1use std::{iter, ops::Range};
2use sum_tree::SumTree;
3use text::{Anchor, BufferSnapshot, Point};
4
5pub use git2 as libgit;
6use libgit::{DiffLineType as GitDiffLineType, DiffOptions as GitOptions, Patch as GitPatch};
7
8#[derive(Debug, Clone, Copy, PartialEq, Eq)]
9pub enum DiffHunkStatus {
10 Added,
11 Modified,
12 Removed,
13}
14
15#[derive(Debug, Clone, PartialEq, Eq)]
16pub struct DiffHunk<T> {
17 pub buffer_range: Range<T>,
18 pub diff_base_byte_range: Range<usize>,
19}
20
21impl DiffHunk<u32> {
22 pub fn status(&self) -> DiffHunkStatus {
23 if self.diff_base_byte_range.is_empty() {
24 DiffHunkStatus::Added
25 } else if self.buffer_range.is_empty() {
26 DiffHunkStatus::Removed
27 } else {
28 DiffHunkStatus::Modified
29 }
30 }
31}
32
33impl sum_tree::Item for DiffHunk<Anchor> {
34 type Summary = DiffHunkSummary;
35
36 fn summary(&self) -> Self::Summary {
37 DiffHunkSummary {
38 buffer_range: self.buffer_range.clone(),
39 }
40 }
41}
42
43#[derive(Debug, Default, Clone)]
44pub struct DiffHunkSummary {
45 buffer_range: Range<Anchor>,
46}
47
48impl sum_tree::Summary for DiffHunkSummary {
49 type Context = text::BufferSnapshot;
50
51 fn add_summary(&mut self, other: &Self, buffer: &Self::Context) {
52 self.buffer_range.start = self
53 .buffer_range
54 .start
55 .min(&other.buffer_range.start, buffer);
56 self.buffer_range.end = self.buffer_range.end.max(&other.buffer_range.end, buffer);
57 }
58}
59
60#[derive(Clone)]
61pub struct BufferDiff {
62 last_buffer_version: Option<clock::Global>,
63 tree: SumTree<DiffHunk<Anchor>>,
64}
65
66impl BufferDiff {
67 pub fn new() -> BufferDiff {
68 BufferDiff {
69 last_buffer_version: None,
70 tree: SumTree::new(),
71 }
72 }
73
74 pub fn hunks_in_row_range<'a>(
75 &'a self,
76 range: Range<u32>,
77 buffer: &'a BufferSnapshot,
78 reversed: bool,
79 ) -> impl 'a + Iterator<Item = DiffHunk<u32>> {
80 let start = buffer.anchor_before(Point::new(range.start, 0));
81 let end = buffer.anchor_after(Point::new(range.end, 0));
82 self.hunks_intersecting_range(start..end, buffer, reversed)
83 }
84
85 pub fn hunks_intersecting_range<'a>(
86 &'a self,
87 range: Range<Anchor>,
88 buffer: &'a BufferSnapshot,
89 reversed: bool,
90 ) -> impl 'a + Iterator<Item = DiffHunk<u32>> {
91 let mut cursor = self.tree.filter::<_, DiffHunkSummary>(move |summary| {
92 let before_start = summary.buffer_range.end.cmp(&range.start, buffer).is_lt();
93 let after_end = summary.buffer_range.start.cmp(&range.end, buffer).is_gt();
94 !before_start && !after_end
95 });
96
97 let anchor_iter = std::iter::from_fn(move || {
98 if reversed {
99 cursor.prev(buffer);
100 } else {
101 cursor.next(buffer);
102 }
103
104 cursor.item()
105 })
106 .flat_map(move |hunk| {
107 [
108 (&hunk.buffer_range.start, hunk.diff_base_byte_range.start),
109 (&hunk.buffer_range.end, hunk.diff_base_byte_range.end),
110 ]
111 .into_iter()
112 });
113
114 let mut summaries = buffer.summaries_for_anchors_with_payload::<Point, _, _>(anchor_iter);
115 iter::from_fn(move || {
116 let (start_point, start_base) = summaries.next()?;
117 let (end_point, end_base) = summaries.next()?;
118
119 let end_row = if end_point.column > 0 {
120 end_point.row + 1
121 } else {
122 end_point.row
123 };
124
125 Some(DiffHunk {
126 buffer_range: start_point.row..end_row,
127 diff_base_byte_range: start_base..end_base,
128 })
129 })
130 }
131
132 pub fn clear(&mut self, buffer: &text::BufferSnapshot) {
133 self.last_buffer_version = Some(buffer.version().clone());
134 self.tree = SumTree::new();
135 }
136
137 pub fn needs_update(&self, buffer: &text::BufferSnapshot) -> bool {
138 match &self.last_buffer_version {
139 Some(last) => buffer.version().changed_since(last),
140 None => true,
141 }
142 }
143
144 pub async fn update(&mut self, diff_base: &str, buffer: &text::BufferSnapshot) {
145 let mut tree = SumTree::new();
146
147 let buffer_text = buffer.as_rope().to_string();
148 let patch = Self::diff(&diff_base, &buffer_text);
149
150 if let Some(patch) = patch {
151 let mut divergence = 0;
152 for hunk_index in 0..patch.num_hunks() {
153 let hunk = Self::process_patch_hunk(&patch, hunk_index, buffer, &mut divergence);
154 tree.push(hunk, buffer);
155 }
156 }
157
158 self.tree = tree;
159 self.last_buffer_version = Some(buffer.version().clone());
160 }
161
162 #[cfg(test)]
163 fn hunks<'a>(&'a self, text: &'a BufferSnapshot) -> impl 'a + Iterator<Item = DiffHunk<u32>> {
164 let start = text.anchor_before(Point::new(0, 0));
165 let end = text.anchor_after(Point::new(u32::MAX, u32::MAX));
166 self.hunks_intersecting_range(start..end, text, false)
167 }
168
169 fn diff<'a>(head: &'a str, current: &'a str) -> Option<GitPatch<'a>> {
170 let mut options = GitOptions::default();
171 options.context_lines(0);
172
173 let patch = GitPatch::from_buffers(
174 head.as_bytes(),
175 None,
176 current.as_bytes(),
177 None,
178 Some(&mut options),
179 );
180
181 match patch {
182 Ok(patch) => Some(patch),
183
184 Err(err) => {
185 log::error!("`GitPatch::from_buffers` failed: {}", err);
186 None
187 }
188 }
189 }
190
191 fn process_patch_hunk<'a>(
192 patch: &GitPatch<'a>,
193 hunk_index: usize,
194 buffer: &text::BufferSnapshot,
195 buffer_row_divergence: &mut i64,
196 ) -> DiffHunk<Anchor> {
197 let line_item_count = patch.num_lines_in_hunk(hunk_index).unwrap();
198 assert!(line_item_count > 0);
199
200 let mut first_deletion_buffer_row: Option<u32> = None;
201 let mut buffer_row_range: Option<Range<u32>> = None;
202 let mut diff_base_byte_range: Option<Range<usize>> = None;
203
204 for line_index in 0..line_item_count {
205 let line = patch.line_in_hunk(hunk_index, line_index).unwrap();
206 let kind = line.origin_value();
207 let content_offset = line.content_offset() as isize;
208 let content_len = line.content().len() as isize;
209
210 if kind == GitDiffLineType::Addition {
211 *buffer_row_divergence += 1;
212 let row = line.new_lineno().unwrap().saturating_sub(1);
213
214 match &mut buffer_row_range {
215 Some(buffer_row_range) => buffer_row_range.end = row + 1,
216 None => buffer_row_range = Some(row..row + 1),
217 }
218 }
219
220 if kind == GitDiffLineType::Deletion {
221 let end = content_offset + content_len;
222
223 match &mut diff_base_byte_range {
224 Some(head_byte_range) => head_byte_range.end = end as usize,
225 None => diff_base_byte_range = Some(content_offset as usize..end as usize),
226 }
227
228 if first_deletion_buffer_row.is_none() {
229 let old_row = line.old_lineno().unwrap().saturating_sub(1);
230 let row = old_row as i64 + *buffer_row_divergence;
231 first_deletion_buffer_row = Some(row as u32);
232 }
233
234 *buffer_row_divergence -= 1;
235 }
236 }
237
238 //unwrap_or deletion without addition
239 let buffer_row_range = buffer_row_range.unwrap_or_else(|| {
240 //we cannot have an addition-less hunk without deletion(s) or else there would be no hunk
241 let row = first_deletion_buffer_row.unwrap();
242 row..row
243 });
244
245 //unwrap_or addition without deletion
246 let diff_base_byte_range = diff_base_byte_range.unwrap_or(0..0);
247
248 let start = Point::new(buffer_row_range.start, 0);
249 let end = Point::new(buffer_row_range.end, 0);
250 let buffer_range = buffer.anchor_before(start)..buffer.anchor_before(end);
251 DiffHunk {
252 buffer_range,
253 diff_base_byte_range,
254 }
255 }
256}
257
258/// Range (crossing new lines), old, new
259#[cfg(any(test, feature = "test-support"))]
260#[track_caller]
261pub fn assert_hunks<Iter>(
262 diff_hunks: Iter,
263 buffer: &BufferSnapshot,
264 diff_base: &str,
265 expected_hunks: &[(Range<u32>, &str, &str)],
266) where
267 Iter: Iterator<Item = DiffHunk<u32>>,
268{
269 let actual_hunks = diff_hunks
270 .map(|hunk| {
271 (
272 hunk.buffer_range.clone(),
273 &diff_base[hunk.diff_base_byte_range],
274 buffer
275 .text_for_range(
276 Point::new(hunk.buffer_range.start, 0)
277 ..Point::new(hunk.buffer_range.end, 0),
278 )
279 .collect::<String>(),
280 )
281 })
282 .collect::<Vec<_>>();
283
284 let expected_hunks: Vec<_> = expected_hunks
285 .iter()
286 .map(|(r, s, h)| (r.clone(), *s, h.to_string()))
287 .collect();
288
289 assert_eq!(actual_hunks, expected_hunks);
290}
291
292#[cfg(test)]
293mod tests {
294 use std::assert_eq;
295
296 use super::*;
297 use text::Buffer;
298 use unindent::Unindent as _;
299
300 #[test]
301 fn test_buffer_diff_simple() {
302 let diff_base = "
303 one
304 two
305 three
306 "
307 .unindent();
308
309 let buffer_text = "
310 one
311 HELLO
312 three
313 "
314 .unindent();
315
316 let mut buffer = Buffer::new(0, 0, buffer_text);
317 let mut diff = BufferDiff::new();
318 smol::block_on(diff.update(&diff_base, &buffer));
319 assert_hunks(
320 diff.hunks(&buffer),
321 &buffer,
322 &diff_base,
323 &[(1..2, "two\n", "HELLO\n")],
324 );
325
326 buffer.edit([(0..0, "point five\n")]);
327 smol::block_on(diff.update(&diff_base, &buffer));
328 assert_hunks(
329 diff.hunks(&buffer),
330 &buffer,
331 &diff_base,
332 &[(0..1, "", "point five\n"), (2..3, "two\n", "HELLO\n")],
333 );
334
335 diff.clear(&buffer);
336 assert_hunks(diff.hunks(&buffer), &buffer, &diff_base, &[]);
337 }
338
339 #[test]
340 fn test_buffer_diff_range() {
341 let diff_base = "
342 one
343 two
344 three
345 four
346 five
347 six
348 seven
349 eight
350 nine
351 ten
352 "
353 .unindent();
354
355 let buffer_text = "
356 A
357 one
358 B
359 two
360 C
361 three
362 HELLO
363 four
364 five
365 SIXTEEN
366 seven
367 eight
368 WORLD
369 nine
370
371 ten
372
373 "
374 .unindent();
375
376 let buffer = Buffer::new(0, 0, buffer_text);
377 let mut diff = BufferDiff::new();
378 smol::block_on(diff.update(&diff_base, &buffer));
379 assert_eq!(diff.hunks(&buffer).count(), 8);
380
381 assert_hunks(
382 diff.hunks_in_row_range(7..12, &buffer, false),
383 &buffer,
384 &diff_base,
385 &[
386 (6..7, "", "HELLO\n"),
387 (9..10, "six\n", "SIXTEEN\n"),
388 (12..13, "", "WORLD\n"),
389 ],
390 );
391 }
392}