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