1use crate::{
2 FormatPromptArgs, PredictionProvider,
3 example::{ActualCursor, Example, ExamplePrompt},
4 headless::EpAppState,
5 progress::{ExampleProgress, Step},
6 retrieve_context::run_context_retrieval,
7};
8use anyhow::{Context as _, Result, anyhow};
9use edit_prediction::{cursor_excerpt::editable_and_context_ranges_for_cursor_position, udiff};
10use gpui::{AppContext, AsyncApp};
11use language::{Buffer, OffsetRangeExt, Point};
12use similar::DiffableStr;
13use std::sync::Arc;
14use std::{fmt::Write as _, ops::Range};
15use zeta_prompt::ZetaFormat;
16use zeta_prompt::format_zeta_prompt;
17
18pub async fn run_format_prompt(
19 example: &mut Example,
20 args: &FormatPromptArgs,
21 app_state: Arc<EpAppState>,
22 example_progress: &ExampleProgress,
23 cx: AsyncApp,
24) -> Result<()> {
25 run_context_retrieval(example, app_state.clone(), example_progress, cx.clone()).await?;
26
27 let step_progress = example_progress.start(Step::FormatPrompt);
28
29 let prompt_inputs = example
30 .prompt_inputs
31 .as_ref()
32 .context("prompt_inputs must be set after context retrieval")?;
33
34 let language = app_state
35 .languages
36 .load_language_for_file_path(&example.spec.cursor_path)
37 .await
38 .ok();
39 let snapshot_fut = cx.update(|cx| {
40 Buffer::build_snapshot(
41 prompt_inputs.content.as_str().into(),
42 language,
43 Some(app_state.languages.clone()),
44 cx,
45 )
46 });
47 let cursor_point = Point::new(prompt_inputs.cursor_row, prompt_inputs.cursor_column);
48 let snapshot = cx.background_spawn(snapshot_fut).await;
49
50 match args.provider {
51 PredictionProvider::Teacher(_) | PredictionProvider::TeacherNonBatching(_) => {
52 step_progress.set_substatus("formatting teacher prompt");
53
54 let (editable_range, context_range) = editable_and_context_ranges_for_cursor_position(
55 cursor_point,
56 &snapshot,
57 edit_prediction::zeta2::max_editable_tokens(ZetaFormat::default()),
58 edit_prediction::zeta2::MAX_CONTEXT_TOKENS,
59 );
60 let editable_range = editable_range.to_offset(&snapshot);
61 let context_range = context_range.to_offset(&snapshot);
62
63 let prompt = TeacherPrompt::format_prompt(example, editable_range, context_range);
64 example.prompt = Some(ExamplePrompt {
65 input: prompt,
66 expected_output: String::new(),
67 rejected_output: None,
68 prefill: None,
69 provider: args.provider,
70 });
71 }
72 PredictionProvider::Zeta2(version) => {
73 step_progress.set_substatus("formatting zeta2 prompt");
74
75 let (editable_range, context_range) = editable_and_context_ranges_for_cursor_position(
76 cursor_point,
77 &snapshot,
78 edit_prediction::zeta2::max_editable_tokens(version),
79 edit_prediction::zeta2::MAX_CONTEXT_TOKENS,
80 );
81 let editable_range = editable_range.to_offset(&snapshot);
82 let context_range = context_range.to_offset(&snapshot);
83
84 let context_start = context_range.start;
85 let cursor_offset_in_excerpt = prompt_inputs.cursor_offset - context_start;
86 let editable_range_in_excerpt =
87 (editable_range.start - context_start)..(editable_range.end - context_start);
88 let input = zeta_prompt::ZetaPromptInput {
89 cursor_path: example.spec.cursor_path.clone(),
90 cursor_excerpt: prompt_inputs.content[context_range].to_string().into(),
91 editable_range_in_excerpt,
92 cursor_offset_in_excerpt,
93 excerpt_start_row: prompt_inputs.excerpt_start_row,
94 events: prompt_inputs.edit_history.clone(),
95 related_files: prompt_inputs.related_files.clone().unwrap_or_default(),
96 excerpt_ranges: None,
97 preferred_model: None,
98 in_open_source_repo: example
99 .spec
100 .captured_prompt_input
101 .as_ref()
102 .map_or(false, |input| input.in_open_source_repo),
103 };
104 let prompt = format_zeta_prompt(&input, version);
105 let prefill = zeta_prompt::get_prefill(&input, version);
106 let (expected_patch, expected_cursor_offset) = example
107 .spec
108 .expected_patches_with_cursor_positions()
109 .into_iter()
110 .next()
111 .context("expected patches is empty")?;
112 let expected_output =
113 zeta2_output_for_patch(&input, &expected_patch, expected_cursor_offset, version)?;
114 let rejected_output = example
115 .spec
116 .rejected_patch
117 .as_ref()
118 .and_then(|patch| zeta2_output_for_patch(&input, patch, None, version).ok());
119
120 example.prompt = Some(ExamplePrompt {
121 input: prompt,
122 expected_output,
123 rejected_output,
124 provider: args.provider,
125 prefill: Some(prefill),
126 });
127 }
128 _ => {
129 panic!("Cannot format prompt for {:?}", args.provider);
130 }
131 };
132 Ok(())
133}
134
135pub fn zeta2_output_for_patch(
136 input: &zeta_prompt::ZetaPromptInput,
137 patch: &str,
138 cursor_offset: Option<usize>,
139 version: ZetaFormat,
140) -> Result<String> {
141 let mut old_editable_region =
142 input.cursor_excerpt[input.editable_range_in_excerpt.clone()].to_string();
143
144 if !old_editable_region.ends_with_newline() {
145 old_editable_region.push('\n');
146 }
147
148 let (mut result, first_hunk_offset) =
149 udiff::apply_diff_to_string_with_hunk_offset(patch, &old_editable_region).with_context(
150 || {
151 format!(
152 "Patch:\n```\n{}```\n\nEditable region:\n```\n{}```",
153 patch, old_editable_region
154 )
155 },
156 )?;
157
158 if let Some(cursor_offset) = cursor_offset {
159 // The cursor_offset is relative to the start of the hunk's new text (context + additions).
160 // We need to add where the hunk context matched in the editable region to compute
161 // the actual cursor position in the result.
162 let hunk_start = first_hunk_offset.unwrap_or(0);
163 let offset = (hunk_start + cursor_offset).min(result.len());
164 result.insert_str(offset, zeta_prompt::CURSOR_MARKER);
165 }
166
167 match version {
168 ZetaFormat::V0120GitMergeMarkers
169 | ZetaFormat::V0131GitMergeMarkersPrefix
170 | ZetaFormat::V0211SeedCoder => {
171 if !result.ends_with('\n') {
172 result.push('\n');
173 }
174 result.push_str(zeta_prompt::v0120_git_merge_markers::END_MARKER);
175 }
176 _ => (),
177 }
178
179 Ok(result)
180}
181
182pub struct TeacherPrompt;
183
184impl TeacherPrompt {
185 pub(crate) const EDITABLE_REGION_START: &str = "<|editable_region_start|>\n";
186 pub(crate) const EDITABLE_REGION_END: &str = "\n<|editable_region_end|>";
187 pub(crate) const USER_CURSOR_MARKER: &str = "<|user_cursor|>";
188 pub(crate) const NO_EDITS: &str = "NO_EDITS";
189
190 /// Truncate edit history to this number of last lines
191 const MAX_HISTORY_LINES: usize = 128;
192
193 pub fn format_prompt(
194 example: &Example,
195 editable_range: Range<usize>,
196 context_range: Range<usize>,
197 ) -> String {
198 let edit_history = Self::format_edit_history(&example.spec.edit_history);
199 let context = Self::format_context(example);
200 let cursor_excerpt = Self::format_cursor_excerpt(example, editable_range, context_range);
201
202 let prompt_template = crate::prompt_assets::get_prompt("teacher.md");
203 let prompt = prompt_template
204 .replace("{{context}}", &context)
205 .replace("{{edit_history}}", &edit_history)
206 .replace("{{cursor_excerpt}}", &cursor_excerpt);
207
208 prompt
209 }
210
211 pub fn parse(example: &Example, response: &str) -> Result<(String, Option<ActualCursor>)> {
212 // Extract updated (new) editable region from the model response.
213 // The model may include editable region markers in its output, so we need to strip them.
214 let new_editable_region = extract_last_codeblock(response);
215
216 // Check if the model indicated no edits are needed
217 if new_editable_region.trim() == Self::NO_EDITS {
218 return Ok((String::new(), None));
219 }
220
221 let new_editable_region = Self::extract_editable_region(&new_editable_region)?;
222 let cursor_offset = new_editable_region.find(Self::USER_CURSOR_MARKER);
223 let mut new_editable_region = new_editable_region.replace(Self::USER_CURSOR_MARKER, "");
224 let old_editable_region = Self::extract_editable_region(
225 &example
226 .prompt
227 .as_ref()
228 .context("example prompt missing")?
229 .input,
230 )?
231 .replace(Self::USER_CURSOR_MARKER, "");
232
233 let prompt_inputs = example
234 .prompt_inputs
235 .as_ref()
236 .context("example is missing prompt inputs")?;
237
238 // Normalize leading newlines: if old starts with newline but new doesn't,
239 // prepend newline to new to preserve whitespace structure.
240 // This handles the case where the model drops the leading blank line.
241 if old_editable_region.starts_with('\n') && !new_editable_region.starts_with('\n') {
242 new_editable_region.insert(0, '\n');
243 }
244
245 let (editable_region_offset, _) = prompt_inputs
246 .content
247 .match_indices(&old_editable_region)
248 .min_by_key(|(index, _)| index.abs_diff(prompt_inputs.cursor_offset))
249 .context("editable region not found in prompt content")?;
250 let editable_region_start_line = prompt_inputs.content[..editable_region_offset]
251 .matches('\n')
252 .count();
253
254 // Use full context so cursor offset (relative to editable region start) aligns with diff content
255 let editable_region_lines = old_editable_region.lines().count() as u32;
256 let diff = language::unified_diff_with_context(
257 &old_editable_region,
258 &new_editable_region,
259 editable_region_start_line as u32,
260 editable_region_start_line as u32,
261 editable_region_lines,
262 );
263
264 let diff = indoc::formatdoc! {"
265 --- a/{path}
266 +++ b/{path}
267 {diff}",
268 path = example.spec.cursor_path.to_string_lossy(),
269 diff = diff,
270 };
271
272 let actual_cursor = cursor_offset.map(|editable_region_cursor_offset| {
273 ActualCursor::from_editable_region(
274 &example.spec.cursor_path,
275 editable_region_cursor_offset,
276 &new_editable_region,
277 &prompt_inputs.content,
278 editable_region_offset,
279 editable_region_start_line,
280 )
281 });
282
283 Ok((diff, actual_cursor))
284 }
285
286 fn format_edit_history(edit_history: &str) -> String {
287 // Strip comments ("garbage lines") from edit history
288 let lines = edit_history
289 .lines()
290 .filter(|&s| Self::is_udiff_content_line(s))
291 .collect::<Vec<_>>();
292
293 let history_lines = if lines.len() > Self::MAX_HISTORY_LINES {
294 &lines[lines.len() - Self::MAX_HISTORY_LINES..]
295 } else {
296 &lines
297 };
298
299 if history_lines.is_empty() {
300 return "(No edit history)".to_string();
301 }
302
303 history_lines.join("\n")
304 }
305
306 pub fn format_context(example: &Example) -> String {
307 let related_files = example
308 .prompt_inputs
309 .as_ref()
310 .and_then(|pi| pi.related_files.as_ref());
311
312 let Some(related_files) = related_files else {
313 return "(No context)".to_string();
314 };
315
316 if related_files.is_empty() {
317 return "(No context)".to_string();
318 }
319
320 let mut prompt = String::new();
321 for file in related_files {
322 let path_str = file.path.to_string_lossy();
323 writeln!(&mut prompt, "`````{path_str}").ok();
324
325 let mut prev_row = 0;
326 for excerpt in &file.excerpts {
327 if excerpt.row_range.start > prev_row {
328 prompt.push_str("…\n");
329 }
330 prompt.push_str(&excerpt.text);
331 prompt.push('\n');
332 prev_row = excerpt.row_range.end;
333 }
334 if prev_row < file.max_row {
335 prompt.push_str("…\n");
336 }
337 prompt.push_str("\n`````\n");
338 }
339
340 prompt
341 }
342
343 fn format_cursor_excerpt(
344 example: &Example,
345 editable_range: Range<usize>,
346 context_range: Range<usize>,
347 ) -> String {
348 let mut result = String::new();
349
350 let prompt_inputs = example.prompt_inputs.as_ref().unwrap();
351
352 let path_str = example.spec.cursor_path.to_string_lossy();
353 result.push_str(&format!("`````{path_str}\n"));
354 result.push_str(&prompt_inputs.content[context_range.start..editable_range.start]);
355 result.push_str(Self::EDITABLE_REGION_START);
356 result.push_str(&prompt_inputs.content[editable_range.start..prompt_inputs.cursor_offset]);
357 result.push_str(Self::USER_CURSOR_MARKER);
358 result.push_str(&prompt_inputs.content[prompt_inputs.cursor_offset..editable_range.end]);
359 result.push_str(Self::EDITABLE_REGION_END);
360 result.push_str(&prompt_inputs.content[editable_range.end..context_range.end]);
361 result.push_str("\n`````");
362
363 result
364 }
365
366 pub fn extract_editable_region(text: &str) -> Result<String> {
367 let start = text
368 .rfind(Self::EDITABLE_REGION_START)
369 .map_or(0, |pos| pos + Self::EDITABLE_REGION_START.len());
370 let end = text.rfind(Self::EDITABLE_REGION_END).unwrap_or(text.len());
371
372 if start >= end {
373 return Err(anyhow!("Invalid editable region markers"));
374 }
375
376 let region = &text[start..end];
377 Ok(region.strip_suffix('\n').unwrap_or(region).to_string())
378 }
379
380 fn is_udiff_content_line(s: &str) -> bool {
381 s.starts_with("-")
382 || s.starts_with("+")
383 || s.starts_with(" ")
384 || s.starts_with("---")
385 || s.starts_with("+++")
386 || s.starts_with("@@")
387 }
388}
389
390/// Extract the cursor excerpt from an example.
391/// First tries to extract from an existing prompt, then falls back to constructing from prompt_inputs.
392pub fn extract_cursor_excerpt_from_example(example: &Example) -> Option<String> {
393 // If we have the original prompt, extract the cursor excerpt from it
394 if let Some(prompt) = &example.prompt {
395 // Find "# 3. Current File" section and extract the content
396 if let Some(start) = prompt.input.find("# 3. Current File") {
397 let content_start = prompt.input[start..].find('`').map(|i| start + i)?;
398 let backtick_count = prompt.input[content_start..]
399 .chars()
400 .take_while(|&c| c == '`')
401 .count();
402 let content_start = content_start + backtick_count;
403
404 // Find the path line and skip it
405 let newline_pos = prompt.input[content_start..].find('\n')?;
406 let text_start = content_start + newline_pos + 1;
407
408 // Find the closing backticks
409 let closing_pattern = "`".repeat(backtick_count);
410 let text_end = prompt.input[text_start..].find(&closing_pattern)?;
411 let cursor_excerpt = &prompt.input[text_start..text_start + text_end];
412
413 let path_str = example.spec.cursor_path.to_string_lossy();
414 return Some(format!("`````{path_str}\n{cursor_excerpt}`````"));
415 }
416 }
417
418 // Fallback: construct from prompt_inputs if available
419 let prompt_inputs = example.prompt_inputs.as_ref()?;
420 let content = &prompt_inputs.content;
421 let cursor_offset = prompt_inputs.cursor_offset;
422
423 // Simple fallback: just show content around cursor with markers
424 let path_str = example.spec.cursor_path.to_string_lossy();
425 let mut result = format!("`````{path_str}\n");
426 result.push_str(TeacherPrompt::EDITABLE_REGION_START);
427 result.push_str(&content[..cursor_offset]);
428 result.push_str(TeacherPrompt::USER_CURSOR_MARKER);
429 result.push_str(&content[cursor_offset..]);
430 result.push_str(TeacherPrompt::EDITABLE_REGION_END);
431 result.push_str("\n`````");
432
433 Some(result)
434}
435
436pub(crate) fn extract_last_codeblock(text: &str) -> String {
437 let mut last_block = None;
438 let mut search_start = 0;
439
440 while let Some(start) = text[search_start..].find("```") {
441 let start = start + search_start;
442 let bytes = text.as_bytes();
443 let mut backtick_end = start;
444
445 while backtick_end < bytes.len() && bytes[backtick_end] == b'`' {
446 backtick_end += 1;
447 }
448
449 let backtick_count = backtick_end - start;
450 let closing_pattern = format!("\n{}", "`".repeat(backtick_count));
451
452 while backtick_end < bytes.len() && bytes[backtick_end] != b'\n' {
453 backtick_end += 1;
454 }
455
456 if let Some(end_pos) = text[backtick_end..].find(&closing_pattern) {
457 let code_block = &text[backtick_end + 1..backtick_end + end_pos + 1];
458 last_block = Some(code_block.to_string());
459 search_start = backtick_end + end_pos + closing_pattern.len();
460 } else {
461 break;
462 }
463 }
464
465 last_block.unwrap_or_else(|| text.to_string())
466}
467
468#[cfg(test)]
469mod tests {
470 use super::*;
471
472 #[test]
473 fn test_extract_last_code_block() {
474 let text = indoc::indoc! {"
475 Some thinking
476
477 ```
478 first block
479 ```
480
481 `````path='something' lines=1:2
482 last block
483 `````
484 "};
485 let last_block = extract_last_codeblock(text);
486 assert_eq!(last_block, "last block\n");
487 }
488
489 #[test]
490 fn test_extract_codeblock_with_nested_fences() {
491 let text = indoc::indoc! {"
492 `````
493 content with ``` inline
494 and ```python nested
495 more content
496 `````
497 "};
498 let last_block = extract_last_codeblock(text);
499 assert_eq!(
500 last_block,
501 "content with ``` inline\nand ```python nested\nmore content\n"
502 );
503 }
504
505 #[test]
506 fn test_extract_codeblock_ignores_inline_backticks() {
507 let text = indoc::indoc! {"
508 `````
509 here is some `code` with inline backticks
510 and here```more```stuff
511 `````
512 "};
513 let last_block = extract_last_codeblock(text);
514 assert_eq!(
515 last_block,
516 "here is some `code` with inline backticks\nand here```more```stuff\n"
517 );
518 }
519
520 #[test]
521 fn test_extract_editable_region() {
522 let text = indoc::indoc! {"
523 some lines
524 are
525 here
526 <|editable_region_start|>
527 one
528 two three
529
530 <|editable_region_end|>
531 more
532 lines here
533 "};
534 let parsed = TeacherPrompt::extract_editable_region(text).unwrap();
535 assert_eq!(
536 parsed,
537 indoc::indoc! {"
538 one
539 two three"}
540 );
541 }
542
543 #[test]
544 fn test_extract_last_codeblock_nested_bibtex() {
545 let text = indoc::indoc! {r#"
546 Looking at the edit history, I can see that a Citation section was just added.
547
548 `````
549 ## Collaborations
550 Our mission is to create a 4D generative model.
551
552 ## Citation
553
554 If you found Unique3D helpful, please cite our report:
555 ```bibtex
556 @misc{wu2024unique3d,
557 title={Unique3D},
558 }
559 ```
560 `````
561 "#};
562 let last_block = extract_last_codeblock(text);
563 assert_eq!(
564 last_block,
565 indoc::indoc! {r#"
566 ## Collaborations
567 Our mission is to create a 4D generative model.
568
569 ## Citation
570
571 If you found Unique3D helpful, please cite our report:
572 ```bibtex
573 @misc{wu2024unique3d,
574 title={Unique3D},
575 }
576 ```
577 "#}
578 );
579 }
580
581 #[test]
582 fn test_extract_editable_region_no_markers() {
583 let text = indoc::indoc! {"
584 one
585 two three"};
586 let parsed = TeacherPrompt::extract_editable_region(text).unwrap();
587 assert_eq!(
588 parsed,
589 indoc::indoc! {"
590 one
591 two three"}
592 );
593 }
594
595 #[test]
596 fn test_parse_no_edits_response() {
597 let response = indoc::indoc! {"
598 The code is already complete. There is no clear next edit to make.
599
600 `````
601 NO_EDITS
602 `````
603 "};
604 let codeblock = extract_last_codeblock(response);
605 assert_eq!(codeblock.trim(), TeacherPrompt::NO_EDITS);
606 }
607}