sweep_ai.rs

  1use crate::{
  2    CurrentEditPrediction, DebugEvent, EditPrediction, EditPredictionFinishedDebugEvent,
  3    EditPredictionId, EditPredictionModelInput, EditPredictionStartedDebugEvent,
  4    EditPredictionStore, UserActionRecord, UserActionType, prediction::EditPredictionResult,
  5};
  6use anyhow::{Result, bail};
  7use client::Client;
  8use edit_prediction_types::SuggestionDisplayType;
  9use futures::{AsyncReadExt as _, channel::mpsc};
 10use gpui::{
 11    App, AppContext as _, Entity, Global, SharedString, Task,
 12    http_client::{self, AsyncBody, Method},
 13};
 14use language::language_settings::all_language_settings;
 15use language::{Anchor, Buffer, BufferSnapshot, Point, ToOffset as _};
 16use language_model::{ApiKeyState, EnvVar, env_var};
 17use lsp::DiagnosticSeverity;
 18use serde::{Deserialize, Serialize};
 19use std::{
 20    fmt::{self, Write as _},
 21    ops::Range,
 22    path::Path,
 23    sync::Arc,
 24    time::Instant,
 25};
 26
 27const SWEEP_API_URL: &str = "https://autocomplete.sweep.dev/backend/next_edit_autocomplete";
 28const SWEEP_METRICS_URL: &str = "https://backend.app.sweep.dev/backend/track_autocomplete_metrics";
 29
 30pub struct SweepAi {
 31    pub api_token: Entity<ApiKeyState>,
 32    pub debug_info: Arc<str>,
 33}
 34
 35impl SweepAi {
 36    pub fn new(cx: &mut App) -> Self {
 37        SweepAi {
 38            api_token: sweep_api_token(cx),
 39            debug_info: debug_info(cx),
 40        }
 41    }
 42
 43    pub fn request_prediction_with_sweep(
 44        &self,
 45        inputs: EditPredictionModelInput,
 46        cx: &mut App,
 47    ) -> Task<Result<Option<EditPredictionResult>>> {
 48        let privacy_mode_enabled = all_language_settings(None, cx)
 49            .edit_predictions
 50            .sweep
 51            .privacy_mode;
 52        let debug_info = self.debug_info.clone();
 53        self.api_token.update(cx, |key_state, cx| {
 54            _ = key_state.load_if_needed(SWEEP_CREDENTIALS_URL, |s| s, cx);
 55        });
 56
 57        let buffer = inputs.buffer.clone();
 58        let debug_tx = inputs.debug_tx.clone();
 59
 60        let Some(api_token) = self.api_token.read(cx).key(&SWEEP_CREDENTIALS_URL) else {
 61            return Task::ready(Ok(None));
 62        };
 63        let full_path: Arc<Path> = inputs
 64            .snapshot
 65            .file()
 66            .map(|file| file.full_path(cx))
 67            .unwrap_or_else(|| "untitled".into())
 68            .into();
 69
 70        let project_file = project::File::from_dyn(inputs.snapshot.file());
 71        let repo_name = project_file
 72            .map(|file| file.worktree.read(cx).root_name_str())
 73            .unwrap_or("untitled")
 74            .into();
 75        let offset = inputs.position.to_offset(&inputs.snapshot);
 76        let buffer_entity_id = inputs.buffer.entity_id();
 77
 78        let recent_buffers = inputs.recent_paths.iter().cloned();
 79        let http_client = cx.http_client();
 80
 81        let recent_buffer_snapshots = recent_buffers
 82            .filter_map(|project_path| {
 83                let buffer = inputs.project.read(cx).get_open_buffer(&project_path, cx)?;
 84                if inputs.buffer == buffer {
 85                    None
 86                } else {
 87                    Some(buffer.read(cx).snapshot())
 88                }
 89            })
 90            .take(3)
 91            .collect::<Vec<_>>();
 92
 93        let buffer_snapshotted_at = Instant::now();
 94
 95        let result = cx.background_spawn(async move {
 96            let text = inputs.snapshot.text();
 97
 98            let mut recent_changes = String::new();
 99            for event in &inputs.events {
100                write_event(event.as_ref(), &mut recent_changes).unwrap();
101            }
102
103            let file_chunks = recent_buffer_snapshots
104                .into_iter()
105                .map(|snapshot| {
106                    let end_point = Point::new(30, 0).min(snapshot.max_point());
107                    FileChunk {
108                        content: snapshot.text_for_range(Point::zero()..end_point).collect(),
109                        file_path: snapshot
110                            .file()
111                            .map(|f| f.path().as_unix_str())
112                            .unwrap_or("untitled")
113                            .to_string(),
114                        start_line: 0,
115                        end_line: end_point.row as usize,
116                        timestamp: snapshot.file().and_then(|file| {
117                            Some(
118                                file.disk_state()
119                                    .mtime()?
120                                    .to_seconds_and_nanos_for_persistence()?
121                                    .0,
122                            )
123                        }),
124                    }
125                })
126                .collect::<Vec<_>>();
127
128            let mut retrieval_chunks: Vec<FileChunk> = inputs
129                .related_files
130                .iter()
131                .flat_map(|related_file| {
132                    related_file.excerpts.iter().map(|excerpt| FileChunk {
133                        file_path: related_file.path.to_string_lossy().to_string(),
134                        start_line: excerpt.row_range.start as usize,
135                        end_line: excerpt.row_range.end as usize,
136                        content: excerpt.text.to_string(),
137                        timestamp: None,
138                    })
139                })
140                .collect();
141
142            let diagnostic_entries = inputs
143                .snapshot
144                .diagnostics_in_range(inputs.diagnostic_search_range, false);
145            let mut diagnostic_content = String::new();
146            let mut diagnostic_count = 0;
147
148            for entry in diagnostic_entries {
149                let start_point: Point = entry.range.start;
150
151                let severity = match entry.diagnostic.severity {
152                    DiagnosticSeverity::ERROR => "error",
153                    DiagnosticSeverity::WARNING => "warning",
154                    DiagnosticSeverity::INFORMATION => "info",
155                    DiagnosticSeverity::HINT => "hint",
156                    _ => continue,
157                };
158
159                diagnostic_count += 1;
160
161                writeln!(
162                    &mut diagnostic_content,
163                    "{}:{}:{}: {}: {}",
164                    full_path.display(),
165                    start_point.row + 1,
166                    start_point.column + 1,
167                    severity,
168                    entry.diagnostic.message
169                )?;
170            }
171
172            if !diagnostic_content.is_empty() {
173                retrieval_chunks.push(FileChunk {
174                    file_path: "diagnostics".to_string(),
175                    start_line: 1,
176                    end_line: diagnostic_count,
177                    content: diagnostic_content,
178                    timestamp: None,
179                });
180            }
181
182            let file_path_str = full_path.display().to_string();
183            let recent_user_actions = inputs
184                .user_actions
185                .iter()
186                .filter(|r| r.buffer_id == buffer_entity_id)
187                .map(|r| to_sweep_user_action(r, &file_path_str))
188                .collect();
189
190            let request_body = AutocompleteRequest {
191                debug_info,
192                repo_name,
193                file_path: full_path.clone(),
194                file_contents: text.clone(),
195                original_file_contents: text,
196                cursor_position: offset,
197                recent_changes: recent_changes.clone(),
198                changes_above_cursor: true,
199                multiple_suggestions: false,
200                branch: None,
201                file_chunks,
202                retrieval_chunks,
203                recent_user_actions,
204                use_bytes: true,
205                privacy_mode_enabled,
206            };
207
208            let mut buf: Vec<u8> = Vec::new();
209            let writer = brotli::CompressorWriter::new(&mut buf, 4096, 1, 22);
210            serde_json::to_writer(writer, &request_body)?;
211            let body: AsyncBody = buf.into();
212
213            let ep_inputs = zeta_prompt::ZetaPromptInput {
214                events: inputs.events,
215                related_files: inputs.related_files.clone(),
216                cursor_path: full_path.clone(),
217                cursor_excerpt: request_body.file_contents.clone().into(),
218                // we actually don't know
219                editable_range_in_excerpt: 0..inputs.snapshot.len(),
220                cursor_offset_in_excerpt: request_body.cursor_position,
221                excerpt_start_row: Some(0),
222                excerpt_ranges: None,
223                preferred_model: None,
224                in_open_source_repo: false,
225            };
226
227            send_started_event(
228                &debug_tx,
229                &buffer,
230                inputs.position,
231                serde_json::to_string(&request_body).unwrap_or_default(),
232            );
233
234            let request = http_client::Request::builder()
235                .uri(SWEEP_API_URL)
236                .header("Content-Type", "application/json")
237                .header("Authorization", format!("Bearer {}", api_token))
238                .header("Connection", "keep-alive")
239                .header("Content-Encoding", "br")
240                .method(Method::POST)
241                .body(body)?;
242
243            let mut response = http_client.send(request).await?;
244
245            let mut body = String::new();
246            response.body_mut().read_to_string(&mut body).await?;
247
248            let response_received_at = Instant::now();
249            if !response.status().is_success() {
250                let message = format!(
251                    "Request failed with status: {:?}\nBody: {}",
252                    response.status(),
253                    body,
254                );
255                send_finished_event(&debug_tx, &buffer, inputs.position, message.clone());
256                bail!(message);
257            };
258
259            let response: AutocompleteResponse = serde_json::from_str(&body)?;
260
261            send_finished_event(&debug_tx, &buffer, inputs.position, body);
262
263            let old_text = inputs
264                .snapshot
265                .text_for_range(response.start_index..response.end_index)
266                .collect::<String>();
267            let edits = language::text_diff(&old_text, &response.completion)
268                .into_iter()
269                .map(|(range, text)| {
270                    (
271                        inputs
272                            .snapshot
273                            .anchor_after(response.start_index + range.start)
274                            ..inputs
275                                .snapshot
276                                .anchor_before(response.start_index + range.end),
277                        text,
278                    )
279                })
280                .collect::<Vec<_>>();
281
282            anyhow::Ok((
283                response.autocomplete_id,
284                edits,
285                inputs.snapshot,
286                response_received_at,
287                ep_inputs,
288            ))
289        });
290
291        let buffer = inputs.buffer.clone();
292
293        cx.spawn(async move |cx| {
294            let (id, edits, old_snapshot, response_received_at, inputs) = result.await?;
295            anyhow::Ok(Some(
296                EditPredictionResult::new(
297                    EditPredictionId(id.into()),
298                    &buffer,
299                    &old_snapshot,
300                    edits.into(),
301                    None,
302                    buffer_snapshotted_at,
303                    response_received_at,
304                    inputs,
305                    cx,
306                )
307                .await,
308            ))
309        })
310    }
311}
312
313fn send_started_event(
314    debug_tx: &Option<mpsc::UnboundedSender<DebugEvent>>,
315    buffer: &Entity<Buffer>,
316    position: Anchor,
317    prompt: String,
318) {
319    if let Some(debug_tx) = debug_tx {
320        _ = debug_tx.unbounded_send(DebugEvent::EditPredictionStarted(
321            EditPredictionStartedDebugEvent {
322                buffer: buffer.downgrade(),
323                position,
324                prompt: Some(prompt),
325            },
326        ));
327    }
328}
329
330fn send_finished_event(
331    debug_tx: &Option<mpsc::UnboundedSender<DebugEvent>>,
332    buffer: &Entity<Buffer>,
333    position: Anchor,
334    model_output: String,
335) {
336    if let Some(debug_tx) = debug_tx {
337        _ = debug_tx.unbounded_send(DebugEvent::EditPredictionFinished(
338            EditPredictionFinishedDebugEvent {
339                buffer: buffer.downgrade(),
340                position,
341                model_output: Some(model_output),
342            },
343        ));
344    }
345}
346
347pub const SWEEP_CREDENTIALS_URL: SharedString =
348    SharedString::new_static("https://autocomplete.sweep.dev");
349pub const SWEEP_CREDENTIALS_USERNAME: &str = "sweep-api-token";
350pub static SWEEP_AI_TOKEN_ENV_VAR: std::sync::LazyLock<EnvVar> = env_var!("SWEEP_AI_TOKEN");
351
352struct GlobalSweepApiKey(Entity<ApiKeyState>);
353
354impl Global for GlobalSweepApiKey {}
355
356pub fn sweep_api_token(cx: &mut App) -> Entity<ApiKeyState> {
357    if let Some(global) = cx.try_global::<GlobalSweepApiKey>() {
358        return global.0.clone();
359    }
360    let entity =
361        cx.new(|_| ApiKeyState::new(SWEEP_CREDENTIALS_URL, SWEEP_AI_TOKEN_ENV_VAR.clone()));
362    cx.set_global(GlobalSweepApiKey(entity.clone()));
363    entity
364}
365
366pub fn load_sweep_api_token(cx: &mut App) -> Task<Result<(), language_model::AuthenticateError>> {
367    sweep_api_token(cx).update(cx, |key_state, cx| {
368        key_state.load_if_needed(SWEEP_CREDENTIALS_URL, |s| s, cx)
369    })
370}
371
372#[derive(Debug, Clone, Serialize)]
373struct AutocompleteRequest {
374    pub debug_info: Arc<str>,
375    pub repo_name: String,
376    pub branch: Option<String>,
377    pub file_path: Arc<Path>,
378    pub file_contents: String,
379    pub recent_changes: String,
380    pub cursor_position: usize,
381    pub original_file_contents: String,
382    pub file_chunks: Vec<FileChunk>,
383    pub retrieval_chunks: Vec<FileChunk>,
384    pub recent_user_actions: Vec<UserAction>,
385    pub multiple_suggestions: bool,
386    pub privacy_mode_enabled: bool,
387    pub changes_above_cursor: bool,
388    pub use_bytes: bool,
389}
390
391#[derive(Debug, Clone, Serialize)]
392struct FileChunk {
393    pub file_path: String,
394    pub start_line: usize,
395    pub end_line: usize,
396    pub content: String,
397    pub timestamp: Option<u64>,
398}
399
400#[derive(Debug, Clone, Serialize)]
401struct UserAction {
402    pub action_type: ActionType,
403    pub line_number: usize,
404    pub offset: usize,
405    pub file_path: String,
406    pub timestamp: u64,
407}
408
409#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize)]
410#[serde(rename_all = "SCREAMING_SNAKE_CASE")]
411enum ActionType {
412    CursorMovement,
413    InsertChar,
414    DeleteChar,
415    InsertSelection,
416    DeleteSelection,
417}
418
419fn to_sweep_user_action(record: &UserActionRecord, file_path: &str) -> UserAction {
420    UserAction {
421        action_type: match record.action_type {
422            UserActionType::InsertChar => ActionType::InsertChar,
423            UserActionType::InsertSelection => ActionType::InsertSelection,
424            UserActionType::DeleteChar => ActionType::DeleteChar,
425            UserActionType::DeleteSelection => ActionType::DeleteSelection,
426            UserActionType::CursorMovement => ActionType::CursorMovement,
427        },
428        line_number: record.line_number as usize,
429        offset: record.offset,
430        file_path: file_path.to_string(),
431        timestamp: record.timestamp_epoch_ms,
432    }
433}
434
435#[derive(Debug, Clone, Deserialize)]
436struct AutocompleteResponse {
437    pub autocomplete_id: String,
438    pub start_index: usize,
439    pub end_index: usize,
440    pub completion: String,
441    #[allow(dead_code)]
442    pub confidence: f64,
443    #[allow(dead_code)]
444    pub logprobs: Option<serde_json::Value>,
445    #[allow(dead_code)]
446    pub finish_reason: Option<String>,
447    #[allow(dead_code)]
448    pub elapsed_time_ms: u64,
449    #[allow(dead_code)]
450    #[serde(default, rename = "completions")]
451    pub additional_completions: Vec<AdditionalCompletion>,
452}
453
454#[allow(dead_code)]
455#[derive(Debug, Clone, Deserialize)]
456struct AdditionalCompletion {
457    pub start_index: usize,
458    pub end_index: usize,
459    pub completion: String,
460    pub confidence: f64,
461    pub autocomplete_id: String,
462    pub logprobs: Option<serde_json::Value>,
463    pub finish_reason: Option<String>,
464}
465
466fn write_event(event: &zeta_prompt::Event, f: &mut impl fmt::Write) -> fmt::Result {
467    match event {
468        zeta_prompt::Event::BufferChange {
469            old_path,
470            path,
471            diff,
472            ..
473        } => {
474            if old_path != path {
475                // TODO confirm how to do this for sweep
476                // writeln!(f, "User renamed {:?} to {:?}\n", old_path, new_path)?;
477            }
478
479            if !diff.is_empty() {
480                write!(f, "File: {}:\n{}\n", path.display(), diff)?
481            }
482
483            fmt::Result::Ok(())
484        }
485    }
486}
487
488fn debug_info(cx: &gpui::App) -> Arc<str> {
489    format!(
490        "Zed v{version} ({sha}) - OS: {os} - Zed v{version}",
491        version = release_channel::AppVersion::global(cx),
492        sha = release_channel::AppCommitSha::try_global(cx)
493            .map_or("unknown".to_string(), |sha| sha.full()),
494        os = client::telemetry::os_name(),
495    )
496    .into()
497}
498
499#[derive(Debug, Clone, Copy, Serialize)]
500#[serde(rename_all = "snake_case")]
501pub enum SweepEventType {
502    AutocompleteSuggestionShown,
503    AutocompleteSuggestionAccepted,
504}
505
506#[derive(Debug, Clone, Copy, Serialize)]
507#[serde(rename_all = "SCREAMING_SNAKE_CASE")]
508pub enum SweepSuggestionType {
509    GhostText,
510    Popup,
511    JumpToEdit,
512}
513
514#[derive(Debug, Clone, Serialize)]
515struct AutocompleteMetricsRequest {
516    event_type: SweepEventType,
517    suggestion_type: SweepSuggestionType,
518    additions: u32,
519    deletions: u32,
520    autocomplete_id: String,
521    edit_tracking: String,
522    edit_tracking_line: Option<u32>,
523    lifespan: Option<u64>,
524    debug_info: Arc<str>,
525    device_id: String,
526    privacy_mode_enabled: bool,
527}
528
529fn send_autocomplete_metrics_request(
530    cx: &App,
531    client: Arc<Client>,
532    api_token: Arc<str>,
533    request_body: AutocompleteMetricsRequest,
534) {
535    let http_client = client.http_client();
536    cx.background_spawn(async move {
537        let body: AsyncBody = serde_json::to_string(&request_body)?.into();
538
539        let request = http_client::Request::builder()
540            .uri(SWEEP_METRICS_URL)
541            .header("Content-Type", "application/json")
542            .header("Authorization", format!("Bearer {}", api_token))
543            .method(Method::POST)
544            .body(body)?;
545
546        let mut response = http_client.send(request).await?;
547
548        if !response.status().is_success() {
549            let mut body = String::new();
550            response.body_mut().read_to_string(&mut body).await?;
551            anyhow::bail!(
552                "Failed to send autocomplete metrics for sweep_ai: {:?}\nBody: {}",
553                response.status(),
554                body,
555            );
556        }
557
558        Ok(())
559    })
560    .detach_and_log_err(cx);
561}
562
563pub(crate) fn edit_prediction_accepted(
564    store: &EditPredictionStore,
565    current_prediction: CurrentEditPrediction,
566    cx: &App,
567) {
568    let Some(api_token) = store
569        .sweep_ai
570        .api_token
571        .read(cx)
572        .key(&SWEEP_CREDENTIALS_URL)
573    else {
574        return;
575    };
576    let debug_info = store.sweep_ai.debug_info.clone();
577
578    let prediction = current_prediction.prediction;
579
580    let (additions, deletions) = compute_edit_metrics(&prediction.edits, &prediction.snapshot);
581    let autocomplete_id = prediction.id.to_string();
582
583    let device_id = store
584        .client
585        .user_id()
586        .as_ref()
587        .map(ToString::to_string)
588        .unwrap_or_default();
589
590    let suggestion_type = match current_prediction.shown_with {
591        Some(SuggestionDisplayType::DiffPopover) => SweepSuggestionType::Popup,
592        Some(SuggestionDisplayType::Jump) => return, // should'nt happen
593        Some(SuggestionDisplayType::GhostText) | None => SweepSuggestionType::GhostText,
594    };
595
596    let request_body = AutocompleteMetricsRequest {
597        event_type: SweepEventType::AutocompleteSuggestionAccepted,
598        suggestion_type,
599        additions,
600        deletions,
601        autocomplete_id,
602        edit_tracking: String::new(),
603        edit_tracking_line: None,
604        lifespan: None,
605        debug_info,
606        device_id,
607        privacy_mode_enabled: false,
608    };
609
610    send_autocomplete_metrics_request(cx, store.client.clone(), api_token, request_body);
611}
612
613pub fn edit_prediction_shown(
614    sweep_ai: &SweepAi,
615    client: Arc<Client>,
616    prediction: &EditPrediction,
617    display_type: SuggestionDisplayType,
618    cx: &App,
619) {
620    let Some(api_token) = sweep_ai.api_token.read(cx).key(&SWEEP_CREDENTIALS_URL) else {
621        return;
622    };
623    let debug_info = sweep_ai.debug_info.clone();
624
625    let (additions, deletions) = compute_edit_metrics(&prediction.edits, &prediction.snapshot);
626    let autocomplete_id = prediction.id.to_string();
627
628    let suggestion_type = match display_type {
629        SuggestionDisplayType::GhostText => SweepSuggestionType::GhostText,
630        SuggestionDisplayType::DiffPopover => SweepSuggestionType::Popup,
631        SuggestionDisplayType::Jump => SweepSuggestionType::JumpToEdit,
632    };
633
634    let request_body = AutocompleteMetricsRequest {
635        event_type: SweepEventType::AutocompleteSuggestionShown,
636        suggestion_type,
637        additions,
638        deletions,
639        autocomplete_id,
640        edit_tracking: String::new(),
641        edit_tracking_line: None,
642        lifespan: None,
643        debug_info,
644        device_id: String::new(),
645        privacy_mode_enabled: false,
646    };
647
648    send_autocomplete_metrics_request(cx, client, api_token, request_body);
649}
650
651fn compute_edit_metrics(
652    edits: &[(Range<Anchor>, Arc<str>)],
653    snapshot: &BufferSnapshot,
654) -> (u32, u32) {
655    let mut additions = 0u32;
656    let mut deletions = 0u32;
657
658    for (range, new_text) in edits {
659        let old_text = snapshot.text_for_range(range.clone());
660        deletions += old_text
661            .map(|chunk| chunk.lines().count())
662            .sum::<usize>()
663            .max(1) as u32;
664        additions += new_text.lines().count().max(1) as u32;
665    }
666
667    (additions, deletions)
668}