self.summary().lines_utf16()
362 }
363 }
364
365 pub fn line_len(&self, row: u32) -> u32 {
366 self.clip_point(Point::new(row, u32::MAX), Bias::Left)
367 .column
368 }
369
370 pub fn fingerprint(&self) -> RopeFingerprint {
371 self.chunks.summary().fingerprint
372 }
373}
374
375impl<'a> From<&'a str> for Rope {
376 fn from(text: &'a str) -> Self {
377 let mut rope = Self::new();
378 rope.push(text);
379 rope
380 }
381}
382
383impl fmt::Display for Rope {
384 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
385 for chunk in self.chunks() {
386 write!(f, "{}", chunk)?;
387 }
388 Ok(())
389 }
390}
391
392pub struct Cursor<'a> {
393 rope: &'a Rope,
394 chunks: sum_tree::Cursor<'a, Chunk, usize>,
395 offset: usize,
396}
397
398impl<'a> Cursor<'a> {
399 pub fn new(rope: &'a Rope, offset: usize) -> Self {
400 let mut chunks = rope.chunks.cursor();
401 chunks.seek(&offset, Bias::Right, &());
402 Self {
403 rope,
404 chunks,
405 offset,
406 }
407 }
408
409 pub fn seek_forward(&mut self, end_offset: usize) {
410 debug_assert!(end_offset >= self.offset);
411
412 self.chunks.seek_forward(&end_offset, Bias::Right, &());
413 self.offset = end_offset;
414 }
415
416 pub fn slice(&mut self, end_offset: usize) -> Rope {
417 debug_assert!(
418 end_offset >= self.offset,
419 "cannot slice backwards from {} to {}",
420 self.offset,
421 end_offset
422 );
423
424 let mut slice = Rope::new();
425 if let Some(start_chunk) = self.chunks.item() {
426 let start_ix = self.offset - self.chunks.start();
427 let end_ix = cmp::min(end_offset, self.chunks.end(&())) - self.chunks.start();
428 slice.push(&start_chunk.0[start_ix..end_ix]);
429 }
430
431 if end_offset > self.chunks.end(&()) {
432 self.chunks.next(&());
433 slice.append(Rope {
434 chunks: self.chunks.slice(&end_offset, Bias::Right, &()),
435 });
436 if let Some(end_chunk) = self.chunks.item() {
437 let end_ix = end_offset - self.chunks.start();
438 slice.push(&end_chunk.0[..end_ix]);
439 }
440 }
441
442 self.offset = end_offset;
443 slice
444 }
445
446 pub fn summary<D: TextDimension>(&mut self, end_offset: usize) -> D {
447 debug_assert!(end_offset >= self.offset);
448
449 let mut summary = D::default();
450 if let Some(start_chunk) = self.chunks.item() {
451 let start_ix = self.offset - self.chunks.start();
452 let end_ix = cmp::min(end_offset, self.chunks.end(&())) - self.chunks.start();
453 summary.add_assign(&D::from_text_summary(&TextSummary::from(
454 &start_chunk.0[start_ix..end_ix],
455 )));
456 }
457
458 if end_offset > self.chunks.end(&()) {
459 self.chunks.next(&());
460 summary.add_assign(&self.chunks.summary(&end_offset, Bias::Right, &()));
461 if let Some(end_chunk) = self.chunks.item() {
462 let end_ix = end_offset - self.chunks.start();
463 summary.add_assign(&D::from_text_summary(&TextSummary::from(
464 &end_chunk.0[..end_ix],
465 )));
466 }
467 }
468
469 self.offset = end_offset;
470 summary
471 }
472
473 pub fn suffix(mut self) -> Rope {
474 self.slice(self.rope.chunks.extent(&()))
475 }
476
477 pub fn offset(&self) -> usize {
478 self.offset
479 }
480}
481
482pub struct Chunks<'a> {
483 chunks: sum_tree::Cursor<'a, Chunk, usize>,
484 range: Range<usize>,
485 reversed: bool,
486}
487
488impl<'a> Chunks<'a> {
489 pub fn new(rope: &'a Rope, range: Range<usize>, reversed: bool) -> Self {
490 let mut chunks = rope.chunks.cursor();
491 if reversed {
492 chunks.seek(&range.end, Bias::Left, &());
493 } else {
494 chunks.seek(&range.start, Bias::Right, &());
495 }
496 Self {
497 chunks,
498 range,
499 reversed,
500 }
501 }
502
503 pub fn offset(&self) -> usize {
504 if self.reversed {
505 self.range.end.min(self.chunks.end(&()))
506 } else {
507 self.range.start.max(*self.chunks.start())
508 }
509 }
510
511 pub fn seek(&mut self, offset: usize) {
512 let bias = if self.reversed {
513 Bias::Left
514 } else {
515 Bias::Right
516 };
517
518 if offset >= self.chunks.end(&()) {
519 self.chunks.seek_forward(&offset, bias, &());
520 } else {
521 self.chunks.seek(&offset, bias, &());
522 }
523
524 if self.reversed {
525 self.range.end = offset;
526 } else {
527 self.range.start = offset;
528 }
529 }
530
531 pub fn peek(&self) -> Option<&'a str> {
532 let chunk = self.chunks.item()?;
533 if self.reversed && self.range.start >= self.chunks.end(&()) {
534 return None;
535 }
536 let chunk_start = *self.chunks.start();
537 if self.range.end <= chunk_start {
538 return None;
539 }
540
541 let start = self.range.start.saturating_sub(chunk_start);
542 let end = self.range.end - chunk_start;
543 Some(&chunk.0[start..chunk.0.len().min(end)])
544 }
545}
546
547impl<'a> Iterator for Chunks<'a> {
548 type Item = &'a str;
549
550 fn next(&mut self) -> Option<Self::Item> {
551 let result = self.peek();
552 if result.is_some() {
553 if self.reversed {
554 self.chunks.prev(&());
555 } else {
556 self.chunks.next(&());
557 }
558 }
559 result
560 }
561}
562
563pub struct Bytes<'a> {
564 chunks: sum_tree::Cursor<'a, Chunk, usize>,
565 range: Range<usize>,
566}
567
568impl<'a> Bytes<'a> {
569 pub fn new(rope: &'a Rope, range: Range<usize>) -> Self {
570 let mut chunks = rope.chunks.cursor();
571 chunks.seek(&range.start, Bias::Right, &());
572 Self { chunks, range }
573 }
574
575 pub fn peek(&self) -> Option<&'a [u8]> {
576 let chunk = self.chunks.item()?;
577 let chunk_start = *self.chunks.start();
578 if self.range.end <= chunk_start {
579 return None;
580 }
581
582 let start = self.range.start.saturating_sub(chunk_start);
583 let end = self.range.end - chunk_start;
584 Some(&chunk.0.as_bytes()[start..chunk.0.len().min(end)])
585 }
586}
587
588impl<'a> Iterator for Bytes<'a> {
589 type Item = &'a [u8];
590
591 fn next(&mut self) -> Option<Self::Item> {
592 let result = self.peek();
593 if result.is_some() {
594 self.chunks.next(&());
595 }
596 result
597 }
598}
599
600impl<'a> io::Read for Bytes<'a> {
601 fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> {
602 if let Some(chunk) = self.peek() {
603 let len = cmp::min(buf.len(), chunk.len());
604 buf[..len].copy_from_slice(&chunk[..len]);
605 self.range.start += len;
606 if len == chunk.len() {
607 self.chunks.next(&());
608 }
609 Ok(len)
610 } else {
611 Ok(0)
612 }
613 }
614}
615
616#[derive(Clone, Debug, Default)]
617struct Chunk(ArrayString<{ 2 * CHUNK_BASE }>);
618
619impl Chunk {
620 fn offset_to_offset_utf16(&self, target: usize) -> OffsetUtf16 {
621 let mut offset = 0;
622 let mut offset_utf16 = OffsetUtf16(0);
623 for ch in self.0.chars() {
624 if offset >= target {
625 break;
626 }
627
628 offset += ch.len_utf8();
629 offset_utf16.0 += ch.len_utf16();
630 }
631 offset_utf16
632 }
633
634 fn offset_utf16_to_offset(&self, target: OffsetUtf16) -> usize {
635 let mut offset_utf16 = OffsetUtf16(0);
636 let mut offset = 0;
637 for ch in self.0.chars() {
638 if offset_utf16 >= target {
639 break;
640 }
641
642 offset += ch.len_utf8();
643 offset_utf16.0 += ch.len_utf16();
644 }
645 offset
646 }
647
648 fn offset_to_point(&self, target: usize) -> Point {
649 let mut offset = 0;
650 let mut point = Point::new(0, 0);
651 for ch in self.0.chars() {
652 if offset >= target {
653 break;
654 }
655
656 if ch == '\n' {
657 point.row += 1;
658 point.column = 0;
659 } else {
660 point.column += ch.len_utf8() as u32;
661 }
662 offset += ch.len_utf8();
663 }
664 point
665 }
666
667 fn offset_to_point_utf16(&self, target: usize) -> PointUtf16 {
668 let mut offset = 0;
669 let mut point = PointUtf16::new(0, 0);
670 for ch in self.0.chars() {
671 if offset >= target {
672 break;
673 }
674
675 if ch == '\n' {
676 point.row += 1;
677 point.column = 0;
678 } else {
679 point.column += ch.len_utf16() as u32;
680 }
681 offset += ch.len_utf8();
682 }
683 point
684 }
685
686 fn point_to_offset(&self, target: Point) -> usize {
687 let mut offset = 0;
688 let mut point = Point::new(0, 0);
689
690 for ch in self.0.chars() {
691 if point >= target {
692 if point > target {
693 debug_panic!("point {target:?} is inside of character {ch:?}");
694 }
695 break;
696 }
697
698 if ch == '\n' {
699 point.row += 1;
700 point.column = 0;
701
702 if point.row > target.row {
703 debug_panic!(
704 "point {target:?} is beyond the end of a line with length {}",
705 point.column
706 );
707 break;
708 }
709 } else {
710 point.column += ch.len_utf8() as u32;
711 }
712
713 offset += ch.len_utf8();
714 }
715
716 offset
717 }
718
719 fn point_to_point_utf16(&self, target: Point) -> PointUtf16 {
720 let mut point = Point::zero();
721 let mut point_utf16 = PointUtf16::new(0, 0);
722 for ch in self.0.chars() {
723 if point >= target {
724 break;
725 }
726
727 if ch == '\n' {
728 point_utf16.row += 1;
729 point_utf16.column = 0;
730 point.row += 1;
731 point.column = 0;
732 } else {
733 point_utf16.column += ch.len_utf16() as u32;
734 point.column += ch.len_utf8() as u32;
735 }
736 }
737 point_utf16
738 }
739
740 fn point_utf16_to_offset(&self, target: PointUtf16, clip: bool) -> usize {
741 let mut offset = 0;
742 let mut point = PointUtf16::new(0, 0);
743
744 for ch in self.0.chars() {
745 if point == target {
746 break;
747 }
748
749 if ch == '\n' {
750 point.row += 1;
751 point.column = 0;
752
753 if point.row > target.row {
754 if !clip {
755 debug_panic!(
756 "point {target:?} is beyond the end of a line with length {}",
757 point.column
758 );
759 }
760 // Return the offset of the newline
761 return offset;
762 }
763 } else {
764 point.column += ch.len_utf16() as u32;
765 }
766
767 if point > target {
768 if !clip {
769 debug_panic!("point {target:?} is inside of codepoint {ch:?}");
770 }
771 // Return the offset of the codepoint which we have landed within, bias left
772 return offset;
773 }
774
775 offset += ch.len_utf8();
776 }
777
778 offset
779 }
780
781 fn unclipped_point_utf16_to_point(&self, target: Unclipped<PointUtf16>) -> Point {
782 let mut point = Point::zero();
783 let mut point_utf16 = PointUtf16::zero();
784
785 for ch in self.0.chars() {
786 if point_utf16 == target.0 {
787 break;
788 }
789
790 if point_utf16 > target.0 {
791 // If the point is past the end of a line or inside of a code point,
792 // return the last valid point before the target.
793 return point;
794 }
795
796 if ch == '\n' {
797 point_utf16 += PointUtf16::new(1, 0);
798 point += Point::new(1, 0);
799 } else {
800 point_utf16 += PointUtf16::new(0, ch.len_utf16() as u32);
801 point += Point::new(0, ch.len_utf8() as u32);
802 }
803 }
804
805 point
806 }
807
808 fn clip_point(&self, target: Point, bias: Bias) -> Point {
809 for (row, line) in self.0.split('\n').enumerate() {
810 if row == target.row as usize {
811 let mut column = target.column.min(line.len() as u32);
812 while !line.is_char_boundary(column as usize) {
813 match bias {
814 Bias::Left => column -= 1,
815 Bias::Right => column += 1,
816 }
817 }
818 return Point::new(row as u32, column);
819 }
820 }
821 unreachable!()
822 }
823
824 fn clip_point_utf16(&self, target: Unclipped<PointUtf16>, bias: Bias) -> PointUtf16 {
825 for (row, line) in self.0.split('\n').enumerate() {
826 if row == target.0.row as usize {
827 let mut code_units = line.encode_utf16();
828 let mut column = code_units.by_ref().take(target.0.column as usize).count();
829 if char::decode_utf16(code_units).next().transpose().is_err() {
830 match bias {
831 Bias::Left => column -= 1,
832 Bias::Right => column += 1,
833 }
834 }
835 return PointUtf16::new(row as u32, column as u32);
836 }
837 }
838 unreachable!()
839 }
840
841 fn clip_offset_utf16(&self, target: OffsetUtf16, bias: Bias) -> OffsetUtf16 {
842 let mut code_units = self.0.encode_utf16();
843 let mut offset = code_units.by_ref().take(target.0 as usize).count();
844 if char::decode_utf16(code_units).next().transpose().is_err() {
845 match bias {
846 Bias::Left => offset -= 1,
847 Bias::Right => offset += 1,
848 }
849 }
850 OffsetUtf16(offset)
851 }
852}
853
854impl sum_tree::Item for Chunk {
855 type Summary = ChunkSummary;
856
857 fn summary(&self) -> Self::Summary {
858 ChunkSummary::from(self.0.as_str())
859 }
860}
861
862#[derive(Clone, Debug, Default, Eq, PartialEq)]
863pub struct ChunkSummary {
864 text: TextSummary,
865 fingerprint: RopeFingerprint,
866}
867
868impl<'a> From<&'a str> for ChunkSummary {
869 fn from(text: &'a str) -> Self {
870 Self {
871 text: TextSummary::from(text),
872 fingerprint: bromberg_sl2::hash_strict(text.as_bytes()),
873 }
874 }
875}
876
877impl sum_tree::Summary for ChunkSummary {
878 type Context = ();
879
880 fn add_summary(&mut self, summary: &Self, _: &()) {
881 self.text += &summary.text;
882 self.fingerprint = self.fingerprint * summary.fingerprint;
883 }
884}
885
886#[derive(Clone, Debug, Default, Eq, PartialEq)]
887pub struct TextSummary {
888 pub len: usize,
889 pub len_utf16: OffsetUtf16,
890 pub lines: Point,
891 pub first_line_chars: u32,
892 pub last_line_chars: u32,
893 pub last_line_len_utf16: u32,
894 pub longest_row: u32,
895 pub longest_row_chars: u32,
896}
897
898impl TextSummary {
899 pub fn lines_utf16(&self) -> PointUtf16 {
900 PointUtf16 {
901 row: self.lines.row,
902 column: self.last_line_len_utf16,
903 }
904 }
905}
906
907impl<'a> From<&'a str> for TextSummary {
908 fn from(text: &'a str) -> Self {
909 let mut len_utf16 = OffsetUtf16(0);
910 let mut lines = Point::new(0, 0);
911 let mut first_line_chars = 0;
912 let mut last_line_chars = 0;
913 let mut last_line_len_utf16 = 0;
914 let mut longest_row = 0;
915 let mut longest_row_chars = 0;
916 for c in text.chars() {
917 len_utf16.0 += c.len_utf16();
918
919 if c == '\n' {
920 lines += Point::new(1, 0);
921 last_line_len_utf16 = 0;
922 last_line_chars = 0;
923 } else {
924 lines.column += c.len_utf8() as u32;
925 last_line_len_utf16 += c.len_utf16() as u32;
926 last_line_chars += 1;
927 }
928
929 if lines.row == 0 {
930 first_line_chars = last_line_chars;
931 }
932
933 if last_line_chars > longest_row_chars {
934 longest_row = lines.row;
935 longest_row_chars = last_line_chars;
936 }
937 }
938
939 TextSummary {
940 len: text.len(),
941 len_utf16,
942 lines,
943 first_line_chars,
944 last_line_chars,
945 last_line_len_utf16,
946 longest_row,
947 longest_row_chars,
948 }
949 }
950}
951
952impl sum_tree::Summary for TextSummary {
953 type Context = ();
954
955 fn add_summary(&mut self, summary: &Self, _: &Self::Context) {
956 *self += summary;
957 }
958}
959
960impl std::ops::Add<Self> for TextSummary {
961 type Output = Self;
962
963 fn add(mut self, rhs: Self) -> Self::Output {
964 AddAssign::add_assign(&mut self, &rhs);
965 self
966 }
967}
968
969impl<'a> std::ops::AddAssign<&'a Self> for TextSummary {
970 fn add_assign(&mut self, other: &'a Self) {
971 let joined_chars = self.last_line_chars + other.first_line_chars;
972 if joined_chars > self.longest_row_chars {
973 self.longest_row = self.lines.row;
974 self.longest_row_chars = joined_chars;
975 }
976 if other.longest_row_chars > self.longest_row_chars {
977 self.longest_row = self.lines.row + other.longest_row;
978 self.longest_row_chars = other.longest_row_chars;
979 }
980
981 if self.lines.row == 0 {
982 self.first_line_chars += other.first_line_chars;
983 }
984
985 if other.lines.row == 0 {
986 self.last_line_chars += other.first_line_chars;
987 self.last_line_len_utf16 += other.last_line_len_utf16;
988 } else {
989 self.last_line_chars = other.last_line_chars;
990 self.last_line_len_utf16 = other.last_line_len_utf16;
991 }
992
993 self.len += other.len;
994 self.len_utf16 += other.len_utf16;
995 self.lines += other.lines;
996 }
997}
998
999impl std::ops::AddAssign<Self> for TextSummary {
1000 fn add_assign(&mut self, other: Self) {
1001 *self += &other;
1002 }
1003}
1004
1005pub trait TextDimension: 'static + for<'a> Dimension<'a, ChunkSummary> {
1006 fn from_text_summary(summary: &TextSummary) -> Self;
1007 fn add_assign(&mut self, other: &Self);
1008}
1009
1010impl<D1: TextDimension, D2: TextDimension> TextDimension for (D1, D2) {
1011 fn from_text_summary(summary: &TextSummary) -> Self {
1012 (
1013 D1::from_text_summary(summary),
1014 D2::from_text_summary(summary),
1015 )
1016 }
1017
1018 fn add_assign(&mut self, other: &Self) {
1019 self.0.add_assign(&other.0);
1020 self.1.add_assign(&other.1);
1021 }
1022}
1023
1024impl<'a> sum_tree::Dimension<'a, ChunkSummary> for TextSummary {
1025 fn add_summary(&mut self, summary: &'a ChunkSummary, _: &()) {
1026 *self += &summary.text;
1027 }
1028}
1029
1030impl TextDimension for TextSummary {
1031 fn from_text_summary(summary: &TextSummary) -> Self {
1032 summary.clone()
1033 }
1034
1035 fn add_assign(&mut self, other: &Self) {
1036 *self += other;
1037 }
1038}
1039
1040impl<'a> sum_tree::Dimension<'a, ChunkSummary> for usize {
1041 fn add_summary(&mut self, summary: &'a ChunkSummary, _: &()) {
1042 *self += summary.text.len;
1043 }
1044}
1045
1046impl TextDimension for usize {
1047 fn from_text_summary(summary: &TextSummary) -> Self {
1048 summary.len
1049 }
1050
1051 fn add_assign(&mut self, other: &Self) {
1052 *self += other;
1053 }
1054}
1055
1056impl<'a> sum_tree::Dimension<'a, ChunkSummary> for OffsetUtf16 {
1057 fn add_summary(&mut self, summary: &'a ChunkSummary, _: &()) {
1058 *self += summary.text.len_utf16;
1059 }
1060}
1061
1062impl TextDimension for OffsetUtf16 {
1063 fn from_text_summary(summary: &TextSummary) -> Self {
1064 summary.len_utf16
1065 }
1066
1067 fn add_assign(&mut self, other: &Self) {
1068 *self += other;
1069 }
1070}
1071
1072impl<'a> sum_tree::Dimension<'a, ChunkSummary> for Point {
1073 fn add_summary(&mut self, summary: &'a ChunkSummary, _: &()) {
1074 *self += summary.text.lines;
1075 }
1076}
1077
1078impl TextDimension for Point {
1079 fn from_text_summary(summary: &TextSummary) -> Self {
1080 summary.lines
1081 }
1082
1083 fn add_assign(&mut self, other: &Self) {
1084 *self += other;
1085 }
1086}
1087
1088impl<'a> sum_tree::Dimension<'a, ChunkSummary> for PointUtf16 {
1089 fn add_summary(&mut self, summary: &'a ChunkSummary, _: &()) {
1090 *self += summary.text.lines_utf16();
1091 }
1092}
1093
1094impl TextDimension for PointUtf16 {
1095 fn from_text_summary(summary: &TextSummary) -> Self {
1096 summary.lines_utf16()
1097 }
1098
1099 fn add_assign(&mut self, other: &Self) {
1100 *self += other;
1101 }
1102}
1103
1104fn find_split_ix(text: &str) -> usize {
1105 let mut ix = text.len() / 2;
1106 while !text.is_char_boundary(ix) {
1107 if ix < 2 * CHUNK_BASE {
1108 ix += 1;
1109 } else {
1110 ix = (text.len() / 2) - 1;
1111 break;
1112 }
1113 }
1114 while !text.is_char_boundary(ix) {
1115 ix -= 1;
1116 }
1117
1118 debug_assert!(ix <= 2 * CHUNK_BASE);
1119 debug_assert!(text.len() - ix <= 2 * CHUNK_BASE);
1120 ix
1121}
1122
1123#[cfg(test)]
1124mod tests {
1125 use super::*;
1126 use rand::prelude::*;
1127 use std::{cmp::Ordering, env, io::Read};
1128 use util::RandomCharIter;
1129 use Bias::{Left, Right};
1130
1131 #[test]
1132 fn test_all_4_byte_chars() {
1133 let mut rope = Rope::new();
1134 let text = "🏀".repeat(256);
1135 rope.push(&text);
1136 assert_eq!(rope.text(), text);
1137 }
1138
1139 #[test]
1140 fn test_clip() {
1141 let rope = Rope::from("🧘");
1142
1143 assert_eq!(rope.clip_offset(1, Bias::Left), 0);
1144 assert_eq!(rope.clip_offset(1, Bias::Right), 4);
1145 assert_eq!(rope.clip_offset(5, Bias::Right), 4);
1146
1147 assert_eq!(
1148 rope.clip_point(Point::new(0, 1), Bias::Left),
1149 Point::new(0, 0)
1150 );
1151 assert_eq!(
1152 rope.clip_point(Point::new(0, 1), Bias::Right),
1153 Point::new(0, 4)
1154 );
1155 assert_eq!(
1156 rope.clip_point(Point::new(0, 5), Bias::Right),
1157 Point::new(0, 4)
1158 );
1159
1160 assert_eq!(
1161 rope.clip_point_utf16(Unclipped(PointUtf16::new(0, 1)), Bias::Left),
1162 PointUtf16::new(0, 0)
1163 );
1164 assert_eq!(
1165 rope.clip_point_utf16(Unclipped(PointUtf16::new(0, 1)), Bias::Right),
1166 PointUtf16::new(0, 2)
1167 );
1168 assert_eq!(
1169 rope.clip_point_utf16(Unclipped(PointUtf16::new(0, 3)), Bias::Right),
1170 PointUtf16::new(0, 2)
1171 );
1172
1173 assert_eq!(
1174 rope.clip_offset_utf16(OffsetUtf16(1), Bias::Left),
1175 OffsetUtf16(0)
1176 );
1177 assert_eq!(
1178 rope.clip_offset_utf16(OffsetUtf16(1), Bias::Right),
1179 OffsetUtf16(2)
1180 );
1181 assert_eq!(
1182 rope.clip_offset_utf16(OffsetUtf16(3), Bias::Right),
1183 OffsetUtf16(2)
1184 );
1185 }
1186
1187 #[gpui::test(iterations = 100)]
1188 fn test_random_rope(mut rng: StdRng) {
1189 let operations = env::var("OPERATIONS")
1190 .map(|i| i.parse().expect("invalid `OPERATIONS` variable"))
1191 .unwrap_or(10);
1192
1193 let mut expected = String::new();
1194 let mut actual = Rope::new();
1195 for _ in 0..operations {
1196 let end_ix = clip_offset(&expected, rng.gen_range(0..=expected.len()), Right);
1197 let start_ix = clip_offset(&expected, rng.gen_range(0..=end_ix), Left);
1198 let len = rng.gen_range(0..=64);
1199 let new_text: String = RandomCharIter::new(&mut rng).take(len).collect();
1200
1201 let mut new_actual = Rope::new();
1202 let mut cursor = actual.cursor(0);
1203 new_actual.append(cursor.slice(start_ix));
1204 new_actual.push(&new_text);
1205 cursor.seek_forward(end_ix);
1206 new_actual.append(cursor.suffix());
1207 actual = new_actual;
1208
1209 expected.replace_range(start_ix..end_ix, &new_text);
1210
1211 assert_eq!(actual.text(), expected);
1212 log::info!("text: {:?}", expected);
1213
1214 for _ in 0..5 {
1215 let end_ix = clip_offset(&expected, rng.gen_range(0..=expected.len()), Right);
1216 let start_ix = clip_offset(&expected, rng.gen_range(0..=end_ix), Left);
1217
1218 let actual_text = actual.chunks_in_range(start_ix..end_ix).collect::<String>();
1219 assert_eq!(actual_text, &expected[start_ix..end_ix]);
1220
1221 let mut actual_text = String::new();
1222 actual
1223 .bytes_in_range(start_ix..end_ix)
1224 .read_to_string(&mut actual_text)
1225 .unwrap();
1226 assert_eq!(actual_text, &expected[start_ix..end_ix]);
1227
1228 assert_eq!(
1229 actual
1230 .reversed_chunks_in_range(start_ix..end_ix)
1231 .collect::<Vec<&str>>()
1232 .into_iter()
1233 .rev()
1234 .collect::<String>(),
1235 &expected[start_ix..end_ix]
1236 );
1237 }
1238
1239 let mut offset_utf16 = OffsetUtf16(0);
1240 let mut point = Point::new(0, 0);
1241 let mut point_utf16 = PointUtf16::new(0, 0);
1242 for (ix, ch) in expected.char_indices().chain(Some((expected.len(), '\0'))) {
1243 assert_eq!(actual.offset_to_point(ix), point, "offset_to_point({})", ix);
1244 assert_eq!(
1245 actual.offset_to_point_utf16(ix),
1246 point_utf16,
1247 "offset_to_point_utf16({})",
1248 ix
1249 );
1250 assert_eq!(
1251 actual.point_to_offset(point),
1252 ix,
1253 "point_to_offset({:?})",
1254 point
1255 );
1256 assert_eq!(
1257 actual.point_utf16_to_offset(point_utf16),
1258 ix,
1259 "point_utf16_to_offset({:?})",
1260 point_utf16
1261 );
1262 assert_eq!(
1263 actual.offset_to_offset_utf16(ix),
1264 offset_utf16,
1265 "offset_to_offset_utf16({:?})",
1266 ix
1267 );
1268 assert_eq!(
1269 actual.offset_utf16_to_offset(offset_utf16),
1270 ix,
1271 "offset_utf16_to_offset({:?})",
1272 offset_utf16
1273 );
1274 if ch == '\n' {
1275 point += Point::new(1, 0);
1276 point_utf16 += PointUtf16::new(1, 0);
1277 } else {
1278 point.column += ch.len_utf8() as u32;
1279 point_utf16.column += ch.len_utf16() as u32;
1280 }
1281 offset_utf16.0 += ch.len_utf16();
1282 }
1283
1284 let mut offset_utf16 = OffsetUtf16(0);
1285 let mut point_utf16 = Unclipped(PointUtf16::zero());
1286 for unit in expected.encode_utf16() {
1287 let left_offset = actual.clip_offset_utf16(offset_utf16, Bias::Left);
1288 let right_offset = actual.clip_offset_utf16(offset_utf16, Bias::Right);
1289 assert!(right_offset >= left_offset);
1290 // Ensure translating UTF-16 offsets to UTF-8 offsets doesn't panic.
1291 actual.offset_utf16_to_offset(left_offset);
1292 actual.offset_utf16_to_offset(right_offset);
1293
1294 let left_point = actual.clip_point_utf16(point_utf16, Bias::Left);
1295 let right_point = actual.clip_point_utf16(point_utf16, Bias::Right);
1296 assert!(right_point >= left_point);
1297 // Ensure translating valid UTF-16 points to offsets doesn't panic.
1298 actual.point_utf16_to_offset(left_point);
1299 actual.point_utf16_to_offset(right_point);
1300
1301 offset_utf16.0 += 1;
1302 if unit == b'\n' as u16 {
1303 point_utf16.0 += PointUtf16::new(1, 0);
1304 } else {
1305 point_utf16.0 += PointUtf16::new(0, 1);
1306 }
1307 }
1308
1309 for _ in 0..5 {
1310 let end_ix = clip_offset(&expected, rng.gen_range(0..=expected.len()), Right);
1311 let start_ix = clip_offset(&expected, rng.gen_range(0..=end_ix), Left);
1312 assert_eq!(
1313 actual.cursor(start_ix).summary::<TextSummary>(end_ix),
1314 TextSummary::from(&expected[start_ix..end_ix])
1315 );
1316 }
1317
1318 let mut expected_longest_rows = Vec::new();
1319 let mut longest_line_len = -1_isize;
1320 for (row, line) in expected.split('\n').enumerate() {
1321 let row = row as u32;
1322 assert_eq!(
1323 actual.line_len(row),
1324 line.len() as u32,
1325 "invalid line len for row {}",
1326 row
1327 );
1328
1329 let line_char_count = line.chars().count() as isize;
1330 match line_char_count.cmp(&longest_line_len) {
1331 Ordering::Less => {}
1332 Ordering::Equal => expected_longest_rows.push(row),
1333 Ordering::Greater => {
1334 longest_line_len = line_char_count;
1335 expected_longest_rows.clear();
1336 expected_longest_rows.push(row);
1337 }
1338 }
1339 }
1340
1341 let longest_row = actual.summary().longest_row;
1342 assert!(
1343 expected_longest_rows.contains(&longest_row),
1344 "incorrect longest row {}. expected {:?} with length {}",
1345 longest_row,
1346 expected_longest_rows,
1347 longest_line_len,
1348 );
1349 }
1350 }
1351
1352 fn clip_offset(text: &str, mut offset: usize, bias: Bias) -> usize {
1353 while !text.is_char_boundary(offset) {
1354 match bias {
1355 Bias::Left => offset -= 1,
1356 Bias::Right => offset += 1,
1357 }
1358 }
1359 offset
1360 }
1361
1362 impl Rope {
1363 fn text(&self) -> String {
1364 let mut text = String::new();
1365 for chunk in self.chunks.cursor::<()>() {
1366 text.push_str(&chunk.0);
1367 }
1368 text
1369 }
1370 }
1371}