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";
30pub const LIVE_ASK: &str = "live_ask";
33pub const EARLY: &str = "early";
35pub const QUEUED_FREE: &str = "queued_free";
37pub const CHANGED: &str = "changed";
39pub const IGNORED: &str = "ignored";
40pub const CORRECTED: &str = "corrected";
41pub const NOT_CORRECTED: &str = "not_corrected";
42pub const REGRETTED: &str = "regretted";
45pub const TESTED: &str = "tested";
48pub const CORRECTION: &str = "correction";
53pub const ANSWERED: &str = "answer";
56pub const VETO: &str = "veto";
59pub const VETO_REPEAT: &str = "veto_repeat";
62pub const LESSON_NEW: &str = "lesson_new";
64pub const LESSON_MERGED: &str = "lesson_merged";
65pub const REVIEW_FINDING: &str = "review_finding";
68pub const NOT_A_LESSON: &str = "not_a_lesson";
70pub const MODEL_CALL: &str = "model_call";
72pub const STEER_ISSUED: &str = "steer_issued";
75pub const STEER_DELIVERED: &str = "steer_delivered";
76pub const STEER_DROPPED: &str = "steer_dropped";
77pub const PAUSE_REFUSED: &str = "pause_refused";
78pub const MSG_SENT: &str = "msg_sent";
82pub const MSG_DELIVERED: &str = "msg_delivered";
83pub const MSG_DROPPED: &str = "msg_dropped";
84pub const QUEEN_QUEUED: &str = "queen_queued";
87pub const QUEEN_DELIVERED: &str = "queen_delivered";
88pub const QUEEN_ANSWERED: &str = "queen_answered";
89pub const QUEEN_SKIPPED: &str = "queen_skipped";
90pub const QUEEN_DISMISSED: &str = "queen_dismissed";
91pub const QUEEN_NO: &str = "queen_no";
95pub const QUEEN_STAMP: &str = "queen_stamp";
96pub const QUEEN_MISSED: &str = "queen_missed";
97pub const ASSIMILATED: &str = "assimilated";
99
100#[derive(Debug, Clone, Default, PartialEq, Serialize, Deserialize)]
101pub struct Event {
102 pub ms: i64,
103 pub kind: String,
104 #[serde(default, skip_serializing_if = "String::is_empty")]
105 pub session: String,
106 #[serde(default, skip_serializing_if = "String::is_empty")]
107 pub engineer: String,
108 #[serde(default, skip_serializing_if = "String::is_empty")]
109 pub repo: String,
110 #[serde(default, skip_serializing_if = "Vec::is_empty")]
111 pub lessons: Vec<String>,
112 #[serde(default, skip_serializing_if = "is_zero")]
113 pub n: u64,
114 #[serde(default, skip_serializing_if = "is_zero_f")]
115 pub usd: f64,
116 #[serde(default, skip_serializing_if = "is_zero")]
117 pub dur_ms: u64,
118 #[serde(default, skip_serializing_if = "String::is_empty")]
119 pub why: String,
120 #[serde(default, skip_serializing_if = "String::is_empty")]
122 pub text: String,
123 #[serde(default, skip_serializing_if = "String::is_empty")]
125 pub mode: String,
126 #[serde(default, skip_serializing_if = "String::is_empty")]
129 pub channel: String,
130 #[serde(default, skip_serializing_if = "String::is_empty")]
134 pub source: String,
135}
136
137pub fn source(harness: &str, version: &str) -> String {
140 format!("{harness} {version}").trim().to_string()
141}
142
143fn is_zero(v: &u64) -> bool {
144 *v == 0
145}
146
147fn is_zero_f(v: &f64) -> bool {
148 *v == 0.0
149}
150
151impl Event {
152 pub fn new(kind: &str, ms: i64) -> Self {
154 let mode = std::env::var("BORG_MODE").map(|_| crate::live::current().name().to_string()).unwrap_or_default();
155 Event { kind: kind.into(), ms, mode, ..Default::default() }
156 }
157}
158
159#[derive(Debug, Clone, Default, PartialEq, Serialize, Deserialize)]
161pub struct Bucket {
162 pub id: String,
163 pub version: u64,
164 pub day: String,
165 pub node: String,
166 pub engineer: String,
167 pub counts: BTreeMap<String, u64>,
169 pub usd: f64,
170 #[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
172 pub usd_by: BTreeMap<String, f64>,
173 pub dur_ms: u64,
174 #[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
176 pub lessons: BTreeMap<String, BTreeMap<String, u64>>,
177 #[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
180 pub usd_mode: BTreeMap<String, f64>,
181 #[serde(default, skip_serializing_if = "BTreeMap::is_empty")]
183 pub channels: BTreeMap<String, u64>,
184}
185
186impl Bucket {
187 fn same(&self, other: &Bucket) -> bool {
189 self.counts == other.counts
190 && self.lessons == other.lessons
191 && self.channels == other.channels
192 && self.dur_ms == other.dur_ms
193 && (self.usd - other.usd).abs() < 1e-9
194 }
195
196 pub fn add(&mut self, other: &Bucket) {
198 for (k, v) in &other.counts {
199 *self.counts.entry(k.clone()).or_default() += v;
200 }
201 for (id, kinds) in &other.lessons {
202 let mine = self.lessons.entry(id.clone()).or_default();
203 for (k, v) in kinds {
204 *mine.entry(k.clone()).or_default() += v;
205 }
206 }
207 for (k, v) in &other.usd_by {
208 *self.usd_by.entry(k.clone()).or_default() += v;
209 }
210 for (k, v) in &other.usd_mode {
211 *self.usd_mode.entry(k.clone()).or_default() += v;
212 }
213 for (k, v) in &other.channels {
214 *self.channels.entry(k.clone()).or_default() += v;
215 }
216 self.usd += other.usd;
217 self.dur_ms += other.dur_ms;
218 }
219}
220
221pub fn events_path(home: &Path) -> PathBuf {
222 home.join("stats").join("events.jsonl")
223}
224
225pub fn rollups_path(home: &Path) -> PathBuf {
226 home.join("stats.jsonl")
227}
228
229pub fn record(home: &Path, ev: &Event) {
231 if !crate::learn::learning() {
232 return;
233 }
234 let _ = crate::learn::append_jsonl(&events_path(home), ev);
235}
236
237pub fn count(home: &Path, kind: &str, session: &str, fill: impl FnOnce(&mut Event)) {
240 let mut ev = Event::new(kind, crate::learn::now_ms());
241 ev.session = session.into();
242 fill(&mut ev);
243 if ev.engineer.is_empty() && !session.is_empty() {
244 ev.engineer = crate::presence::read(home, session).map(|p| p.engineer).unwrap_or_default();
245 }
246 record(home, &ev);
247}
248
249pub const FEED_CHARS: usize = 160;
251
252pub fn excerpt(text: &str, max: usize) -> String {
254 let one = text.split_whitespace().collect::<Vec<_>>().join(" ");
255 match one.char_indices().nth(max) {
256 Some((i, _)) => format!("{}…", &one[..i]),
257 None => one,
258 }
259}
260
261static PURPOSE: std::sync::OnceLock<&'static str> = std::sync::OnceLock::new();
262
263pub fn set_purpose(purpose: &'static str) {
265 let _ = PURPOSE.set(purpose);
266}
267
268pub fn purpose() -> &'static str {
269 PURPOSE.get().copied().unwrap_or("model")
270}
271
272fn read_events(path: &Path) -> Vec<Event> {
273 std::fs::read_to_string(path).unwrap_or_default().lines().filter_map(|l| serde_json::from_str(l).ok()).collect()
274}
275
276pub fn events(home: &Path) -> Vec<Event> {
278 let mut all = read_events(&backfill_path(home));
279 all.extend(read_events(&events_path(home)));
280 all
281}
282
283pub fn live(home: &Path) -> Vec<Event> {
285 read_events(&events_path(home))
286}
287
288pub fn live_since(home: &Path) -> Option<i64> {
292 live(home).iter().filter(|e| e.kind != VETO_REPEAT).map(|e| e.ms).min()
293}
294
295pub fn backfill_path(home: &Path) -> PathBuf {
296 home.join("stats").join("backfill.jsonl")
297}
298
299fn lines(path: &Path) -> Vec<serde_json::Value> {
300 std::fs::read_to_string(path).unwrap_or_default().lines().filter_map(|l| serde_json::from_str(l).ok()).collect()
301}
302
303pub fn backfill(home: &Path) -> usize {
309 let path = backfill_path(home);
310 if path.exists() {
311 return 0;
312 }
313 let cutoff = live_since(home).unwrap_or_else(crate::learn::now_ms);
314 let mut out = derive_history(home);
315 out.retain(|e| e.ms > 0 && e.ms < cutoff);
316 out.sort_by_key(|e| e.ms);
317 let text: String = out.iter().map(|e| serde_json::to_string(e).unwrap_or_default() + "\n").collect();
318 if std::fs::create_dir_all(path.parent().unwrap_or(home)).is_err() || std::fs::write(&path, text).is_err() {
319 return 0;
320 }
321 out.len()
322}
323
324fn derive_history(home: &Path) -> Vec<Event> {
325 use serde_json::Value;
326 let s = |v: &Value, k: &str| v[k].as_str().unwrap_or_default().to_string();
327 let ms = |v: &Value, k: &str| v[k].as_i64().unwrap_or(0);
328 let ids = |v: &Value, k: &str| v[k].as_array().into_iter().flatten().filter_map(Value::as_str).map(String::from).collect::<Vec<_>>();
329 let mut out = Vec::new();
330 let mut who: std::collections::HashMap<String, String> = std::collections::HashMap::new();
332 for e in lines(&home.join("episodes.jsonl")) {
333 let tool = s(&e, "rejected_tool");
334 if !s(&e, "engineer").is_empty() {
335 who.insert(s(&e, "session"), s(&e, "engineer"));
336 }
337 let snippet = excerpt(&s(&e, "correction"), FEED_CHARS);
338 if s(&e, "kind") == crate::veto::KIND {
340 out.push(Event {
341 ms: ms(&e, "ms"),
342 kind: VETO.into(),
343 session: s(&e, "session"),
344 engineer: s(&e, "engineer"),
345 repo: s(&e, "repo"),
346 text: snippet,
347 ..Default::default()
348 });
349 continue;
350 }
351 let answer = s(&e, "via") == crate::learn::ANSWER;
352 out.push(Event {
353 ms: ms(&e, "ms"),
354 kind: CORRECTION.into(),
355 session: s(&e, "session"),
356 engineer: s(&e, "engineer"),
357 repo: s(&e, "repo"),
358 why: if answer {
359 ANSWERED.into()
360 } else if tool.is_empty() {
361 "pushback".into()
362 } else {
363 "rejected_tool".into()
364 },
365 text: if tool.is_empty() || answer { snippet } else { format!("{tool}: {snippet}") },
366 source: source(&s(&e, "harness"), &s(&e, "harness_version")),
367 ..Default::default()
368 });
369 }
370 let engineer = |session: &str| who.get(session).cloned().unwrap_or_default();
371 let mut call: std::collections::HashMap<(String, &str, String, i64), usize> = std::collections::HashMap::new();
375 for o in lines(&home.join("outcomes.jsonl")) {
376 if o["channel"] == crate::outcome::PLACEBO {
378 continue;
379 }
380 let live = o["channel"] == "live";
381 let kind = match o["kind"].as_str().unwrap_or_default() {
382 crate::outcome::ASKED if o["channel"] == "early" || o["channel"] == "queued" => continue,
384 crate::outcome::ASKED if live => LIVE_ASK,
385 crate::outcome::ASKED => BLOCK,
386 crate::outcome::CHANGED => CHANGED,
387 crate::outcome::IGNORED => IGNORED,
388 crate::outcome::CORRECTED => CORRECTED,
389 crate::outcome::NOT_CORRECTED => NOT_CORRECTED,
390 crate::outcome::REGRETTED => REGRETTED,
391 crate::outcome::TESTED => TESTED,
392 _ => continue,
393 };
394 let lessons = ids(&o, "lessons");
395 let session = s(&o, "session");
396 let asks = kind == BLOCK || kind == LIVE_ASK;
397 let key = (session.clone(), kind, s(&o, "channel"), ms(&o, "ms"));
398 let merges = [BLOCK, LIVE_ASK, CHANGED, IGNORED].contains(&kind);
399 if merges
400 && let Some(&i) = call.get(&key)
401 && let Some(e) = out.get_mut(i)
402 {
403 e.n += if asks { lessons.len() as u64 } else { 0 };
404 e.lessons.extend(lessons);
405 continue;
406 }
407 if merges {
408 call.insert(key, out.len());
409 }
410 out.push(Event {
411 ms: ms(&o, "ms"),
412 kind: kind.into(),
413 engineer: engineer(&session),
414 session,
415 repo: s(&o, "repo"),
416 n: if asks { lessons.len() as u64 } else { 0 },
417 lessons,
418 channel: s(&o, "channel"),
419 ..Default::default()
420 });
421 }
422 let mut seen = std::collections::HashSet::new();
424 for l in lines(&home.join("lessons.jsonl")) {
425 let id = s(&l, "id");
426 if id.is_empty() || !seen.insert(id.clone()) {
427 continue;
428 }
429 let source = Some(s(&l, "source")).filter(|x| !x.is_empty()).unwrap_or_else(|| "session".into());
430 out.push(Event {
431 ms: ms(&l, "origin_ms"),
432 kind: LESSON_NEW.into(),
433 engineer: ids(&l, "engineers").into_iter().next().unwrap_or_default(),
434 repo: s(&l, "repo"),
435 lessons: vec![id],
436 why: source,
437 ..Default::default()
438 });
439 }
440 let mut queen: BTreeMap<String, Value> = BTreeMap::new();
441 for q in lines(&home.join("queen").join("queue.jsonl")) {
442 queen.insert(s(&q, "id"), q);
443 }
444 for q in queen.values() {
445 let base = || Event { engineer: s(q, "engineer"), session: s(q, "session"), repo: s(q, "repo"), lessons: vec![s(q, "id")], ..Default::default() };
446 out.push(Event { ms: ms(q, "ms"), kind: QUEEN_QUEUED.into(), text: excerpt(&s(q, "question"), FEED_CHARS), ..base() });
447 if let Some(at) = q["answered_ms"].as_i64() {
448 let dur = u64::try_from(at - ms(q, "ms")).unwrap_or(0);
449 out.push(Event {
450 ms: at,
451 kind: QUEEN_ANSWERED.into(),
452 dur_ms: dur,
453 engineer: s(q, "answered_by"),
454 text: excerpt(&s(q, "answer"), FEED_CHARS),
455 ..base()
456 });
457 }
458 if let Some(at) = q["dismissed_ms"].as_i64() {
459 out.push(Event { ms: at, kind: QUEEN_DISMISSED.into(), text: s(q, "dismissed"), ..base() });
460 }
461 }
462 for (file, kind) in [("delivered.jsonl", QUEEN_DELIVERED), ("skips.jsonl", QUEEN_SKIPPED)] {
463 for m in lines(&home.join("queen").join(file)) {
464 let session = s(&m, "session");
465 out.push(Event { ms: ms(&m, "ms"), kind: kind.into(), engineer: engineer(&session), session, lessons: vec![s(&m, "id")], ..Default::default() });
466 }
467 }
468 for a in lines(&home.join("assimilated.jsonl")) {
469 out.push(Event {
470 ms: ms(&a, "ms"),
471 kind: ASSIMILATED.into(),
472 repo: s(&a, "source"),
473 n: ids(&a, "lessons").len() as u64,
474 usd: a["cost_usd"].as_f64().unwrap_or(0.0),
475 ..Default::default()
476 });
477 }
478 for (file, why) in [("experts.jsonl", "expert"), ("facets.jsonl", "facets")] {
479 for c in lines(&home.join(file)) {
480 let session = s(&c, "session");
481 out.push(Event {
482 ms: ms(&c, "ms"),
483 kind: MODEL_CALL.into(),
484 engineer: engineer(&session),
485 session,
486 repo: s(&c, "repo"),
487 usd: c["cost_usd"].as_f64().unwrap_or(0.0),
488 why: why.into(),
489 ..Default::default()
490 });
491 }
492 }
493 for (file, kind) in [(home.join("mesh").join("directives.jsonl"), STEER_ISSUED), (home.join("steer").join("inbox.jsonl"), STEER_DELIVERED)] {
496 for d in lines(&file) {
497 out.push(Event {
498 ms: ms(&d, "ms"),
499 kind: kind.into(),
500 engineer: s(&d, "engineer"),
501 why: s(&d, "kind"),
502 text: excerpt(&s(&d, "text"), FEED_CHARS),
503 ..Default::default()
504 });
505 }
506 }
507 for d in lines(&crate::steer::dropped_path(home)) {
508 out.push(Event { ms: ms(&d, "ms"), kind: STEER_DROPPED.into(), engineer: s(&d, "engineer"), why: s(&d, "why"), ..Default::default() });
509 }
510 out.extend(turns_from_state(&home.join("state"), &who));
511 out
512}
513
514fn turns_from_state(dir: &Path, who: &std::collections::HashMap<String, String>) -> Vec<Event> {
517 let mut out = Vec::new();
518 let Ok(rd) = std::fs::read_dir(dir) else { return out };
519 for f in rd.filter_map(Result::ok).map(|e| e.path()).filter(|p| p.extension().is_some_and(|x| x == "ctx")) {
520 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 {
521 continue;
522 };
523 let session = f.file_stem().map(|s| s.to_string_lossy().to_string()).unwrap_or_default();
524 let base =
525 || Event { ms: at.as_millis() as i64, session: session.clone(), engineer: who.get(&session).cloned().unwrap_or_default(), ..Default::default() };
526 for line in std::fs::read_to_string(&f).unwrap_or_default().lines() {
527 if let Some(score) = line.strip_prefix("#borg-skip ") {
528 out.push(Event { kind: TURN_END.into(), ..base() });
529 out.push(Event { kind: GATE_SKIP.into(), why: score.trim().into(), ..base() });
530 } else if let Some(cost) = line.strip_prefix("#borg-predict-cost ") {
531 out.push(Event { kind: TURN_END.into(), ..base() });
532 out.push(Event { kind: CHECK.into(), usd: cost.trim().parse().unwrap_or(0.0), ..base() });
533 }
534 }
535 }
536 out
537}
538
539pub fn node(home: &Path) -> String {
542 let path = home.join("stats").join("node");
543 if let Ok(id) = std::fs::read_to_string(&path)
544 && !id.trim().is_empty()
545 {
546 return id.trim().to_string();
547 }
548 let id = fresh_id();
552 let _ = std::fs::create_dir_all(path.parent().unwrap_or(home));
553 let tmp = path.with_extension(format!("{id}.tmp"));
554 let linked = std::fs::write(&tmp, &id).and_then(|()| std::fs::hard_link(&tmp, &path));
555 let _ = std::fs::remove_file(&tmp);
556 match linked {
557 Err(e) if e.kind() == std::io::ErrorKind::AlreadyExists => {
558 std::fs::read_to_string(&path).map(|s| s.trim().to_string()).ok().filter(|s| !s.is_empty()).unwrap_or(id)
559 }
560 _ => id,
561 }
562}
563
564fn fresh_id() -> String {
565 let mut bytes = [0u8; 8];
566 if let Ok(mut f) = std::fs::File::open("/dev/urandom") {
567 use std::io::Read;
568 let _ = f.read_exact(&mut bytes);
569 }
570 if bytes == [0; 8] {
571 use std::hash::{BuildHasher, Hasher};
572 let mut h = std::collections::hash_map::RandomState::new().build_hasher();
573 h.write_i64(crate::learn::now_ms());
574 h.write_u32(std::process::id());
575 bytes = h.finish().to_le_bytes();
576 }
577 bytes.iter().map(|b| format!("{b:02x}")).collect()
578}
579
580pub fn day(ms: i64) -> String {
582 let s = crate::install::utc(ms);
583 format!("{}-{}-{}", &s[0..4], &s[4..6], &s[6..8])
584}
585
586pub fn days_between(first: &str, last: &str) -> Vec<String> {
588 let (Some(a), Some(b)) = (day_number(first), day_number(last)) else { return Vec::new() };
589 (a..=b).map(|d| day(d * crate::clock::DAY_MS)).collect()
590}
591
592fn day_number(date: &str) -> Option<i64> {
594 let mut it = date.split('-').map(|p| p.parse::<i64>().ok());
595 let (y, m, d) = (it.next()??, it.next()??, it.next()??);
596 Some(crate::clock::days_from_civil(y, m, d))
597}
598
599#[derive(Debug, Clone, Copy, Default, PartialEq, Eq)]
601pub struct Said {
602 pub corrections: u64,
606 pub vetoes: u64,
608}
609
610pub fn per_session(events: &[Event]) -> BTreeMap<String, Said> {
614 let mut out: BTreeMap<String, Said> = BTreeMap::new();
615 for e in events.iter().filter(|e| !e.session.is_empty()) {
616 match e.kind.as_str() {
617 CORRECTION if e.why != "regret" && e.why != crate::learn::ANSWER => out.entry(e.session.clone()).or_default().corrections += 1,
618 VETO => out.entry(e.session.clone()).or_default().vetoes += 1,
619 _ => {}
620 }
621 }
622 out
623}
624
625pub fn fold(node: &str, events: &[Event]) -> BTreeMap<String, Bucket> {
627 let mut out: BTreeMap<String, Bucket> = BTreeMap::new();
628 for e in events {
629 let day = day(e.ms);
630 let id = format!("stat:{node}:{}:{day}", e.engineer);
631 let b = out.entry(id.clone()).or_insert_with(|| Bucket { id, day, node: node.into(), engineer: e.engineer.clone(), ..Default::default() });
632 let kind = if e.kind == CORRECTION && e.why == crate::learn::ANSWER { ANSWERED } else { e.kind.as_str() };
634 *b.counts.entry(kind.to_string()).or_default() += 1;
635 if e.n > 0 {
636 *b.counts.entry(format!("{}_n", e.kind)).or_default() += e.n;
637 }
638 b.usd += e.usd;
639 if e.usd > 0.0 {
640 *b.usd_by.entry(e.kind.clone()).or_default() += e.usd;
641 }
642 if e.usd > 0.0 && (e.kind == CHECK || e.kind == LIVE_CHECK) {
644 let mode = if e.mode.is_empty() { "hook" } else { e.mode.as_str() };
645 *b.usd_mode.entry(mode.into()).or_default() += e.usd;
646 }
647 if !e.channel.is_empty() {
648 *b.channels.entry(format!("{}@{}", e.kind, e.channel)).or_default() += 1;
649 }
650 b.dur_ms += e.dur_ms;
651 for l in &e.lessons {
652 *b.lessons.entry(l.clone()).or_default().entry(e.kind.clone()).or_default() += 1;
653 }
654 }
655 out
656}
657
658pub fn rollups(home: &Path) -> BTreeMap<String, Bucket> {
662 let mut out: BTreeMap<String, (Bucket, [u8; 32])> = BTreeMap::new();
663 for l in std::fs::read_to_string(rollups_path(home)).unwrap_or_default().lines() {
664 let Ok(b) = serde_json::from_str::<Bucket>(l) else { continue };
665 let h = *blake3::hash(l.as_bytes()).as_bytes();
666 if out.get(&b.id).is_none_or(|(old, oh)| (b.version, h) > (old.version, *oh)) {
667 out.insert(b.id.clone(), (b, h));
668 }
669 }
670 out.into_iter().map(|(id, (b, _))| (id, b)).collect()
671}
672
673pub fn rollup(home: &Path) -> usize {
676 backfill(home);
677 let evs = events(home);
679 if evs.is_empty() {
680 return 0;
681 }
682 let me = node(home);
683 let published = rollups(home);
684 let fresh: Vec<Bucket> = fold(&me, &evs)
685 .into_values()
686 .filter_map(|mut b| {
687 let old = published.get(&b.id);
688 if old.is_some_and(|o| o.same(&b)) {
689 return None;
690 }
691 b.version = old.map_or(1, |o| o.version + 1);
692 Some(b)
693 })
694 .collect();
695 if fresh.is_empty() {
696 return 0;
697 }
698 let lines: String = fresh.iter().map(|b| serde_json::to_string(b).unwrap_or_default() + "\n").collect();
699 let path = rollups_path(home);
700 let ok = std::fs::OpenOptions::new().create(true).append(true).open(&path).and_then(|mut f| f.write_all(lines.as_bytes()));
701 if ok.is_ok() { fresh.len() } else { 0 }
702}
703
704pub fn current(home: &Path) -> Vec<Bucket> {
707 backfill(home);
708 let me = node(home);
709 let mut all: BTreeMap<String, Bucket> = rollups(home).into_iter().filter(|(_, b)| b.node != me).collect();
710 all.extend(fold(&me, &events(home)));
711 all.into_values().collect()
712}
713
714#[cfg(test)]
715mod tests {
716 use super::*;
717
718 #[test]
719 fn corrections_and_vetoes_count_per_session_merged_or_not() {
720 let ev = |kind: &str, session: &str, why: &str| Event { kind: kind.into(), session: session.into(), why: why.into(), ..Default::default() };
721 let events = [
722 ev(CORRECTION, "a", "pushback"),
723 ev(CORRECTION, "a", "rejected_tool"),
724 ev(CORRECTION, "a", "regret"),
725 ev(CORRECTION, "a", crate::learn::ANSWER),
726 ev(VETO, "a", "prompt"),
727 ev(VETO, "b", "cli"),
728 ev(CHANGED, "b", ""),
729 ev(CORRECTION, "", "pushback"),
730 ];
731 let got = per_session(&events);
732 assert_eq!(got["a"], Said { corrections: 2, vetoes: 1 }, "an answer to the drone's own question is no correction");
733 assert_eq!(got["b"], Said { corrections: 0, vetoes: 1 });
734 assert_eq!(got.len(), 2);
735 let buckets = fold("n", &events);
737 let b = buckets.values().next().unwrap();
738 assert_eq!((b.counts[CORRECTION], b.counts[ANSWERED]), (4, 1));
739 }
740
741 #[test]
742 fn the_history_derived_from_episodes_tells_answers_apart() {
743 let home = std::env::temp_dir().join(format!("borg-stats-answer-{}", std::process::id()));
744 let _ = std::fs::remove_dir_all(&home);
745 std::fs::create_dir_all(&home).unwrap();
746 let eps = [
747 serde_json::json!({"id":"e1","ms":1,"repo":"r","session":"s","task":"t","agent":"Which backoff?","rejected_tool":null,"correction":"Exponential","via":"answer"}),
748 serde_json::json!({"id":"e2","ms":2,"repo":"r","session":"s","task":"t","agent":"a","rejected_tool":null,"correction":"no, not like that"}),
749 ];
750 std::fs::write(home.join("episodes.jsonl"), eps.iter().map(|e| e.to_string() + "\n").collect::<String>()).unwrap();
751 let whys: Vec<String> = derive_history(&home).into_iter().filter(|e| e.kind == CORRECTION).map(|e| e.why).collect();
752 assert_eq!(whys, [ANSWERED, "pushback"]);
753 let _ = std::fs::remove_dir_all(&home);
754 }
755
756 #[test]
757 fn concurrent_first_calls_agree_on_one_node() {
758 for round in 0..20 {
761 let home = std::env::temp_dir().join(format!("borg-stats-node-{}-{round}", std::process::id()));
762 let _ = std::fs::remove_dir_all(&home);
763 let ids: Vec<String> = std::thread::scope(|s| {
764 let hs: Vec<_> = (0..16).map(|_| s.spawn(|| node(&home))).collect();
765 hs.into_iter().map(|h| h.join().unwrap()).collect()
766 });
767 assert!(ids.iter().all(|id| *id == ids[0] && id.len() == 16), "{ids:?}");
768 assert_eq!(node(&home), ids[0]);
769 let _ = std::fs::remove_dir_all(&home);
770 }
771 }
772
773 fn tmp(tag: &str) -> PathBuf {
774 let d = std::env::temp_dir().join(format!("borg-stats-{tag}-{}", std::process::id()));
775 let _ = std::fs::remove_dir_all(&d);
776 std::fs::create_dir_all(&d).unwrap();
777 d
778 }
779
780 fn ev(kind: &str, ms: i64, engineer: &str, lessons: &[&str], n: u64) -> Event {
781 Event { engineer: engineer.into(), lessons: lessons.iter().map(|s| s.to_string()).collect(), n, ..Event::new(kind, ms) }
782 }
783
784 const DAY: i64 = crate::clock::DAY_MS;
786
787 #[test]
788 fn days_between_fills_every_day() {
789 assert_eq!(days_between("2026-02-27", "2026-03-02"), ["2026-02-27", "2026-02-28", "2026-03-01", "2026-03-02"]);
790 assert_eq!(days_between("1970-01-01", "1970-01-01"), ["1970-01-01"]);
791 assert!(days_between("x", "1970-01-01").is_empty());
792 }
793
794 #[test]
795 fn day_is_utc_date() {
796 assert_eq!(day(0), "1970-01-01");
797 assert_eq!(day(DAY - 1), "1970-01-01");
798 assert_eq!(day(DAY), "1970-01-02");
799 }
800
801 #[test]
802 fn fold_counts_per_engineer_and_day() {
803 let evs = [
804 ev(TURN_END, 10, "a@x", &[], 0),
805 ev(CHECK, 20, "a@x", &[], 7),
806 ev(BLOCK, 30, "a@x", &["L1", "L2"], 2),
807 ev(TURN_END, DAY + 5, "a@x", &[], 0),
808 ev(TURN_END, 40, "b@x", &[], 0),
809 ];
810 let b = fold("n1", &evs);
811 assert_eq!(b.len(), 3, "two days for a, one for b");
812 let a = &b["stat:n1:a@x:1970-01-01"];
813 assert_eq!((a.counts[TURN_END], a.counts[CHECK], a.counts["check_n"], a.counts["block_n"]), (1, 1, 7, 2));
814 assert_eq!(a.lessons["L1"][BLOCK], 1);
815 }
816
817 #[test]
818 fn fold_splits_check_cost_by_mode_and_asks_by_channel() {
819 let with = |kind: &str, usd: f64, mode: &str, channel: &str| Event { usd, mode: mode.into(), channel: channel.into(), ..ev(kind, 10, "a@x", &[], 0) };
820 let evs = [
821 with(CHECK, 0.05, "", ""),
822 with(CHECK, 0.04, "live", ""),
823 with(LIVE_CHECK, 0.03, "live", ""),
824 with(MODEL_CALL, 0.5, "live", ""),
825 with(BLOCK, 0.0, "live", "stop"),
826 with(LIVE_ASK, 0.0, "live", "live"),
827 with(CHANGED, 0.0, "", "live"),
828 ];
829 let b = &fold("n1", &evs)["stat:n1:a@x:1970-01-01"];
830 assert!((b.usd_mode["hook"] - 0.05).abs() < 1e-9 && (b.usd_mode["live"] - 0.07).abs() < 1e-9, "{:?}", b.usd_mode);
831 assert_eq!(b.usd_mode.len(), 2, "only the conscience's checks: {:?}", b.usd_mode);
832 assert_eq!((b.channels["block@stop"], b.channels["live_ask@live"], b.channels["changed@live"]), (1, 1, 1));
833 assert_eq!((b.counts[CHECK], b.counts[LIVE_CHECK]), (2, 1), "live checks are counted apart from turn-end checks");
834 }
835
836 #[test]
837 fn rollup_versions_only_on_change_and_is_idempotent() {
838 let home = tmp("rollup");
839 record(&home, &ev(TURN_END, 10, "a@x", &[], 0));
840 assert_eq!(rollup(&home), 1);
841 assert_eq!(rollup(&home), 0, "nothing changed: nothing appended");
842 record(&home, &ev(GATE_SKIP, 11, "a@x", &[], 0));
843 assert_eq!(rollup(&home), 1);
844 let r = rollups(&home);
845 let b = r.values().next().unwrap();
846 assert_eq!((b.version, b.counts[TURN_END], b.counts[GATE_SKIP]), (2, 1, 1));
847 let text = std::fs::read_to_string(rollups_path(&home)).unwrap();
848 assert!(!text.contains("session"), "no session ids leave the machine");
849 }
850
851 #[test]
852 fn current_prefers_own_fresh_counts_and_keeps_peers() {
853 let home = tmp("current");
854 record(&home, &ev(TURN_END, 10, "a@x", &[], 0));
855 rollup(&home);
856 record(&home, &ev(TURN_END, 12, "a@x", &[], 0));
857 let peer = Bucket {
858 id: "stat:other:b@x:1970-01-01".into(),
859 version: 3,
860 day: "1970-01-01".into(),
861 node: "other".into(),
862 engineer: "b@x".into(),
863 ..Default::default()
864 };
865 let mut f = std::fs::OpenOptions::new().append(true).open(rollups_path(&home)).unwrap();
866 writeln!(f, "{}", serde_json::to_string(&peer).unwrap()).unwrap();
867 let cur = current(&home);
868 assert_eq!(cur.len(), 2);
869 let mine = cur.iter().find(|b| b.engineer == "a@x").unwrap();
870 assert_eq!(mine.counts[TURN_END], 2, "own counts are folded fresh, not the stale rollup");
871 }
872
873 #[test]
876 fn rollups_break_version_ties_by_hash_whatever_the_order() {
877 let bucket = |n: u64| {
878 let mut b = Bucket { id: "stat:n:a@x:1970-01-01".into(), version: 2, ..Default::default() };
879 b.counts.insert(TURN_END.into(), n);
880 serde_json::to_string(&b).unwrap()
881 };
882 let (a, b) = (bucket(1), bucket(2));
883 let shown = |name: &str, first: &str, second: &str| {
884 let home = tmp(name);
885 std::fs::write(rollups_path(&home), format!("{first}\n{second}\n")).unwrap();
886 rollups(&home).into_values().next().unwrap().counts[TURN_END]
887 };
888 assert_eq!(shown("ties-ab", &a, &b), shown("ties-ba", &b, &a));
889 }
890
891 #[test]
892 fn record_fails_open() {
893 let home = tmp("open");
894 std::fs::write(home.join("stats"), "a file where the directory should be").unwrap();
895 record(&home, &ev(TURN_END, 1, "a@x", &[], 0));
896 assert!(events(&home).is_empty());
897 }
898
899 #[test]
900 fn backfill_derives_history_before_live_counting_once() {
901 let home = tmp("backfill");
902 let w = |f: &str, lines: &[serde_json::Value]| {
903 let p = home.join(f);
904 std::fs::create_dir_all(p.parent().unwrap()).unwrap();
905 std::fs::write(p, lines.iter().map(|l| l.to_string() + "\n").collect::<String>()).unwrap();
906 };
907 use serde_json::json;
908 w(
909 "episodes.jsonl",
910 &[
911 json!({"id":"e1","ms":100,"repo":"app","session":"s1","task":"t","agent":"a","rejected_tool":"Bash","correction":"no, not like that","engineer":"a@x"}),
912 json!({"id":"e2","ms":5000,"repo":"app","session":"s2","task":"t","agent":"a","rejected_tool":null,"correction":"after live counting began","engineer":"a@x"}),
913 ],
914 );
915 w(
916 "outcomes.jsonl",
917 &[
918 json!({"kind":"asked","session":"s1","lessons":["L1","L2"],"ms":110,"repo":"app"}),
919 json!({"kind":"asked","session":"s1","lessons":["L3"],"ms":112,"repo":"app","channel":"early"}),
921 json!({"kind":"asked","session":"s1","lessons":["L4"],"ms":114,"repo":"app","channel":"queued"}),
922 json!({"kind":"asked_then_changed","session":"s1","lessons":["L1"],"ms":120,"repo":"app"}),
923 ],
924 );
925 w(
926 "lessons.jsonl",
927 &[json!({"id":"L1","origin_ms":90,"question":"q","version":1,"engineers":["a@x"]}), json!({"id":"L1","origin_ms":90,"question":"q2","version":2})],
928 );
929 w("queen/queue.jsonl", &[json!({"id":"Q1","ms":130,"question":"which?","engineer":"a@x","answered_ms":190,"answer":"this","answered_by":"a@x"})]);
930 record(&home, &ev(TURN_END, 1000, "a@x", &[], 0));
931 let n = backfill(&home);
932 let all = read_events(&backfill_path(&home));
933 assert_eq!(n, all.len());
934 let kinds: Vec<&str> = all.iter().map(|e| e.kind.as_str()).collect();
935 assert_eq!(kinds, [LESSON_NEW, CORRECTION, BLOCK, CHANGED, QUEEN_QUEUED, QUEEN_ANSWERED], "ordered by time, the later correction left out");
936 let c = &all[1];
937 assert_eq!((c.why.as_str(), c.text.as_str(), c.engineer.as_str()), ("rejected_tool", "Bash: no, not like that", "a@x"));
938 assert_eq!(all[3].engineer, "a@x", "outcomes take the session's engineer from its corrections");
939 assert_eq!(all[5].dur_ms, 60, "time to answer");
940 assert_eq!(backfill(&home), 0, "once");
941 assert_eq!(events(&home).len(), all.len() + 1, "history and live events together");
942 }
943
944 #[test]
949 fn backfill_counts_one_event_per_call_as_live_counting_does() {
950 let home = tmp("backfill-call");
951 use serde_json::json;
952 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});
953 let log = [
954 rec("asked", &["L1", "L2"], 110, "stop", "c1"),
955 rec("asked", &["L3"], 110, "stop", "c2"),
956 rec("asked", &["L4"], 120, "live", "c3"),
957 rec("asked", &["L5"], 120, "live", "c4"),
958 rec("asked_then_changed", &["L1"], 130, "stop", "c1"),
959 rec("asked_then_changed", &["L3"], 130, "stop", "c2"),
960 rec("asked", &["L6"], 140, "stop", "c5"),
961 rec("asked_not_corrected", &["L2"], 150, "stop", "c1"),
962 rec("asked_not_corrected", &["L6"], 150, "stop", "c5"),
963 ];
964 std::fs::write(home.join("outcomes.jsonl"), log.iter().map(|l| l.to_string() + "\n").collect::<String>()).unwrap();
965 record(&home, &ev(TURN_END, 1000, "a@x", &[], 0));
966 backfill(&home);
967 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();
968 let row = |kind: &str, ms: i64, n: u64, lessons: &[&str]| (kind.to_string(), ms, n, lessons.iter().map(|l| l.to_string()).collect::<Vec<_>>());
969 assert_eq!(
970 all,
971 [
972 row(BLOCK, 110, 3, &["L1", "L2", "L3"]),
973 row(LIVE_ASK, 120, 2, &["L4", "L5"]),
974 row(CHANGED, 130, 0, &["L1", "L3"]),
975 row(BLOCK, 140, 1, &["L6"]),
976 row(NOT_CORRECTED, 150, 0, &["L2"]),
977 row(NOT_CORRECTED, 150, 0, &["L6"]),
978 ]
979 );
980 }
981
982 #[test]
983 fn node_id_is_stable() {
984 let home = tmp("node");
985 let a = node(&home);
986 assert_eq!(a.len(), 16);
987 assert_eq!(node(&home), a);
988 }
989}