1mod chunk;
2mod offset_utf16;
3mod point;
4mod point_utf16;
5mod unclipped;
6
7use chunk::Chunk;
8use rayon::iter::{IntoParallelIterator, ParallelIterator as _};
9use smallvec::SmallVec;
10use std::{
11 cmp, fmt, io, mem,
12 ops::{self, AddAssign, Range},
13 str,
14};
15use sum_tree::{Bias, Dimension, Dimensions, SumTree};
16
17pub use chunk::ChunkSlice;
18pub use offset_utf16::OffsetUtf16;
19pub use point::Point;
20pub use point_utf16::PointUtf16;
21pub use unclipped::Unclipped;
22
23#[derive(Clone, Default)]
24pub struct Rope {
25 chunks: SumTree<Chunk>,
26}
27
28impl Rope {
29 pub fn new() -> Self {
30 Self::default()
31 }
32
33 pub fn append(&mut self, rope: Rope) {
34 if let Some(chunk) = rope.chunks.first()
35 && (self
36 .chunks
37 .last()
38 .is_some_and(|c| c.text.len() < chunk::MIN_BASE)
39 || chunk.text.len() < chunk::MIN_BASE)
40 {
41 self.push_chunk(chunk.as_slice());
42
43 let mut chunks = rope.chunks.cursor::<()>(&());
44 chunks.next();
45 chunks.next();
46 self.chunks.append(chunks.suffix(), &());
47 self.check_invariants();
48 return;
49 }
50
51 self.chunks.append(rope.chunks.clone(), &());
52 self.check_invariants();
53 }
54
55 pub fn replace(&mut self, range: Range<usize>, text: &str) {
56 let mut new_rope = Rope::new();
57 let mut cursor = self.cursor(0);
58 new_rope.append(cursor.slice(range.start));
59 cursor.seek_forward(range.end);
60 new_rope.push(text);
61 new_rope.append(cursor.suffix());
62 *self = new_rope;
63 }
64
65 pub fn slice(&self, range: Range<usize>) -> Rope {
66 let mut cursor = self.cursor(0);
67 cursor.seek_forward(range.start);
68 cursor.slice(range.end)
69 }
70
71 pub fn slice_rows(&self, range: Range<u32>) -> Rope {
72 // This would be more efficient with a forward advance after the first, but it's fine.
73 let start = self.point_to_offset(Point::new(range.start, 0));
74 let end = self.point_to_offset(Point::new(range.end, 0));
75 self.slice(start..end)
76 }
77
78 pub fn push(&mut self, mut text: &str) {
79 self.chunks.update_last(
80 |last_chunk| {
81 let split_ix = if last_chunk.text.len() + text.len() <= chunk::MAX_BASE {
82 text.len()
83 } else {
84 let mut split_ix = cmp::min(
85 chunk::MIN_BASE.saturating_sub(last_chunk.text.len()),
86 text.len(),
87 );
88 while !text.is_char_boundary(split_ix) {
89 split_ix += 1;
90 }
91 split_ix
92 };
93
94 let (suffix, remainder) = text.split_at(split_ix);
95 last_chunk.push_str(suffix);
96 text = remainder;
97 },
98 &(),
99 );
100
101 if text.len() > 2048 {
102 return self.push_large(text);
103 }
104 let mut new_chunks = SmallVec::<[_; 16]>::new();
105
106 while !text.is_empty() {
107 let mut split_ix = cmp::min(chunk::MAX_BASE, text.len());
108 while !text.is_char_boundary(split_ix) {
109 split_ix -= 1;
110 }
111 let (chunk, remainder) = text.split_at(split_ix);
112 new_chunks.push(chunk);
113 text = remainder;
114 }
115
116 #[cfg(test)]
117 const PARALLEL_THRESHOLD: usize = 4;
118 #[cfg(not(test))]
119 const PARALLEL_THRESHOLD: usize = 4 * (2 * sum_tree::TREE_BASE);
120
121 if new_chunks.len() >= PARALLEL_THRESHOLD {
122 self.chunks
123 .par_extend(new_chunks.into_vec().into_par_iter().map(Chunk::new), &());
124 } else {
125 self.chunks
126 .extend(new_chunks.into_iter().map(Chunk::new), &());
127 }
128
129 self.check_invariants();
130 }
131
132 /// A copy of `push` specialized for working with large quantities of text.
133 fn push_large(&mut self, mut text: &str) {
134 // To avoid frequent reallocs when loading large swaths of file contents,
135 // we estimate worst-case `new_chunks` capacity;
136 // Chunk is a fixed-capacity buffer. If a character falls on
137 // chunk boundary, we push it off to the following chunk (thus leaving a small bit of capacity unfilled in current chunk).
138 // Worst-case chunk count when loading a file is then a case where every chunk ends up with that unused capacity.
139 // Since we're working with UTF-8, each character is at most 4 bytes wide. It follows then that the worst case is where
140 // a chunk ends with 3 bytes of a 4-byte character. These 3 bytes end up being stored in the following chunk, thus wasting
141 // 3 bytes of storage in current chunk.
142 // For example, a 1024-byte string can occupy between 32 (full ASCII, 1024/32) and 36 (full 4-byte UTF-8, 1024 / 29 rounded up) chunks.
143 const MIN_CHUNK_SIZE: usize = chunk::MAX_BASE - 3;
144
145 // We also round up the capacity up by one, for a good measure; we *really* don't want to realloc here, as we assume that the # of characters
146 // we're working with there is large.
147 let capacity = text.len().div_ceil(MIN_CHUNK_SIZE);
148 let mut new_chunks = Vec::with_capacity(capacity);
149
150 while !text.is_empty() {
151 let mut split_ix = cmp::min(chunk::MAX_BASE, text.len());
152 while !text.is_char_boundary(split_ix) {
153 split_ix -= 1;
154 }
155 let (chunk, remainder) = text.split_at(split_ix);
156 new_chunks.push(chunk);
157 text = remainder;
158 }
159
160 #[cfg(test)]
161 const PARALLEL_THRESHOLD: usize = 4;
162 #[cfg(not(test))]
163 const PARALLEL_THRESHOLD: usize = 4 * (2 * sum_tree::TREE_BASE);
164
165 if new_chunks.len() >= PARALLEL_THRESHOLD {
166 self.chunks
167 .par_extend(new_chunks.into_par_iter().map(Chunk::new), &());
168 } else {
169 self.chunks
170 .extend(new_chunks.into_iter().map(Chunk::new), &());
171 }
172
173 self.check_invariants();
174 }
175
176 fn push_chunk(&mut self, mut chunk: ChunkSlice) {
177 self.chunks.update_last(
178 |last_chunk| {
179 let split_ix = if last_chunk.text.len() + chunk.len() <= chunk::MAX_BASE {
180 chunk.len()
181 } else {
182 let mut split_ix = cmp::min(
183 chunk::MIN_BASE.saturating_sub(last_chunk.text.len()),
184 chunk.len(),
185 );
186 while !chunk.is_char_boundary(split_ix) {
187 split_ix += 1;
188 }
189 split_ix
190 };
191
192 let (suffix, remainder) = chunk.split_at(split_ix);
193 last_chunk.append(suffix);
194 chunk = remainder;
195 },
196 &(),
197 );
198
199 if !chunk.is_empty() {
200 self.chunks.push(chunk.into(), &());
201 }
202 }
203
204 pub fn push_front(&mut self, text: &str) {
205 let suffix = mem::replace(self, Rope::from(text));
206 self.append(suffix);
207 }
208
209 fn check_invariants(&self) {
210 #[cfg(test)]
211 {
212 // Ensure all chunks except maybe the last one are not underflowing.
213 // Allow some wiggle room for multibyte characters at chunk boundaries.
214 let mut chunks = self.chunks.cursor::<()>(&()).peekable();
215 while let Some(chunk) = chunks.next() {
216 if chunks.peek().is_some() {
217 assert!(chunk.text.len() + 3 >= chunk::MIN_BASE);
218 }
219 }
220 }
221 }
222
223 pub fn summary(&self) -> TextSummary {
224 self.chunks.summary().text
225 }
226
227 pub fn len(&self) -> usize {
228 self.chunks.extent(&())
229 }
230
231 pub fn is_empty(&self) -> bool {
232 self.len() == 0
233 }
234
235 pub fn max_point(&self) -> Point {
236 self.chunks.extent(&())
237 }
238
239 pub fn max_point_utf16(&self) -> PointUtf16 {
240 self.chunks.extent(&())
241 }
242
243 pub fn cursor(&self, offset: usize) -> Cursor<'_> {
244 Cursor::new(self, offset)
245 }
246
247 pub fn chars(&self) -> impl Iterator<Item = char> + '_ {
248 self.chars_at(0)
249 }
250
251 pub fn chars_at(&self, start: usize) -> impl Iterator<Item = char> + '_ {
252 self.chunks_in_range(start..self.len()).flat_map(str::chars)
253 }
254
255 pub fn reversed_chars_at(&self, start: usize) -> impl Iterator<Item = char> + '_ {
256 self.reversed_chunks_in_range(0..start)
257 .flat_map(|chunk| chunk.chars().rev())
258 }
259
260 pub fn bytes_in_range(&self, range: Range<usize>) -> Bytes<'_> {
261 Bytes::new(self, range, false)
262 }
263
264 pub fn reversed_bytes_in_range(&self, range: Range<usize>) -> Bytes<'_> {
265 Bytes::new(self, range, true)
266 }
267
268 pub fn chunks(&self) -> Chunks<'_> {
269 self.chunks_in_range(0..self.len())
270 }
271
272 pub fn chunks_in_range(&self, range: Range<usize>) -> Chunks<'_> {
273 Chunks::new(self, range, false)
274 }
275
276 pub fn reversed_chunks_in_range(&self, range: Range<usize>) -> Chunks<'_> {
277 Chunks::new(self, range, true)
278 }
279
280 pub fn offset_to_offset_utf16(&self, offset: usize) -> OffsetUtf16 {
281 if offset >= self.summary().len {
282 return self.summary().len_utf16;
283 }
284 let mut cursor = self.chunks.cursor::<Dimensions<usize, OffsetUtf16>>(&());
285 cursor.seek(&offset, Bias::Left);
286 let overshoot = offset - cursor.start().0;
287 cursor.start().1
288 + cursor.item().map_or(Default::default(), |chunk| {
289 chunk.as_slice().offset_to_offset_utf16(overshoot)
290 })
291 }
292
293 pub fn offset_utf16_to_offset(&self, offset: OffsetUtf16) -> usize {
294 if offset >= self.summary().len_utf16 {
295 return self.summary().len;
296 }
297 let mut cursor = self.chunks.cursor::<Dimensions<OffsetUtf16, usize>>(&());
298 cursor.seek(&offset, Bias::Left);
299 let overshoot = offset - cursor.start().0;
300 cursor.start().1
301 + cursor.item().map_or(Default::default(), |chunk| {
302 chunk.as_slice().offset_utf16_to_offset(overshoot)
303 })
304 }
305
306 pub fn offset_to_point(&self, offset: usize) -> Point {
307 if offset >= self.summary().len {
308 return self.summary().lines;
309 }
310 let mut cursor = self.chunks.cursor::<Dimensions<usize, Point>>(&());
311 cursor.seek(&offset, Bias::Left);
312 let overshoot = offset - cursor.start().0;
313 cursor.start().1
314 + cursor.item().map_or(Point::zero(), |chunk| {
315 chunk.as_slice().offset_to_point(overshoot)
316 })
317 }
318
319 pub fn offset_to_point_utf16(&self, offset: usize) -> PointUtf16 {
320 if offset >= self.summary().len {
321 return self.summary().lines_utf16();
322 }
323 let mut cursor = self.chunks.cursor::<Dimensions<usize, PointUtf16>>(&());
324 cursor.seek(&offset, Bias::Left);
325 let overshoot = offset - cursor.start().0;
326 cursor.start().1
327 + cursor.item().map_or(PointUtf16::zero(), |chunk| {
328 chunk.as_slice().offset_to_point_utf16(overshoot)
329 })
330 }
331
332 pub fn point_to_point_utf16(&self, point: Point) -> PointUtf16 {
333 if point >= self.summary().lines {
334 return self.summary().lines_utf16();
335 }
336 let mut cursor = self.chunks.cursor::<Dimensions<Point, PointUtf16>>(&());
337 cursor.seek(&point, Bias::Left);
338 let overshoot = point - cursor.start().0;
339 cursor.start().1
340 + cursor.item().map_or(PointUtf16::zero(), |chunk| {
341 chunk.as_slice().point_to_point_utf16(overshoot)
342 })
343 }
344
345 pub fn point_to_offset(&self, point: Point) -> usize {
346 if point >= self.summary().lines {
347 return self.summary().len;
348 }
349 let mut cursor = self.chunks.cursor::<Dimensions<Point, usize>>(&());
350 cursor.seek(&point, Bias::Left);
351 let overshoot = point - cursor.start().0;
352 cursor.start().1
353 + cursor
354 .item()
355 .map_or(0, |chunk| chunk.as_slice().point_to_offset(overshoot))
356 }
357
358 pub fn point_utf16_to_offset(&self, point: PointUtf16) -> usize {
359 self.point_utf16_to_offset_impl(point, false)
360 }
361
362 pub fn unclipped_point_utf16_to_offset(&self, point: Unclipped<PointUtf16>) -> usize {
363 self.point_utf16_to_offset_impl(point.0, true)
364 }
365
366 fn point_utf16_to_offset_impl(&self, point: PointUtf16, clip: bool) -> usize {
367 if point >= self.summary().lines_utf16() {
368 return self.summary().len;
369 }
370 let mut cursor = self.chunks.cursor::<Dimensions<PointUtf16, usize>>(&());
371 cursor.seek(&point, Bias::Left);
372 let overshoot = point - cursor.start().0;
373 cursor.start().1
374 + cursor.item().map_or(0, |chunk| {
375 chunk.as_slice().point_utf16_to_offset(overshoot, clip)
376 })
377 }
378
379 pub fn unclipped_point_utf16_to_point(&self, point: Unclipped<PointUtf16>) -> Point {
380 if point.0 >= self.summary().lines_utf16() {
381 return self.summary().lines;
382 }
383 let mut cursor = self.chunks.cursor::<Dimensions<PointUtf16, Point>>(&());
384 cursor.seek(&point.0, Bias::Left);
385 let overshoot = Unclipped(point.0 - cursor.start().0);
386 cursor.start().1
387 + cursor.item().map_or(Point::zero(), |chunk| {
388 chunk.as_slice().unclipped_point_utf16_to_point(overshoot)
389 })
390 }
391
392 pub fn clip_offset(&self, mut offset: usize, bias: Bias) -> usize {
393 let mut cursor = self.chunks.cursor::<usize>(&());
394 cursor.seek(&offset, Bias::Left);
395 if let Some(chunk) = cursor.item() {
396 let mut ix = offset - cursor.start();
397 while !chunk.text.is_char_boundary(ix) {
398 match bias {
399 Bias::Left => {
400 ix -= 1;
401 offset -= 1;
402 }
403 Bias::Right => {
404 ix += 1;
405 offset += 1;
406 }
407 }
408 }
409 offset
410 } else {
411 self.summary().len
412 }
413 }
414
415 pub fn clip_offset_utf16(&self, offset: OffsetUtf16, bias: Bias) -> OffsetUtf16 {
416 let mut cursor = self.chunks.cursor::<OffsetUtf16>(&());
417 cursor.seek(&offset, Bias::Right);
418 if let Some(chunk) = cursor.item() {
419 let overshoot = offset - cursor.start();
420 *cursor.start() + chunk.as_slice().clip_offset_utf16(overshoot, bias)
421 } else {
422 self.summary().len_utf16
423 }
424 }
425
426 pub fn clip_point(&self, point: Point, bias: Bias) -> Point {
427 let mut cursor = self.chunks.cursor::<Point>(&());
428 cursor.seek(&point, Bias::Right);
429 if let Some(chunk) = cursor.item() {
430 let overshoot = point - cursor.start();
431 *cursor.start() + chunk.as_slice().clip_point(overshoot, bias)
432 } else {
433 self.summary().lines
434 }
435 }
436
437 pub fn clip_point_utf16(&self, point: Unclipped<PointUtf16>, bias: Bias) -> PointUtf16 {
438 let mut cursor = self.chunks.cursor::<PointUtf16>(&());
439 cursor.seek(&point.0, Bias::Right);
440 if let Some(chunk) = cursor.item() {
441 let overshoot = Unclipped(point.0 - cursor.start());
442 *cursor.start() + chunk.as_slice().clip_point_utf16(overshoot, bias)
443 } else {
444 self.summary().lines_utf16()
445 }
446 }
447
448 pub fn line_len(&self, row: u32) -> u32 {
449 self.clip_point(Point::new(row, u32::MAX), Bias::Left)
450 .column
451 }
452}
453
454impl<'a> From<&'a str> for Rope {
455 fn from(text: &'a str) -> Self {
456 let mut rope = Self::new();
457 rope.push(text);
458 rope
459 }
460}
461
462impl<'a> FromIterator<&'a str> for Rope {
463 fn from_iter<T: IntoIterator<Item = &'a str>>(iter: T) -> Self {
464 let mut rope = Rope::new();
465 for chunk in iter {
466 rope.push(chunk);
467 }
468 rope
469 }
470}
471
472impl From<String> for Rope {
473 #[inline(always)]
474 fn from(text: String) -> Self {
475 Rope::from(text.as_str())
476 }
477}
478
479impl From<&String> for Rope {
480 #[inline(always)]
481 fn from(text: &String) -> Self {
482 Rope::from(text.as_str())
483 }
484}
485
486impl fmt::Display for Rope {
487 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
488 for chunk in self.chunks() {
489 write!(f, "{}", chunk)?;
490 }
491 Ok(())
492 }
493}
494
495impl fmt::Debug for Rope {
496 fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
497 use std::fmt::Write as _;
498
499 write!(f, "\"")?;
500 let mut format_string = String::new();
501 for chunk in self.chunks() {
502 write!(&mut format_string, "{:?}", chunk)?;
503 write!(f, "{}", &format_string[1..format_string.len() - 1])?;
504 format_string.clear();
505 }
506 write!(f, "\"")?;
507 Ok(())
508 }
509}
510
511pub struct Cursor<'a> {
512 rope: &'a Rope,
513 chunks: sum_tree::Cursor<'a, Chunk, usize>,
514 offset: usize,
515}
516
517impl<'a> Cursor<'a> {
518 pub fn new(rope: &'a Rope, offset: usize) -> Self {
519 let mut chunks = rope.chunks.cursor(&());
520 chunks.seek(&offset, Bias::Right);
521 Self {
522 rope,
523 chunks,
524 offset,
525 }
526 }
527
528 pub fn seek_forward(&mut self, end_offset: usize) {
529 debug_assert!(end_offset >= self.offset);
530
531 self.chunks.seek_forward(&end_offset, Bias::Right);
532 self.offset = end_offset;
533 }
534
535 pub fn slice(&mut self, end_offset: usize) -> Rope {
536 debug_assert!(
537 end_offset >= self.offset,
538 "cannot slice backwards from {} to {}",
539 self.offset,
540 end_offset
541 );
542
543 let mut slice = Rope::new();
544 if let Some(start_chunk) = self.chunks.item() {
545 let start_ix = self.offset - self.chunks.start();
546 let end_ix = cmp::min(end_offset, self.chunks.end()) - self.chunks.start();
547 slice.push_chunk(start_chunk.slice(start_ix..end_ix));
548 }
549
550 if end_offset > self.chunks.end() {
551 self.chunks.next();
552 slice.append(Rope {
553 chunks: self.chunks.slice(&end_offset, Bias::Right),
554 });
555 if let Some(end_chunk) = self.chunks.item() {
556 let end_ix = end_offset - self.chunks.start();
557 slice.push_chunk(end_chunk.slice(0..end_ix));
558 }
559 }
560
561 self.offset = end_offset;
562 slice
563 }
564
565 pub fn summary<D: TextDimension>(&mut self, end_offset: usize) -> D {
566 debug_assert!(end_offset >= self.offset);
567
568 let mut summary = D::zero(&());
569 if let Some(start_chunk) = self.chunks.item() {
570 let start_ix = self.offset - self.chunks.start();
571 let end_ix = cmp::min(end_offset, self.chunks.end()) - self.chunks.start();
572 summary.add_assign(&D::from_chunk(start_chunk.slice(start_ix..end_ix)));
573 }
574
575 if end_offset > self.chunks.end() {
576 self.chunks.next();
577 summary.add_assign(&self.chunks.summary(&end_offset, Bias::Right));
578 if let Some(end_chunk) = self.chunks.item() {
579 let end_ix = end_offset - self.chunks.start();
580 summary.add_assign(&D::from_chunk(end_chunk.slice(0..end_ix)));
581 }
582 }
583
584 self.offset = end_offset;
585 summary
586 }
587
588 pub fn suffix(mut self) -> Rope {
589 self.slice(self.rope.chunks.extent(&()))
590 }
591
592 pub fn offset(&self) -> usize {
593 self.offset
594 }
595}
596
597#[derive(Clone)]
598pub struct Chunks<'a> {
599 chunks: sum_tree::Cursor<'a, Chunk, usize>,
600 range: Range<usize>,
601 offset: usize,
602 reversed: bool,
603}
604
605impl<'a> Chunks<'a> {
606 pub fn new(rope: &'a Rope, range: Range<usize>, reversed: bool) -> Self {
607 let mut chunks = rope.chunks.cursor(&());
608 let offset = if reversed {
609 chunks.seek(&range.end, Bias::Left);
610 range.end
611 } else {
612 chunks.seek(&range.start, Bias::Right);
613 range.start
614 };
615 Self {
616 chunks,
617 range,
618 offset,
619 reversed,
620 }
621 }
622
623 fn offset_is_valid(&self) -> bool {
624 if self.reversed {
625 if self.offset <= self.range.start || self.offset > self.range.end {
626 return false;
627 }
628 } else if self.offset < self.range.start || self.offset >= self.range.end {
629 return false;
630 }
631
632 true
633 }
634
635 pub fn offset(&self) -> usize {
636 self.offset
637 }
638
639 pub fn seek(&mut self, mut offset: usize) {
640 offset = offset.clamp(self.range.start, self.range.end);
641
642 if self.reversed {
643 if offset > self.chunks.end() {
644 self.chunks.seek_forward(&offset, Bias::Left);
645 } else if offset <= *self.chunks.start() {
646 self.chunks.seek(&offset, Bias::Left);
647 }
648 } else {
649 if offset >= self.chunks.end() {
650 self.chunks.seek_forward(&offset, Bias::Right);
651 } else if offset < *self.chunks.start() {
652 self.chunks.seek(&offset, Bias::Right);
653 }
654 };
655
656 self.offset = offset;
657 }
658
659 pub fn set_range(&mut self, range: Range<usize>) {
660 self.range = range.clone();
661 self.seek(range.start);
662 }
663
664 /// Moves this cursor to the start of the next line in the rope.
665 ///
666 /// This method advances the cursor to the beginning of the next line.
667 /// If the cursor is already at the end of the rope, this method does nothing.
668 /// Reversed chunks iterators are not currently supported and will panic.
669 ///
670 /// Returns `true` if the cursor was successfully moved to the next line start,
671 /// or `false` if the cursor was already at the end of the rope.
672 pub fn next_line(&mut self) -> bool {
673 assert!(!self.reversed);
674
675 let mut found = false;
676 if let Some(chunk) = self.peek() {
677 if let Some(newline_ix) = chunk.find('\n') {
678 self.offset += newline_ix + 1;
679 found = self.offset <= self.range.end;
680 } else {
681 self.chunks
682 .search_forward(|summary| summary.text.lines.row > 0);
683 self.offset = *self.chunks.start();
684
685 if let Some(newline_ix) = self.peek().and_then(|chunk| chunk.find('\n')) {
686 self.offset += newline_ix + 1;
687 found = self.offset <= self.range.end;
688 } else {
689 self.offset = self.chunks.end();
690 }
691 }
692
693 if self.offset == self.chunks.end() {
694 self.next();
695 }
696 }
697
698 if self.offset > self.range.end {
699 self.offset = cmp::min(self.offset, self.range.end);
700 self.chunks.seek(&self.offset, Bias::Right);
701 }
702
703 found
704 }
705
706 /// Move this cursor to the preceding position in the rope that starts a new line.
707 /// Reversed chunks iterators are not currently supported and will panic.
708 ///
709 /// If this cursor is not on the start of a line, it will be moved to the start of
710 /// its current line. Otherwise it will be moved to the start of the previous line.
711 /// It updates the cursor's position and returns true if a previous line was found,
712 /// or false if the cursor was already at the start of the rope.
713 pub fn prev_line(&mut self) -> bool {
714 assert!(!self.reversed);
715
716 let initial_offset = self.offset;
717
718 if self.offset == *self.chunks.start() {
719 self.chunks.prev();
720 }
721
722 if let Some(chunk) = self.chunks.item() {
723 let mut end_ix = self.offset - *self.chunks.start();
724 if chunk.text.as_bytes()[end_ix - 1] == b'\n' {
725 end_ix -= 1;
726 }
727
728 if let Some(newline_ix) = chunk.text[..end_ix].rfind('\n') {
729 self.offset = *self.chunks.start() + newline_ix + 1;
730 if self.offset_is_valid() {
731 return true;
732 }
733 }
734 }
735
736 self.chunks
737 .search_backward(|summary| summary.text.lines.row > 0);
738 self.offset = *self.chunks.start();
739 if let Some(chunk) = self.chunks.item()
740 && let Some(newline_ix) = chunk.text.rfind('\n')
741 {
742 self.offset += newline_ix + 1;
743 if self.offset_is_valid() {
744 if self.offset == self.chunks.end() {
745 self.chunks.next();
746 }
747
748 return true;
749 }
750 }
751
752 if !self.offset_is_valid() || self.chunks.item().is_none() {
753 self.offset = self.range.start;
754 self.chunks.seek(&self.offset, Bias::Right);
755 }
756
757 self.offset < initial_offset && self.offset == 0
758 }
759
760 pub fn peek(&self) -> Option<&'a str> {
761 if !self.offset_is_valid() {
762 return None;
763 }
764
765 let chunk = self.chunks.item()?;
766 let chunk_start = *self.chunks.start();
767 let slice_range = if self.reversed {
768 let slice_start = cmp::max(chunk_start, self.range.start) - chunk_start;
769 let slice_end = self.offset - chunk_start;
770 slice_start..slice_end
771 } else {
772 let slice_start = self.offset - chunk_start;
773 let slice_end = cmp::min(self.chunks.end(), self.range.end) - chunk_start;
774 slice_start..slice_end
775 };
776
777 Some(&chunk.text[slice_range])
778 }
779
780 pub fn lines(self) -> Lines<'a> {
781 let reversed = self.reversed;
782 Lines {
783 chunks: self,
784 current_line: String::new(),
785 done: false,
786 reversed,
787 }
788 }
789
790 pub fn equals_str(&self, other: &str) -> bool {
791 let chunk = self.clone();
792 if chunk.reversed {
793 let mut offset = other.len();
794 for chunk in chunk {
795 if other[0..offset].ends_with(chunk) {
796 offset -= chunk.len();
797 } else {
798 return false;
799 }
800 }
801 if offset != 0 {
802 return false;
803 }
804 } else {
805 let mut offset = 0;
806 for chunk in chunk {
807 if offset >= other.len() {
808 return false;
809 }
810 if other[offset..].starts_with(chunk) {
811 offset += chunk.len();
812 } else {
813 return false;
814 }
815 }
816 if offset != other.len() {
817 return false;
818 }
819 }
820
821 true
822 }
823}
824
825impl<'a> Iterator for Chunks<'a> {
826 type Item = &'a str;
827
828 fn next(&mut self) -> Option<Self::Item> {
829 let chunk = self.peek()?;
830 if self.reversed {
831 self.offset -= chunk.len();
832 if self.offset <= *self.chunks.start() {
833 self.chunks.prev();
834 }
835 } else {
836 self.offset += chunk.len();
837 if self.offset >= self.chunks.end() {
838 self.chunks.next();
839 }
840 }
841
842 Some(chunk)
843 }
844}
845
846pub struct Bytes<'a> {
847 chunks: sum_tree::Cursor<'a, Chunk, usize>,
848 range: Range<usize>,
849 reversed: bool,
850}
851
852impl<'a> Bytes<'a> {
853 pub fn new(rope: &'a Rope, range: Range<usize>, reversed: bool) -> Self {
854 let mut chunks = rope.chunks.cursor(&());
855 if reversed {
856 chunks.seek(&range.end, Bias::Left);
857 } else {
858 chunks.seek(&range.start, Bias::Right);
859 }
860 Self {
861 chunks,
862 range,
863 reversed,
864 }
865 }
866
867 pub fn peek(&self) -> Option<&'a [u8]> {
868 let chunk = self.chunks.item()?;
869 if self.reversed && self.range.start >= self.chunks.end() {
870 return None;
871 }
872 let chunk_start = *self.chunks.start();
873 if self.range.end <= chunk_start {
874 return None;
875 }
876 let start = self.range.start.saturating_sub(chunk_start);
877 let end = self.range.end - chunk_start;
878 Some(&chunk.text.as_bytes()[start..chunk.text.len().min(end)])
879 }
880}
881
882impl<'a> Iterator for Bytes<'a> {
883 type Item = &'a [u8];
884
885 fn next(&mut self) -> Option<Self::Item> {
886 let result = self.peek();
887 if result.is_some() {
888 if self.reversed {
889 self.chunks.prev();
890 } else {
891 self.chunks.next();
892 }
893 }
894 result
895 }
896}
897
898impl io::Read for Bytes<'_> {
899 fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> {
900 if let Some(chunk) = self.peek() {
901 let len = cmp::min(buf.len(), chunk.len());
902 if self.reversed {
903 buf[..len].copy_from_slice(&chunk[chunk.len() - len..]);
904 buf[..len].reverse();
905 self.range.end -= len;
906 } else {
907 buf[..len].copy_from_slice(&chunk[..len]);
908 self.range.start += len;
909 }
910
911 if len == chunk.len() {
912 if self.reversed {
913 self.chunks.prev();
914 } else {
915 self.chunks.next();
916 }
917 }
918 Ok(len)
919 } else {
920 Ok(0)
921 }
922 }
923}
924
925pub struct Lines<'a> {
926 chunks: Chunks<'a>,
927 current_line: String,
928 done: bool,
929 reversed: bool,
930}
931
932impl Lines<'_> {
933 pub fn next(&mut self) -> Option<&str> {
934 if self.done {
935 return None;
936 }
937
938 self.current_line.clear();
939
940 while let Some(chunk) = self.chunks.peek() {
941 let chunk_lines = chunk.split('\n');
942 if self.reversed {
943 let mut chunk_lines = chunk_lines.rev().peekable();
944 if let Some(chunk_line) = chunk_lines.next() {
945 let done = chunk_lines.peek().is_some();
946 if done {
947 self.chunks
948 .seek(self.chunks.offset() - chunk_line.len() - "\n".len());
949 if self.current_line.is_empty() {
950 return Some(chunk_line);
951 }
952 }
953 self.current_line.insert_str(0, chunk_line);
954 if done {
955 return Some(&self.current_line);
956 }
957 }
958 } else {
959 let mut chunk_lines = chunk_lines.peekable();
960 if let Some(chunk_line) = chunk_lines.next() {
961 let done = chunk_lines.peek().is_some();
962 if done {
963 self.chunks
964 .seek(self.chunks.offset() + chunk_line.len() + "\n".len());
965 if self.current_line.is_empty() {
966 return Some(chunk_line);
967 }
968 }
969 self.current_line.push_str(chunk_line);
970 if done {
971 return Some(&self.current_line);
972 }
973 }
974 }
975
976 self.chunks.next();
977 }
978
979 self.done = true;
980 Some(&self.current_line)
981 }
982
983 pub fn seek(&mut self, offset: usize) {
984 self.chunks.seek(offset);
985 self.current_line.clear();
986 self.done = false;
987 }
988
989 pub fn offset(&self) -> usize {
990 self.chunks.offset()
991 }
992}
993
994impl sum_tree::Item for Chunk {
995 type Summary = ChunkSummary;
996
997 fn summary(&self, _cx: &()) -> Self::Summary {
998 ChunkSummary {
999 text: self.as_slice().text_summary(),
1000 }
1001 }
1002}
1003
1004#[derive(Clone, Debug, Default, Eq, PartialEq)]
1005pub struct ChunkSummary {
1006 text: TextSummary,
1007}
1008
1009impl sum_tree::Summary for ChunkSummary {
1010 type Context = ();
1011
1012 fn zero(_cx: &()) -> Self {
1013 Default::default()
1014 }
1015
1016 fn add_summary(&mut self, summary: &Self, _: &()) {
1017 self.text += &summary.text;
1018 }
1019}
1020
1021/// Summary of a string of text.
1022#[derive(Copy, Clone, Debug, Default, Eq, PartialEq)]
1023pub struct TextSummary {
1024 /// Length in bytes.
1025 pub len: usize,
1026 /// Length in UTF-8.
1027 pub chars: usize,
1028 /// Length in UTF-16 code units
1029 pub len_utf16: OffsetUtf16,
1030 /// A point representing the number of lines and the length of the last line.
1031 ///
1032 /// In other words, it marks the point after the last byte in the text, (if
1033 /// EOF was a character, this would be its position).
1034 pub lines: Point,
1035 /// How many `char`s are in the first line
1036 pub first_line_chars: u32,
1037 /// How many `char`s are in the last line
1038 pub last_line_chars: u32,
1039 /// How many UTF-16 code units are in the last line
1040 pub last_line_len_utf16: u32,
1041 /// The row idx of the longest row
1042 pub longest_row: u32,
1043 /// How many `char`s are in the longest row
1044 pub longest_row_chars: u32,
1045}
1046
1047impl TextSummary {
1048 pub fn lines_utf16(&self) -> PointUtf16 {
1049 PointUtf16 {
1050 row: self.lines.row,
1051 column: self.last_line_len_utf16,
1052 }
1053 }
1054
1055 pub fn newline() -> Self {
1056 Self {
1057 len: 1,
1058 chars: 1,
1059 len_utf16: OffsetUtf16(1),
1060 first_line_chars: 0,
1061 last_line_chars: 0,
1062 last_line_len_utf16: 0,
1063 lines: Point::new(1, 0),
1064 longest_row: 0,
1065 longest_row_chars: 0,
1066 }
1067 }
1068
1069 pub fn add_newline(&mut self) {
1070 self.len += 1;
1071 self.len_utf16 += OffsetUtf16(self.len_utf16.0 + 1);
1072 self.last_line_chars = 0;
1073 self.last_line_len_utf16 = 0;
1074 self.lines += Point::new(1, 0);
1075 }
1076}
1077
1078impl<'a> From<&'a str> for TextSummary {
1079 fn from(text: &'a str) -> Self {
1080 let mut len_utf16 = OffsetUtf16(0);
1081 let mut lines = Point::new(0, 0);
1082 let mut first_line_chars = 0;
1083 let mut last_line_chars = 0;
1084 let mut last_line_len_utf16 = 0;
1085 let mut longest_row = 0;
1086 let mut longest_row_chars = 0;
1087 let mut chars = 0;
1088 for c in text.chars() {
1089 chars += 1;
1090 len_utf16.0 += c.len_utf16();
1091
1092 if c == '\n' {
1093 lines += Point::new(1, 0);
1094 last_line_len_utf16 = 0;
1095 last_line_chars = 0;
1096 } else {
1097 lines.column += c.len_utf8() as u32;
1098 last_line_len_utf16 += c.len_utf16() as u32;
1099 last_line_chars += 1;
1100 }
1101
1102 if lines.row == 0 {
1103 first_line_chars = last_line_chars;
1104 }
1105
1106 if last_line_chars > longest_row_chars {
1107 longest_row = lines.row;
1108 longest_row_chars = last_line_chars;
1109 }
1110 }
1111
1112 TextSummary {
1113 len: text.len(),
1114 chars,
1115 len_utf16,
1116 lines,
1117 first_line_chars,
1118 last_line_chars,
1119 last_line_len_utf16,
1120 longest_row,
1121 longest_row_chars,
1122 }
1123 }
1124}
1125
1126impl sum_tree::Summary for TextSummary {
1127 type Context = ();
1128
1129 fn zero(_cx: &()) -> Self {
1130 Default::default()
1131 }
1132
1133 fn add_summary(&mut self, summary: &Self, _: &Self::Context) {
1134 *self += summary;
1135 }
1136}
1137
1138impl ops::Add<Self> for TextSummary {
1139 type Output = Self;
1140
1141 fn add(mut self, rhs: Self) -> Self::Output {
1142 AddAssign::add_assign(&mut self, &rhs);
1143 self
1144 }
1145}
1146
1147impl<'a> ops::AddAssign<&'a Self> for TextSummary {
1148 fn add_assign(&mut self, other: &'a Self) {
1149 let joined_chars = self.last_line_chars + other.first_line_chars;
1150 if joined_chars > self.longest_row_chars {
1151 self.longest_row = self.lines.row;
1152 self.longest_row_chars = joined_chars;
1153 }
1154 if other.longest_row_chars > self.longest_row_chars {
1155 self.longest_row = self.lines.row + other.longest_row;
1156 self.longest_row_chars = other.longest_row_chars;
1157 }
1158
1159 if self.lines.row == 0 {
1160 self.first_line_chars += other.first_line_chars;
1161 }
1162
1163 if other.lines.row == 0 {
1164 self.last_line_chars += other.first_line_chars;
1165 self.last_line_len_utf16 += other.last_line_len_utf16;
1166 } else {
1167 self.last_line_chars = other.last_line_chars;
1168 self.last_line_len_utf16 = other.last_line_len_utf16;
1169 }
1170
1171 self.chars += other.chars;
1172 self.len += other.len;
1173 self.len_utf16 += other.len_utf16;
1174 self.lines += other.lines;
1175 }
1176}
1177
1178impl ops::AddAssign<Self> for TextSummary {
1179 fn add_assign(&mut self, other: Self) {
1180 *self += &other;
1181 }
1182}
1183
1184pub trait TextDimension:
1185 'static + Clone + Copy + Default + for<'a> Dimension<'a, ChunkSummary> + std::fmt::Debug
1186{
1187 fn from_text_summary(summary: &TextSummary) -> Self;
1188 fn from_chunk(chunk: ChunkSlice) -> Self;
1189 fn add_assign(&mut self, other: &Self);
1190}
1191
1192impl<D1: TextDimension, D2: TextDimension> TextDimension for Dimensions<D1, D2, ()> {
1193 fn from_text_summary(summary: &TextSummary) -> Self {
1194 Dimensions(
1195 D1::from_text_summary(summary),
1196 D2::from_text_summary(summary),
1197 (),
1198 )
1199 }
1200
1201 fn from_chunk(chunk: ChunkSlice) -> Self {
1202 Dimensions(D1::from_chunk(chunk), D2::from_chunk(chunk), ())
1203 }
1204
1205 fn add_assign(&mut self, other: &Self) {
1206 self.0.add_assign(&other.0);
1207 self.1.add_assign(&other.1);
1208 }
1209}
1210
1211impl<'a> sum_tree::Dimension<'a, ChunkSummary> for TextSummary {
1212 fn zero(_cx: &()) -> Self {
1213 Default::default()
1214 }
1215
1216 fn add_summary(&mut self, summary: &'a ChunkSummary, _: &()) {
1217 *self += &summary.text;
1218 }
1219}
1220
1221impl TextDimension for TextSummary {
1222 fn from_text_summary(summary: &TextSummary) -> Self {
1223 *summary
1224 }
1225
1226 fn from_chunk(chunk: ChunkSlice) -> Self {
1227 chunk.text_summary()
1228 }
1229
1230 fn add_assign(&mut self, other: &Self) {
1231 *self += other;
1232 }
1233}
1234
1235impl<'a> sum_tree::Dimension<'a, ChunkSummary> for usize {
1236 fn zero(_cx: &()) -> Self {
1237 Default::default()
1238 }
1239
1240 fn add_summary(&mut self, summary: &'a ChunkSummary, _: &()) {
1241 *self += summary.text.len;
1242 }
1243}
1244
1245impl TextDimension for usize {
1246 fn from_text_summary(summary: &TextSummary) -> Self {
1247 summary.len
1248 }
1249
1250 fn from_chunk(chunk: ChunkSlice) -> Self {
1251 chunk.len()
1252 }
1253
1254 fn add_assign(&mut self, other: &Self) {
1255 *self += other;
1256 }
1257}
1258
1259impl<'a> sum_tree::Dimension<'a, ChunkSummary> for OffsetUtf16 {
1260 fn zero(_cx: &()) -> Self {
1261 Default::default()
1262 }
1263
1264 fn add_summary(&mut self, summary: &'a ChunkSummary, _: &()) {
1265 *self += summary.text.len_utf16;
1266 }
1267}
1268
1269impl TextDimension for OffsetUtf16 {
1270 fn from_text_summary(summary: &TextSummary) -> Self {
1271 summary.len_utf16
1272 }
1273
1274 fn from_chunk(chunk: ChunkSlice) -> Self {
1275 chunk.len_utf16()
1276 }
1277
1278 fn add_assign(&mut self, other: &Self) {
1279 *self += other;
1280 }
1281}
1282
1283impl<'a> sum_tree::Dimension<'a, ChunkSummary> for Point {
1284 fn zero(_cx: &()) -> Self {
1285 Default::default()
1286 }
1287
1288 fn add_summary(&mut self, summary: &'a ChunkSummary, _: &()) {
1289 *self += summary.text.lines;
1290 }
1291}
1292
1293impl TextDimension for Point {
1294 fn from_text_summary(summary: &TextSummary) -> Self {
1295 summary.lines
1296 }
1297
1298 fn from_chunk(chunk: ChunkSlice) -> Self {
1299 chunk.lines()
1300 }
1301
1302 fn add_assign(&mut self, other: &Self) {
1303 *self += other;
1304 }
1305}
1306
1307impl<'a> sum_tree::Dimension<'a, ChunkSummary> for PointUtf16 {
1308 fn zero(_cx: &()) -> Self {
1309 Default::default()
1310 }
1311
1312 fn add_summary(&mut self, summary: &'a ChunkSummary, _: &()) {
1313 *self += summary.text.lines_utf16();
1314 }
1315}
1316
1317impl TextDimension for PointUtf16 {
1318 fn from_text_summary(summary: &TextSummary) -> Self {
1319 summary.lines_utf16()
1320 }
1321
1322 fn from_chunk(chunk: ChunkSlice) -> Self {
1323 PointUtf16 {
1324 row: chunk.lines().row,
1325 column: chunk.last_line_len_utf16(),
1326 }
1327 }
1328
1329 fn add_assign(&mut self, other: &Self) {
1330 *self += other;
1331 }
1332}
1333
1334/// A pair of text dimensions in which only the first dimension is used for comparison,
1335/// but both dimensions are updated during addition and subtraction.
1336#[derive(Clone, Copy, Debug)]
1337pub struct DimensionPair<K, V> {
1338 pub key: K,
1339 pub value: Option<V>,
1340}
1341
1342impl<K: Default, V: Default> Default for DimensionPair<K, V> {
1343 fn default() -> Self {
1344 Self {
1345 key: Default::default(),
1346 value: Some(Default::default()),
1347 }
1348 }
1349}
1350
1351impl<K, V> cmp::Ord for DimensionPair<K, V>
1352where
1353 K: cmp::Ord,
1354{
1355 fn cmp(&self, other: &Self) -> cmp::Ordering {
1356 self.key.cmp(&other.key)
1357 }
1358}
1359
1360impl<K, V> cmp::PartialOrd for DimensionPair<K, V>
1361where
1362 K: cmp::PartialOrd,
1363{
1364 fn partial_cmp(&self, other: &Self) -> Option<cmp::Ordering> {
1365 self.key.partial_cmp(&other.key)
1366 }
1367}
1368
1369impl<K, V> cmp::PartialEq for DimensionPair<K, V>
1370where
1371 K: cmp::PartialEq,
1372{
1373 fn eq(&self, other: &Self) -> bool {
1374 self.key.eq(&other.key)
1375 }
1376}
1377
1378impl<K, V> ops::Sub for DimensionPair<K, V>
1379where
1380 K: ops::Sub<K, Output = K>,
1381 V: ops::Sub<V, Output = V>,
1382{
1383 type Output = Self;
1384
1385 fn sub(self, rhs: Self) -> Self::Output {
1386 Self {
1387 key: self.key - rhs.key,
1388 value: self.value.zip(rhs.value).map(|(a, b)| a - b),
1389 }
1390 }
1391}
1392
1393impl<K, V> cmp::Eq for DimensionPair<K, V> where K: cmp::Eq {}
1394
1395impl<'a, K, V> sum_tree::Dimension<'a, ChunkSummary> for DimensionPair<K, V>
1396where
1397 K: sum_tree::Dimension<'a, ChunkSummary>,
1398 V: sum_tree::Dimension<'a, ChunkSummary>,
1399{
1400 fn zero(_cx: &()) -> Self {
1401 Self {
1402 key: K::zero(_cx),
1403 value: Some(V::zero(_cx)),
1404 }
1405 }
1406
1407 fn add_summary(&mut self, summary: &'a ChunkSummary, _cx: &()) {
1408 self.key.add_summary(summary, _cx);
1409 if let Some(value) = &mut self.value {
1410 value.add_summary(summary, _cx);
1411 }
1412 }
1413}
1414
1415impl<K, V> TextDimension for DimensionPair<K, V>
1416where
1417 K: TextDimension,
1418 V: TextDimension,
1419{
1420 fn add_assign(&mut self, other: &Self) {
1421 self.key.add_assign(&other.key);
1422 if let Some(value) = &mut self.value {
1423 if let Some(other_value) = other.value.as_ref() {
1424 value.add_assign(other_value);
1425 } else {
1426 self.value.take();
1427 }
1428 }
1429 }
1430
1431 fn from_chunk(chunk: ChunkSlice) -> Self {
1432 Self {
1433 key: K::from_chunk(chunk),
1434 value: Some(V::from_chunk(chunk)),
1435 }
1436 }
1437
1438 fn from_text_summary(summary: &TextSummary) -> Self {
1439 Self {
1440 key: K::from_text_summary(summary),
1441 value: Some(V::from_text_summary(summary)),
1442 }
1443 }
1444}
1445
1446#[cfg(test)]
1447mod tests {
1448 use super::*;
1449 use Bias::{Left, Right};
1450 use rand::prelude::*;
1451 use std::{cmp::Ordering, env, io::Read};
1452 use util::RandomCharIter;
1453
1454 #[ctor::ctor]
1455 fn init_logger() {
1456 zlog::init_test();
1457 }
1458
1459 #[test]
1460 fn test_all_4_byte_chars() {
1461 let mut rope = Rope::new();
1462 let text = "🏀".repeat(256);
1463 rope.push(&text);
1464 assert_eq!(rope.text(), text);
1465 }
1466
1467 #[test]
1468 fn test_clip() {
1469 let rope = Rope::from("🧘");
1470
1471 assert_eq!(rope.clip_offset(1, Bias::Left), 0);
1472 assert_eq!(rope.clip_offset(1, Bias::Right), 4);
1473 assert_eq!(rope.clip_offset(5, Bias::Right), 4);
1474
1475 assert_eq!(
1476 rope.clip_point(Point::new(0, 1), Bias::Left),
1477 Point::new(0, 0)
1478 );
1479 assert_eq!(
1480 rope.clip_point(Point::new(0, 1), Bias::Right),
1481 Point::new(0, 4)
1482 );
1483 assert_eq!(
1484 rope.clip_point(Point::new(0, 5), Bias::Right),
1485 Point::new(0, 4)
1486 );
1487
1488 assert_eq!(
1489 rope.clip_point_utf16(Unclipped(PointUtf16::new(0, 1)), Bias::Left),
1490 PointUtf16::new(0, 0)
1491 );
1492 assert_eq!(
1493 rope.clip_point_utf16(Unclipped(PointUtf16::new(0, 1)), Bias::Right),
1494 PointUtf16::new(0, 2)
1495 );
1496 assert_eq!(
1497 rope.clip_point_utf16(Unclipped(PointUtf16::new(0, 3)), Bias::Right),
1498 PointUtf16::new(0, 2)
1499 );
1500
1501 assert_eq!(
1502 rope.clip_offset_utf16(OffsetUtf16(1), Bias::Left),
1503 OffsetUtf16(0)
1504 );
1505 assert_eq!(
1506 rope.clip_offset_utf16(OffsetUtf16(1), Bias::Right),
1507 OffsetUtf16(2)
1508 );
1509 assert_eq!(
1510 rope.clip_offset_utf16(OffsetUtf16(3), Bias::Right),
1511 OffsetUtf16(2)
1512 );
1513 }
1514
1515 #[test]
1516 fn test_prev_next_line() {
1517 let rope = Rope::from("abc\ndef\nghi\njkl");
1518
1519 let mut chunks = rope.chunks();
1520 assert_eq!(chunks.peek().unwrap().chars().next().unwrap(), 'a');
1521
1522 assert!(chunks.next_line());
1523 assert_eq!(chunks.peek().unwrap().chars().next().unwrap(), 'd');
1524
1525 assert!(chunks.next_line());
1526 assert_eq!(chunks.peek().unwrap().chars().next().unwrap(), 'g');
1527
1528 assert!(chunks.next_line());
1529 assert_eq!(chunks.peek().unwrap().chars().next().unwrap(), 'j');
1530
1531 assert!(!chunks.next_line());
1532 assert_eq!(chunks.peek(), None);
1533
1534 assert!(chunks.prev_line());
1535 assert_eq!(chunks.peek().unwrap().chars().next().unwrap(), 'j');
1536
1537 assert!(chunks.prev_line());
1538 assert_eq!(chunks.peek().unwrap().chars().next().unwrap(), 'g');
1539
1540 assert!(chunks.prev_line());
1541 assert_eq!(chunks.peek().unwrap().chars().next().unwrap(), 'd');
1542
1543 assert!(chunks.prev_line());
1544 assert_eq!(chunks.peek().unwrap().chars().next().unwrap(), 'a');
1545
1546 assert!(!chunks.prev_line());
1547 assert_eq!(chunks.peek().unwrap().chars().next().unwrap(), 'a');
1548
1549 // Only return true when the cursor has moved to the start of a line
1550 let mut chunks = rope.chunks_in_range(5..7);
1551 chunks.seek(6);
1552 assert!(!chunks.prev_line());
1553 assert_eq!(chunks.peek().unwrap().chars().next().unwrap(), 'e');
1554
1555 assert!(!chunks.next_line());
1556 assert_eq!(chunks.peek(), None);
1557 }
1558
1559 #[test]
1560 fn test_lines() {
1561 let rope = Rope::from("abc\ndefg\nhi");
1562 let mut lines = rope.chunks().lines();
1563 assert_eq!(lines.next(), Some("abc"));
1564 assert_eq!(lines.next(), Some("defg"));
1565 assert_eq!(lines.next(), Some("hi"));
1566 assert_eq!(lines.next(), None);
1567
1568 let rope = Rope::from("abc\ndefg\nhi\n");
1569 let mut lines = rope.chunks().lines();
1570 assert_eq!(lines.next(), Some("abc"));
1571 assert_eq!(lines.next(), Some("defg"));
1572 assert_eq!(lines.next(), Some("hi"));
1573 assert_eq!(lines.next(), Some(""));
1574 assert_eq!(lines.next(), None);
1575
1576 let rope = Rope::from("abc\ndefg\nhi");
1577 let mut lines = rope.reversed_chunks_in_range(0..rope.len()).lines();
1578 assert_eq!(lines.next(), Some("hi"));
1579 assert_eq!(lines.next(), Some("defg"));
1580 assert_eq!(lines.next(), Some("abc"));
1581 assert_eq!(lines.next(), None);
1582
1583 let rope = Rope::from("abc\ndefg\nhi\n");
1584 let mut lines = rope.reversed_chunks_in_range(0..rope.len()).lines();
1585 assert_eq!(lines.next(), Some(""));
1586 assert_eq!(lines.next(), Some("hi"));
1587 assert_eq!(lines.next(), Some("defg"));
1588 assert_eq!(lines.next(), Some("abc"));
1589 assert_eq!(lines.next(), None);
1590
1591 let rope = Rope::from("abc\nlonger line test\nhi");
1592 let mut lines = rope.chunks().lines();
1593 assert_eq!(lines.next(), Some("abc"));
1594 assert_eq!(lines.next(), Some("longer line test"));
1595 assert_eq!(lines.next(), Some("hi"));
1596 assert_eq!(lines.next(), None);
1597
1598 let rope = Rope::from("abc\nlonger line test\nhi");
1599 let mut lines = rope.reversed_chunks_in_range(0..rope.len()).lines();
1600 assert_eq!(lines.next(), Some("hi"));
1601 assert_eq!(lines.next(), Some("longer line test"));
1602 assert_eq!(lines.next(), Some("abc"));
1603 assert_eq!(lines.next(), None);
1604 }
1605
1606 #[gpui::test(iterations = 100)]
1607 fn test_random_rope(mut rng: StdRng) {
1608 let operations = env::var("OPERATIONS")
1609 .map(|i| i.parse().expect("invalid `OPERATIONS` variable"))
1610 .unwrap_or(10);
1611
1612 let mut expected = String::new();
1613 let mut actual = Rope::new();
1614 for _ in 0..operations {
1615 let end_ix = clip_offset(&expected, rng.random_range(0..=expected.len()), Right);
1616 let start_ix = clip_offset(&expected, rng.random_range(0..=end_ix), Left);
1617 let len = rng.random_range(0..=64);
1618 let new_text: String = RandomCharIter::new(&mut rng).take(len).collect();
1619
1620 let mut new_actual = Rope::new();
1621 let mut cursor = actual.cursor(0);
1622 new_actual.append(cursor.slice(start_ix));
1623 new_actual.push(&new_text);
1624 cursor.seek_forward(end_ix);
1625 new_actual.append(cursor.suffix());
1626 actual = new_actual;
1627
1628 expected.replace_range(start_ix..end_ix, &new_text);
1629
1630 assert_eq!(actual.text(), expected);
1631 log::info!("text: {:?}", expected);
1632
1633 for _ in 0..5 {
1634 let end_ix = clip_offset(&expected, rng.random_range(0..=expected.len()), Right);
1635 let start_ix = clip_offset(&expected, rng.random_range(0..=end_ix), Left);
1636
1637 let actual_text = actual.chunks_in_range(start_ix..end_ix).collect::<String>();
1638 assert_eq!(actual_text, &expected[start_ix..end_ix]);
1639
1640 let mut actual_text = String::new();
1641 actual
1642 .bytes_in_range(start_ix..end_ix)
1643 .read_to_string(&mut actual_text)
1644 .unwrap();
1645 assert_eq!(actual_text, &expected[start_ix..end_ix]);
1646
1647 assert_eq!(
1648 actual
1649 .reversed_chunks_in_range(start_ix..end_ix)
1650 .collect::<Vec<&str>>()
1651 .into_iter()
1652 .rev()
1653 .collect::<String>(),
1654 &expected[start_ix..end_ix]
1655 );
1656
1657 let mut expected_line_starts: Vec<_> = expected[start_ix..end_ix]
1658 .match_indices('\n')
1659 .map(|(index, _)| start_ix + index + 1)
1660 .collect();
1661
1662 let mut chunks = actual.chunks_in_range(start_ix..end_ix);
1663
1664 let mut actual_line_starts = Vec::new();
1665 while chunks.next_line() {
1666 actual_line_starts.push(chunks.offset());
1667 }
1668 assert_eq!(
1669 actual_line_starts,
1670 expected_line_starts,
1671 "actual line starts != expected line starts when using next_line() for {:?} ({:?})",
1672 &expected[start_ix..end_ix],
1673 start_ix..end_ix
1674 );
1675
1676 if start_ix < end_ix
1677 && (start_ix == 0 || expected.as_bytes()[start_ix - 1] == b'\n')
1678 {
1679 expected_line_starts.insert(0, start_ix);
1680 }
1681 // Remove the last index if it starts at the end of the range.
1682 if expected_line_starts.last() == Some(&end_ix) {
1683 expected_line_starts.pop();
1684 }
1685
1686 let mut actual_line_starts = Vec::new();
1687 while chunks.prev_line() {
1688 actual_line_starts.push(chunks.offset());
1689 }
1690 actual_line_starts.reverse();
1691 assert_eq!(
1692 actual_line_starts,
1693 expected_line_starts,
1694 "actual line starts != expected line starts when using prev_line() for {:?} ({:?})",
1695 &expected[start_ix..end_ix],
1696 start_ix..end_ix
1697 );
1698
1699 // Check that next_line/prev_line work correctly from random positions
1700 let mut offset = rng.random_range(start_ix..=end_ix);
1701 while !expected.is_char_boundary(offset) {
1702 offset -= 1;
1703 }
1704 chunks.seek(offset);
1705
1706 for _ in 0..5 {
1707 if rng.random() {
1708 let expected_next_line_start = expected[offset..end_ix]
1709 .find('\n')
1710 .map(|newline_ix| offset + newline_ix + 1);
1711
1712 let moved = chunks.next_line();
1713 assert_eq!(
1714 moved,
1715 expected_next_line_start.is_some(),
1716 "unexpected result from next_line after seeking to {} in range {:?} ({:?})",
1717 offset,
1718 start_ix..end_ix,
1719 &expected[start_ix..end_ix]
1720 );
1721 if let Some(expected_next_line_start) = expected_next_line_start {
1722 assert_eq!(
1723 chunks.offset(),
1724 expected_next_line_start,
1725 "invalid position after seeking to {} in range {:?} ({:?})",
1726 offset,
1727 start_ix..end_ix,
1728 &expected[start_ix..end_ix]
1729 );
1730 } else {
1731 assert_eq!(
1732 chunks.offset(),
1733 end_ix,
1734 "invalid position after seeking to {} in range {:?} ({:?})",
1735 offset,
1736 start_ix..end_ix,
1737 &expected[start_ix..end_ix]
1738 );
1739 }
1740 } else {
1741 let search_end = if offset > 0 && expected.as_bytes()[offset - 1] == b'\n' {
1742 offset - 1
1743 } else {
1744 offset
1745 };
1746
1747 let expected_prev_line_start = expected[..search_end]
1748 .rfind('\n')
1749 .and_then(|newline_ix| {
1750 let line_start_ix = newline_ix + 1;
1751 if line_start_ix >= start_ix {
1752 Some(line_start_ix)
1753 } else {
1754 None
1755 }
1756 })
1757 .or({
1758 if offset > 0 && start_ix == 0 {
1759 Some(0)
1760 } else {
1761 None
1762 }
1763 });
1764
1765 let moved = chunks.prev_line();
1766 assert_eq!(
1767 moved,
1768 expected_prev_line_start.is_some(),
1769 "unexpected result from prev_line after seeking to {} in range {:?} ({:?})",
1770 offset,
1771 start_ix..end_ix,
1772 &expected[start_ix..end_ix]
1773 );
1774 if let Some(expected_prev_line_start) = expected_prev_line_start {
1775 assert_eq!(
1776 chunks.offset(),
1777 expected_prev_line_start,
1778 "invalid position after seeking to {} in range {:?} ({:?})",
1779 offset,
1780 start_ix..end_ix,
1781 &expected[start_ix..end_ix]
1782 );
1783 } else {
1784 assert_eq!(
1785 chunks.offset(),
1786 start_ix,
1787 "invalid position after seeking to {} in range {:?} ({:?})",
1788 offset,
1789 start_ix..end_ix,
1790 &expected[start_ix..end_ix]
1791 );
1792 }
1793 }
1794
1795 assert!((start_ix..=end_ix).contains(&chunks.offset()));
1796 if rng.random() {
1797 offset = rng.random_range(start_ix..=end_ix);
1798 while !expected.is_char_boundary(offset) {
1799 offset -= 1;
1800 }
1801 chunks.seek(offset);
1802 } else {
1803 chunks.next();
1804 offset = chunks.offset();
1805 assert!((start_ix..=end_ix).contains(&chunks.offset()));
1806 }
1807 }
1808 }
1809
1810 let mut offset_utf16 = OffsetUtf16(0);
1811 let mut point = Point::new(0, 0);
1812 let mut point_utf16 = PointUtf16::new(0, 0);
1813 for (ix, ch) in expected.char_indices().chain(Some((expected.len(), '\0'))) {
1814 assert_eq!(actual.offset_to_point(ix), point, "offset_to_point({})", ix);
1815 assert_eq!(
1816 actual.offset_to_point_utf16(ix),
1817 point_utf16,
1818 "offset_to_point_utf16({})",
1819 ix
1820 );
1821 assert_eq!(
1822 actual.point_to_offset(point),
1823 ix,
1824 "point_to_offset({:?})",
1825 point
1826 );
1827 assert_eq!(
1828 actual.point_utf16_to_offset(point_utf16),
1829 ix,
1830 "point_utf16_to_offset({:?})",
1831 point_utf16
1832 );
1833 assert_eq!(
1834 actual.offset_to_offset_utf16(ix),
1835 offset_utf16,
1836 "offset_to_offset_utf16({:?})",
1837 ix
1838 );
1839 assert_eq!(
1840 actual.offset_utf16_to_offset(offset_utf16),
1841 ix,
1842 "offset_utf16_to_offset({:?})",
1843 offset_utf16
1844 );
1845 if ch == '\n' {
1846 point += Point::new(1, 0);
1847 point_utf16 += PointUtf16::new(1, 0);
1848 } else {
1849 point.column += ch.len_utf8() as u32;
1850 point_utf16.column += ch.len_utf16() as u32;
1851 }
1852 offset_utf16.0 += ch.len_utf16();
1853 }
1854
1855 let mut offset_utf16 = OffsetUtf16(0);
1856 let mut point_utf16 = Unclipped(PointUtf16::zero());
1857 for unit in expected.encode_utf16() {
1858 let left_offset = actual.clip_offset_utf16(offset_utf16, Bias::Left);
1859 let right_offset = actual.clip_offset_utf16(offset_utf16, Bias::Right);
1860 assert!(right_offset >= left_offset);
1861 // Ensure translating UTF-16 offsets to UTF-8 offsets doesn't panic.
1862 actual.offset_utf16_to_offset(left_offset);
1863 actual.offset_utf16_to_offset(right_offset);
1864
1865 let left_point = actual.clip_point_utf16(point_utf16, Bias::Left);
1866 let right_point = actual.clip_point_utf16(point_utf16, Bias::Right);
1867 assert!(right_point >= left_point);
1868 // Ensure translating valid UTF-16 points to offsets doesn't panic.
1869 actual.point_utf16_to_offset(left_point);
1870 actual.point_utf16_to_offset(right_point);
1871
1872 offset_utf16.0 += 1;
1873 if unit == b'\n' as u16 {
1874 point_utf16.0 += PointUtf16::new(1, 0);
1875 } else {
1876 point_utf16.0 += PointUtf16::new(0, 1);
1877 }
1878 }
1879
1880 for _ in 0..5 {
1881 let end_ix = clip_offset(&expected, rng.random_range(0..=expected.len()), Right);
1882 let start_ix = clip_offset(&expected, rng.random_range(0..=end_ix), Left);
1883 assert_eq!(
1884 actual.cursor(start_ix).summary::<TextSummary>(end_ix),
1885 TextSummary::from(&expected[start_ix..end_ix])
1886 );
1887 }
1888
1889 let mut expected_longest_rows = Vec::new();
1890 let mut longest_line_len = -1_isize;
1891 for (row, line) in expected.split('\n').enumerate() {
1892 let row = row as u32;
1893 assert_eq!(
1894 actual.line_len(row),
1895 line.len() as u32,
1896 "invalid line len for row {}",
1897 row
1898 );
1899
1900 let line_char_count = line.chars().count() as isize;
1901 match line_char_count.cmp(&longest_line_len) {
1902 Ordering::Less => {}
1903 Ordering::Equal => expected_longest_rows.push(row),
1904 Ordering::Greater => {
1905 longest_line_len = line_char_count;
1906 expected_longest_rows.clear();
1907 expected_longest_rows.push(row);
1908 }
1909 }
1910 }
1911
1912 let longest_row = actual.summary().longest_row;
1913 assert!(
1914 expected_longest_rows.contains(&longest_row),
1915 "incorrect longest row {}. expected {:?} with length {}",
1916 longest_row,
1917 expected_longest_rows,
1918 longest_line_len,
1919 );
1920 }
1921 }
1922
1923 #[test]
1924 fn test_chunks_equals_str() {
1925 let text = "This is a multi-chunk\n& multi-line test string!";
1926 let rope = Rope::from(text);
1927 for start in 0..text.len() {
1928 for end in start..text.len() {
1929 let range = start..end;
1930 let correct_substring = &text[start..end];
1931
1932 // Test that correct range returns true
1933 assert!(
1934 rope.chunks_in_range(range.clone())
1935 .equals_str(correct_substring)
1936 );
1937 assert!(
1938 rope.reversed_chunks_in_range(range.clone())
1939 .equals_str(correct_substring)
1940 );
1941
1942 // Test that all other ranges return false (unless they happen to match)
1943 for other_start in 0..text.len() {
1944 for other_end in other_start..text.len() {
1945 if other_start == start && other_end == end {
1946 continue;
1947 }
1948 let other_substring = &text[other_start..other_end];
1949
1950 // Only assert false if the substrings are actually different
1951 if other_substring == correct_substring {
1952 continue;
1953 }
1954 assert!(
1955 !rope
1956 .chunks_in_range(range.clone())
1957 .equals_str(other_substring)
1958 );
1959 assert!(
1960 !rope
1961 .reversed_chunks_in_range(range.clone())
1962 .equals_str(other_substring)
1963 );
1964 }
1965 }
1966 }
1967 }
1968
1969 let rope = Rope::from("");
1970 assert!(rope.chunks_in_range(0..0).equals_str(""));
1971 assert!(rope.reversed_chunks_in_range(0..0).equals_str(""));
1972 assert!(!rope.chunks_in_range(0..0).equals_str("foo"));
1973 assert!(!rope.reversed_chunks_in_range(0..0).equals_str("foo"));
1974 }
1975
1976 fn clip_offset(text: &str, mut offset: usize, bias: Bias) -> usize {
1977 while !text.is_char_boundary(offset) {
1978 match bias {
1979 Bias::Left => offset -= 1,
1980 Bias::Right => offset += 1,
1981 }
1982 }
1983 offset
1984 }
1985
1986 impl Rope {
1987 fn text(&self) -> String {
1988 let mut text = String::new();
1989 for chunk in self.chunks.cursor::<()>(&()) {
1990 text.push_str(&chunk.text);
1991 }
1992 text
1993 }
1994 }
1995}