1use crate::streaming_diff::{Hunk, StreamingDiff};
2use ai::completion::{CompletionProvider, OpenAIRequest};
3use anyhow::Result;
4use editor::{multi_buffer, Anchor, MultiBuffer, MultiBufferSnapshot, ToOffset, ToPoint};
5use futures::{channel::mpsc, SinkExt, Stream, StreamExt};
6use gpui::{Entity, ModelContext, ModelHandle, Task};
7use language::{Rope, TransactionId};
8use std::{cmp, future, ops::Range, sync::Arc};
9
10pub enum Event {
11 Finished,
12 Undone,
13}
14
15#[derive(Clone)]
16pub enum CodegenKind {
17 Transform { range: Range<Anchor> },
18 Generate { position: Anchor },
19}
20
21pub struct Codegen {
22 provider: Arc<dyn CompletionProvider>,
23 buffer: ModelHandle<MultiBuffer>,
24 snapshot: MultiBufferSnapshot,
25 kind: CodegenKind,
26 last_equal_ranges: Vec<Range<Anchor>>,
27 transaction_id: Option<TransactionId>,
28 error: Option<anyhow::Error>,
29 generation: Task<()>,
30 idle: bool,
31 _subscription: gpui::Subscription,
32}
33
34impl Entity for Codegen {
35 type Event = Event;
36}
37
38impl Codegen {
39 pub fn new(
40 buffer: ModelHandle<MultiBuffer>,
41 kind: CodegenKind,
42 provider: Arc<dyn CompletionProvider>,
43 cx: &mut ModelContext<Self>,
44 ) -> Self {
45 let snapshot = buffer.read(cx).snapshot(cx);
46 Self {
47 provider,
48 buffer: buffer.clone(),
49 snapshot,
50 kind,
51 last_equal_ranges: Default::default(),
52 transaction_id: Default::default(),
53 error: Default::default(),
54 idle: true,
55 generation: Task::ready(()),
56 _subscription: cx.subscribe(&buffer, Self::handle_buffer_event),
57 }
58 }
59
60 fn handle_buffer_event(
61 &mut self,
62 _buffer: ModelHandle<MultiBuffer>,
63 event: &multi_buffer::Event,
64 cx: &mut ModelContext<Self>,
65 ) {
66 if let multi_buffer::Event::TransactionUndone { transaction_id } = event {
67 if self.transaction_id == Some(*transaction_id) {
68 self.transaction_id = None;
69 self.generation = Task::ready(());
70 cx.emit(Event::Undone);
71 }
72 }
73 }
74
75 pub fn range(&self) -> Range<Anchor> {
76 match &self.kind {
77 CodegenKind::Transform { range } => range.clone(),
78 CodegenKind::Generate { position } => position.bias_left(&self.snapshot)..*position,
79 }
80 }
81
82 pub fn kind(&self) -> &CodegenKind {
83 &self.kind
84 }
85
86 pub fn last_equal_ranges(&self) -> &[Range<Anchor>] {
87 &self.last_equal_ranges
88 }
89
90 pub fn idle(&self) -> bool {
91 self.idle
92 }
93
94 pub fn error(&self) -> Option<&anyhow::Error> {
95 self.error.as_ref()
96 }
97
98 pub fn start(&mut self, prompt: OpenAIRequest, cx: &mut ModelContext<Self>) {
99 let range = self.range();
100 let snapshot = self.snapshot.clone();
101 let selected_text = snapshot
102 .text_for_range(range.start..range.end)
103 .collect::<Rope>();
104
105 let selection_start = range.start.to_point(&snapshot);
106 let suggested_line_indent = snapshot
107 .suggested_indents(selection_start.row..selection_start.row + 1, cx)
108 .into_values()
109 .next()
110 .unwrap_or_else(|| snapshot.indent_size_for_line(selection_start.row));
111
112 let response = self.provider.complete(prompt);
113 self.generation = cx.spawn_weak(|this, mut cx| {
114 async move {
115 let generate = async {
116 let mut edit_start = range.start.to_offset(&snapshot);
117
118 let (mut hunks_tx, mut hunks_rx) = mpsc::channel(1);
119 let diff = cx.background().spawn(async move {
120 let chunks = strip_markdown_codeblock(response.await?);
121 futures::pin_mut!(chunks);
122 let mut diff = StreamingDiff::new(selected_text.to_string());
123
124 let mut new_text = String::new();
125 let mut base_indent = None;
126 let mut line_indent = None;
127 let mut first_line = true;
128
129 while let Some(chunk) = chunks.next().await {
130 let chunk = chunk?;
131
132 let mut lines = chunk.split('\n').peekable();
133 while let Some(line) = lines.next() {
134 new_text.push_str(line);
135 if line_indent.is_none() {
136 if let Some(non_whitespace_ch_ix) =
137 new_text.find(|ch: char| !ch.is_whitespace())
138 {
139 line_indent = Some(non_whitespace_ch_ix);
140 base_indent = base_indent.or(line_indent);
141
142 let line_indent = line_indent.unwrap();
143 let base_indent = base_indent.unwrap();
144 let indent_delta = line_indent as i32 - base_indent as i32;
145 let mut corrected_indent_len = cmp::max(
146 0,
147 suggested_line_indent.len as i32 + indent_delta,
148 )
149 as usize;
150 if first_line {
151 corrected_indent_len = corrected_indent_len
152 .saturating_sub(selection_start.column as usize);
153 }
154
155 let indent_char = suggested_line_indent.char();
156 let mut indent_buffer = [0; 4];
157 let indent_str =
158 indent_char.encode_utf8(&mut indent_buffer);
159 new_text.replace_range(
160 ..line_indent,
161 &indent_str.repeat(corrected_indent_len),
162 );
163 }
164 }
165
166 if line_indent.is_some() {
167 hunks_tx.send(diff.push_new(&new_text)).await?;
168 new_text.clear();
169 }
170
171 if lines.peek().is_some() {
172 hunks_tx.send(diff.push_new("\n")).await?;
173 line_indent = None;
174 first_line = false;
175 }
176 }
177 }
178 hunks_tx.send(diff.push_new(&new_text)).await?;
179 hunks_tx.send(diff.finish()).await?;
180
181 anyhow::Ok(())
182 });
183
184 while let Some(hunks) = hunks_rx.next().await {
185 let this = if let Some(this) = this.upgrade(&cx) {
186 this
187 } else {
188 break;
189 };
190
191 this.update(&mut cx, |this, cx| {
192 this.last_equal_ranges.clear();
193
194 let transaction = this.buffer.update(cx, |buffer, cx| {
195 // Avoid grouping assistant edits with user edits.
196 buffer.finalize_last_transaction(cx);
197
198 buffer.start_transaction(cx);
199 buffer.edit(
200 hunks.into_iter().filter_map(|hunk| match hunk {
201 Hunk::Insert { text } => {
202 let edit_start = snapshot.anchor_after(edit_start);
203 Some((edit_start..edit_start, text))
204 }
205 Hunk::Remove { len } => {
206 let edit_end = edit_start + len;
207 let edit_range = snapshot.anchor_after(edit_start)
208 ..snapshot.anchor_before(edit_end);
209 edit_start = edit_end;
210 Some((edit_range, String::new()))
211 }
212 Hunk::Keep { len } => {
213 let edit_end = edit_start + len;
214 let edit_range = snapshot.anchor_after(edit_start)
215 ..snapshot.anchor_before(edit_end);
216 edit_start = edit_end;
217 this.last_equal_ranges.push(edit_range);
218 None
219 }
220 }),
221 None,
222 cx,
223 );
224
225 buffer.end_transaction(cx)
226 });
227
228 if let Some(transaction) = transaction {
229 if let Some(first_transaction) = this.transaction_id {
230 // Group all assistant edits into the first transaction.
231 this.buffer.update(cx, |buffer, cx| {
232 buffer.merge_transactions(
233 transaction,
234 first_transaction,
235 cx,
236 )
237 });
238 } else {
239 this.transaction_id = Some(transaction);
240 this.buffer.update(cx, |buffer, cx| {
241 buffer.finalize_last_transaction(cx)
242 });
243 }
244 }
245
246 cx.notify();
247 });
248 }
249
250 diff.await?;
251 anyhow::Ok(())
252 };
253
254 let result = generate.await;
255 if let Some(this) = this.upgrade(&cx) {
256 this.update(&mut cx, |this, cx| {
257 this.last_equal_ranges.clear();
258 this.idle = true;
259 if let Err(error) = result {
260 this.error = Some(error);
261 }
262 cx.emit(Event::Finished);
263 cx.notify();
264 });
265 }
266 }
267 });
268 self.error.take();
269 self.idle = false;
270 cx.notify();
271 }
272
273 pub fn undo(&mut self, cx: &mut ModelContext<Self>) {
274 if let Some(transaction_id) = self.transaction_id {
275 self.buffer
276 .update(cx, |buffer, cx| buffer.undo_transaction(transaction_id, cx));
277 }
278 }
279}
280
281fn strip_markdown_codeblock(
282 stream: impl Stream<Item = Result<String>>,
283) -> impl Stream<Item = Result<String>> {
284 let mut first_line = true;
285 let mut buffer = String::new();
286 let mut starts_with_fenced_code_block = false;
287 stream.filter_map(move |chunk| {
288 let chunk = match chunk {
289 Ok(chunk) => chunk,
290 Err(err) => return future::ready(Some(Err(err))),
291 };
292 buffer.push_str(&chunk);
293
294 if first_line {
295 if buffer == "" || buffer == "`" || buffer == "``" {
296 return future::ready(None);
297 } else if buffer.starts_with("```") {
298 starts_with_fenced_code_block = true;
299 if let Some(newline_ix) = buffer.find('\n') {
300 buffer.replace_range(..newline_ix + 1, "");
301 first_line = false;
302 } else {
303 return future::ready(None);
304 }
305 }
306 }
307
308 let text = if starts_with_fenced_code_block {
309 buffer
310 .strip_suffix("\n```\n")
311 .or_else(|| buffer.strip_suffix("\n```"))
312 .or_else(|| buffer.strip_suffix("\n``"))
313 .or_else(|| buffer.strip_suffix("\n`"))
314 .or_else(|| buffer.strip_suffix('\n'))
315 .unwrap_or(&buffer)
316 } else {
317 &buffer
318 };
319
320 if text.contains('\n') {
321 first_line = false;
322 }
323
324 let remainder = buffer.split_off(text.len());
325 let result = if buffer.is_empty() {
326 None
327 } else {
328 Some(Ok(buffer.clone()))
329 };
330 buffer = remainder;
331 future::ready(result)
332 })
333}
334
335#[cfg(test)]
336mod tests {
337 use super::*;
338 use futures::{
339 future::BoxFuture,
340 stream::{self, BoxStream},
341 };
342 use gpui::{executor::Deterministic, TestAppContext};
343 use indoc::indoc;
344 use language::{language_settings, tree_sitter_rust, Buffer, Language, LanguageConfig, Point};
345 use parking_lot::Mutex;
346 use rand::prelude::*;
347 use settings::SettingsStore;
348 use smol::future::FutureExt;
349
350 #[gpui::test(iterations = 10)]
351 async fn test_transform_autoindent(
352 cx: &mut TestAppContext,
353 mut rng: StdRng,
354 deterministic: Arc<Deterministic>,
355 ) {
356 cx.set_global(cx.read(SettingsStore::test));
357 cx.update(language_settings::init);
358
359 let text = indoc! {"
360 fn main() {
361 let x = 0;
362 for _ in 0..10 {
363 x += 1;
364 }
365 }
366 "};
367 let buffer =
368 cx.add_model(|cx| Buffer::new(0, 0, text).with_language(Arc::new(rust_lang()), cx));
369 let buffer = cx.add_model(|cx| MultiBuffer::singleton(buffer, cx));
370 let range = buffer.read_with(cx, |buffer, cx| {
371 let snapshot = buffer.snapshot(cx);
372 snapshot.anchor_before(Point::new(1, 0))..snapshot.anchor_after(Point::new(4, 5))
373 });
374 let provider = Arc::new(TestCompletionProvider::new());
375 let codegen = cx.add_model(|cx| {
376 Codegen::new(
377 buffer.clone(),
378 CodegenKind::Transform { range },
379 provider.clone(),
380 cx,
381 )
382 });
383 codegen.update(cx, |codegen, cx| codegen.start(Default::default(), cx));
384
385 let mut new_text = concat!(
386 " let mut x = 0;\n",
387 " while x < 10 {\n",
388 " x += 1;\n",
389 " }",
390 );
391 while !new_text.is_empty() {
392 let max_len = cmp::min(new_text.len(), 10);
393 let len = rng.gen_range(1..=max_len);
394 let (chunk, suffix) = new_text.split_at(len);
395 provider.send_completion(chunk);
396 new_text = suffix;
397 deterministic.run_until_parked();
398 }
399 provider.finish_completion();
400 deterministic.run_until_parked();
401
402 assert_eq!(
403 buffer.read_with(cx, |buffer, cx| buffer.snapshot(cx).text()),
404 indoc! {"
405 fn main() {
406 let mut x = 0;
407 while x < 10 {
408 x += 1;
409 }
410 }
411 "}
412 );
413 }
414
415 #[gpui::test(iterations = 10)]
416 async fn test_autoindent_when_generating_past_indentation(
417 cx: &mut TestAppContext,
418 mut rng: StdRng,
419 deterministic: Arc<Deterministic>,
420 ) {
421 cx.set_global(cx.read(SettingsStore::test));
422 cx.update(language_settings::init);
423
424 let text = indoc! {"
425 fn main() {
426 le
427 }
428 "};
429 let buffer =
430 cx.add_model(|cx| Buffer::new(0, 0, text).with_language(Arc::new(rust_lang()), cx));
431 let buffer = cx.add_model(|cx| MultiBuffer::singleton(buffer, cx));
432 let position = buffer.read_with(cx, |buffer, cx| {
433 let snapshot = buffer.snapshot(cx);
434 snapshot.anchor_before(Point::new(1, 6))
435 });
436 let provider = Arc::new(TestCompletionProvider::new());
437 let codegen = cx.add_model(|cx| {
438 Codegen::new(
439 buffer.clone(),
440 CodegenKind::Generate { position },
441 provider.clone(),
442 cx,
443 )
444 });
445 codegen.update(cx, |codegen, cx| codegen.start(Default::default(), cx));
446
447 let mut new_text = concat!(
448 "t mut x = 0;\n",
449 "while x < 10 {\n",
450 " x += 1;\n",
451 "}", //
452 );
453 while !new_text.is_empty() {
454 let max_len = cmp::min(new_text.len(), 10);
455 let len = rng.gen_range(1..=max_len);
456 let (chunk, suffix) = new_text.split_at(len);
457 provider.send_completion(chunk);
458 new_text = suffix;
459 deterministic.run_until_parked();
460 }
461 provider.finish_completion();
462 deterministic.run_until_parked();
463
464 assert_eq!(
465 buffer.read_with(cx, |buffer, cx| buffer.snapshot(cx).text()),
466 indoc! {"
467 fn main() {
468 let mut x = 0;
469 while x < 10 {
470 x += 1;
471 }
472 }
473 "}
474 );
475 }
476
477 #[gpui::test(iterations = 10)]
478 async fn test_autoindent_when_generating_before_indentation(
479 cx: &mut TestAppContext,
480 mut rng: StdRng,
481 deterministic: Arc<Deterministic>,
482 ) {
483 cx.set_global(cx.read(SettingsStore::test));
484 cx.update(language_settings::init);
485
486 let text = concat!(
487 "fn main() {\n",
488 " \n",
489 "}\n" //
490 );
491 let buffer =
492 cx.add_model(|cx| Buffer::new(0, 0, text).with_language(Arc::new(rust_lang()), cx));
493 let buffer = cx.add_model(|cx| MultiBuffer::singleton(buffer, cx));
494 let position = buffer.read_with(cx, |buffer, cx| {
495 let snapshot = buffer.snapshot(cx);
496 snapshot.anchor_before(Point::new(1, 2))
497 });
498 let provider = Arc::new(TestCompletionProvider::new());
499 let codegen = cx.add_model(|cx| {
500 Codegen::new(
501 buffer.clone(),
502 CodegenKind::Generate { position },
503 provider.clone(),
504 cx,
505 )
506 });
507 codegen.update(cx, |codegen, cx| codegen.start(Default::default(), cx));
508
509 let mut new_text = concat!(
510 "let mut x = 0;\n",
511 "while x < 10 {\n",
512 " x += 1;\n",
513 "}", //
514 );
515 while !new_text.is_empty() {
516 let max_len = cmp::min(new_text.len(), 10);
517 let len = rng.gen_range(1..=max_len);
518 let (chunk, suffix) = new_text.split_at(len);
519 provider.send_completion(chunk);
520 new_text = suffix;
521 deterministic.run_until_parked();
522 }
523 provider.finish_completion();
524 deterministic.run_until_parked();
525
526 assert_eq!(
527 buffer.read_with(cx, |buffer, cx| buffer.snapshot(cx).text()),
528 indoc! {"
529 fn main() {
530 let mut x = 0;
531 while x < 10 {
532 x += 1;
533 }
534 }
535 "}
536 );
537 }
538
539 #[gpui::test]
540 async fn test_strip_markdown_codeblock() {
541 assert_eq!(
542 strip_markdown_codeblock(chunks("Lorem ipsum dolor", 2))
543 .map(|chunk| chunk.unwrap())
544 .collect::<String>()
545 .await,
546 "Lorem ipsum dolor"
547 );
548 assert_eq!(
549 strip_markdown_codeblock(chunks("```\nLorem ipsum dolor", 2))
550 .map(|chunk| chunk.unwrap())
551 .collect::<String>()
552 .await,
553 "Lorem ipsum dolor"
554 );
555 assert_eq!(
556 strip_markdown_codeblock(chunks("```\nLorem ipsum dolor\n```", 2))
557 .map(|chunk| chunk.unwrap())
558 .collect::<String>()
559 .await,
560 "Lorem ipsum dolor"
561 );
562 assert_eq!(
563 strip_markdown_codeblock(chunks("```\nLorem ipsum dolor\n```\n", 2))
564 .map(|chunk| chunk.unwrap())
565 .collect::<String>()
566 .await,
567 "Lorem ipsum dolor"
568 );
569 assert_eq!(
570 strip_markdown_codeblock(chunks("```html\n```js\nLorem ipsum dolor\n```\n```", 2))
571 .map(|chunk| chunk.unwrap())
572 .collect::<String>()
573 .await,
574 "```js\nLorem ipsum dolor\n```"
575 );
576 assert_eq!(
577 strip_markdown_codeblock(chunks("``\nLorem ipsum dolor\n```", 2))
578 .map(|chunk| chunk.unwrap())
579 .collect::<String>()
580 .await,
581 "``\nLorem ipsum dolor\n```"
582 );
583
584 fn chunks(text: &str, size: usize) -> impl Stream<Item = Result<String>> {
585 stream::iter(
586 text.chars()
587 .collect::<Vec<_>>()
588 .chunks(size)
589 .map(|chunk| Ok(chunk.iter().collect::<String>()))
590 .collect::<Vec<_>>(),
591 )
592 }
593 }
594
595 struct TestCompletionProvider {
596 last_completion_tx: Mutex<Option<mpsc::Sender<String>>>,
597 }
598
599 impl TestCompletionProvider {
600 fn new() -> Self {
601 Self {
602 last_completion_tx: Mutex::new(None),
603 }
604 }
605
606 fn send_completion(&self, completion: impl Into<String>) {
607 let mut tx = self.last_completion_tx.lock();
608 tx.as_mut().unwrap().try_send(completion.into()).unwrap();
609 }
610
611 fn finish_completion(&self) {
612 self.last_completion_tx.lock().take().unwrap();
613 }
614 }
615
616 impl CompletionProvider for TestCompletionProvider {
617 fn complete(
618 &self,
619 _prompt: OpenAIRequest,
620 ) -> BoxFuture<'static, Result<BoxStream<'static, Result<String>>>> {
621 let (tx, rx) = mpsc::channel(1);
622 *self.last_completion_tx.lock() = Some(tx);
623 async move { Ok(rx.map(|rx| Ok(rx)).boxed()) }.boxed()
624 }
625 }
626
627 fn rust_lang() -> Language {
628 Language::new(
629 LanguageConfig {
630 name: "Rust".into(),
631 path_suffixes: vec!["rs".to_string()],
632 ..Default::default()
633 },
634 Some(tree_sitter_rust::language()),
635 )
636 .with_indents_query(
637 r#"
638 (call_expression) @indent
639 (field_expression) @indent
640 (_ "(" ")" @end) @indent
641 (_ "{" "}" @end) @indent
642 "#,
643 )
644 .unwrap()
645 }
646}