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::zeta::max_editable_tokens(ZetaFormat::default()),
58 edit_prediction::zeta::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::zeta::max_editable_tokens(version),
79 edit_prediction::zeta::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 can_collect_data: false,
104 };
105 let prompt = format_zeta_prompt(&input, version);
106 let prefill = zeta_prompt::get_prefill(&input, version);
107 let (expected_patch, expected_cursor_offset) = example
108 .spec
109 .expected_patches_with_cursor_positions()
110 .into_iter()
111 .next()
112 .context("expected patches is empty")?;
113 let expected_output =
114 zeta2_output_for_patch(&input, &expected_patch, expected_cursor_offset, version)?;
115 let rejected_output = example
116 .spec
117 .rejected_patch
118 .as_ref()
119 .and_then(|patch| zeta2_output_for_patch(&input, patch, None, version).ok());
120
121 example.prompt = Some(ExamplePrompt {
122 input: prompt,
123 expected_output,
124 rejected_output,
125 provider: args.provider,
126 prefill: Some(prefill),
127 });
128 }
129 _ => {
130 panic!("Cannot format prompt for {:?}", args.provider);
131 }
132 };
133 Ok(())
134}
135
136pub fn zeta2_output_for_patch(
137 input: &zeta_prompt::ZetaPromptInput,
138 patch: &str,
139 cursor_offset: Option<usize>,
140 version: ZetaFormat,
141) -> Result<String> {
142 let mut old_editable_region =
143 input.cursor_excerpt[input.editable_range_in_excerpt.clone()].to_string();
144
145 if !old_editable_region.ends_with_newline() {
146 old_editable_region.push('\n');
147 }
148
149 let (mut result, first_hunk_offset) =
150 udiff::apply_diff_to_string_with_hunk_offset(patch, &old_editable_region).with_context(
151 || {
152 format!(
153 "Patch:\n```\n{}```\n\nEditable region:\n```\n{}```",
154 patch, old_editable_region
155 )
156 },
157 )?;
158
159 if let Some(cursor_offset) = cursor_offset {
160 // The cursor_offset is relative to the start of the hunk's new text (context + additions).
161 // We need to add where the hunk context matched in the editable region to compute
162 // the actual cursor position in the result.
163 let hunk_start = first_hunk_offset.unwrap_or(0);
164 let offset = result.floor_char_boundary((hunk_start + cursor_offset).min(result.len()));
165 result.insert_str(offset, zeta_prompt::CURSOR_MARKER);
166 }
167
168 match version {
169 ZetaFormat::V0120GitMergeMarkers
170 | ZetaFormat::V0131GitMergeMarkersPrefix
171 | ZetaFormat::V0211SeedCoder => {
172 if !result.ends_with('\n') {
173 result.push('\n');
174 }
175 result.push_str(zeta_prompt::v0120_git_merge_markers::END_MARKER);
176 }
177 _ => (),
178 }
179
180 Ok(result)
181}
182
183pub struct TeacherPrompt;
184
185impl TeacherPrompt {
186 pub(crate) const EDITABLE_REGION_START: &str = "<|editable_region_start|>\n";
187 pub(crate) const EDITABLE_REGION_END: &str = "\n<|editable_region_end|>";
188 pub(crate) const USER_CURSOR_MARKER: &str = "<|user_cursor|>";
189 pub(crate) const NO_EDITS: &str = "NO_EDITS";
190
191 /// Truncate edit history to this number of last lines
192 const MAX_HISTORY_LINES: usize = 128;
193
194 pub fn format_prompt(
195 example: &Example,
196 editable_range: Range<usize>,
197 context_range: Range<usize>,
198 ) -> String {
199 let edit_history = Self::format_edit_history(&example.spec.edit_history);
200 let context = Self::format_context(example);
201 let cursor_excerpt = Self::format_cursor_excerpt(example, editable_range, context_range);
202
203 let prompt_template = crate::prompt_assets::get_prompt("teacher.md");
204 let prompt = prompt_template
205 .replace("{{context}}", &context)
206 .replace("{{edit_history}}", &edit_history)
207 .replace("{{cursor_excerpt}}", &cursor_excerpt);
208
209 prompt
210 }
211
212 pub fn parse(example: &Example, response: &str) -> Result<(String, Option<ActualCursor>)> {
213 // Check if the model indicated no edits are needed
214 let no_edits = (String::new(), None);
215 if let Some(last_codeblock) = extract_last_codeblock(&response) {
216 if last_codeblock.trim() == Self::NO_EDITS {
217 return Ok(no_edits);
218 }
219 }
220
221 if response.trim().ends_with(Self::NO_EDITS) {
222 return Ok(no_edits);
223 }
224
225 // Extract updated (new) editable region from the model response.
226 let new_editable_region = Self::extract_editable_region(&response)?;
227 let cursor_offset = new_editable_region.find(Self::USER_CURSOR_MARKER);
228 let mut new_editable_region = new_editable_region.replace(Self::USER_CURSOR_MARKER, "");
229 let old_editable_region = Self::extract_editable_region(
230 &example
231 .prompt
232 .as_ref()
233 .context("example prompt missing")?
234 .input,
235 )?
236 .replace(Self::USER_CURSOR_MARKER, "");
237
238 let prompt_inputs = example
239 .prompt_inputs
240 .as_ref()
241 .context("example is missing prompt inputs")?;
242
243 // Normalize leading newlines: if old starts with newline but new doesn't,
244 // prepend newline to new to preserve whitespace structure.
245 // This handles the case where the model drops the leading blank line.
246 if old_editable_region.starts_with('\n') && !new_editable_region.starts_with('\n') {
247 new_editable_region.insert(0, '\n');
248 }
249
250 let (editable_region_offset, _) = prompt_inputs
251 .content
252 .match_indices(&old_editable_region)
253 .min_by_key(|(index, _)| index.abs_diff(prompt_inputs.cursor_offset))
254 .context("editable region not found in prompt content")?;
255 let editable_region_start_line = prompt_inputs.content[..editable_region_offset]
256 .matches('\n')
257 .count();
258
259 // Use full context so cursor offset (relative to editable region start) aligns with diff content
260 let editable_region_lines = old_editable_region.lines().count() as u32;
261 let diff = language::unified_diff_with_context(
262 &old_editable_region,
263 &new_editable_region,
264 editable_region_start_line as u32,
265 editable_region_start_line as u32,
266 editable_region_lines,
267 );
268
269 let diff = indoc::formatdoc! {"
270 --- a/{path}
271 +++ b/{path}
272 {diff}",
273 path = example.spec.cursor_path.to_string_lossy(),
274 diff = diff,
275 };
276
277 let actual_cursor = cursor_offset.map(|editable_region_cursor_offset| {
278 ActualCursor::from_editable_region(
279 &example.spec.cursor_path,
280 editable_region_cursor_offset,
281 &new_editable_region,
282 &prompt_inputs.content,
283 editable_region_offset,
284 editable_region_start_line,
285 )
286 });
287
288 Ok((diff, actual_cursor))
289 }
290
291 fn format_edit_history(edit_history: &str) -> String {
292 let lines: Vec<&str> = edit_history.lines().collect();
293
294 if lines.is_empty() {
295 return "(No edit history)".to_string();
296 }
297
298 if lines.len() > Self::MAX_HISTORY_LINES {
299 let truncated = lines[lines.len() - Self::MAX_HISTORY_LINES..].join("\n");
300 format!("{truncated}\n[...truncated...]")
301 } else {
302 lines.join("\n")
303 }
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
381/// Extract the cursor excerpt from an example.
382/// First tries to extract from an existing prompt, then falls back to constructing from prompt_inputs.
383pub fn extract_cursor_excerpt_from_example(example: &Example) -> Option<String> {
384 // If we have the original prompt, extract the cursor excerpt from it
385 if let Some(prompt) = &example.prompt {
386 // Find "# 3. Current File" section and extract the content
387 if let Some(start) = prompt.input.find("# 3. Current File") {
388 let content_start = prompt.input[start..].find('`').map(|i| start + i)?;
389 let backtick_count = prompt.input[content_start..]
390 .chars()
391 .take_while(|&c| c == '`')
392 .count();
393 let content_start = content_start + backtick_count;
394
395 // Find the path line and skip it
396 let newline_pos = prompt.input[content_start..].find('\n')?;
397 let text_start = content_start + newline_pos + 1;
398
399 // Find the closing backticks
400 let closing_pattern = "`".repeat(backtick_count);
401 let text_end = prompt.input[text_start..].find(&closing_pattern)?;
402 let cursor_excerpt = &prompt.input[text_start..text_start + text_end];
403
404 let path_str = example.spec.cursor_path.to_string_lossy();
405 return Some(format!("`````{path_str}\n{cursor_excerpt}`````"));
406 }
407 }
408
409 // Fallback: construct from prompt_inputs if available
410 let prompt_inputs = example.prompt_inputs.as_ref()?;
411 let content = &prompt_inputs.content;
412 let cursor_offset = prompt_inputs.cursor_offset;
413
414 // Simple fallback: just show content around cursor with markers
415 let path_str = example.spec.cursor_path.to_string_lossy();
416 let mut result = format!("`````{path_str}\n");
417 result.push_str(TeacherPrompt::EDITABLE_REGION_START);
418 result.push_str(&content[..cursor_offset]);
419 result.push_str(TeacherPrompt::USER_CURSOR_MARKER);
420 result.push_str(&content[cursor_offset..]);
421 result.push_str(TeacherPrompt::EDITABLE_REGION_END);
422 result.push_str("\n`````");
423
424 Some(result)
425}
426
427pub(crate) fn extract_last_codeblock(text: &str) -> Option<String> {
428 let lines: Vec<&str> = text.lines().collect();
429
430 // Search from the end for a closing fence (line containing only backticks, 3+)
431 let mut closing_line_idx = None;
432 let mut backtick_count = 0;
433
434 for i in (0..lines.len()).rev() {
435 let line = lines[i].trim();
436 if line.len() >= 3 && line.chars().all(|c| c == '`') {
437 closing_line_idx = Some(i);
438 backtick_count = line.len();
439 break;
440 }
441 }
442
443 let closing_idx = closing_line_idx?;
444
445 // Search backwards for matching opening fence
446 // Opening fence starts with same backtick count, possibly followed by language/metadata
447 let opening_pattern = "`".repeat(backtick_count);
448
449 for i in (0..closing_idx).rev() {
450 let line = lines[i];
451 if line.starts_with(&opening_pattern) {
452 // Ensure it's exactly the right number of backticks (not more)
453 let rest = &line[backtick_count..];
454 if rest.is_empty() || !rest.starts_with('`') {
455 // Found matching opening fence
456 // Extract content between opening and closing (exclusive)
457 if closing_idx > i + 1 {
458 let content = lines[i + 1..closing_idx].join("\n");
459 // Preserve trailing newline to match previous behavior
460 return Some(format!("{}\n", content));
461 } else {
462 // Empty block
463 return Some(String::new());
464 }
465 }
466 }
467 }
468
469 None
470}
471
472#[cfg(test)]
473mod tests {
474 use super::*;
475
476 #[test]
477 fn test_extract_last_code_block() {
478 let text = indoc::indoc! {"
479 Some thinking
480
481 ```
482 first block
483 ```
484
485 `````path='something' lines=1:2
486 last block
487 `````
488 "};
489 let last_block = extract_last_codeblock(text).unwrap();
490 assert_eq!(last_block, "last block\n");
491 }
492
493 #[test]
494 fn test_extract_codeblock_with_nested_fences() {
495 let text = indoc::indoc! {"
496 `````
497 content with ``` inline
498 and ```python nested
499 more content
500 `````
501 "};
502 let last_block = extract_last_codeblock(text).unwrap();
503 assert_eq!(
504 last_block,
505 "content with ``` inline\nand ```python nested\nmore content\n"
506 );
507 }
508
509 #[test]
510 fn test_extract_codeblock_ignores_inline_backticks() {
511 let text = indoc::indoc! {"
512 `````
513 here is some `code` with inline backticks
514 and here```more```stuff
515 `````
516 "};
517 let last_block = extract_last_codeblock(text).unwrap();
518 assert_eq!(
519 last_block,
520 "here is some `code` with inline backticks\nand here```more```stuff\n"
521 );
522 }
523
524 #[test]
525 fn test_extract_editable_region() {
526 let text = indoc::indoc! {"
527 some lines
528 are
529 here
530 <|editable_region_start|>
531 one
532 two three
533
534 <|editable_region_end|>
535 more
536 lines here
537 "};
538 let parsed = TeacherPrompt::extract_editable_region(text).unwrap();
539 assert_eq!(
540 parsed,
541 indoc::indoc! {"
542 one
543 two three"}
544 );
545 }
546
547 #[test]
548 fn test_extract_last_codeblock_nested_bibtex() {
549 let text = indoc::indoc! {r#"
550 Looking at the edit history, I can see that a Citation section was just added.
551
552 `````
553 ## Collaborations
554 Our mission is to create a 4D generative model.
555
556 ## Citation
557
558 If you found Unique3D helpful, please cite our report:
559 ```bibtex
560 @misc{wu2024unique3d,
561 title={Unique3D},
562 }
563 ```
564 `````
565 "#};
566 let last_block = extract_last_codeblock(text).unwrap();
567 assert_eq!(
568 last_block,
569 indoc::indoc! {r#"
570 ## Collaborations
571 Our mission is to create a 4D generative model.
572
573 ## Citation
574
575 If you found Unique3D helpful, please cite our report:
576 ```bibtex
577 @misc{wu2024unique3d,
578 title={Unique3D},
579 }
580 ```
581 "#}
582 );
583 }
584
585 #[test]
586 fn test_extract_editable_region_no_markers() {
587 let text = indoc::indoc! {"
588 one
589 two three"};
590 let parsed = TeacherPrompt::extract_editable_region(text).unwrap();
591 assert_eq!(
592 parsed,
593 indoc::indoc! {"
594 one
595 two three"}
596 );
597 }
598
599 #[test]
600 fn test_parse_no_edits_response() {
601 let response = indoc::indoc! {"
602 The code is already complete. There is no clear next edit to make.
603
604 `````
605 NO_EDITS
606 `````
607 "};
608 let codeblock = extract_last_codeblock(response).unwrap();
609 assert_eq!(codeblock.trim(), TeacherPrompt::NO_EDITS);
610 }
611
612 #[test]
613 fn test_extract_codeblock_no_valid_block() {
614 // Text with no code blocks should return None
615 let text = "Just some plain text without any code blocks";
616 assert!(extract_last_codeblock(text).is_none());
617
618 // Unclosed code block should return None
619 let text = indoc::indoc! {"
620 ```
621 unclosed block
622 "};
623 assert!(extract_last_codeblock(text).is_none());
624
625 // Analysis text with nested markdown but no proper outer block
626 let text = indoc::indoc! {"
627 # Analysis
628 Looking at this:
629 ```
630 some code
631 ```
632 But then more analysis without wrapping block
633 "};
634 // This should find the inner block
635 let result = extract_last_codeblock(text).unwrap();
636 assert_eq!(result, "some code\n");
637 }
638
639 #[test]
640 fn test_extract_codeblock_no_trailing_newline() {
641 // Text ending without trailing newline after closing fence
642 let text = "`````\ncontent here\n`````";
643 let result = extract_last_codeblock(text).unwrap();
644 assert_eq!(result, "content here\n");
645 }
646}