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                cursor_offset_in_excerpt: request_body.cursor_position,
219                excerpt_start_row: Some(0),
220                excerpt_ranges: zeta_prompt::ExcerptRanges {
221                    editable_150: 0..inputs.snapshot.len(),
222                    editable_180: 0..inputs.snapshot.len(),
223                    editable_350: 0..inputs.snapshot.len(),
224                    editable_150_context_350: 0..inputs.snapshot.len(),
225                    editable_180_context_350: 0..inputs.snapshot.len(),
226                    editable_350_context_150: 0..inputs.snapshot.len(),
227                    ..Default::default()
228                },
229                experiment: None,
230                in_open_source_repo: false,
231                can_collect_data: false,
232            };
233
234            send_started_event(
235                &debug_tx,
236                &buffer,
237                inputs.position,
238                serde_json::to_string(&request_body).unwrap_or_default(),
239            );
240
241            let request = http_client::Request::builder()
242                .uri(SWEEP_API_URL)
243                .header("Content-Type", "application/json")
244                .header("Authorization", format!("Bearer {}", api_token))
245                .header("Connection", "keep-alive")
246                .header("Content-Encoding", "br")
247                .method(Method::POST)
248                .body(body)?;
249
250            let mut response = http_client.send(request).await?;
251
252            let mut body = String::new();
253            response.body_mut().read_to_string(&mut body).await?;
254
255            let response_received_at = Instant::now();
256            if !response.status().is_success() {
257                let message = format!(
258                    "Request failed with status: {:?}\nBody: {}",
259                    response.status(),
260                    body,
261                );
262                send_finished_event(&debug_tx, &buffer, inputs.position, message.clone());
263                bail!(message);
264            };
265
266            let response: AutocompleteResponse = serde_json::from_str(&body)?;
267
268            send_finished_event(&debug_tx, &buffer, inputs.position, body);
269
270            let old_text = inputs
271                .snapshot
272                .text_for_range(response.start_index..response.end_index)
273                .collect::<String>();
274            let edits = language::text_diff(&old_text, &response.completion)
275                .into_iter()
276                .map(|(range, text)| {
277                    (
278                        inputs
279                            .snapshot
280                            .anchor_after(response.start_index + range.start)
281                            ..inputs
282                                .snapshot
283                                .anchor_before(response.start_index + range.end),
284                        text,
285                    )
286                })
287                .collect::<Vec<_>>();
288
289            anyhow::Ok((
290                response.autocomplete_id,
291                edits,
292                inputs.snapshot,
293                response_received_at,
294                ep_inputs,
295            ))
296        });
297
298        let buffer = inputs.buffer.clone();
299
300        cx.spawn(async move |cx| {
301            let (id, edits, old_snapshot, response_received_at, inputs) = result.await?;
302            anyhow::Ok(Some(
303                EditPredictionResult::new(
304                    EditPredictionId(id.into()),
305                    &buffer,
306                    &old_snapshot,
307                    edits.into(),
308                    None,
309                    buffer_snapshotted_at,
310                    response_received_at,
311                    inputs,
312                    None,
313                    cx,
314                )
315                .await,
316            ))
317        })
318    }
319}
320
321fn send_started_event(
322    debug_tx: &Option<mpsc::UnboundedSender<DebugEvent>>,
323    buffer: &Entity<Buffer>,
324    position: Anchor,
325    prompt: String,
326) {
327    if let Some(debug_tx) = debug_tx {
328        _ = debug_tx.unbounded_send(DebugEvent::EditPredictionStarted(
329            EditPredictionStartedDebugEvent {
330                buffer: buffer.downgrade(),
331                position,
332                prompt: Some(prompt),
333            },
334        ));
335    }
336}
337
338fn send_finished_event(
339    debug_tx: &Option<mpsc::UnboundedSender<DebugEvent>>,
340    buffer: &Entity<Buffer>,
341    position: Anchor,
342    model_output: String,
343) {
344    if let Some(debug_tx) = debug_tx {
345        _ = debug_tx.unbounded_send(DebugEvent::EditPredictionFinished(
346            EditPredictionFinishedDebugEvent {
347                buffer: buffer.downgrade(),
348                position,
349                model_output: Some(model_output),
350            },
351        ));
352    }
353}
354
355pub const SWEEP_CREDENTIALS_URL: SharedString =
356    SharedString::new_static("https://autocomplete.sweep.dev");
357pub const SWEEP_CREDENTIALS_USERNAME: &str = "sweep-api-token";
358pub static SWEEP_AI_TOKEN_ENV_VAR: std::sync::LazyLock<EnvVar> = env_var!("SWEEP_AI_TOKEN");
359
360struct GlobalSweepApiKey(Entity<ApiKeyState>);
361
362impl Global for GlobalSweepApiKey {}
363
364pub fn sweep_api_token(cx: &mut App) -> Entity<ApiKeyState> {
365    if let Some(global) = cx.try_global::<GlobalSweepApiKey>() {
366        return global.0.clone();
367    }
368    let entity =
369        cx.new(|_| ApiKeyState::new(SWEEP_CREDENTIALS_URL, SWEEP_AI_TOKEN_ENV_VAR.clone()));
370    cx.set_global(GlobalSweepApiKey(entity.clone()));
371    entity
372}
373
374pub fn load_sweep_api_token(cx: &mut App) -> Task<Result<(), language_model::AuthenticateError>> {
375    sweep_api_token(cx).update(cx, |key_state, cx| {
376        key_state.load_if_needed(SWEEP_CREDENTIALS_URL, |s| s, cx)
377    })
378}
379
380#[derive(Debug, Clone, Serialize)]
381struct AutocompleteRequest {
382    pub debug_info: Arc<str>,
383    pub repo_name: String,
384    pub branch: Option<String>,
385    pub file_path: Arc<Path>,
386    pub file_contents: String,
387    pub recent_changes: String,
388    pub cursor_position: usize,
389    pub original_file_contents: String,
390    pub file_chunks: Vec<FileChunk>,
391    pub retrieval_chunks: Vec<FileChunk>,
392    pub recent_user_actions: Vec<UserAction>,
393    pub multiple_suggestions: bool,
394    pub privacy_mode_enabled: bool,
395    pub changes_above_cursor: bool,
396    pub use_bytes: bool,
397}
398
399#[derive(Debug, Clone, Serialize)]
400struct FileChunk {
401    pub file_path: String,
402    pub start_line: usize,
403    pub end_line: usize,
404    pub content: String,
405    pub timestamp: Option<u64>,
406}
407
408#[derive(Debug, Clone, Serialize)]
409struct UserAction {
410    pub action_type: ActionType,
411    pub line_number: usize,
412    pub offset: usize,
413    pub file_path: String,
414    pub timestamp: u64,
415}
416
417#[derive(Debug, Clone, Copy, PartialEq, Eq, Serialize)]
418#[serde(rename_all = "SCREAMING_SNAKE_CASE")]
419enum ActionType {
420    CursorMovement,
421    InsertChar,
422    DeleteChar,
423    InsertSelection,
424    DeleteSelection,
425}
426
427fn to_sweep_user_action(record: &UserActionRecord, file_path: &str) -> UserAction {
428    UserAction {
429        action_type: match record.action_type {
430            UserActionType::InsertChar => ActionType::InsertChar,
431            UserActionType::InsertSelection => ActionType::InsertSelection,
432            UserActionType::DeleteChar => ActionType::DeleteChar,
433            UserActionType::DeleteSelection => ActionType::DeleteSelection,
434            UserActionType::CursorMovement => ActionType::CursorMovement,
435        },
436        line_number: record.line_number as usize,
437        offset: record.offset,
438        file_path: file_path.to_string(),
439        timestamp: record.timestamp_epoch_ms,
440    }
441}
442
443#[derive(Debug, Clone, Deserialize)]
444struct AutocompleteResponse {
445    pub autocomplete_id: String,
446    pub start_index: usize,
447    pub end_index: usize,
448    pub completion: String,
449    #[allow(dead_code)]
450    pub confidence: f64,
451    #[allow(dead_code)]
452    pub logprobs: Option<serde_json::Value>,
453    #[allow(dead_code)]
454    pub finish_reason: Option<String>,
455    #[allow(dead_code)]
456    pub elapsed_time_ms: u64,
457    #[allow(dead_code)]
458    #[serde(default, rename = "completions")]
459    pub additional_completions: Vec<AdditionalCompletion>,
460}
461
462#[allow(dead_code)]
463#[derive(Debug, Clone, Deserialize)]
464struct AdditionalCompletion {
465    pub start_index: usize,
466    pub end_index: usize,
467    pub completion: String,
468    pub confidence: f64,
469    pub autocomplete_id: String,
470    pub logprobs: Option<serde_json::Value>,
471    pub finish_reason: Option<String>,
472}
473
474fn write_event(event: &zeta_prompt::Event, f: &mut impl fmt::Write) -> fmt::Result {
475    match event {
476        zeta_prompt::Event::BufferChange {
477            old_path,
478            path,
479            diff,
480            ..
481        } => {
482            if old_path != path {
483                // TODO confirm how to do this for sweep
484                // writeln!(f, "User renamed {:?} to {:?}\n", old_path, new_path)?;
485            }
486
487            if !diff.is_empty() {
488                write!(f, "File: {}:\n{}\n", path.display(), diff)?
489            }
490
491            fmt::Result::Ok(())
492        }
493    }
494}
495
496fn debug_info(cx: &gpui::App) -> Arc<str> {
497    format!(
498        "Zed v{version} ({sha}) - OS: {os} - Zed v{version}",
499        version = release_channel::AppVersion::global(cx),
500        sha = release_channel::AppCommitSha::try_global(cx)
501            .map_or("unknown".to_string(), |sha| sha.full()),
502        os = client::telemetry::os_name(),
503    )
504    .into()
505}
506
507#[derive(Debug, Clone, Copy, Serialize)]
508#[serde(rename_all = "snake_case")]
509pub enum SweepEventType {
510    AutocompleteSuggestionShown,
511    AutocompleteSuggestionAccepted,
512}
513
514#[derive(Debug, Clone, Copy, Serialize)]
515#[serde(rename_all = "SCREAMING_SNAKE_CASE")]
516pub enum SweepSuggestionType {
517    GhostText,
518    Popup,
519    JumpToEdit,
520}
521
522#[derive(Debug, Clone, Serialize)]
523struct AutocompleteMetricsRequest {
524    event_type: SweepEventType,
525    suggestion_type: SweepSuggestionType,
526    additions: u32,
527    deletions: u32,
528    autocomplete_id: String,
529    edit_tracking: String,
530    edit_tracking_line: Option<u32>,
531    lifespan: Option<u64>,
532    debug_info: Arc<str>,
533    device_id: String,
534    privacy_mode_enabled: bool,
535}
536
537fn send_autocomplete_metrics_request(
538    cx: &App,
539    client: Arc<Client>,
540    api_token: Arc<str>,
541    request_body: AutocompleteMetricsRequest,
542) {
543    let http_client = client.http_client();
544    cx.background_spawn(async move {
545        let body: AsyncBody = serde_json::to_string(&request_body)?.into();
546
547        let request = http_client::Request::builder()
548            .uri(SWEEP_METRICS_URL)
549            .header("Content-Type", "application/json")
550            .header("Authorization", format!("Bearer {}", api_token))
551            .method(Method::POST)
552            .body(body)?;
553
554        let mut response = http_client.send(request).await?;
555
556        if !response.status().is_success() {
557            let mut body = String::new();
558            response.body_mut().read_to_string(&mut body).await?;
559            anyhow::bail!(
560                "Failed to send autocomplete metrics for sweep_ai: {:?}\nBody: {}",
561                response.status(),
562                body,
563            );
564        }
565
566        Ok(())
567    })
568    .detach_and_log_err(cx);
569}
570
571pub(crate) fn edit_prediction_accepted(
572    store: &EditPredictionStore,
573    current_prediction: CurrentEditPrediction,
574    cx: &App,
575) {
576    let Some(api_token) = store
577        .sweep_ai
578        .api_token
579        .read(cx)
580        .key(&SWEEP_CREDENTIALS_URL)
581    else {
582        return;
583    };
584    let debug_info = store.sweep_ai.debug_info.clone();
585
586    let prediction = current_prediction.prediction;
587
588    let (additions, deletions) = compute_edit_metrics(&prediction.edits, &prediction.snapshot);
589    let autocomplete_id = prediction.id.to_string();
590
591    let device_id = store
592        .client
593        .user_id()
594        .as_ref()
595        .map(ToString::to_string)
596        .unwrap_or_default();
597
598    let suggestion_type = match current_prediction.shown_with {
599        Some(SuggestionDisplayType::DiffPopover) => SweepSuggestionType::Popup,
600        Some(SuggestionDisplayType::Jump) => return, // should'nt happen
601        Some(SuggestionDisplayType::GhostText) | None => SweepSuggestionType::GhostText,
602    };
603
604    let request_body = AutocompleteMetricsRequest {
605        event_type: SweepEventType::AutocompleteSuggestionAccepted,
606        suggestion_type,
607        additions,
608        deletions,
609        autocomplete_id,
610        edit_tracking: String::new(),
611        edit_tracking_line: None,
612        lifespan: None,
613        debug_info,
614        device_id,
615        privacy_mode_enabled: false,
616    };
617
618    send_autocomplete_metrics_request(cx, store.client.clone(), api_token, request_body);
619}
620
621pub fn edit_prediction_shown(
622    sweep_ai: &SweepAi,
623    client: Arc<Client>,
624    prediction: &EditPrediction,
625    display_type: SuggestionDisplayType,
626    cx: &App,
627) {
628    let Some(api_token) = sweep_ai.api_token.read(cx).key(&SWEEP_CREDENTIALS_URL) else {
629        return;
630    };
631    let debug_info = sweep_ai.debug_info.clone();
632
633    let (additions, deletions) = compute_edit_metrics(&prediction.edits, &prediction.snapshot);
634    let autocomplete_id = prediction.id.to_string();
635
636    let suggestion_type = match display_type {
637        SuggestionDisplayType::GhostText => SweepSuggestionType::GhostText,
638        SuggestionDisplayType::DiffPopover => SweepSuggestionType::Popup,
639        SuggestionDisplayType::Jump => SweepSuggestionType::JumpToEdit,
640    };
641
642    let request_body = AutocompleteMetricsRequest {
643        event_type: SweepEventType::AutocompleteSuggestionShown,
644        suggestion_type,
645        additions,
646        deletions,
647        autocomplete_id,
648        edit_tracking: String::new(),
649        edit_tracking_line: None,
650        lifespan: None,
651        debug_info,
652        device_id: String::new(),
653        privacy_mode_enabled: false,
654    };
655
656    send_autocomplete_metrics_request(cx, client, api_token, request_body);
657}
658
659fn compute_edit_metrics(
660    edits: &[(Range<Anchor>, Arc<str>)],
661    snapshot: &BufferSnapshot,
662) -> (u32, u32) {
663    let mut additions = 0u32;
664    let mut deletions = 0u32;
665
666    for (range, new_text) in edits {
667        let old_text = snapshot.text_for_range(range.clone());
668        deletions += old_text
669            .map(|chunk| chunk.lines().count())
670            .sum::<usize>()
671            .max(1) as u32;
672        additions += new_text.lines().count().max(1) as u32;
673    }
674
675    (additions, deletions)
676}