| 1 | //! The events service: the bus every state change in g1t is published on, |
| 2 | //! and its durable log. |
| 3 | //! |
| 4 | //! Publishing puts events on a queue and returns. The queue consumer writes |
| 5 | //! them to the log and passes each batch on to each subscriber's own |
| 6 | //! queue, so a slow or failing subscriber holds up nobody else. Each |
| 7 | //! subscriber is sent only the types it acts on |
| 8 | //! (`g1t_contracts::subscribers`), in batches the queues take (fanout.rs). |
| 9 | //! |
| 10 | //! It keeps two more things beside the log: the audit log (audit.rs) and |
| 11 | //! each person's inbox (inbox.rs), written as events arrive, with who |
| 12 | //! follows what (subscriptions.rs). |
| 13 | //! |
| 14 | //! Other services reach it over `POST /rpc/<method>`; see |
| 15 | //! `g1t_contracts::events`, `audit` and `inbox` for the methods and their |
| 16 | //! arguments. `activity_digest` (digest.rs) counts a span of the log for |
| 17 | //! Home. Security alerts that are not urgent wait for each person's |
| 18 | //! security digest (security_digest.rs), sent by a cron of its own. |
| 19 | //! |
| 20 | //! Repository pages open in someone's browser refresh themselves: each |
| 21 | //! batch's events that such a page shows are passed to the notify service |
| 22 | //! (live.rs), which tells those watching the repository. |
| 23 | |
| 24 | mod audit; |
| 25 | mod digest; |
| 26 | mod fanout; |
| 27 | mod inbox; |
| 28 | mod live; |
| 29 | #[cfg(test)] |
| 30 | mod schema; |
| 31 | mod security_digest; |
| 32 | mod subscriptions; |
| 33 | mod work_items; |
| 34 | |
| 35 | use g1t_contracts::events::{Event, ListArgs, MAX_LIST_REPOS, PublishArgs}; |
| 36 | use g1t_contracts::new_id; |
| 37 | use g1t_contracts::time::rfc3339; |
| 38 | use g1t_kit::{args, js, now_ms, reply, rpc_method}; |
| 39 | use serde::Deserialize; |
| 40 | use std::collections::{HashMap, HashSet}; |
| 41 | use worker::js_sys::{Array, Object}; |
| 42 | use worker::wasm_bindgen::{JsCast, JsValue}; |
| 43 | use worker::{Context, D1Database, Env, MessageBatch, Request, Response, Result, event}; |
| 44 | |
| 45 | const DEFAULT_PAGE: u32 = 50; |
| 46 | const MAX_PAGE: u32 = 200; |
| 47 | /// Statuses and check runs reported on commits: delivered to webhooks, and |
| 48 | /// left out of a timeline unless asked for by type. |
| 49 | const REPORTING: [&str; 7] = [ |
| 50 | "status.created", |
| 51 | "check_run.created", |
| 52 | "check_run.completed", |
| 53 | "check_run.rerequested", |
| 54 | "check_run.requested_action", |
| 55 | "check_suite.completed", |
| 56 | "check_suite.rerequested", |
| 57 | ]; |
| 58 | /// Every binding whose name starts with this is a queue that receives |
| 59 | /// events: one per subscribing service, sent the types it routes. |
| 60 | const SUBSCRIBER_PREFIX: &str = "SUBSCRIBER_"; |
| 61 | /// How long a record of which queues a batch reached is kept for its retries. |
| 62 | const FANOUT_KEEP_MS: u64 = 24 * 60 * 60 * 1000; |
| 63 | |
| 64 | /// An event as the log's table holds it. |
| 65 | #[derive(Deserialize)] |
| 66 | pub(crate) struct EventRow { |
| 67 | id: String, |
| 68 | #[serde(rename = "type")] |
| 69 | kind: String, |
| 70 | source: String, |
| 71 | time: String, |
| 72 | repo_id: Option<String>, |
| 73 | actor: Option<String>, |
| 74 | /// JSON. |
| 75 | data: String, |
| 76 | } |
| 77 | |
| 78 | impl From<EventRow> for Event { |
| 79 | fn from(row: EventRow) -> Self { |
| 80 | Event { |
| 81 | id: row.id, |
| 82 | kind: row.kind, |
| 83 | source: row.source, |
| 84 | time: row.time, |
| 85 | repo_id: row.repo_id, |
| 86 | actor: row.actor, |
| 87 | data: serde_json::from_str(&row.data).unwrap_or_default(), |
| 88 | } |
| 89 | } |
| 90 | } |
| 91 | |
| 92 | fn optional(value: &Option<String>) -> JsValue { |
| 93 | value.as_deref().map_or(JsValue::NULL, JsValue::from) |
| 94 | } |
| 95 | |
| 96 | #[derive(Deserialize)] |
| 97 | struct FanoutRow { |
| 98 | event_id: String, |
| 99 | /// JSON array of binding names. |
| 100 | bindings: String, |
| 101 | } |
| 102 | |
| 103 | /// Sends `events` to a queue binding, one message each, in as many |
| 104 | /// `sendBatch` calls as the queue's limits need (fanout.rs). Events are |
| 105 | /// expected to have been fit to a message. |
| 106 | async fn send(queue: &JsValue, events: &[&Event]) -> Result<()> { |
| 107 | for chunk in fanout::chunks(events) { |
| 108 | let messages = Array::new(); |
| 109 | for event in chunk { |
| 110 | let message = Object::new(); |
| 111 | js::set(&message, "body", &js::to_js(event)?); |
| 112 | messages.push(&message); |
| 113 | } |
| 114 | js::call(queue, "sendBatch", &[messages.into()]).await?; |
| 115 | } |
| 116 | Ok(()) |
| 117 | } |
| 118 | |
| 119 | struct Events { |
| 120 | db: D1Database, |
| 121 | env: Env, |
| 122 | } |
| 123 | |
| 124 | impl Events { |
| 125 | /// Assigns each event its id and time and puts it on the bus. |
| 126 | async fn publish(&self, a: PublishArgs) -> Result<()> { |
| 127 | if a.events.is_empty() { |
| 128 | return Ok(()); |
| 129 | } |
| 130 | let now = now_ms(); |
| 131 | let events: Vec<Event> = a |
| 132 | .events |
| 133 | .into_iter() |
| 134 | .map(|event| { |
| 135 | let mut event = Event { |
| 136 | id: new_id("evt", now), |
| 137 | kind: event.kind, |
| 138 | source: event.source, |
| 139 | time: rfc3339(now), |
| 140 | repo_id: event.repo_id, |
| 141 | actor: event.actor, |
| 142 | data: event.data, |
| 143 | }; |
| 144 | // Too large for one message: its long text is shortened. |
| 145 | fanout::fit(&mut event); |
| 146 | event |
| 147 | }) |
| 148 | .collect(); |
| 149 | let events: Vec<&Event> = events.iter().collect(); |
| 150 | send(&js::binding(&self.env, "BUS")?, &events).await |
| 151 | } |
| 152 | |
| 153 | /// Newest first, or with `after` oldest first in the order the log |
| 154 | /// wrote them. Callers must have checked that the viewer may see the |
| 155 | /// repository asked about. |
| 156 | async fn list(&self, a: ListArgs) -> Result<Vec<Event>> { |
| 157 | let (filter, binds) = list_filter(&a); |
| 158 | let mut values: Vec<JsValue> = binds |
| 159 | .into_iter() |
| 160 | .map(|bind| match bind { |
| 161 | Bind::Text(text) => JsValue::from(text), |
| 162 | Bind::Number(number) => JsValue::from(number), |
| 163 | }) |
| 164 | .collect(); |
| 165 | values.push(a.limit.unwrap_or(DEFAULT_PAGE).min(MAX_PAGE).into()); |
| 166 | let order = if a.after.is_some() { "seq ASC" } else { "id DESC" }; |
| 167 | let rows = self |
| 168 | .db |
| 169 | .prepare(format!( |
| 170 | "SELECT * FROM events {filter} ORDER BY {order} LIMIT ?" |
| 171 | )) |
| 172 | .bind(&values)? |
| 173 | .all() |
| 174 | .await? |
| 175 | .results::<EventRow>()?; |
| 176 | Ok(rows.into_iter().map(Event::from).collect()) |
| 177 | } |
| 178 | |
| 179 | /// Writes a batch from the bus to the log, notes who now follows what |
| 180 | /// (inbox.rs), hands each subscriber the events it routes, then tells |
| 181 | /// the people it concerns. |
| 182 | /// |
| 183 | /// Everything before the hand-off is safe to repeat. A retried batch |
| 184 | /// skips the queues that already have it: when a hand-off fails part |
| 185 | /// way, which queues each event reached is written down first. |
| 186 | async fn deliver(&self, events: &[Event]) -> Result<()> { |
| 187 | let mut statements = Vec::with_capacity(events.len() + 1); |
| 188 | for event in events { |
| 189 | statements.push( |
| 190 | self.db |
| 191 | .prepare( |
| 192 | // Redelivered batches must not duplicate log rows. |
| 193 | // `seq` is the order of writing, which `after` follows |
| 194 | // (0010_events_seq.sql): the log has one writer. |
| 195 | "INSERT OR IGNORE INTO events (id, type, source, time, repo_id, actor, data, seq) |
| 196 | VALUES (?, ?, ?, ?, ?, ?, ?, (SELECT COALESCE(MAX(seq), 0) + 1 FROM events))", |
| 197 | ) |
| 198 | .bind(&[ |
| 199 | event.id.as_str().into(), |
| 200 | event.kind.as_str().into(), |
| 201 | event.source.as_str().into(), |
| 202 | event.time.as_str().into(), |
| 203 | optional(&event.repo_id), |
| 204 | optional(&event.actor), |
| 205 | serde_json::to_string(&event.data)?.into(), |
| 206 | ])?, |
| 207 | ); |
| 208 | } |
| 209 | let ids: Vec<&str> = events.iter().map(|event| event.id.as_str()).collect(); |
| 210 | // Read in the same round trip: nearly always nothing. |
| 211 | statements.push( |
| 212 | self.db |
| 213 | .prepare("SELECT event_id, bindings FROM fanout_sent WHERE event_id IN (SELECT value FROM json_each(?))") |
| 214 | .bind(&[serde_json::to_string(&ids)?.into()])?, |
| 215 | ); |
| 216 | let results = self.db.batch(statements).await?; |
| 217 | let mut sent: HashMap<String, HashSet<String>> = HashMap::new(); |
| 218 | if let Some(rows) = results.last() { |
| 219 | for row in rows.results::<FanoutRow>()? { |
| 220 | let bindings: Vec<String> = serde_json::from_str(&row.bindings).unwrap_or_default(); |
| 221 | sent.insert(row.event_id, bindings.into_iter().collect()); |
| 222 | } |
| 223 | } |
| 224 | audit::follow_renames(&self.db, events).await?; |
| 225 | // Before the hand-off, so its failing does not send the batch to |
| 226 | // every queue again. |
| 227 | inbox::follow(&self.db, events).await?; |
| 228 | |
| 229 | let bindings: &JsValue = self.env.as_ref(); |
| 230 | let mut reached: Vec<String> = Vec::new(); |
| 231 | let mut failed = None; |
| 232 | for name in Object::keys(bindings.unchecked_ref::<Object>()).iter() { |
| 233 | let Some(name) = name |
| 234 | .as_string() |
| 235 | .filter(|name| name.starts_with(SUBSCRIBER_PREFIX)) |
| 236 | else { |
| 237 | continue; |
| 238 | }; |
| 239 | let routed: Vec<&Event> = events |
| 240 | .iter() |
| 241 | .filter(|event| g1t_contracts::subscribers::routed(&name, &event.kind)) |
| 242 | .filter(|event| !sent.get(&event.id).is_some_and(|reached| reached.contains(&name))) |
| 243 | .collect(); |
| 244 | if routed.is_empty() { |
| 245 | continue; |
| 246 | } |
| 247 | match send(&js::binding(&self.env, &name)?, &routed).await { |
| 248 | Ok(()) => reached.push(name), |
| 249 | Err(error) => { |
| 250 | worker::console_error!("events: passing {} events to {name} failed: {error}", routed.len()); |
| 251 | failed = Some(error); |
| 252 | } |
| 253 | } |
| 254 | } |
| 255 | if let Some(error) = failed { |
| 256 | self.note_reached(events, &sent, &reached).await?; |
| 257 | return Err(error); |
| 258 | } |
| 259 | let (work, repos, identity) = ( |
| 260 | self.env.service("WORK")?, |
| 261 | self.env.service("REPOS")?, |
| 262 | self.env.service("IDENTITY")?, |
| 263 | ); |
| 264 | // Optional: without it the inbox still fills, nobody is told live. |
| 265 | let notify = self.env.service("NOTIFY").ok(); |
| 266 | // Pages open on the repositories: told first, as nobody waits longer. |
| 267 | if let Some(notify) = notify.as_ref() { |
| 268 | live::tell(notify, events).await; |
| 269 | } |
| 270 | let sources = inbox::Sources { |
| 271 | work: &work, |
| 272 | repos: &repos, |
| 273 | identity: &identity, |
| 274 | notify: notify.as_ref(), |
| 275 | }; |
| 276 | inbox::deliver(&self.db, &sources, events).await; |
| 277 | Ok(()) |
| 278 | } |
| 279 | |
| 280 | /// Writes down which queues each event of a batch has reached, with |
| 281 | /// those it had reached before, for the batch's retry. |
| 282 | async fn note_reached( |
| 283 | &self, |
| 284 | events: &[Event], |
| 285 | sent: &HashMap<String, HashSet<String>>, |
| 286 | reached: &[String], |
| 287 | ) -> Result<()> { |
| 288 | let now = rfc3339(now_ms()); |
| 289 | let mut statements = Vec::with_capacity(events.len()); |
| 290 | for event in events { |
| 291 | let mut bindings: Vec<&str> = reached.iter().map(String::as_str).collect(); |
| 292 | if let Some(before) = sent.get(&event.id) { |
| 293 | bindings.extend(before.iter().map(String::as_str)); |
| 294 | } |
| 295 | bindings.sort_unstable(); |
| 296 | bindings.dedup(); |
| 297 | statements.push( |
| 298 | self.db |
| 299 | .prepare( |
| 300 | "INSERT INTO fanout_sent (event_id, bindings, created_at) VALUES (?, ?, ?) |
| 301 | ON CONFLICT (event_id) DO UPDATE SET bindings = excluded.bindings", |
| 302 | ) |
| 303 | .bind(&[event.id.as_str().into(), serde_json::to_string(&bindings)?.into(), now.as_str().into()])?, |
| 304 | ); |
| 305 | } |
| 306 | self.db.batch(statements).await?; |
| 307 | Ok(()) |
| 308 | } |
| 309 | } |
| 310 | |
| 311 | #[event(fetch)] |
| 312 | async fn fetch(mut request: Request, env: Env, _ctx: Context) -> Result<Response> { |
| 313 | let Some(method) = rpc_method(&request) else { |
| 314 | return Response::error("Not found", 404); |
| 315 | }; |
| 316 | let body: serde_json::Value = request.json().await?; |
| 317 | let events = Events { |
| 318 | db: env.d1("DB")?, |
| 319 | env, |
| 320 | }; |
| 321 | match method.as_str() { |
| 322 | "publish" => reply(&events.publish(args(body)?).await?), |
| 323 | "list" => reply(&events.list(args(body)?).await?), |
| 324 | // Home's digest of a span: counted, by who did it (digest.rs). |
| 325 | "activity_digest" => reply(&digest::activity_digest(&events.db, args(body)?).await?), |
| 326 | "audit_record" => reply(&audit::record(&events.db, args(body)?).await?), |
| 327 | "audit_list" => reply(&audit::list(&events.db, args(body)?).await?), |
| 328 | "inbox_list" => { |
| 329 | let repos = events.env.service("REPOS")?; |
| 330 | reply(&inbox::list(&events.db, &repos, args(body)?).await?) |
| 331 | } |
| 332 | "inbox_counts" => reply(&inbox::counts(&events.db, args(body)?).await?), |
| 333 | "inbox_mark" => { |
| 334 | let marked: g1t_contracts::inbox::MarkInboxArgs = args(body)?; |
| 335 | let username = marked.username.clone(); |
| 336 | let changed = inbox::mark(&events.db, marked).await?; |
| 337 | // Notify: the new count in every open tab, whoever marked (the site, the API, MCP). |
| 338 | if changed > 0 |
| 339 | && let Ok(notify) = events.env.service("NOTIFY") |
| 340 | { |
| 341 | inbox::tell_inbox_count(&events.db, ¬ify, &username).await; |
| 342 | } |
| 343 | reply(&changed) |
| 344 | } |
| 345 | "inbox_thread" => { |
| 346 | let (repos, work) = (events.env.service("REPOS")?, events.env.service("WORK")?); |
| 347 | reply(&inbox::thread(&events.db, &repos, &work, args(body)?).await?) |
| 348 | } |
| 349 | "inbox_subscription" => { |
| 350 | let work = events.env.service("WORK")?; |
| 351 | reply(&subscriptions::subscription(&events.db, &work, args(body)?).await?) |
| 352 | } |
| 353 | "inbox_subscribe" => { |
| 354 | let work = events.env.service("WORK")?; |
| 355 | reply(&subscriptions::subscribe(&events.db, &work, args(body)?).await?) |
| 356 | } |
| 357 | "inbox_watching" => reply(&subscriptions::watching(&events.db, args(body)?).await?), |
| 358 | "inbox_watch" => reply(&subscriptions::watch(&events.db, args(body)?).await?), |
| 359 | "inbox_watchers" => reply(&subscriptions::watchers_count(&events.db, args(body)?).await?), |
| 360 | "inbox_watched" => reply(&subscriptions::watched(&events.db, args(body)?).await?), |
| 361 | "inbox_settings" => reply(&subscriptions::settings(&events.db, args(body)?).await?), |
| 362 | "inbox_update_settings" => reply(&subscriptions::update_settings(&events.db, args(body)?).await?), |
| 363 | _ => Response::error("Unknown method", 404), |
| 364 | } |
| 365 | } |
| 366 | |
| 367 | /// Events from the bus. A batch that fails is retried whole, which is safe |
| 368 | /// because the log ignores an event it has seen, and queues that already |
| 369 | /// have it are skipped (`deliver`). After its retries it goes to the |
| 370 | /// dead-letter queue (wrangler.jsonc). |
| 371 | #[event(queue)] |
| 372 | async fn queue(batch: MessageBatch<Event>, env: Env, _ctx: Context) -> Result<()> { |
| 373 | let events = Events { |
| 374 | db: env.d1("DB")?, |
| 375 | env, |
| 376 | }; |
| 377 | let delivered: Vec<Event> = batch |
| 378 | .messages()? |
| 379 | .into_iter() |
| 380 | .map(|message| message.into_body()) |
| 381 | .collect(); |
| 382 | events.deliver(&delivered).await?; |
| 383 | batch.ack_all(); |
| 384 | Ok(()) |
| 385 | } |
| 386 | |
| 387 | /// Once a day, old audit entries are removed: first everything older than |
| 388 | /// any workspace keeps (`AUDIT_MAX_DAYS`, 400 days), then each workspace's |
| 389 | /// entries older than its own plan keeps, as billing says (7 days free, 90 |
| 390 | /// on the plan, or what staff set). Workspaces with nothing older than the |
| 391 | /// shortest (`AUDIT_MIN_DAYS`, 7) are left alone. |
| 392 | #[event(scheduled)] |
| 393 | async fn scheduled(event: worker::ScheduledEvent, env: Env, _ctx: worker::ScheduleContext) { |
| 394 | // The morning cron sends the security digests that are due, and nothing else. |
| 395 | if event.cron() == security_digest::DIGEST_CRON { |
| 396 | let (Ok(db), Ok(identity)) = (env.d1("DB"), env.service("IDENTITY")) else { return }; |
| 397 | let notify = env.service("NOTIFY").ok(); |
| 398 | match security_digest::send_due(&db, &identity, notify.as_ref(), now_ms()).await { |
| 399 | Ok(sent) if sent > 0 => worker::console_log!("sent {sent} security digests"), |
| 400 | Ok(_) => {} |
| 401 | Err(error) => worker::console_error!("security digests not sent: {error}"), |
| 402 | } |
| 403 | return; |
| 404 | } |
| 405 | let days = |name: &str, default: u32| { |
| 406 | env.var(name) |
| 407 | .ok() |
| 408 | .and_then(|v| v.to_string().trim().parse::<u32>().ok()) |
| 409 | .filter(|days| *days > 0) |
| 410 | .unwrap_or(default) |
| 411 | }; |
| 412 | let max_days = days("AUDIT_MAX_DAYS", audit::DEFAULT_MAX_DAYS); |
| 413 | let min_days = days("AUDIT_MIN_DAYS", audit::DEFAULT_MIN_DAYS).min(max_days); |
| 414 | let Ok(db) = env.d1("DB") else { return }; |
| 415 | let now = now_ms(); |
| 416 | let cutoff = rfc3339(now.saturating_sub(FANOUT_KEEP_MS)); |
| 417 | let purged = match db.prepare("DELETE FROM fanout_sent WHERE created_at < ?").bind(&[cutoff.into()]) { |
| 418 | Ok(statement) => statement.run().await.map(|_| ()), |
| 419 | Err(error) => Err(error), |
| 420 | }; |
| 421 | if let Err(error) = purged { |
| 422 | worker::console_error!("could not remove old fan-out records: {error}"); |
| 423 | } |
| 424 | match inbox::purge(&db, now).await { |
| 425 | Ok(removed) if removed > 0 => worker::console_log!("removed {removed} old inbox items"), |
| 426 | Ok(_) => {} |
| 427 | Err(error) => worker::console_error!("could not remove old inbox items: {error}"), |
| 428 | } |
| 429 | match audit::purge(&db, &audit::keep_from(now, max_days), 20).await { |
| 430 | Ok(removed) if removed > 0 => { |
| 431 | worker::console_log!("removed {removed} audit entries older than {max_days} days") |
| 432 | } |
| 433 | Ok(_) => {} |
| 434 | Err(error) => worker::console_error!("could not remove old audit entries: {error}"), |
| 435 | } |
| 436 | // Without billing nobody's plan is known, so nothing younger than the |
| 437 | // ceiling is removed. |
| 438 | let billing = match env.service("BILLING") { |
| 439 | Ok(billing) => billing, |
| 440 | Err(error) => { |
| 441 | worker::console_error!( |
| 442 | "audit entries kept past their plan's days: no billing: {error}" |
| 443 | ); |
| 444 | return; |
| 445 | } |
| 446 | }; |
| 447 | match audit::purge_by_plan(&db, &billing, now, min_days, max_days).await { |
| 448 | Ok(removed) if removed > 0 => { |
| 449 | worker::console_log!("removed {removed} audit entries older than their plan keeps") |
| 450 | } |
| 451 | Ok(_) => {} |
| 452 | Err(error) => worker::console_error!("audit entries kept past their plan's days: {error}"), |
| 453 | } |
| 454 | } |
| 455 | |
| 456 | /// The condition of an `after` cursor, binding its id twice. |
| 457 | const AFTER: &str = concat!( |
| 458 | "seq > COALESCE(", |
| 459 | "(SELECT seq FROM events WHERE id = ?), ", |
| 460 | "(SELECT seq FROM events WHERE id > ? AND seq IS NOT NULL ORDER BY id LIMIT 1) - 1, ", |
| 461 | "(SELECT MAX(seq) FROM events), 0)" |
| 462 | ); |
| 463 | |
| 464 | /// A value bound to a `?` in [`list_filter`]'s clause. |
| 465 | #[derive(Debug, PartialEq)] |
| 466 | enum Bind { |
| 467 | Text(String), |
| 468 | Number(f64), |
| 469 | } |
| 470 | |
| 471 | /// The `WHERE` clause `list` reads with, and what it binds, in order. |
| 472 | fn list_filter(a: &ListArgs) -> (String, Vec<Bind>) { |
| 473 | let mut conditions = Vec::new(); |
| 474 | let mut values = Vec::new(); |
| 475 | if let Some(repo_id) = &a.repo_id { |
| 476 | conditions.push("repo_id = ?".to_owned()); |
| 477 | values.push(Bind::Text(repo_id.clone())); |
| 478 | } |
| 479 | if !a.repo_ids.is_empty() { |
| 480 | // A workspace's activity across its repositories: one list, bound |
| 481 | // as JSON so the statement stays small however many there are. |
| 482 | let ids: Vec<&String> = a.repo_ids.iter().take(MAX_LIST_REPOS).collect(); |
| 483 | conditions.push("repo_id IN (SELECT value FROM json_each(?))".to_owned()); |
| 484 | values.push(Bind::Text(serde_json::to_string(&ids).unwrap_or_else(|_| "[]".to_owned()))); |
| 485 | } |
| 486 | if !a.types.is_empty() { |
| 487 | let marks = vec!["?"; a.types.len()].join(", "); |
| 488 | conditions.push(format!("type IN ({marks})")); |
| 489 | values.extend(a.types.iter().map(|kind| Bind::Text(kind.clone()))); |
| 490 | } else { |
| 491 | // What CI and integrations report on commits goes to webhooks, |
| 492 | // and is read from each commit's checks; a timeline asked for |
| 493 | // everything would be little else on a busy repository. |
| 494 | let marks = vec!["?"; REPORTING.len()].join(", "); |
| 495 | conditions.push(format!("type NOT IN ({marks})")); |
| 496 | values.extend(REPORTING.iter().map(|kind| Bind::Text((*kind).to_owned()))); |
| 497 | } |
| 498 | if let Some(actor) = &a.actor { |
| 499 | conditions.push("actor = ?".to_owned()); |
| 500 | values.push(Bind::Text(actor.clone())); |
| 501 | } |
| 502 | if !a.numbers.is_empty() { |
| 503 | // A work item's or pull request's own events name it as `number`; a |
| 504 | // comment, review or link on it names it as `item`. |
| 505 | let marks = vec!["?"; a.numbers.len()].join(", "); |
| 506 | conditions.push(format!( |
| 507 | "(json_extract(data, '$.number') IN ({marks}) OR json_extract(data, '$.item') IN ({marks}))" |
| 508 | )); |
| 509 | for _ in 0..2 { |
| 510 | values.extend(a.numbers.iter().map(|number| Bind::Number(f64::from(*number)))); |
| 511 | } |
| 512 | } |
| 513 | if let Some(since) = &a.since { |
| 514 | conditions.push("time >= ?".to_owned()); |
| 515 | values.push(Bind::Text(since.clone())); |
| 516 | } |
| 517 | if let Some(before) = &a.before { |
| 518 | conditions.push("id < ?".to_owned()); |
| 519 | values.push(Bind::Text(before.clone())); |
| 520 | } |
| 521 | if let Some(after) = &a.after { |
| 522 | // What the log wrote after the cursor's event. A cursor the log |
| 523 | // does not have (yet: an id from a webhook can arrive before its |
| 524 | // row is written) starts at the first event with a later id, and |
| 525 | // one past them all at the end of the log. |
| 526 | conditions.push(AFTER.to_owned()); |
| 527 | values.push(Bind::Text(after.clone())); |
| 528 | values.push(Bind::Text(after.clone())); |
| 529 | } |
| 530 | let filter = if conditions.is_empty() { |
| 531 | String::new() |
| 532 | } else { |
| 533 | format!("WHERE {}", conditions.join(" AND ")) |
| 534 | }; |
| 535 | (filter, values) |
| 536 | } |
| 537 | |
| 538 | #[cfg(test)] |
| 539 | mod tests { |
| 540 | use super::*; |
| 541 | |
| 542 | #[test] |
| 543 | fn a_plain_list_leaves_out_what_is_reported_on_commits() { |
| 544 | let (filter, binds) = list_filter(&ListArgs { repo_id: Some("rep_1".into()), ..ListArgs::default() }); |
| 545 | assert!(filter.starts_with("WHERE repo_id = ? AND type NOT IN (")); |
| 546 | assert_eq!(binds[0], Bind::Text("rep_1".into())); |
| 547 | assert_eq!(binds.len(), 1 + REPORTING.len()); |
| 548 | } |
| 549 | |
| 550 | #[test] |
| 551 | fn a_workspaces_repositories_are_read_as_one_list() { |
| 552 | let many: Vec<String> = (0..150).map(|n| format!("rep_{n}")).collect(); |
| 553 | let (filter, binds) = list_filter(&ListArgs { |
| 554 | repo_ids: many, |
| 555 | types: vec!["pull.merged".into(), "release.published".into()], |
| 556 | since: Some("2026-10-12T00:00:00Z".into()), |
| 557 | ..ListArgs::default() |
| 558 | }); |
| 559 | assert_eq!(filter, "WHERE repo_id IN (SELECT value FROM json_each(?)) AND type IN (?, ?) AND time >= ?"); |
| 560 | let Bind::Text(ids) = &binds[0] else { panic!("the ids are one JSON text") }; |
| 561 | let ids: Vec<String> = serde_json::from_str(ids).unwrap(); |
| 562 | assert_eq!(ids.len(), MAX_LIST_REPOS); |
| 563 | assert_eq!(ids[0], "rep_0"); |
| 564 | } |
| 565 | |
| 566 | #[test] |
| 567 | fn numbers_match_an_item_or_what_is_said_on_it_since_a_time() { |
| 568 | let (filter, binds) = list_filter(&ListArgs { |
| 569 | repo_id: Some("rep_1".into()), |
| 570 | types: vec!["work_item.opened".into()], |
| 571 | numbers: vec![4, 9], |
| 572 | since: Some("2026-10-01T00:00:00Z".into()), |
| 573 | actor: Some("usr_g1t_agent".into()), |
| 574 | ..ListArgs::default() |
| 575 | }); |
| 576 | assert_eq!( |
| 577 | filter, |
| 578 | "WHERE repo_id = ? AND type IN (?) AND actor = ? AND \ |
| 579 | (json_extract(data, '$.number') IN (?, ?) OR json_extract(data, '$.item') IN (?, ?)) AND time >= ?" |
| 580 | ); |
| 581 | assert_eq!( |
| 582 | binds, |
| 583 | vec![ |
| 584 | Bind::Text("rep_1".into()), |
| 585 | Bind::Text("work_item.opened".into()), |
| 586 | Bind::Text("usr_g1t_agent".into()), |
| 587 | Bind::Number(4.0), |
| 588 | Bind::Number(9.0), |
| 589 | Bind::Number(4.0), |
| 590 | Bind::Number(9.0), |
| 591 | Bind::Text("2026-10-01T00:00:00Z".into()), |
| 592 | ] |
| 593 | ); |
| 594 | } |
| 595 | |
| 596 | #[test] |
| 597 | fn an_after_cursor_reads_what_the_log_wrote_later_with_every_other_filter() { |
| 598 | let (filter, binds) = list_filter(&ListArgs { |
| 599 | repo_id: Some("rep_1".into()), |
| 600 | types: vec!["work.claimed".into(), "integration.failed".into()], |
| 601 | numbers: vec![7], |
| 602 | since: Some("2026-10-12T00:00:00Z".into()), |
| 603 | after: Some("evt_9".into()), |
| 604 | ..ListArgs::default() |
| 605 | }); |
| 606 | assert_eq!( |
| 607 | filter, |
| 608 | format!( |
| 609 | "WHERE repo_id = ? AND type IN (?, ?) AND \ |
| 610 | (json_extract(data, '$.number') IN (?) OR json_extract(data, '$.item') IN (?)) AND time >= ? AND {AFTER}" |
| 611 | ) |
| 612 | ); |
| 613 | // The cursor's id is bound once to find its place, and once for a |
| 614 | // cursor the log does not have. |
| 615 | assert_eq!(&binds[binds.len() - 2..], &[Bind::Text("evt_9".into()), Bind::Text("evt_9".into())]); |
| 616 | assert_eq!(binds.len(), 8); |
| 617 | assert_eq!(AFTER.matches('?').count(), 2); |
| 618 | } |
| 619 | } |