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                can_collect_data: false,
226            };
227
228            send_started_event(
229                &debug_tx,
230                &buffer,
231                inputs.position,
232                serde_json::to_string(&request_body).unwrap_or_default(),
233            );
234
235            let request = http_client::Request::builder()
236                .uri(SWEEP_API_URL)
237                .header("Content-Type", "application/json")
238                .header("Authorization", format!("Bearer {}", api_token))
239                .header("Connection", "keep-alive")
240                .header("Content-Encoding", "br")
241                .method(Method::POST)
242                .body(body)?;
243
244            let mut response = http_client.send(request).await?;
245
246            let mut body = String::new();
247            response.body_mut().read_to_string(&mut body).await?;
248
249            let response_received_at = Instant::now();
250            if !response.status().is_success() {
251                let message = format!(
252                    "Request failed with status: {:?}\nBody: {}",
253                    response.status(),
254                    body,
255                );
256                send_finished_event(&debug_tx, &buffer, inputs.position, message.clone());
257                bail!(message);
258            };
259
260            let response: AutocompleteResponse = serde_json::from_str(&body)?;
261
262            send_finished_event(&debug_tx, &buffer, inputs.position, body);
263
264            let old_text = inputs
265                .snapshot
266                .text_for_range(response.start_index..response.end_index)
267                .collect::<String>();
268            let edits = language::text_diff(&old_text, &response.completion)
269                .into_iter()
270                .map(|(range, text)| {
271                    (
272                        inputs
273                            .snapshot
274                            .anchor_after(response.start_index + range.start)
275                            ..inputs
276                                .snapshot
277                                .anchor_before(response.start_index + range.end),
278                        text,
279                    )
280                })
281                .collect::<Vec<_>>();
282
283            anyhow::Ok((
284                response.autocomplete_id,
285                edits,
286                inputs.snapshot,
287                response_received_at,
288                ep_inputs,
289            ))
290        });
291
292        let buffer = inputs.buffer.clone();
293
294        cx.spawn(async move |cx| {
295            let (id, edits, old_snapshot, response_received_at, inputs) = result.await?;
296            anyhow::Ok(Some(
297                EditPredictionResult::new(
298                    EditPredictionId(id.into()),
299                    &buffer,
300                    &old_snapshot,
301                    edits.into(),
302                    None,
303                    buffer_snapshotted_at,
304                    response_received_at,
305                    inputs,
306                    None,
307                    cx,
308                )
309                .await,
310            ))
311        })
312    }
313}
314
315fn send_started_event(
316    debug_tx: &Option<mpsc::UnboundedSender<DebugEvent>>,
317    buffer: &Entity<Buffer>,
318    position: Anchor,
319    prompt: String,
320) {
321    if let Some(debug_tx) = debug_tx {
322        _ = debug_tx.unbounded_send(DebugEvent::EditPredictionStarted(
323            EditPredictionStartedDebugEvent {
324                buffer: buffer.downgrade(),
325                position,
326                prompt: Some(prompt),
327            },
328        ));
329    }
330}
331
332fn send_finished_event(
333    debug_tx: &Option<mpsc::UnboundedSender<DebugEvent>>,
334    buffer: &Entity<Buffer>,
335    position: Anchor,
336    model_output: String,
337) {
338    if let Some(debug_tx) = debug_tx {
339        _ = debug_tx.unbounded_send(DebugEvent::EditPredictionFinished(
340            EditPredictionFinishedDebugEvent {
341                buffer: buffer.downgrade(),
342                position,
343                model_output: Some(model_output),
344            },
345        ));
346    }
347}
348
349pub const SWEEP_CREDENTIALS_URL: SharedString =
350    SharedString::new_static("https://autocomplete.sweep.dev");
351pub const SWEEP_CREDENTIALS_USERNAME: &str = "sweep-api-token";
352pub static SWEEP_AI_TOKEN_ENV_VAR: std::sync::LazyLock<EnvVar> = env_var!("SWEEP_AI_TOKEN");
353
354struct GlobalSweepApiKey(Entity<ApiKeyState>);
355
356impl Global for GlobalSweepApiKey {}
357
358pub fn sweep_api_token(cx: &mut App) -> Entity<ApiKeyState> {
359    if let Some(global) = cx.try_global::<GlobalSweepApiKey>() {
360        return global.0.clone();
361    }
362    let entity =
363        cx.new(|_| ApiKeyState::new(SWEEP_CREDENTIALS_URL, SWEEP_AI_TOKEN_ENV_VAR.clone()));
364    cx.set_global(GlobalSweepApiKey(entity.clone()));
365    entity
366}
367
368pub fn load_sweep_api_token(cx: &mut App) -> Task<Result<(), language_model::AuthenticateError>> {
369    sweep_api_token(cx).update(cx, |key_state, cx| {
370        key_state.load_if_needed(SWEEP_CREDENTIALS_URL, |s| s, cx)
371    })
372}
373
374#[derive(Debug, Clone, Serialize)]
375struct AutocompleteRequest {
376    pub debug_info: Arc<str>,
377    pub repo_name: String,
378    pub branch: Option<String>,
379    pub file_path: Arc<Path>,
380    pub file_contents: String,
381    pub recent_changes: String,
382    pub cursor_position: usize,
383    pub original_file_contents: String,
384    pub file_chunks: Vec<FileChunk>,
385    pub retrieval_chunks: Vec<FileChunk>,
386    pub recent_user_actions: Vec<UserAction>,
387    pub multiple_suggestions: bool,
388    pub privacy_mode_enabled: bool,
389    pub changes_above_cursor: bool,
390    pub use_bytes: bool,
391}
392
393#[derive(Debug, Clone, Serialize)]
394struct FileChunk {
395    pub file_path: String,
396    pub start_line: usize,
397    pub end_line: usize,
398    pub content: String,
399    pub timestamp: Option<u64>,
400}
401
402#[derive(Debug, Clone, Serialize)]
403struct UserAction {
404    pub action_type: ActionType,
405    pub line_number: usize,
406    pub offset: usize,
407    pub file_path: String,
408    pub timestamp: u64,
409}
410
411#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize)]
412#[serde(rename_all = "SCREAMING_SNAKE_CASE")]
413enum ActionType {
414    CursorMovement,
415    InsertChar,
416    DeleteChar,
417    InsertSelection,
418    DeleteSelection,
419}
420
421fn to_sweep_user_action(record: &UserActionRecord, file_path: &str) -> UserAction {
422    UserAction {
423        action_type: match record.action_type {
424            UserActionType::InsertChar => ActionType::InsertChar,
425            UserActionType::InsertSelection => ActionType::InsertSelection,
426            UserActionType::DeleteChar => ActionType::DeleteChar,
427            UserActionType::DeleteSelection => ActionType::DeleteSelection,
428            UserActionType::CursorMovement => ActionType::CursorMovement,
429        },
430        line_number: record.line_number as usize,
431        offset: record.offset,
432        file_path: file_path.to_string(),
433        timestamp: record.timestamp_epoch_ms,
434    }
435}
436
437#[derive(Debug, Clone, Deserialize)]
438struct AutocompleteResponse {
439    pub autocomplete_id: String,
440    pub start_index: usize,
441    pub end_index: usize,
442    pub completion: String,
443    #[allow(dead_code)]
444    pub confidence: f64,
445    #[allow(dead_code)]
446    pub logprobs: Option<serde_json::Value>,
447    #[allow(dead_code)]
448    pub finish_reason: Option<String>,
449    #[allow(dead_code)]
450    pub elapsed_time_ms: u64,
451    #[allow(dead_code)]
452    #[serde(default, rename = "completions")]
453    pub additional_completions: Vec<AdditionalCompletion>,
454}
455
456#[allow(dead_code)]
457#[derive(Debug, Clone, Deserialize)]
458struct AdditionalCompletion {
459    pub start_index: usize,
460    pub end_index: usize,
461    pub completion: String,
462    pub confidence: f64,
463    pub autocomplete_id: String,
464    pub logprobs: Option<serde_json::Value>,
465    pub finish_reason: Option<String>,
466}
467
468fn write_event(event: &zeta_prompt::Event, f: &mut impl fmt::Write) -> fmt::Result {
469    match event {
470        zeta_prompt::Event::BufferChange {
471            old_path,
472            path,
473            diff,
474            ..
475        } => {
476            if old_path != path {
477                // TODO confirm how to do this for sweep
478                // writeln!(f, "User renamed {:?} to {:?}\n", old_path, new_path)?;
479            }
480
481            if !diff.is_empty() {
482                write!(f, "File: {}:\n{}\n", path.display(), diff)?
483            }
484
485            fmt::Result::Ok(())
486        }
487    }
488}
489
490fn debug_info(cx: &gpui::App) -> Arc<str> {
491    format!(
492        "Zed v{version} ({sha}) - OS: {os} - Zed v{version}",
493        version = release_channel::AppVersion::global(cx),
494        sha = release_channel::AppCommitSha::try_global(cx)
495            .map_or("unknown".to_string(), |sha| sha.full()),
496        os = client::telemetry::os_name(),
497    )
498    .into()
499}
500
501#[derive(Debug, Clone, Copy, Serialize)]
502#[serde(rename_all = "snake_case")]
503pub enum SweepEventType {
504    AutocompleteSuggestionShown,
505    AutocompleteSuggestionAccepted,
506}
507
508#[derive(Debug, Clone, Copy, Serialize)]
509#[serde(rename_all = "SCREAMING_SNAKE_CASE")]
510pub enum SweepSuggestionType {
511    GhostText,
512    Popup,
513    JumpToEdit,
514}
515
516#[derive(Debug, Clone, Serialize)]
517struct AutocompleteMetricsRequest {
518    event_type: SweepEventType,
519    suggestion_type: SweepSuggestionType,
520    additions: u32,
521    deletions: u32,
522    autocomplete_id: String,
523    edit_tracking: String,
524    edit_tracking_line: Option<u32>,
525    lifespan: Option<u64>,
526    debug_info: Arc<str>,
527    device_id: String,
528    privacy_mode_enabled: bool,
529}
530
531fn send_autocomplete_metrics_request(
532    cx: &App,
533    client: Arc<Client>,
534    api_token: Arc<str>,
535    request_body: AutocompleteMetricsRequest,
536) {
537    let http_client = client.http_client();
538    cx.background_spawn(async move {
539        let body: AsyncBody = serde_json::to_string(&request_body)?.into();
540
541        let request = http_client::Request::builder()
542            .uri(SWEEP_METRICS_URL)
543            .header("Content-Type", "application/json")
544            .header("Authorization", format!("Bearer {}", api_token))
545            .method(Method::POST)
546            .body(body)?;
547
548        let mut response = http_client.send(request).await?;
549
550        if !response.status().is_success() {
551            let mut body = String::new();
552            response.body_mut().read_to_string(&mut body).await?;
553            anyhow::bail!(
554                "Failed to send autocomplete metrics for sweep_ai: {:?}\nBody: {}",
555                response.status(),
556                body,
557            );
558        }
559
560        Ok(())
561    })
562    .detach_and_log_err(cx);
563}
564
565pub(crate) fn edit_prediction_accepted(
566    store: &EditPredictionStore,
567    current_prediction: CurrentEditPrediction,
568    cx: &App,
569) {
570    let Some(api_token) = store
571        .sweep_ai
572        .api_token
573        .read(cx)
574        .key(&SWEEP_CREDENTIALS_URL)
575    else {
576        return;
577    };
578    let debug_info = store.sweep_ai.debug_info.clone();
579
580    let prediction = current_prediction.prediction;
581
582    let (additions, deletions) = compute_edit_metrics(&prediction.edits, &prediction.snapshot);
583    let autocomplete_id = prediction.id.to_string();
584
585    let device_id = store
586        .client
587        .user_id()
588        .as_ref()
589        .map(ToString::to_string)
590        .unwrap_or_default();
591
592    let suggestion_type = match current_prediction.shown_with {
593        Some(SuggestionDisplayType::DiffPopover) => SweepSuggestionType::Popup,
594        Some(SuggestionDisplayType::Jump) => return, // should'nt happen
595        Some(SuggestionDisplayType::GhostText) | None => SweepSuggestionType::GhostText,
596    };
597
598    let request_body = AutocompleteMetricsRequest {
599        event_type: SweepEventType::AutocompleteSuggestionAccepted,
600        suggestion_type,
601        additions,
602        deletions,
603        autocomplete_id,
604        edit_tracking: String::new(),
605        edit_tracking_line: None,
606        lifespan: None,
607        debug_info,
608        device_id,
609        privacy_mode_enabled: false,
610    };
611
612    send_autocomplete_metrics_request(cx, store.client.clone(), api_token, request_body);
613}
614
615pub fn edit_prediction_shown(
616    sweep_ai: &SweepAi,
617    client: Arc<Client>,
618    prediction: &EditPrediction,
619    display_type: SuggestionDisplayType,
620    cx: &App,
621) {
622    let Some(api_token) = sweep_ai.api_token.read(cx).key(&SWEEP_CREDENTIALS_URL) else {
623        return;
624    };
625    let debug_info = sweep_ai.debug_info.clone();
626
627    let (additions, deletions) = compute_edit_metrics(&prediction.edits, &prediction.snapshot);
628    let autocomplete_id = prediction.id.to_string();
629
630    let suggestion_type = match display_type {
631        SuggestionDisplayType::GhostText => SweepSuggestionType::GhostText,
632        SuggestionDisplayType::DiffPopover => SweepSuggestionType::Popup,
633        SuggestionDisplayType::Jump => SweepSuggestionType::JumpToEdit,
634    };
635
636    let request_body = AutocompleteMetricsRequest {
637        event_type: SweepEventType::AutocompleteSuggestionShown,
638        suggestion_type,
639        additions,
640        deletions,
641        autocomplete_id,
642        edit_tracking: String::new(),
643        edit_tracking_line: None,
644        lifespan: None,
645        debug_info,
646        device_id: String::new(),
647        privacy_mode_enabled: false,
648    };
649
650    send_autocomplete_metrics_request(cx, client, api_token, request_body);
651}
652
653fn compute_edit_metrics(
654    edits: &[(Range<Anchor>, Arc<str>)],
655    snapshot: &BufferSnapshot,
656) -> (u32, u32) {
657    let mut additions = 0u32;
658    let mut deletions = 0u32;
659
660    for (range, new_text) in edits {
661        let old_text = snapshot.text_for_range(range.clone());
662        deletions += old_text
663            .map(|chunk| chunk.lines().count())
664            .sum::<usize>()
665            .max(1) as u32;
666        additions += new_text.lines().count().max(1) as u32;
667    }
668
669    (additions, deletions)
670}