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            };
222
223            send_started_event(
224                &debug_tx,
225                &buffer,
226                inputs.position,
227                serde_json::to_string(&request_body).unwrap_or_default(),
228            );
229
230            let request = http_client::Request::builder()
231                .uri(SWEEP_API_URL)
232                .header("Content-Type", "application/json")
233                .header("Authorization", format!("Bearer {}", api_token))
234                .header("Connection", "keep-alive")
235                .header("Content-Encoding", "br")
236                .method(Method::POST)
237                .body(body)?;
238
239            let mut response = http_client.send(request).await?;
240
241            let mut body = String::new();
242            response.body_mut().read_to_string(&mut body).await?;
243
244            let response_received_at = Instant::now();
245            if !response.status().is_success() {
246                let message = format!(
247                    "Request failed with status: {:?}\nBody: {}",
248                    response.status(),
249                    body,
250                );
251                send_finished_event(&debug_tx, &buffer, inputs.position, message.clone());
252                bail!(message);
253            };
254
255            let response: AutocompleteResponse = serde_json::from_str(&body)?;
256
257            send_finished_event(&debug_tx, &buffer, inputs.position, body);
258
259            let old_text = inputs
260                .snapshot
261                .text_for_range(response.start_index..response.end_index)
262                .collect::<String>();
263            let edits = language::text_diff(&old_text, &response.completion)
264                .into_iter()
265                .map(|(range, text)| {
266                    (
267                        inputs
268                            .snapshot
269                            .anchor_after(response.start_index + range.start)
270                            ..inputs
271                                .snapshot
272                                .anchor_before(response.start_index + range.end),
273                        text,
274                    )
275                })
276                .collect::<Vec<_>>();
277
278            anyhow::Ok((
279                response.autocomplete_id,
280                edits,
281                inputs.snapshot,
282                response_received_at,
283                ep_inputs,
284            ))
285        });
286
287        let buffer = inputs.buffer.clone();
288
289        cx.spawn(async move |cx| {
290            let (id, edits, old_snapshot, response_received_at, inputs) = result.await?;
291            anyhow::Ok(Some(
292                EditPredictionResult::new(
293                    EditPredictionId(id.into()),
294                    &buffer,
295                    &old_snapshot,
296                    edits.into(),
297                    buffer_snapshotted_at,
298                    response_received_at,
299                    inputs,
300                    cx,
301                )
302                .await,
303            ))
304        })
305    }
306}
307
308fn send_started_event(
309    debug_tx: &Option<mpsc::UnboundedSender<DebugEvent>>,
310    buffer: &Entity<Buffer>,
311    position: Anchor,
312    prompt: String,
313) {
314    if let Some(debug_tx) = debug_tx {
315        _ = debug_tx.unbounded_send(DebugEvent::EditPredictionStarted(
316            EditPredictionStartedDebugEvent {
317                buffer: buffer.downgrade(),
318                position,
319                prompt: Some(prompt),
320            },
321        ));
322    }
323}
324
325fn send_finished_event(
326    debug_tx: &Option<mpsc::UnboundedSender<DebugEvent>>,
327    buffer: &Entity<Buffer>,
328    position: Anchor,
329    model_output: String,
330) {
331    if let Some(debug_tx) = debug_tx {
332        _ = debug_tx.unbounded_send(DebugEvent::EditPredictionFinished(
333            EditPredictionFinishedDebugEvent {
334                buffer: buffer.downgrade(),
335                position,
336                model_output: Some(model_output),
337            },
338        ));
339    }
340}
341
342pub const SWEEP_CREDENTIALS_URL: SharedString =
343    SharedString::new_static("https://autocomplete.sweep.dev");
344pub const SWEEP_CREDENTIALS_USERNAME: &str = "sweep-api-token";
345pub static SWEEP_AI_TOKEN_ENV_VAR: std::sync::LazyLock<EnvVar> = env_var!("SWEEP_AI_TOKEN");
346
347struct GlobalSweepApiKey(Entity<ApiKeyState>);
348
349impl Global for GlobalSweepApiKey {}
350
351pub fn sweep_api_token(cx: &mut App) -> Entity<ApiKeyState> {
352    if let Some(global) = cx.try_global::<GlobalSweepApiKey>() {
353        return global.0.clone();
354    }
355    let entity =
356        cx.new(|_| ApiKeyState::new(SWEEP_CREDENTIALS_URL, SWEEP_AI_TOKEN_ENV_VAR.clone()));
357    cx.set_global(GlobalSweepApiKey(entity.clone()));
358    entity
359}
360
361pub fn load_sweep_api_token(cx: &mut App) -> Task<Result<(), language_model::AuthenticateError>> {
362    sweep_api_token(cx).update(cx, |key_state, cx| {
363        key_state.load_if_needed(SWEEP_CREDENTIALS_URL, |s| s, cx)
364    })
365}
366
367#[derive(Debug, Clone, Serialize)]
368struct AutocompleteRequest {
369    pub debug_info: Arc<str>,
370    pub repo_name: String,
371    pub branch: Option<String>,
372    pub file_path: Arc<Path>,
373    pub file_contents: String,
374    pub recent_changes: String,
375    pub cursor_position: usize,
376    pub original_file_contents: String,
377    pub file_chunks: Vec<FileChunk>,
378    pub retrieval_chunks: Vec<FileChunk>,
379    pub recent_user_actions: Vec<UserAction>,
380    pub multiple_suggestions: bool,
381    pub privacy_mode_enabled: bool,
382    pub changes_above_cursor: bool,
383    pub use_bytes: bool,
384}
385
386#[derive(Debug, Clone, Serialize)]
387struct FileChunk {
388    pub file_path: String,
389    pub start_line: usize,
390    pub end_line: usize,
391    pub content: String,
392    pub timestamp: Option<u64>,
393}
394
395#[derive(Debug, Clone, Serialize)]
396struct UserAction {
397    pub action_type: ActionType,
398    pub line_number: usize,
399    pub offset: usize,
400    pub file_path: String,
401    pub timestamp: u64,
402}
403
404#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize)]
405#[serde(rename_all = "SCREAMING_SNAKE_CASE")]
406enum ActionType {
407    CursorMovement,
408    InsertChar,
409    DeleteChar,
410    InsertSelection,
411    DeleteSelection,
412}
413
414fn to_sweep_user_action(record: &UserActionRecord, file_path: &str) -> UserAction {
415    UserAction {
416        action_type: match record.action_type {
417            UserActionType::InsertChar => ActionType::InsertChar,
418            UserActionType::InsertSelection => ActionType::InsertSelection,
419            UserActionType::DeleteChar => ActionType::DeleteChar,
420            UserActionType::DeleteSelection => ActionType::DeleteSelection,
421            UserActionType::CursorMovement => ActionType::CursorMovement,
422        },
423        line_number: record.line_number as usize,
424        offset: record.offset,
425        file_path: file_path.to_string(),
426        timestamp: record.timestamp_epoch_ms,
427    }
428}
429
430#[derive(Debug, Clone, Deserialize)]
431struct AutocompleteResponse {
432    pub autocomplete_id: String,
433    pub start_index: usize,
434    pub end_index: usize,
435    pub completion: String,
436    #[allow(dead_code)]
437    pub confidence: f64,
438    #[allow(dead_code)]
439    pub logprobs: Option<serde_json::Value>,
440    #[allow(dead_code)]
441    pub finish_reason: Option<String>,
442    #[allow(dead_code)]
443    pub elapsed_time_ms: u64,
444    #[allow(dead_code)]
445    #[serde(default, rename = "completions")]
446    pub additional_completions: Vec<AdditionalCompletion>,
447}
448
449#[allow(dead_code)]
450#[derive(Debug, Clone, Deserialize)]
451struct AdditionalCompletion {
452    pub start_index: usize,
453    pub end_index: usize,
454    pub completion: String,
455    pub confidence: f64,
456    pub autocomplete_id: String,
457    pub logprobs: Option<serde_json::Value>,
458    pub finish_reason: Option<String>,
459}
460
461fn write_event(event: &zeta_prompt::Event, f: &mut impl fmt::Write) -> fmt::Result {
462    match event {
463        zeta_prompt::Event::BufferChange {
464            old_path,
465            path,
466            diff,
467            ..
468        } => {
469            if old_path != path {
470                // TODO confirm how to do this for sweep
471                // writeln!(f, "User renamed {:?} to {:?}\n", old_path, new_path)?;
472            }
473
474            if !diff.is_empty() {
475                write!(f, "File: {}:\n{}\n", path.display(), diff)?
476            }
477
478            fmt::Result::Ok(())
479        }
480    }
481}
482
483fn debug_info(cx: &gpui::App) -> Arc<str> {
484    format!(
485        "Zed v{version} ({sha}) - OS: {os} - Zed v{version}",
486        version = release_channel::AppVersion::global(cx),
487        sha = release_channel::AppCommitSha::try_global(cx)
488            .map_or("unknown".to_string(), |sha| sha.full()),
489        os = client::telemetry::os_name(),
490    )
491    .into()
492}
493
494#[derive(Debug, Clone, Copy, Serialize)]
495#[serde(rename_all = "snake_case")]
496pub enum SweepEventType {
497    AutocompleteSuggestionShown,
498    AutocompleteSuggestionAccepted,
499}
500
501#[derive(Debug, Clone, Copy, Serialize)]
502#[serde(rename_all = "SCREAMING_SNAKE_CASE")]
503pub enum SweepSuggestionType {
504    GhostText,
505    Popup,
506    JumpToEdit,
507}
508
509#[derive(Debug, Clone, Serialize)]
510struct AutocompleteMetricsRequest {
511    event_type: SweepEventType,
512    suggestion_type: SweepSuggestionType,
513    additions: u32,
514    deletions: u32,
515    autocomplete_id: String,
516    edit_tracking: String,
517    edit_tracking_line: Option<u32>,
518    lifespan: Option<u64>,
519    debug_info: Arc<str>,
520    device_id: String,
521    privacy_mode_enabled: bool,
522}
523
524fn send_autocomplete_metrics_request(
525    cx: &App,
526    client: Arc<Client>,
527    api_token: Arc<str>,
528    request_body: AutocompleteMetricsRequest,
529) {
530    let http_client = client.http_client();
531    cx.background_spawn(async move {
532        let body: AsyncBody = serde_json::to_string(&request_body)?.into();
533
534        let request = http_client::Request::builder()
535            .uri(SWEEP_METRICS_URL)
536            .header("Content-Type", "application/json")
537            .header("Authorization", format!("Bearer {}", api_token))
538            .method(Method::POST)
539            .body(body)?;
540
541        let mut response = http_client.send(request).await?;
542
543        if !response.status().is_success() {
544            let mut body = String::new();
545            response.body_mut().read_to_string(&mut body).await?;
546            anyhow::bail!(
547                "Failed to send autocomplete metrics for sweep_ai: {:?}\nBody: {}",
548                response.status(),
549                body,
550            );
551        }
552
553        Ok(())
554    })
555    .detach_and_log_err(cx);
556}
557
558pub(crate) fn edit_prediction_accepted(
559    store: &EditPredictionStore,
560    current_prediction: CurrentEditPrediction,
561    cx: &App,
562) {
563    let Some(api_token) = store
564        .sweep_ai
565        .api_token
566        .read(cx)
567        .key(&SWEEP_CREDENTIALS_URL)
568    else {
569        return;
570    };
571    let debug_info = store.sweep_ai.debug_info.clone();
572
573    let prediction = current_prediction.prediction;
574
575    let (additions, deletions) = compute_edit_metrics(&prediction.edits, &prediction.snapshot);
576    let autocomplete_id = prediction.id.to_string();
577
578    let device_id = store
579        .client
580        .user_id()
581        .as_ref()
582        .map(ToString::to_string)
583        .unwrap_or_default();
584
585    let suggestion_type = match current_prediction.shown_with {
586        Some(SuggestionDisplayType::DiffPopover) => SweepSuggestionType::Popup,
587        Some(SuggestionDisplayType::Jump) => return, // should'nt happen
588        Some(SuggestionDisplayType::GhostText) | None => SweepSuggestionType::GhostText,
589    };
590
591    let request_body = AutocompleteMetricsRequest {
592        event_type: SweepEventType::AutocompleteSuggestionAccepted,
593        suggestion_type,
594        additions,
595        deletions,
596        autocomplete_id,
597        edit_tracking: String::new(),
598        edit_tracking_line: None,
599        lifespan: None,
600        debug_info,
601        device_id,
602        privacy_mode_enabled: false,
603    };
604
605    send_autocomplete_metrics_request(cx, store.client.clone(), api_token, request_body);
606}
607
608pub fn edit_prediction_shown(
609    sweep_ai: &SweepAi,
610    client: Arc<Client>,
611    prediction: &EditPrediction,
612    display_type: SuggestionDisplayType,
613    cx: &App,
614) {
615    let Some(api_token) = sweep_ai.api_token.read(cx).key(&SWEEP_CREDENTIALS_URL) else {
616        return;
617    };
618    let debug_info = sweep_ai.debug_info.clone();
619
620    let (additions, deletions) = compute_edit_metrics(&prediction.edits, &prediction.snapshot);
621    let autocomplete_id = prediction.id.to_string();
622
623    let suggestion_type = match display_type {
624        SuggestionDisplayType::GhostText => SweepSuggestionType::GhostText,
625        SuggestionDisplayType::DiffPopover => SweepSuggestionType::Popup,
626        SuggestionDisplayType::Jump => SweepSuggestionType::JumpToEdit,
627    };
628
629    let request_body = AutocompleteMetricsRequest {
630        event_type: SweepEventType::AutocompleteSuggestionShown,
631        suggestion_type,
632        additions,
633        deletions,
634        autocomplete_id,
635        edit_tracking: String::new(),
636        edit_tracking_line: None,
637        lifespan: None,
638        debug_info,
639        device_id: String::new(),
640        privacy_mode_enabled: false,
641    };
642
643    send_autocomplete_metrics_request(cx, client, api_token, request_body);
644}
645
646fn compute_edit_metrics(
647    edits: &[(Range<Anchor>, Arc<str>)],
648    snapshot: &BufferSnapshot,
649) -> (u32, u32) {
650    let mut additions = 0u32;
651    let mut deletions = 0u32;
652
653    for (range, new_text) in edits {
654        let old_text = snapshot.text_for_range(range.clone());
655        deletions += old_text
656            .map(|chunk| chunk.lines().count())
657            .sum::<usize>()
658            .max(1) as u32;
659        additions += new_text.lines().count().max(1) as u32;
660    }
661
662    (additions, deletions)
663}