1use std::collections::BTreeMap;
15use std::io::Write;
16use std::path::{Path, PathBuf};
17
18use serde::{Deserialize, Serialize};
19
20pub const TURN_END: &str = "turn_end";
22pub const GATE_SKIP: &str = "gate_skip";
24pub const CHECK: &str = "check";
26pub const BLOCK: &str = "block";
28pub const LIVE_CHECK: &str = "live_check";
31pub const LIVE_ASK: &str = "live_ask";
34pub const EARLY: &str = "early";
36pub const QUEUED_FREE: &str = "queued_free";
38pub const CHANGED: &str = "changed";
40pub const IGNORED: &str = "ignored";
41pub const CORRECTED: &str = "corrected";
42pub const NOT_CORRECTED: &str = "not_corrected";
43pub const REGRETTED: &str = "regretted";
46pub const TESTED: &str = "tested";
49pub const CORRECTION: &str = "correction";
54pub const ANSWERED: &str = "answer";
57pub const VETO: &str = "veto";
60pub const VETO_REPEAT: &str = "veto_repeat";
63pub const LESSON_NEW: &str = "lesson_new";
65pub const LESSON_MERGED: &str = "lesson_merged";
66pub const REVIEW_FINDING: &str = "review_finding";
69pub const NOT_A_LESSON: &str = "not_a_lesson";
71pub const MODEL_CALL: &str = "model_call";
74pub const STEER_ISSUED: &str = "steer_issued";
77pub const STEER_DELIVERED: &str = "steer_delivered";
78pub const STEER_DROPPED: &str = "steer_dropped";
79pub const PAUSE_REFUSED: &str = "pause_refused";
80pub const MSG_SENT: &str = "msg_sent";
84pub const MSG_DELIVERED: &str = "msg_delivered";
85pub const MSG_DROPPED: &str = "msg_dropped";
86pub const QUEEN_QUEUED: &str = "queen_queued";
89pub const QUEEN_DELIVERED: &str = "queen_delivered";
90pub const QUEEN_ANSWERED: &str = "queen_answered";
91pub const QUEEN_SKIPPED: &str = "queen_skipped";
92pub const QUEEN_DISMISSED: &str = "queen_dismissed";
93pub const QUEEN_NO: &str = "queen_no";
97pub const QUEEN_STAMP: &str = "queen_stamp";
98pub const QUEEN_MISSED: &str = "queen_missed";
99pub const ASSIMILATED: &str = "assimilated";
101
102#[derive(Debug, Clone, Default, PartialEq, Serialize, Deserialize)]
103pub struct Event {
104 pub ms: i64,
105 pub kind: String,
106 #[serde(default, skip_serializing_if = "String::is_empty")]
107 pub session: String,
108 #[serde(default, skip_serializing_if = "String::is_empty")]
109 pub engineer: String,
110 #[serde(default, skip_serializing_if = "String::is_empty")]
111 pub repo: String,
112 #[serde(default, skip_serializing_if = "Vec::is_empty")]
113 pub lessons: Vec<String>,
114 #[serde(default, skip_serializing_if = "is_zero")]
115 pub n: u64,
116 #[serde(default, skip_serializing_if = "is_zero_f")]
117 pub usd: f64,
118 #[serde(default, skip_serializing_if = "is_zero")]
119 pub dur_ms: u64,
120 #[serde(default, skip_serializing_if = "String::is_empty")]
121 pub why: String,
122 #[serde(default, skip_serializing_if = "String::is_empty")]
124 pub text: String,
125 #[serde(default, skip_serializing_if = "String::is_empty")]
128 pub mode: String,
129 #[serde(default, skip_serializing_if = "String::is_empty")]
133 pub channel: String,
134 #[serde(default, skip_serializing_if = "String::is_empty")]
138 pub source: String,
139}
140
141pub fn source(harness: &str, version: &str) -> String {
144 format!("{harness} {version}").trim().to_string()
145}
146
147fn is_zero(v: &u64) -> bool {
148 *v == 0
149}
150
151fn is_zero_f(v: &f64) -> bool {
152 *v == 0.0
153}
154
155impl Event {
156 pub fn new(kind: &str, ms: i64) -> Self {
158 Event { kind: kind.into(), ms, ..Default::default() }
159 }
160}
161
162#[derive(Debug, Clone, Default, PartialEq, Serialize, Deserialize)]
164pub struct Bucket {
165 pub id: String,
166 pub version: u64,
167 pub day: String,
168 pub node: String,
169 pub engineer: String,
170 pub counts: BTreeMap<String, u64>,
172 pub usd: f64,
173 #[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
175 pub usd_by: BTreeMap<String, f64>,
176 pub dur_ms: u64,
177 #[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
179 pub lessons: BTreeMap<String, BTreeMap<String, u64>>,
180 #[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
184 pub usd_mode: BTreeMap<String, f64>,
185 #[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
187 pub channels: BTreeMap<String, u64>,
188}
189
190impl Bucket {
191 fn same(&self, other: &Bucket) -> bool {
193 self.counts == other.counts
194 && self.lessons == other.lessons
195 && self.channels == other.channels
196 && self.dur_ms == other.dur_ms
197 && (self.usd - other.usd).abs() < 1e-9
198 }
199
200 pub fn add(&mut self, other: &Bucket) {
202 for (k, v) in &other.counts {
203 *self.counts.entry(k.clone()).or_default() += v;
204 }
205 for (id, kinds) in &other.lessons {
206 let mine = self.lessons.entry(id.clone()).or_default();
207 for (k, v) in kinds {
208 *mine.entry(k.clone()).or_default() += v;
209 }
210 }
211 for (k, v) in &other.usd_by {
212 *self.usd_by.entry(k.clone()).or_default() += v;
213 }
214 for (k, v) in &other.usd_mode {
215 *self.usd_mode.entry(k.clone()).or_default() += v;
216 }
217 for (k, v) in &other.channels {
218 *self.channels.entry(k.clone()).or_default() += v;
219 }
220 self.usd += other.usd;
221 self.dur_ms += other.dur_ms;
222 }
223}
224
225pub fn events_path(home: &Path) -> PathBuf {
226 home.join("stats").join("events.jsonl")
227}
228
229pub fn rollups_path(home: &Path) -> PathBuf {
230 home.join("stats.jsonl")
231}
232
233pub fn record(home: &Path, ev: &Event) {
235 if !crate::learn::learning() {
236 return;
237 }
238 let _ = crate::learn::append_jsonl(&events_path(home), ev);
239}
240
241pub fn count(home: &Path, kind: &str, session: &str, fill: impl FnOnce(&mut Event)) {
244 let mut ev = Event::new(kind, crate::learn::now_ms());
245 ev.session = session.into();
246 fill(&mut ev);
247 if ev.engineer.is_empty() && !session.is_empty() {
248 ev.engineer = crate::presence::read(home, session).map(|p| p.engineer).unwrap_or_default();
249 }
250 record(home, &ev);
251}
252
253pub const FEED_CHARS: usize = 160;
255
256pub fn excerpt(text: &str, max: usize) -> String {
258 let one = text.split_whitespace().collect::<Vec<_>>().join(" ");
259 match one.char_indices().nth(max) {
260 Some((i, _)) => format!("{}…", &one[..i]),
261 None => one,
262 }
263}
264
265static PURPOSE: std::sync::OnceLock<&'static str> = std::sync::OnceLock::new();
266
267pub fn set_purpose(purpose: &'static str) {
269 let _ = PURPOSE.set(purpose);
270}
271
272pub fn purpose() -> &'static str {
273 PURPOSE.get().copied().unwrap_or("model")
274}
275
276fn read_events(path: &Path) -> Vec<Event> {
277 std::fs::read_to_string(path).unwrap_or_default().lines().filter_map(|l| serde_json::from_str(l).ok()).collect()
278}
279
280pub fn events(home: &Path) -> Vec<Event> {
282 let mut all = read_events(&backfill_path(home));
283 all.extend(read_events(&events_path(home)));
284 all
285}
286
287pub fn live(home: &Path) -> Vec<Event> {
289 read_events(&events_path(home))
290}
291
292pub fn live_since(home: &Path) -> Option<i64> {
296 live(home).iter().filter(|e| e.kind != VETO_REPEAT).map(|e| e.ms).min()
297}
298
299pub fn backfill_path(home: &Path) -> PathBuf {
300 home.join("stats").join("backfill.jsonl")
301}
302
303fn lines(path: &Path) -> Vec<serde_json::Value> {
304 std::fs::read_to_string(path).unwrap_or_default().lines().filter_map(|l| serde_json::from_str(l).ok()).collect()
305}
306
307pub fn backfill(home: &Path) -> usize {
313 let path = backfill_path(home);
314 if path.exists() {
315 return 0;
316 }
317 let cutoff = live_since(home).unwrap_or_else(crate::learn::now_ms);
318 let mut out = derive_history(home);
319 out.retain(|e| e.ms > 0 && e.ms < cutoff);
320 out.sort_by_key(|e| e.ms);
321 let text: String = out.iter().map(|e| serde_json::to_string(e).unwrap_or_default() + "\n").collect();
322 if std::fs::create_dir_all(path.parent().unwrap_or(home)).is_err() || std::fs::write(&path, text).is_err() {
323 return 0;
324 }
325 out.len()
326}
327
328fn derive_history(home: &Path) -> Vec<Event> {
329 use serde_json::Value;
330 let s = |v: &Value, k: &str| v[k].as_str().unwrap_or_default().to_string();
331 let ms = |v: &Value, k: &str| v[k].as_i64().unwrap_or(0);
332 let ids = |v: &Value, k: &str| v[k].as_array().into_iter().flatten().filter_map(Value::as_str).map(String::from).collect::<Vec<_>>();
333 let mut out = Vec::new();
334 let mut who: std::collections::HashMap<String, String> = std::collections::HashMap::new();
336 for e in lines(&home.join("episodes.jsonl")) {
337 let tool = s(&e, "rejected_tool");
338 if !s(&e, "engineer").is_empty() {
339 who.insert(s(&e, "session"), s(&e, "engineer"));
340 }
341 let snippet = excerpt(&s(&e, "correction"), FEED_CHARS);
342 if s(&e, "kind") == crate::veto::KIND {
344 out.push(Event {
345 ms: ms(&e, "ms"),
346 kind: VETO.into(),
347 session: s(&e, "session"),
348 engineer: s(&e, "engineer"),
349 repo: s(&e, "repo"),
350 text: snippet,
351 ..Default::default()
352 });
353 continue;
354 }
355 let answer = s(&e, "via") == crate::learn::ANSWER;
356 out.push(Event {
357 ms: ms(&e, "ms"),
358 kind: CORRECTION.into(),
359 session: s(&e, "session"),
360 engineer: s(&e, "engineer"),
361 repo: s(&e, "repo"),
362 why: if answer {
363 ANSWERED.into()
364 } else if tool.is_empty() {
365 "pushback".into()
366 } else {
367 "rejected_tool".into()
368 },
369 text: if tool.is_empty() || answer { snippet } else { format!("{tool}: {snippet}") },
370 source: source(&s(&e, "harness"), &s(&e, "harness_version")),
371 ..Default::default()
372 });
373 }
374 let engineer = |session: &str| who.get(session).cloned().unwrap_or_default();
375 let mut call: std::collections::HashMap<(String, &str, String, i64), usize> = std::collections::HashMap::new();
379 for o in lines(&home.join("outcomes.jsonl")) {
380 if o["channel"] == crate::outcome::PLACEBO {
382 continue;
383 }
384 let live = o["channel"] == "live";
385 let kind = match o["kind"].as_str().unwrap_or_default() {
386 crate::outcome::ASKED if o["channel"] == "early" || o["channel"] == "queued" => continue,
388 crate::outcome::ASKED if live => LIVE_ASK,
389 crate::outcome::ASKED => BLOCK,
390 crate::outcome::CHANGED => CHANGED,
391 crate::outcome::IGNORED => IGNORED,
392 crate::outcome::CORRECTED => CORRECTED,
393 crate::outcome::NOT_CORRECTED => NOT_CORRECTED,
394 crate::outcome::REGRETTED => REGRETTED,
395 crate::outcome::TESTED => TESTED,
396 crate::outcome::UNCLEAR | crate::outcome::REMAINS | crate::outcome::NOT_ARISEN => continue,
398 _ => continue,
399 };
400 let lessons = ids(&o, "lessons");
401 let session = s(&o, "session");
402 let asks = kind == BLOCK || kind == LIVE_ASK;
403 let key = (session.clone(), kind, s(&o, "channel"), ms(&o, "ms"));
404 let merges = [BLOCK, LIVE_ASK, CHANGED, IGNORED].contains(&kind);
405 if merges
406 && let Some(&i) = call.get(&key)
407 && let Some(e) = out.get_mut(i)
408 {
409 e.n += if asks { lessons.len() as u64 } else { 0 };
410 e.lessons.extend(lessons);
411 continue;
412 }
413 if merges {
414 call.insert(key, out.len());
415 }
416 out.push(Event {
417 ms: ms(&o, "ms"),
418 kind: kind.into(),
419 engineer: engineer(&session),
420 session,
421 repo: s(&o, "repo"),
422 n: if asks { lessons.len() as u64 } else { 0 },
423 lessons,
424 channel: s(&o, "channel"),
425 ..Default::default()
426 });
427 }
428 let mut seen = std::collections::HashSet::new();
430 for l in lines(&home.join("lessons.jsonl")) {
431 let id = s(&l, "id");
432 if id.is_empty() || !seen.insert(id.clone()) {
433 continue;
434 }
435 let source = Some(s(&l, "source")).filter(|x| !x.is_empty()).unwrap_or_else(|| "session".into());
436 out.push(Event {
437 ms: ms(&l, "origin_ms"),
438 kind: LESSON_NEW.into(),
439 engineer: ids(&l, "engineers").into_iter().next().unwrap_or_default(),
440 repo: s(&l, "repo"),
441 lessons: vec![id],
442 why: source,
443 ..Default::default()
444 });
445 }
446 let mut queen: BTreeMap<String, Value> = BTreeMap::new();
447 for q in lines(&home.join("queen").join("queue.jsonl")) {
448 queen.insert(s(&q, "id"), q);
449 }
450 for q in queen.values() {
451 let base = || Event { engineer: s(q, "engineer"), session: s(q, "session"), repo: s(q, "repo"), lessons: vec![s(q, "id")], ..Default::default() };
452 out.push(Event { ms: ms(q, "ms"), kind: QUEEN_QUEUED.into(), text: excerpt(&s(q, "question"), FEED_CHARS), ..base() });
453 if let Some(at) = q["answered_ms"].as_i64() {
454 let dur = u64::try_from(at - ms(q, "ms")).unwrap_or(0);
455 out.push(Event {
456 ms: at,
457 kind: QUEEN_ANSWERED.into(),
458 dur_ms: dur,
459 engineer: s(q, "answered_by"),
460 text: excerpt(&s(q, "answer"), FEED_CHARS),
461 ..base()
462 });
463 }
464 if let Some(at) = q["dismissed_ms"].as_i64() {
465 out.push(Event { ms: at, kind: QUEEN_DISMISSED.into(), text: s(q, "dismissed"), ..base() });
466 }
467 }
468 for (file, kind) in [("delivered.jsonl", QUEEN_DELIVERED), ("skips.jsonl", QUEEN_SKIPPED)] {
469 for m in lines(&home.join("queen").join(file)) {
470 let session = s(&m, "session");
471 out.push(Event { ms: ms(&m, "ms"), kind: kind.into(), engineer: engineer(&session), session, lessons: vec![s(&m, "id")], ..Default::default() });
472 }
473 }
474 for a in lines(&home.join("assimilated.jsonl")) {
475 out.push(Event {
476 ms: ms(&a, "ms"),
477 kind: ASSIMILATED.into(),
478 repo: s(&a, "source"),
479 n: ids(&a, "lessons").len() as u64,
480 usd: a["cost_usd"].as_f64().unwrap_or(0.0),
481 ..Default::default()
482 });
483 }
484 for (file, why) in [("experts.jsonl", "expert"), ("facets.jsonl", "facets")] {
485 for c in lines(&home.join(file)) {
486 let session = s(&c, "session");
487 out.push(Event {
488 ms: ms(&c, "ms"),
489 kind: MODEL_CALL.into(),
490 engineer: engineer(&session),
491 session,
492 repo: s(&c, "repo"),
493 usd: c["cost_usd"].as_f64().unwrap_or(0.0),
494 why: why.into(),
495 ..Default::default()
496 });
497 }
498 }
499 for (file, kind) in [(home.join("mesh").join("directives.jsonl"), STEER_ISSUED), (home.join("steer").join("inbox.jsonl"), STEER_DELIVERED)] {
502 for d in lines(&file) {
503 out.push(Event {
504 ms: ms(&d, "ms"),
505 kind: kind.into(),
506 engineer: s(&d, "engineer"),
507 why: s(&d, "kind"),
508 text: excerpt(&s(&d, "text"), FEED_CHARS),
509 ..Default::default()
510 });
511 }
512 }
513 for d in lines(&crate::steer::dropped_path(home)) {
514 out.push(Event { ms: ms(&d, "ms"), kind: STEER_DROPPED.into(), engineer: s(&d, "engineer"), why: s(&d, "why"), ..Default::default() });
515 }
516 out.extend(turns_from_state(&home.join("state"), &who));
517 out
518}
519
520fn turns_from_state(dir: &Path, who: &std::collections::HashMap<String, String>) -> Vec<Event> {
523 let mut out = Vec::new();
524 let Ok(rd) = std::fs::read_dir(dir) else { return out };
525 for f in rd.filter_map(Result::ok).map(|e| e.path()).filter(|p| p.extension().is_some_and(|x| x == "ctx")) {
526 let Some(at) = std::fs::metadata(&f).ok().and_then(|m| m.modified().ok()).and_then(|t| t.duration_since(std::time::UNIX_EPOCH).ok()) else {
527 continue;
528 };
529 let session = f.file_stem().map(|s| s.to_string_lossy().to_string()).unwrap_or_default();
530 let base =
531 || Event { ms: at.as_millis() as i64, session: session.clone(), engineer: who.get(&session).cloned().unwrap_or_default(), ..Default::default() };
532 for line in std::fs::read_to_string(&f).unwrap_or_default().lines() {
533 if let Some(score) = line.strip_prefix("#borg-skip ") {
534 out.push(Event { kind: TURN_END.into(), ..base() });
535 out.push(Event { kind: GATE_SKIP.into(), why: score.trim().into(), ..base() });
536 } else if let Some(cost) = line.strip_prefix("#borg-predict-cost ") {
537 out.push(Event { kind: TURN_END.into(), ..base() });
538 out.push(Event { kind: CHECK.into(), usd: cost.trim().parse().unwrap_or(0.0), ..base() });
539 }
540 }
541 }
542 out
543}
544
545pub fn node(home: &Path) -> String {
548 let path = home.join("stats").join("node");
549 if let Ok(id) = std::fs::read_to_string(&path)
550 && !id.trim().is_empty()
551 {
552 return id.trim().to_string();
553 }
554 let id = fresh_id();
558 let _ = std::fs::create_dir_all(path.parent().unwrap_or(home));
559 let tmp = path.with_extension(format!("{id}.tmp"));
560 let linked = std::fs::write(&tmp, &id).and_then(|()| std::fs::hard_link(&tmp, &path));
561 let _ = std::fs::remove_file(&tmp);
562 match linked {
563 Err(e) if e.kind() == std::io::ErrorKind::AlreadyExists => {
564 std::fs::read_to_string(&path).map(|s| s.trim().to_string()).ok().filter(|s| !s.is_empty()).unwrap_or(id)
565 }
566 _ => id,
567 }
568}
569
570fn fresh_id() -> String {
571 let mut bytes = [0u8; 8];
572 if let Ok(mut f) = std::fs::File::open("/dev/urandom") {
573 use std::io::Read;
574 let _ = f.read_exact(&mut bytes);
575 }
576 if bytes == [0; 8] {
577 use std::hash::{BuildHasher, Hasher};
578 let mut h = std::collections::hash_map::RandomState::new().build_hasher();
579 h.write_i64(crate::learn::now_ms());
580 h.write_u32(std::process::id());
581 bytes = h.finish().to_le_bytes();
582 }
583 bytes.iter().map(|b| format!("{b:02x}")).collect()
584}
585
586pub fn day(ms: i64) -> String {
588 let s = crate::install::utc(ms);
589 format!("{}-{}-{}", &s[0..4], &s[4..6], &s[6..8])
590}
591
592pub fn days_between(first: &str, last: &str) -> Vec<String> {
594 let (Some(a), Some(b)) = (day_number(first), day_number(last)) else { return Vec::new() };
595 (a..=b).map(|d| day(d * crate::clock::DAY_MS)).collect()
596}
597
598fn day_number(date: &str) -> Option<i64> {
600 let mut it = date.split('-').map(|p| p.parse::<i64>().ok());
601 let (y, m, d) = (it.next()??, it.next()??, it.next()??);
602 Some(crate::clock::days_from_civil(y, m, d))
603}
604
605#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
607pub struct Said {
608 pub corrections: u64,
612 pub vetoes: u64,
614}
615
616pub fn per_session(events: &[Event]) -> BTreeMap<String, Said> {
620 let mut out: BTreeMap<String, Said> = BTreeMap::new();
621 for e in events.iter().filter(|e| !e.session.is_empty()) {
622 match e.kind.as_str() {
623 CORRECTION if e.why != "regret" && e.why != crate::learn::ANSWER => out.entry(e.session.clone()).or_default().corrections += 1,
624 VETO => out.entry(e.session.clone()).or_default().vetoes += 1,
625 _ => {}
626 }
627 }
628 out
629}
630
631pub fn fold(node: &str, events: &[Event]) -> BTreeMap<String, Bucket> {
633 let mut out: BTreeMap<String, Bucket> = BTreeMap::new();
634 for e in events {
635 let day = day(e.ms);
636 let id = format!("stat:{node}:{}:{day}", e.engineer);
637 let b = out.entry(id.clone()).or_insert_with(|| Bucket { id, day, node: node.into(), engineer: e.engineer.clone(), ..Default::default() });
638 let kind = if e.kind == CORRECTION && e.why == crate::learn::ANSWER { ANSWERED } else { e.kind.as_str() };
640 *b.counts.entry(kind.to_string()).or_default() += 1;
641 if e.n > 0 {
642 *b.counts.entry(format!("{}_n", e.kind)).or_default() += e.n;
643 }
644 b.usd += e.usd;
645 if e.usd > 0.0 {
646 *b.usd_by.entry(e.kind.clone()).or_default() += e.usd;
647 }
648 if e.usd > 0.0 && (e.kind == CHECK || e.kind == LIVE_CHECK) {
650 let mode = if e.mode.is_empty() { "hook" } else { e.mode.as_str() };
651 *b.usd_mode.entry(mode.into()).or_default() += e.usd;
652 }
653 if !e.channel.is_empty() {
654 *b.channels.entry(format!("{}@{}", e.kind, e.channel)).or_default() += 1;
655 }
656 b.dur_ms += e.dur_ms;
657 for l in &e.lessons {
658 *b.lessons.entry(l.clone()).or_default().entry(e.kind.clone()).or_default() += 1;
659 }
660 }
661 out
662}
663
664pub fn rollups(home: &Path) -> BTreeMap<String, Bucket> {
668 let mut out: BTreeMap<String, (Bucket, [u8; 32])> = BTreeMap::new();
669 for l in std::fs::read_to_string(rollups_path(home)).unwrap_or_default().lines() {
670 let Ok(b) = serde_json::from_str::<Bucket>(l) else { continue };
671 let h = *blake3::hash(l.as_bytes()).as_bytes();
672 if out.get(&b.id).is_none_or(|(old, oh)| (b.version, h) > (old.version, *oh)) {
673 out.insert(b.id.clone(), (b, h));
674 }
675 }
676 out.into_iter().map(|(id, (b, _))| (id, b)).collect()
677}
678
679pub fn rollup(home: &Path) -> usize {
682 backfill(home);
683 let evs = events(home);
685 if evs.is_empty() {
686 return 0;
687 }
688 let me = node(home);
689 let published = rollups(home);
690 let personal = crate::reach::personal_ids(home);
692 let fresh: Vec<Bucket> = fold(&me, &evs)
693 .into_values()
694 .filter_map(|mut b| {
695 b.lessons.retain(|id, _| !personal.contains(id));
696 let old = published.get(&b.id);
697 if old.is_some_and(|o| o.same(&b)) {
698 return None;
699 }
700 b.version = old.map_or(1, |o| o.version + 1);
701 Some(b)
702 })
703 .collect();
704 if fresh.is_empty() {
705 return 0;
706 }
707 let lines: String = fresh.iter().map(|b| serde_json::to_string(b).unwrap_or_default() + "\n").collect();
708 let path = rollups_path(home);
709 let ok = std::fs::OpenOptions::new().create(true).append(true).open(&path).and_then(|mut f| f.write_all(lines.as_bytes()));
710 if ok.is_ok() { fresh.len() } else { 0 }
711}
712
713pub fn current(home: &Path) -> Vec<Bucket> {
716 backfill(home);
717 let me = node(home);
718 let mut all: BTreeMap<String, Bucket> = rollups(home).into_iter().filter(|(_, b)| b.node != me).collect();
719 all.extend(fold(&me, &events(home)));
720 all.into_values().collect()
721}
722
723#[cfg(test)]
724mod tests {
725 use super::*;
726
727 #[test]
728 fn corrections_and_vetoes_count_per_session_merged_or_not() {
729 let ev = |kind: &str, session: &str, why: &str| Event { kind: kind.into(), session: session.into(), why: why.into(), ..Default::default() };
730 let events = [
731 ev(CORRECTION, "a", "pushback"),
732 ev(CORRECTION, "a", "rejected_tool"),
733 ev(CORRECTION, "a", "regret"),
734 ev(CORRECTION, "a", crate::learn::ANSWER),
735 ev(VETO, "a", "prompt"),
736 ev(VETO, "b", "cli"),
737 ev(CHANGED, "b", ""),
738 ev(CORRECTION, "", "pushback"),
739 ];
740 let got = per_session(&events);
741 assert_eq!(got["a"], Said { corrections: 2, vetoes: 1 }, "an answer to the drone's own question is no correction");
742 assert_eq!(got["b"], Said { corrections: 0, vetoes: 1 });
743 assert_eq!(got.len(), 2);
744 let buckets = fold("n", &events);
746 let b = buckets.values().next().unwrap();
747 assert_eq!((b.counts[CORRECTION], b.counts[ANSWERED]), (4, 1));
748 }
749
750 #[test]
751 fn the_history_derived_from_episodes_tells_answers_apart() {
752 let home = std::env::temp_dir().join(format!("borg-stats-answer-{}", std::process::id()));
753 let _ = std::fs::remove_dir_all(&home);
754 std::fs::create_dir_all(&home).unwrap();
755 let eps = [
756 serde_json::json!({"id":"e1","ms":1,"repo":"r","session":"s","task":"t","agent":"Which backoff?","rejected_tool":null,"correction":"Exponential","via":"answer"}),
757 serde_json::json!({"id":"e2","ms":2,"repo":"r","session":"s","task":"t","agent":"a","rejected_tool":null,"correction":"no, not like that"}),
758 ];
759 std::fs::write(home.join("episodes.jsonl"), eps.iter().map(|e| e.to_string() + "\n").collect::<String>()).unwrap();
760 let whys: Vec<String> = derive_history(&home).into_iter().filter(|e| e.kind == CORRECTION).map(|e| e.why).collect();
761 assert_eq!(whys, [ANSWERED, "pushback"]);
762 let _ = std::fs::remove_dir_all(&home);
763 }
764
765 #[test]
766 fn concurrent_first_calls_agree_on_one_node() {
767 for round in 0..20 {
770 let home = std::env::temp_dir().join(format!("borg-stats-node-{}-{round}", std::process::id()));
771 let _ = std::fs::remove_dir_all(&home);
772 let ids: Vec<String> = std::thread::scope(|s| {
773 let hs: Vec<_> = (0..16).map(|_| s.spawn(|| node(&home))).collect();
774 hs.into_iter().map(|h| h.join().unwrap()).collect()
775 });
776 assert!(ids.iter().all(|id| *id == ids[0] && id.len() == 16), "{ids:?}");
777 assert_eq!(node(&home), ids[0]);
778 let _ = std::fs::remove_dir_all(&home);
779 }
780 }
781
782 fn tmp(tag: &str) -> PathBuf {
783 let d = std::env::temp_dir().join(format!("borg-stats-{tag}-{}", std::process::id()));
784 let _ = std::fs::remove_dir_all(&d);
785 std::fs::create_dir_all(&d).unwrap();
786 d
787 }
788
789 fn ev(kind: &str, ms: i64, engineer: &str, lessons: &[&str], n: u64) -> Event {
790 Event { engineer: engineer.into(), lessons: lessons.iter().map(|s| s.to_string()).collect(), n, ..Event::new(kind, ms) }
791 }
792
793 const DAY: i64 = crate::clock::DAY_MS;
795
796 #[test]
797 fn days_between_fills_every_day() {
798 assert_eq!(days_between("2026-02-27", "2026-03-02"), ["2026-02-27", "2026-02-28", "2026-03-01", "2026-03-02"]);
799 assert_eq!(days_between("1970-01-01", "1970-01-01"), ["1970-01-01"]);
800 assert!(days_between("x", "1970-01-01").is_empty());
801 }
802
803 #[test]
804 fn day_is_utc_date() {
805 assert_eq!(day(0), "1970-01-01");
806 assert_eq!(day(DAY - 1), "1970-01-01");
807 assert_eq!(day(DAY), "1970-01-02");
808 }
809
810 #[test]
811 fn fold_counts_per_engineer_and_day() {
812 let evs = [
813 ev(TURN_END, 10, "a@x", &[], 0),
814 ev(CHECK, 20, "a@x", &[], 7),
815 ev(BLOCK, 30, "a@x", &["L1", "L2"], 2),
816 ev(TURN_END, DAY + 5, "a@x", &[], 0),
817 ev(TURN_END, 40, "b@x", &[], 0),
818 ];
819 let b = fold("n1", &evs);
820 assert_eq!(b.len(), 3, "two days for a, one for b");
821 let a = &b["stat:n1:a@x:1970-01-01"];
822 assert_eq!((a.counts[TURN_END], a.counts[CHECK], a.counts["check_n"], a.counts["block_n"]), (1, 1, 7, 2));
823 assert_eq!(a.lessons["L1"][BLOCK], 1);
824 }
825
826 #[test]
827 fn fold_splits_check_cost_by_mode_and_asks_by_channel() {
828 let with = |kind: &str, usd: f64, mode: &str, channel: &str| Event { usd, mode: mode.into(), channel: channel.into(), ..ev(kind, 10, "a@x", &[], 0) };
829 let evs = [
830 with(CHECK, 0.05, "", ""),
831 with(CHECK, 0.04, "live", ""),
832 with(LIVE_CHECK, 0.03, "live", ""),
833 with(MODEL_CALL, 0.5, "live", ""),
834 with(BLOCK, 0.0, "live", "stop"),
835 with(LIVE_ASK, 0.0, "live", "live"),
836 with(CHANGED, 0.0, "", "live"),
837 ];
838 let b = &fold("n1", &evs)["stat:n1:a@x:1970-01-01"];
839 assert!((b.usd_mode["hook"] - 0.05).abs() < 1e-9 && (b.usd_mode["live"] - 0.07).abs() < 1e-9, "{:?}", b.usd_mode);
840 assert_eq!(b.usd_mode.len(), 2, "only the conscience's checks: {:?}", b.usd_mode);
841 assert_eq!((b.channels["block@stop"], b.channels["live_ask@live"], b.channels["changed@live"]), (1, 1, 1));
842 assert_eq!((b.counts[CHECK], b.counts[LIVE_CHECK]), (2, 1), "live checks are counted apart from turn-end checks");
843 }
844
845 #[test]
846 fn rollup_versions_only_on_change_and_is_idempotent() {
847 let home = tmp("rollup");
848 record(&home, &ev(TURN_END, 10, "a@x", &[], 0));
849 assert_eq!(rollup(&home), 1);
850 assert_eq!(rollup(&home), 0, "nothing changed: nothing appended");
851 record(&home, &ev(GATE_SKIP, 11, "a@x", &[], 0));
852 assert_eq!(rollup(&home), 1);
853 let r = rollups(&home);
854 let b = r.values().next().unwrap();
855 assert_eq!((b.version, b.counts[TURN_END], b.counts[GATE_SKIP]), (2, 1, 1));
856 let text = std::fs::read_to_string(rollups_path(&home)).unwrap();
857 assert!(!text.contains("session"), "no session ids leave the machine");
858 }
859
860 #[test]
861 fn current_prefers_own_fresh_counts_and_keeps_peers() {
862 let home = tmp("current");
863 record(&home, &ev(TURN_END, 10, "a@x", &[], 0));
864 rollup(&home);
865 record(&home, &ev(TURN_END, 12, "a@x", &[], 0));
866 let peer = Bucket {
867 id: "stat:other:b@x:1970-01-01".into(),
868 version: 3,
869 day: "1970-01-01".into(),
870 node: "other".into(),
871 engineer: "b@x".into(),
872 ..Default::default()
873 };
874 let mut f = std::fs::OpenOptions::new().append(true).open(rollups_path(&home)).unwrap();
875 writeln!(f, "{}", serde_json::to_string(&peer).unwrap()).unwrap();
876 let cur = current(&home);
877 assert_eq!(cur.len(), 2);
878 let mine = cur.iter().find(|b| b.engineer == "a@x").unwrap();
879 assert_eq!(mine.counts[TURN_END], 2, "own counts are folded fresh, not the stale rollup");
880 }
881
882 #[test]
885 fn rollups_break_version_ties_by_hash_whatever_the_order() {
886 let bucket = |n: u64| {
887 let mut b = Bucket { id: "stat:n:a@x:1970-01-01".into(), version: 2, ..Default::default() };
888 b.counts.insert(TURN_END.into(), n);
889 serde_json::to_string(&b).unwrap()
890 };
891 let (a, b) = (bucket(1), bucket(2));
892 let shown = |name: &str, first: &str, second: &str| {
893 let home = tmp(name);
894 std::fs::write(rollups_path(&home), format!("{first}\n{second}\n")).unwrap();
895 rollups(&home).into_values().next().unwrap().counts[TURN_END]
896 };
897 assert_eq!(shown("ties-ab", &a, &b), shown("ties-ba", &b, &a));
898 }
899
900 #[test]
901 fn record_fails_open() {
902 let home = tmp("open");
903 std::fs::write(home.join("stats"), "a file where the directory should be").unwrap();
904 record(&home, &ev(TURN_END, 1, "a@x", &[], 0));
905 assert!(events(&home).is_empty());
906 }
907
908 #[test]
909 fn backfill_derives_history_before_live_counting_once() {
910 let home = tmp("backfill");
911 let w = |f: &str, lines: &[serde_json::Value]| {
912 let p = home.join(f);
913 std::fs::create_dir_all(p.parent().unwrap()).unwrap();
914 std::fs::write(p, lines.iter().map(|l| l.to_string() + "\n").collect::<String>()).unwrap();
915 };
916 use serde_json::json;
917 w(
918 "episodes.jsonl",
919 &[
920 json!({"id":"e1","ms":100,"repo":"app","session":"s1","task":"t","agent":"a","rejected_tool":"Bash","correction":"no, not like that","engineer":"a@x"}),
921 json!({"id":"e2","ms":5000,"repo":"app","session":"s2","task":"t","agent":"a","rejected_tool":null,"correction":"after live counting began","engineer":"a@x"}),
922 ],
923 );
924 w(
925 "outcomes.jsonl",
926 &[
927 json!({"kind":"asked","session":"s1","lessons":["L1","L2"],"ms":110,"repo":"app"}),
928 json!({"kind":"asked","session":"s1","lessons":["L3"],"ms":112,"repo":"app","channel":"early"}),
930 json!({"kind":"asked","session":"s1","lessons":["L4"],"ms":114,"repo":"app","channel":"queued"}),
931 json!({"kind":"asked_then_changed","session":"s1","lessons":["L1"],"ms":120,"repo":"app"}),
932 ],
933 );
934 w(
935 "lessons.jsonl",
936 &[json!({"id":"L1","origin_ms":90,"question":"q","version":1,"engineers":["a@x"]}), json!({"id":"L1","origin_ms":90,"question":"q2","version":2})],
937 );
938 w("queen/queue.jsonl", &[json!({"id":"Q1","ms":130,"question":"which?","engineer":"a@x","answered_ms":190,"answer":"this","answered_by":"a@x"})]);
939 record(&home, &ev(TURN_END, 1000, "a@x", &[], 0));
940 let n = backfill(&home);
941 let all = read_events(&backfill_path(&home));
942 assert_eq!(n, all.len());
943 let kinds: Vec<&str> = all.iter().map(|e| e.kind.as_str()).collect();
944 assert_eq!(kinds, [LESSON_NEW, CORRECTION, BLOCK, CHANGED, QUEEN_QUEUED, QUEEN_ANSWERED], "ordered by time, the later correction left out");
945 let c = &all[1];
946 assert_eq!((c.why.as_str(), c.text.as_str(), c.engineer.as_str()), ("rejected_tool", "Bash: no, not like that", "a@x"));
947 assert_eq!(all[3].engineer, "a@x", "outcomes take the session's engineer from its corrections");
948 assert_eq!(all[5].dur_ms, 60, "time to answer");
949 assert_eq!(backfill(&home), 0, "once");
950 assert_eq!(events(&home).len(), all.len() + 1, "history and live events together");
951 }
952
953 #[test]
958 fn backfill_counts_one_event_per_call_as_live_counting_does() {
959 let home = tmp("backfill-call");
960 use serde_json::json;
961 let rec = |kind: &str, lessons: &[&str], ms: i64, channel: &str, check: &str| json!({"kind": kind, "session": "s1", "lessons": lessons, "ms": ms, "repo": "app", "channel": channel, "check": check});
962 let log = [
963 rec("asked", &["L1", "L2"], 110, "stop", "c1"),
964 rec("asked", &["L3"], 110, "stop", "c2"),
965 rec("asked", &["L4"], 120, "live", "c3"),
966 rec("asked", &["L5"], 120, "live", "c4"),
967 rec("asked_then_changed", &["L1"], 130, "stop", "c1"),
968 rec("asked_then_changed", &["L3"], 130, "stop", "c2"),
969 rec("asked", &["L6"], 140, "stop", "c5"),
970 rec("asked_not_corrected", &["L2"], 150, "stop", "c1"),
971 rec("asked_not_corrected", &["L6"], 150, "stop", "c5"),
972 ];
973 std::fs::write(home.join("outcomes.jsonl"), log.iter().map(|l| l.to_string() + "\n").collect::<String>()).unwrap();
974 record(&home, &ev(TURN_END, 1000, "a@x", &[], 0));
975 backfill(&home);
976 let all: Vec<(String, i64, u64, Vec<String>)> = read_events(&backfill_path(&home)).into_iter().map(|e| (e.kind, e.ms, e.n, e.lessons)).collect();
977 let row = |kind: &str, ms: i64, n: u64, lessons: &[&str]| (kind.to_string(), ms, n, lessons.iter().map(|l| l.to_string()).collect::<Vec<_>>());
978 assert_eq!(
979 all,
980 [
981 row(BLOCK, 110, 3, &["L1", "L2", "L3"]),
982 row(LIVE_ASK, 120, 2, &["L4", "L5"]),
983 row(CHANGED, 130, 0, &["L1", "L3"]),
984 row(BLOCK, 140, 1, &["L6"]),
985 row(NOT_CORRECTED, 150, 0, &["L2"]),
986 row(NOT_CORRECTED, 150, 0, &["L6"]),
987 ]
988 );
989 }
990
991 #[test]
992 fn node_id_is_stable() {
993 let home = tmp("node");
994 let a = node(&home);
995 assert_eq!(a.len(), 16);
996 assert_eq!(node(&home), a);
997 }
998}