@chatpanel/events 0.102.0 → 0.103.1

package/entity.js

@@ -1,354 +1 @@
-// ENTITY IDENTITY — forty mentions of one person are one subject, or they are nothing.
-//
-// Extraction we already have: `meeting-people.js` names the speakers on a call,
-// `extraction.js` pulls topics and ENTITIES, `tags.js` folds the filing vocabulary. What
-// none of them answer is the next question: are "Alex Rivera", "alex rivera" and the
-// "Alex" who spoke in yesterday's standup the SAME subject? Without that answer a derived
-// layer cannot exist — a brief is by definition an accumulation across records, so it needs
-// something to accumulate *about*.
-//
-// Two rules do almost all the work, and both are deliberately conservative:
-//
-//   • CANONICAL FORM IS LOSSY, IDENTITY IS NOT. `subjectKey()` folds case, punctuation and
-//     spacing — filing noise, exactly as in tags.js — but never folds two different names
-//     together. The display name stays whatever the corpus said most often, so the user
-//     reads "Alex Rivera" and not "alex-rivera".
-//   • AN ALIAS IS ONLY AN ALIAS WHEN IT IS UNAMBIGUOUS. "Alex" folds into "Alex Rivera"
-//     when Alex Rivera is the only Alex in the corpus. The moment a second Alex appears,
-//     the bare token stops resolving to either — for good, and retroactively. Guessing here
-//     is how a brief ends up attributing one person's decisions to another, which is the
-//     single most expensive mistake this layer can make.
-//
-// Pure input → output: no storage, no clock, no model. The extension, the gateway and a
-// future mobile client must agree on what "the same subject" is, and three implementations
-// would mean three answers — the argument tags.js already makes for tags.
-
-import { blockedPairs, editDistance } from './distance.js';
-// The placeholder recognizer lives on its own (see that file for why). Re-exported so every
-// existing caller of entity.js is unchanged.
-import { isRedactionToken } from './redaction-tokens.js';
-
-export { REDACTION_TOKEN_TYPES, isRedactionToken } from './redaction-tokens.js';
-
-// Naming and sizing a subject live next door, so a caller that needs only those does not
-// drag alias resolution and a Levenshtein along. Re-exported: no caller of entity.js changes.
-import {
-  DEFAULT_THRESHOLD, MAX_SUBJECTS, MAX_SUBJECT_CHARS, SELF_LABELS,
-  isSelfLabel, normalizeSubject, stripQualifiers,
-} from './subject-name.js';
-import { SUBJECT_KINDS } from './subject-kinds.js';
-
-export { SUBJECT_KINDS } from './subject-kinds.js';
-export {
-  DEFAULT_THRESHOLD, MAX_SUBJECTS, MAX_SUBJECT_CHARS, SELF_LABELS, isSelfLabel,
-  normalizeSubject, stripQualifiers, subjectKey, subjectTokens,
-} from './subject-name.js';
-
-
-// Words that are never a subject on their own. A one-token candidate has to survive this
-// list before it can become a page, because "notes", "meeting" and "update" appear in every
-// record and would each clear any threshold instantly.
-const STOP_SUBJECTS = new Set([
-  'chat', 'chats', 'note', 'notes', 'meeting', 'meetings', 'call', 'calls', 'update',
-  'updates', 'summary', 'summaries', 'agenda', 'todo', 'todos', 'task', 'tasks', 'item',
-  'items', 'thing', 'things', 'stuff', 'misc', 'other', 'general', 'test', 'testing',
-  'untitled', 'draft', 'drafts', 'new', 'old', 'today', 'yesterday', 'tomorrow', 'week',
-  'day', 'month', 'year', 'time', 'people', 'person', 'team', 'work',
-]);
-
-/**
- * Is this string worth considering as a subject at all?
- *
- * Rejects blanks, over-long phrases, pure numbers and the stop list above. A multi-token
- * phrase is allowed even when one of its tokens is a stop word ("design review" is a real
- * topic, "review" alone is not).
- */
-export function isSubjectCandidate(name, { kind = 'topic' } = {}) {
-  // Length is judged BEFORE folding: normalizeSubject truncates at MAX_SUBJECT_CHARS, so a
-  // check on its output can never fire and a whole sentence would slip through as a subject.
-  if (String(name ?? '').trim().length > MAX_SUBJECT_CHARS) return false;
-  // Checked on the RAW value, before folding: normalizeSubject lowercases, and the pattern
-  // is upper-case by construction.
-  if (isRedactionToken(name)) return false;
-  // "You" is a pronoun, not a person. Without a `self` name to fold it into (resolveSubjects
-  // takes one), it must not become a subject of its own — every meeting has a "You" and they
-  // are not all the same participant.
-  if (kind === 'person' && isSelfLabel(name)) return false;
-  const norm = normalizeSubject(name);
-  if (!norm || norm.length < 2) return false;
-  const tokens = norm.split(' ').filter(Boolean);
-  if (!tokens.length) return false;
-  if (tokens.every((t) => /^\p{N}+$/u.test(t))) return false;
-  if (tokens.length === 1 && STOP_SUBJECTS.has(tokens[0])) return false;
-  // A person needs at least two characters of actual letters — "j r" is initials, not an
-  // identity we can accumulate against.
-  if (kind === 'person' && !/\p{L}{2}/u.test(norm)) return false;
-  return true;
-}
-
-/**
- * Resolve short forms to full names, but ONLY when the corpus leaves no doubt.
- *
- * Given the canonical names seen for one kind, returns `alias -> canonical`. A bare token
- * maps to a multi-token name when it is that name's first or last token AND exactly one
- * name in the corpus claims it. Two people called Alex means neither owns "alex", so the
- * bare token maps to nothing and stays a subject of its own — visible, unmerged, and
- * therefore checkable, which is the failure mode we want.
- */
-export function aliasMap(names) {
-  const canonical = new Set();
-  for (const n of names || []) {
-    const norm = normalizeSubject(n);
-    if (norm && norm.includes(' ')) canonical.add(norm);
-  }
-  const claims = new Map(); // token -> Set(full names claiming it)
-  for (const full of canonical) {
-    const tokens = full.split(' ');
-    for (const t of [tokens[0], tokens[tokens.length - 1]]) {
-      if (!t || t.length < 2 || STOP_SUBJECTS.has(t)) continue;
-      if (!claims.has(t)) claims.set(t, new Set());
-      claims.get(t).add(full);
-    }
-  }
-  const out = new Map();
-  for (const [token, owners] of claims) {
-    if (owners.size !== 1) continue; // ambiguous — resolve to nothing, on purpose
-    out.set(token, [...owners][0]);
-  }
-  return out;
-}
-
-/**
- * Fold a list of raw mentions into subjects.
- *
- * `mentions` is `[{ kind, name, recordId }]` — whatever the corpus said, in whatever form.
- * The result is one entry per identity, carrying every surface form that reached it, the
- * distinct records it appeared in, and the display name the corpus used most often.
- *
- * Aliases are resolved per KIND: two topics can share a word without being the same topic,
- * and the person rule above must not leak into tags.
- */
-export function resolveSubjects(mentions = [], { merges = null, self = '' } = {}) {
-  // The user's own name, however they told us. Everything the platforms call "you" folds
-  // into it — and with no name supplied, nothing does.
-  const selfName = String(self ?? '').trim();
-  const selfCanonical = normalizeSubject(selfName);
-  // User-authored merges: `alias canonical form -> the canonical form it belongs to`. These
-  // are an INPUT to derivation, never an edit to its output, which is what keeps I-K2 true —
-  // a rebuild that dropped the user's corrections would teach them not to make any.
-  const merged = new Map();
-  for (const [from, to] of merges instanceof Map ? merges : Object.entries(merges || {})) {
-    const a = normalizeSubject(from); const b = normalizeSubject(to);
-    if (a && b && a !== b) merged.set(a, b);
-  }
-  // One hop only. A chain (a→b, b→c) is resolved here rather than at read time, and a cycle
-  // simply stops, because a merge loop must not hang a rebuild.
-  const resolveMerge = (key) => {
-    let cur = key;
-    for (let i = 0; i < 8 && merged.has(cur); i += 1) {
-      const next = merged.get(cur);
-      if (next === cur) break;
-      cur = next;
-    }
-    return cur;
-  };
-
-  const byKind = new Map();
-  for (const m of mentions) {
-    const kind = m?.kind;
-    if (!SUBJECT_KINDS.includes(kind)) continue;
-    // A self-label survives candidacy ONLY when there is a name to fold it into. Otherwise
-    // it is a pronoun, and every meeting's "You" would pile into one fictional participant.
-    const isSelf = kind === 'person' && !!selfCanonical && isSelfLabel(m.name);
-    if (!isSelf && !isSubjectCandidate(m.name, { kind })) continue;
-    if (!byKind.has(kind)) byKind.set(kind, []);
-    byKind.get(kind).push(m);
-  }
-
-  const subjects = new Map();
-  for (const [kind, list] of byKind) {
-    // Only PERSON names carry the short-form rule. A topic named "design" is not the
-    // "design review" topic, and folding them would silently merge two pages.
-    const aliases = kind === 'person'
-      ? aliasMap(list.map((m) => stripQualifiers(m.name)).filter((n) => !isSelfLabel(n)))
-      : new Map();
-    for (const m of list) {
-      const raw = String(m.name ?? '').trim();
-      // For a PERSON, the qualifier is decoration and the platform's "You" is the user.
-      // Both are resolved before the alias rule, so "Alex Rivera (ACME)" and "You" reach the
-      // same identity that a bare "Alex" does.
-      let norm = normalizeSubject(kind === 'person' ? stripQualifiers(raw) : raw);
-      if (kind === 'person' && selfCanonical && SELF_LABELS.includes(norm)) norm = selfCanonical;
-      const canonical = resolveMerge(aliases.get(norm) || norm);
-      const key = `${kind}:${canonical}`;
-      let s = subjects.get(key);
-      if (!s) {
-        s = { key, kind, name: '', canonical, aliases: [], records: new Set(), mentions: 0, forms: new Map() };
-        subjects.set(key, s);
-      }
-      s.mentions += 1;
-      if (m.recordId) s.records.add(m.recordId);
-      // The SURFACE FORM a person is displayed under is the stripped one. The canonical was
-      // already right — "Sam Okonkwo [ACME - Platform]" resolved to person:sam okonkwo — but the
-      // raw string was recorded as the display form, so a subject seen only that way got a
-      // page titled with the directory's decoration. A qualifier is not part of a name.
-      const display = String(kind === 'person' ? stripQualifiers(raw) : raw).normalize('NFKC').trim();
-      if (display) s.forms.set(display, (s.forms.get(display) || 0) + 1);
-      if (norm !== canonical && !s.aliases.includes(norm)) s.aliases.push(norm);
-    }
-  }
-
-  for (const s of subjects.values()) {
-    // The name the user reads is a surface form of the CANONICAL identity, most common
-    // first, ties broken alphabetically so the choice is stable across runs rather than
-    // insertion-ordered. Forms that only reached this subject through an alias are ranked
-    // last: `person:alex rivera` displayed as "Alex" because the short form happened to be
-    // one mention commoner would be a page whose title is not the subject's name.
-    const forms = [...s.forms.entries()]
-      .sort((a, b) => b[1] - a[1] || a[0].localeCompare(b[0]));
-    s.name = forms.find(([f]) => normalizeSubject(f) === s.canonical)?.[0]
-      || forms[0]?.[0]
-      || s.canonical;
-    s.aliases.sort();
-    delete s.forms;
-  }
-  return subjects;
-}
-
-/** Does this subject have enough evidence to deserve a page? */
-export function earnsBrief(subject, threshold = DEFAULT_THRESHOLD) {
-  if (!subject) return false;
-  const records = subject.records instanceof Set ? subject.records.size : (subject.records?.length || 0);
-  const mentions = Number(subject.mentions) || 0;
-  return records >= (threshold.records ?? DEFAULT_THRESHOLD.records)
-      && mentions >= (threshold.mentions ?? DEFAULT_THRESHOLD.mentions);
-}
-
-/**
- * Rank subjects and keep the ones that earned a page, strongest first.
- *
- * `limit` is the I-K4 count ceiling made concrete: a corpus with 4000 qualifying subjects
- * does not get 4000 briefs, it gets the best `limit` of them, and the rest stay subjects
- * without pages until the evidence moves.
- */
-export function rankSubjects(subjects, { threshold = DEFAULT_THRESHOLD, limit = MAX_SUBJECTS } = {}) {
-  const list = [...(subjects instanceof Map ? subjects.values() : subjects || [])];
-  return list
-    .filter((s) => earnsBrief(s, threshold))
-    .map((s) => ({ ...s, recordCount: s.records instanceof Set ? s.records.size : (s.records?.length || 0) }))
-    .sort((a, b) => b.recordCount - a.recordCount || b.mentions - a.mentions || a.key.localeCompare(b.key))
-    .slice(0, Math.max(0, limit));
-}
-
-/**
- * PROPOSE merges; never apply them.
- *
- * The alias rule folds what it can prove. Everything it cannot is left visible — "A. Rivera"
- * beside "Alex Rivera", an initialism beside a full name, a second spelling of a project —
- * and a user staring at two pages for one person has no way to say so. This is that way: a
- * deterministic, model-free list of pairs worth asking about, ranked by how likely they are.
- *
- * PROPOSING is the whole design. The pairs below are exactly the ones the alias rule refuses
- * to decide on its own, because deciding wrongly merges two people permanently and silently.
- * A human answers in one click, the answer is stored as a merge rule, and every later rebuild
- * applies it — so the correction survives I-K2 rather than being erased by the next pass.
- *
- * Three signals, strongest first:
- *   • initials — "A. Rivera" against "Alex Rivera"
- *   • containment — one name's tokens are a subset of the other's
- *   • near-spelling — one edit apart, or a transposition, which catches the common typos
- *
- * A shared SURNAME is deliberately not a signal: two people with one last name are usually
- * two people, and proposing every such pair would bury the real suggestions.
- */
-export function suggestMerges(subjects, { limit = 40, distance = 1 } = {}) {
-  const list = [...(subjects instanceof Map ? subjects.values() : subjects || [])]
-    .filter((s) => s && s.canonical);
-
-  // BLOCKED, not pairwise. Comparing every pair did not finish 12,000 subjects in two
-  // minutes — 6.8s at 2,000, 32s at 6,000, 141s at 12,000 — and a corpus with a few thousand
-  // distinct people, topics and tags reaches that easily. This pass runs on the UI thread,
-  // so that is a hung page, not a slow one.
-  //
-  // All three signals need either a shared end or a shared last token, which is precisely
-  // what `blockKeys` files on, so the findings survive the change: an abbreviated first name
-  // shares the surname, a contained name shares a prefix, a typo shares whichever end it is
-  // not in.
-  const byCanonical = new Map();
-  for (const s of list) {
-    if (!byCanonical.has(s.canonical)) byCanonical.set(s.canonical, []);
-    byCanonical.get(s.canonical).push(s);
-  }
-
-  const out = [];
-  for (const [ca, cb] of blockedPairs([...byCanonical.keys()])) {
-    // One reason per canonical PAIR, computed before the (rare) fan-out over subjects that
-    // share a canonical form across kinds.
-    for (const a of byCanonical.get(ca)) {
-      for (const b of byCanonical.get(cb)) {
-        if (a.kind !== b.kind) continue; // a person and a topic are never the same subject
-        const reason = mergeReason(a.canonical, b.canonical, a.kind, distance);
-        if (!reason) continue;
-        // The better-evidenced side is proposed as the survivor: it has more records behind
-        // it and is more likely the name the user actually thinks in.
-        const [keep, drop] = countOf(a) >= countOf(b) ? [a, b] : [b, a];
-        out.push({ kind: a.kind, keep: keep.key, keepName: keep.name, drop: drop.key, dropName: drop.name, reason });
-      }
-    }
-  }
-
-  const rank = { initials: 0, containment: 1, spelling: 2 };
-  return out
-    .sort((x, y) => (rank[x.reason] - rank[y.reason]) || x.keepName.localeCompare(y.keepName))
-    .slice(0, Math.max(0, limit));
-}
-
-function countOf(s) {
-  return s.records instanceof Set ? s.records.size : (s.records?.length || 0);
-}
-
-const sortedChars = (s) => [...s.replace(/ /g, '')].sort().join('');
-// "q3 planning" and "q4 planning" are one edit apart and are not the same subject; nor are
-// "phase 2" and "phase 3", or "atlas sync 1" and "atlas sync 2". If masking the digits makes
-// two names identical, the difference IS the digits, and that is a SERIES. duplicateTitles
-// makes the same argument for trailing numbers; this is the in-word case.
-const digitSeries = (a, b) => a !== b && a.replace(/\d/g, '#') === b.replace(/\d/g, '#');
-
-function mergeReason(a, b, kind, distance) {
-  if (a === b) return null;
-  const ta = a.split(' ').filter(Boolean);
-  const tb = b.split(' ').filter(Boolean);
-
-  // "a. rivera" / "ar" against "alex rivera" — same last token, first token abbreviates.
-  const last = ta[ta.length - 1] === tb[tb.length - 1];
-  if (last && ta.length > 1 && tb.length > 1) {
-    const [fa, fb] = [ta[0], tb[0]];
-    if (fa !== fb && (fa.startsWith(fb) || fb.startsWith(fa))) return 'initials';
-  }
-  if (ta.length === 1 && tb.length === 1 && ta[0] !== tb[0]) {
-    const [short, long] = ta[0].length <= tb[0].length ? [ta[0], tb[0]] : [tb[0], ta[0]];
-    if (short.length >= 2 && long.startsWith(short)) return 'initials';
-  }
-
-  // One name's tokens are all present in the other's.
-  const setA = new Set(ta); const setB = new Set(tb);
-  const [small, big] = setA.size <= setB.size ? [setA, setB] : [setB, setA];
-  if (small.size && [...small].every((t) => big.has(t)) && small.size !== big.size) return 'containment';
-
-  if (digitSeries(a, b)) return null;
-
-  // Too short for an edit to mean anything — see duplicateTitles for the same guard.
-  if (Math.min(a.length, b.length) > distance * 3) {
-    if (editDistance(a, b, distance) <= distance) return 'spelling';
-    // A transposition costs TWO edits in Levenshtein, and "atals" for "atlas" is the single
-    // commonest typo there is. Admitted only when the two are anagrams, so widening the
-    // budget cannot also admit "q3 planning" against "q4 planning".
-    if (editDistance(a, b, 2) <= 2 && sortedChars(a) === sortedChars(b)) return 'spelling';
-  }
-
-  // A SHARED SURNAME IS NOT A SIGNAL, and it was tempting. Two people with the same last
-  // name are usually two people — colleagues, relatives — so proposing every such pair on
-  // every rebuild would bury the three real suggestions under thirty. A list nobody reads
-  // is worse than no list, which is the same finding that keeps briefs bounded.
-  return null;
-}
+import{blockedPairs as L,editDistance as T}from"./distance.js";import{isRedactionToken as z}from"./redaction-tokens.js";import{REDACTION_TOKEN_TYPES as j,isRedactionToken as Q}from"./redaction-tokens.js";import{DEFAULT_THRESHOLD as h,MAX_SUBJECTS as A,MAX_SUBJECT_CHARS as N,SELF_LABELS as x,isSelfLabel as w,normalizeSubject as S,stripQualifiers as E}from"./subject-name.js";import{SUBJECT_KINDS as U}from"./subject-kinds.js";import{SUBJECT_KINDS as q}from"./subject-kinds.js";import{DEFAULT_THRESHOLD as V,MAX_SUBJECTS as Z,MAX_SUBJECT_CHARS as ee,SELF_LABELS as ne,isSelfLabel as te,normalizeSubject as oe,stripQualifiers as se,subjectKey as re,subjectTokens as ie}from"./subject-name.js";const y=new Set(["chat","chats","note","notes","meeting","meetings","call","calls","update","updates","summary","summaries","agenda","todo","todos","task","tasks","item","items","thing","things","stuff","misc","other","general","test","testing","untitled","draft","drafts","new","old","today","yesterday","tomorrow","week","day","month","year","time","people","person","team","work"]);function D(n,{kind:r="topic"}={}){if(String(n??"").trim().length>N||z(n)||r==="person"&&w(n))return!1;const l=S(n);if(!l||l.length<2)return!1;const f=l.split(" ").filter(Boolean);return!(!f.length||f.every(e=>new RegExp("^\\p{N}+$","u").test(e))||f.length===1&&y.has(f[0])||r==="person"&&!new RegExp("\\p{L}{2}","u").test(l))}function R(n){const r=new Set;for(const e of n||[]){const o=S(e);o&&o.includes(" ")&&r.add(o)}const l=new Map;for(const e of r){const o=e.split(" ");for(const m of[o[0],o[o.length-1]])!m||m.length<2||y.has(m)||(l.has(m)||l.set(m,new Set),l.get(m).add(e))}const f=new Map;for(const[e,o]of l)o.size===1&&f.set(e,[...o][0]);return f}function b(n=[],{merges:r=null,self:l=""}={}){const f=String(l??"").trim(),e=S(f),o=new Map;for(const[t,s]of r instanceof Map?r:Object.entries(r||{})){const i=S(t),a=S(s);i&&a&&i!==a&&o.set(i,a)}const m=t=>{let s=t;for(let i=0;i<8&&o.has(s);i+=1){const a=o.get(s);if(a===s)break;s=a}return s},c=new Map;for(const t of n){const s=t?.kind;!U.includes(s)||!(s==="person"&&e&&w(t.name))&&!D(t.name,{kind:s})||(c.has(s)||c.set(s,[]),c.get(s).push(t))}const u=new Map;for(const[t,s]of c){const i=t==="person"?R(s.map(a=>E(a.name)).filter(a=>!w(a))):new Map;for(const a of s){const d=String(a.name??"").trim();let g=S(t==="person"?E(d):d);t==="person"&&e&&x.includes(g)&&(g=e);const k=m(i.get(g)||g),C=`${t}:${k}`;let p=u.get(C);p||(p={key:C,kind:t,name:"",canonical:k,aliases:[],records:new Set,mentions:0,forms:new Map},u.set(C,p)),p.mentions+=1,a.recordId&&p.records.add(a.recordId);const M=String(t==="person"?E(d):d).normalize("NFKC").trim();M&&p.forms.set(M,(p.forms.get(M)||0)+1),g!==k&&!p.aliases.includes(g)&&p.aliases.push(g)}}for(const t of u.values()){const s=[...t.forms.entries()].sort((i,a)=>a[1]-i[1]||i[0].localeCompare(a[0]));t.name=s.find(([i])=>S(i)===t.canonical)?.[0]||s[0]?.[0]||t.canonical,t.aliases.sort(),delete t.forms}return u}function v(n,r=h){if(!n)return!1;const l=n.records instanceof Set?n.records.size:n.records?.length||0,f=Number(n.mentions)||0;return l>=(r.records??h.records)&&f>=(r.mentions??h.mentions)}function X(n,{threshold:r=h,limit:l=A}={}){return[...n instanceof Map?n.values():n||[]].filter(e=>v(e,r)).map(e=>({...e,recordCount:e.records instanceof Set?e.records.size:e.records?.length||0})).sort((e,o)=>o.recordCount-e.recordCount||o.mentions-e.mentions||e.key.localeCompare(o.key)).slice(0,Math.max(0,l))}function P(n,{limit:r=40,distance:l=1}={}){const f=[...n instanceof Map?n.values():n||[]].filter(c=>c&&c.canonical),e=new Map;for(const c of f)e.has(c.canonical)||e.set(c.canonical,[]),e.get(c.canonical).push(c);const o=[];for(const[c,u]of L([...e.keys()]))for(const t of e.get(c))for(const s of e.get(u)){if(t.kind!==s.kind)continue;const i=O(t.canonical,s.canonical,t.kind,l);if(!i)continue;const[a,d]=_(t)>=_(s)?[t,s]:[s,t];o.push({kind:t.kind,keep:a.key,keepName:a.name,drop:d.key,dropName:d.name,reason:i})}const m={initials:0,containment:1,spelling:2};return o.sort((c,u)=>m[c.reason]-m[u.reason]||c.keepName.localeCompare(u.keepName)).slice(0,Math.max(0,r))}function _(n){return n.records instanceof Set?n.records.size:n.records?.length||0}const B=n=>[...n.replace(/ /g,"")].sort().join(""),J=(n,r)=>n!==r&&n.replace(/\d/g,"#")===r.replace(/\d/g,"#");function O(n,r,l,f){if(n===r)return null;const e=n.split(" ").filter(Boolean),o=r.split(" ").filter(Boolean);if(e[e.length-1]===o[o.length-1]&&e.length>1&&o.length>1){const[i,a]=[e[0],o[0]];if(i!==a&&(i.startsWith(a)||a.startsWith(i)))return"initials"}if(e.length===1&&o.length===1&&e[0]!==o[0]){const[i,a]=e[0].length<=o[0].length?[e[0],o[0]]:[o[0],e[0]];if(i.length>=2&&a.startsWith(i))return"initials"}const c=new Set(e),u=new Set(o),[t,s]=c.size<=u.size?[c,u]:[u,c];return t.size&&[...t].every(i=>s.has(i))&&t.size!==s.size?"containment":J(n,r)?null:Math.min(n.length,r.length)>f*3&&(T(n,r,f)<=f||T(n,r,2)<=2&&B(n)===B(r))?"spelling":null}export{V as DEFAULT_THRESHOLD,Z as MAX_SUBJECTS,ee as MAX_SUBJECT_CHARS,j as REDACTION_TOKEN_TYPES,ne as SELF_LABELS,q as SUBJECT_KINDS,R as aliasMap,v as earnsBrief,Q as isRedactionToken,te as isSelfLabel,D as isSubjectCandidate,oe as normalizeSubject,X as rankSubjects,b as resolveSubjects,se as stripQualifiers,re as subjectKey,ie as subjectTokens,P as suggestMerges};

package/event.js

@@ -1,175 +1 @@
-// The event envelope — durable facts only.
-//
-// ORDER WITHOUT CLOCKS. `at` is wall time and is ADVISORY: it is shown to humans and
-// never consulted for ordering. `(host, seq)` plus `causes` are the authority, because
-// once two hosts append concurrently a timestamp is not an order (see order.js).
-//
-// EPHEMERAL STREAMS ARE NOT THIS TYPE. Market ticks, live captions and DOM mutations
-// live in an in-memory ring buffer with no id, no seq and no persistence; only the
-// windowed aggregate that entered a model request or crossed the device boundary is
-// promoted to an Event. That is structural rather than a rule to remember: durably
-// logging a caption stream is ~3 MB per meeting, ~5.5 GB/year, and there is no Event
-// type that would accept one.
-//
-// METADATA ONLY. Payloads carry Refs and counts, never content. `privacy.redacted`
-// carries how many of each entity type were redacted and never the values — a log of
-// what was redacted must not itself contain the redacted data.
-
-import { isRef } from './ref.js';
-
-export const CURRENT_VERSION = 1;
-
-/** The seven families of v1. Adding is cheap; removing is not. */
-export const EVENT_TYPES = Object.freeze({
-  turn: ['started', 'ended'],
-  // What the model was SHOWN and what it SAID. Deliberately arriving as refs, not text —
-  // see the note on the validators below.
-  assistant: ['prompted', 'message', 'reasoning'],
-  context: ['assembled', 'attached', 'expanded'],
-  capability: ['offered', 'granted', 'denied', 'activated', 'revoked', 'invoked', 'resulted'],
-  privacy: ['redacted', 'egress'],
-  policy: ['changed', 'guard_denied'],
-  data: ['created', 'updated', 'deleted'],
-  automation: ['fired', 'suppressed'],
-});
-
-export const ALL_TYPES = Object.freeze(
-  Object.entries(EVENT_TYPES).flatMap(([fam, kinds]) => kinds.map((k) => `${fam}.${k}`)),
-);
-
-// 'channel' is a message arriving from a paired external surface — Telegram, WhatsApp — that
-// drives a turn the same way a person pressing send does. It is turn-independent for the same
-// reason 'schedule' and 'agent' are: nobody is sitting in the panel when it fires, so consent
-// and reach have to be settled at pairing time, not at the keystroke. The actor.id carries the
-// surface and the sender, e.g. 'telegram:8412…'. See chatpanel-channels for the invoker.
-export const ACTOR_KINDS = Object.freeze(['user', 'rule', 'schedule', 'model', 'agent', 'channel']);
-export const SCOPE_KINDS = Object.freeze(['global', 'site', 'tab', 'session', 'agent']);
-export const CLASSES = Object.freeze(['R', 'M', 'L', 'C', 'A', 'X', 'H']);
-export const EFFECTS = Object.freeze(['pure', 'idempotent', 'replay-safe', 'non-replayable']);
-export const EGRESS = Object.freeze(['none', 'redacted', 'delegated']);
-
-export class EventError extends Error {
-  constructor(code, message, detail = null) {
-    super(message);
-    this.name = 'EventError';
-    this.code = code;
-    this.detail = detail;
-  }
-}
-
-const str = (v) => typeof v === 'string' && v.length > 0;
-const num = (v) => typeof v === 'number' && Number.isFinite(v);
-const arr = (v) => Array.isArray(v);
-
-function actorOk(a) { return !!a && ACTOR_KINDS.includes(a.kind) && str(a.id); }
-function scopeOk(s) { return !!s && SCOPE_KINDS.includes(s.kind) && str(s.id); }
-
-// Per-type payload requirements. Deliberately light — enough that a malformed event
-// cannot enter an append-only log, not a full schema language.
-const PAYLOAD = {
-  'turn.started': (p) => str(p.turnId),
-  'turn.ended': (p) => str(p.turnId),
-
-  // ── assistant ─────────────────────────────────────────────────────────────
-  // "Model-visible means logged" was true of the toolset and false of the conversation:
-  // the log could say a turn happened and what it cost, but never what was asked or
-  // answered. That is the half a trajectory view needs, and the half replay cannot be
-  // checked without.
-  //
-  // CONTENT IS NEVER IN THE EVENT. Each of these carries a Ref — a content hash — and the
-  // bytes live in the blob store. Three reasons, in order of how much they matter:
-  //   1. The log stays metadata, so exporting or replicating it does not export the user's
-  //      conversations by accident. That property is the whole reason an event log is safe
-  //      to keep at all.
-  //   2. Deletion stays honest: an append-only log cannot unsay a message, but a blob can
-  //      be crypto-shredded and the ref then resolves to verified-but-unavailable. Inlined
-  //      text would make "delete my data" a lie the schema enforces forever.
-  //   3. Repeated content (the same system prompt on every turn) is stored once.
-  'assistant.prompted': (p) => str(p.turnId) && isRef(p.ref),
-  'assistant.message': (p) => str(p.turnId) && isRef(p.ref),
-  'assistant.reasoning': (p) => str(p.turnId) && isRef(p.ref),
-
-  'context.assembled': (p) => num(p.budget) && num(p.used) && !!p.parts && typeof p.parts === 'object'
-    && arr(p.resident) && p.resident.every(isRef) && num(p.reachableCount),
-  'context.attached': (p) => isRef(p.ref),
-  'context.expanded': (p) => isRef(p.ref) && num(p.tokens),
-
-  'capability.offered': (p) => str(p.capability) && str(p.reason),
-  'capability.granted': (p) => str(p.capability) && actorOk(p.actor),
-  'capability.denied': (p) => str(p.capability) && actorOk(p.actor),
-  'capability.activated': (p) => str(p.capability) && CLASSES.includes(p.classUsed),
-  'capability.revoked': (p) => str(p.capability) && str(p.cause),
-  // I3 is enforced here structurally: a non-pure invocation without a key is not a
-  // valid event, so it cannot reach the log at all.
-  'capability.invoked': (p) => str(p.capability) && actorOk(p.actor) && scopeOk(p.scope)
-    && EFFECTS.includes(p.effects) && (p.effects === 'pure' || str(p.idempotencyKey)),
-  'capability.resulted': (p) => str(p.capability) && typeof p.ok === 'boolean'
-    && CLASSES.includes(p.classUsed) && !!p.cost && num(p.cost.ms),
-
-  // counts only — never values
-  'privacy.redacted': (p) => !!p.counts && typeof p.counts === 'object'
-    && Object.values(p.counts).every(num),
-  'privacy.egress': (p) => str(p.host) && typeof p.redacted === 'boolean'
-    && typeof p.controlled === 'boolean',
-
-  'policy.changed': (p) => str(p.dial) && actorOk(p.actor) && 'from' in p && 'to' in p,
-  'policy.guard_denied': (p) => str(p.capability) && str(p.reason),
-
-  'data.created': (p) => isRef(p.ref),
-  'data.updated': (p) => isRef(p.ref),
-  'data.deleted': (p) => isRef(p.ref) && typeof p.shredded === 'boolean',
-
-  'automation.fired': (p) => str(p.ruleId) && CLASSES.includes(p.classUsed),
-  'automation.suppressed': (p) => str(p.ruleId) && str(p.reason),
-};
-
-/** Validate a durable event. Throws EventError; never mutates. */
-export function validateEvent(e) {
-  if (!e || typeof e !== 'object') throw new EventError('SHAPE', 'event must be an object');
-  if (e.v !== CURRENT_VERSION) throw new EventError('VERSION', `expected v=${CURRENT_VERSION}, got ${e.v} — upcast first`, e.v);
-  if (!str(e.id)) throw new EventError('SHAPE', 'id required');
-  if (!str(e.host)) throw new EventError('SHAPE', 'host required');
-  if (!Number.isInteger(e.seq) || e.seq < 0) throw new EventError('SHAPE', 'seq must be a non-negative integer');
-  if (!arr(e.causes) || !e.causes.every(str)) throw new EventError('SHAPE', 'causes must be string[]');
-  if (!num(e.at)) throw new EventError('SHAPE', 'at required (advisory wall clock)');
-  if (!ALL_TYPES.includes(e.type)) throw new EventError('TYPE', `unknown type ${e.type}`, e.type);
-  if (!e.payload || typeof e.payload !== 'object') throw new EventError('SHAPE', 'payload required');
-  const check = PAYLOAD[e.type];
-  if (check && !check(e.payload)) throw new EventError('PAYLOAD', `payload invalid for ${e.type}`, e.type);
-  return e;
-}
-
-export function isValidEvent(e) {
-  try { validateEvent(e); return true; } catch { return false; }
-}
-
-/**
- * A per-host append cursor. Owns `seq` — the only monotonic thing in the system — so
- * callers cannot skip or reuse one.
- *
- * `now` and `newId` are injected so tests are deterministic and so the package holds no
- * ambient dependency on a clock or a crypto implementation.
- */
-export function createAppender({ host, seq = 0, now = () => Date.now(), newId }) {
-  if (!str(host)) throw new EventError('SHAPE', 'host required');
-  const genId = newId || (() => globalThis.crypto.randomUUID());
-  let n = seq;
-  return {
-    get host() { return host; },
-    get seq() { return n; },
-    /** Build + validate the next event for this host. Does not persist — the caller stores it. */
-    append(type, payload, causes = []) {
-      const e = Object.freeze({
-        v: CURRENT_VERSION,
-        id: genId(),
-        host,
-        seq: n++,
-        causes: Object.freeze([...causes]),
-        at: now(),
-        type,
-        payload,
-      });
-      return validateEvent(e);
-    },
-  };
-}
+import{isRef as a}from"./ref.js";const c=1,m=Object.freeze({turn:["started","ended"],assistant:["prompted","message","reasoning"],context:["assembled","attached","expanded"],capability:["offered","granted","denied","activated","revoked","invoked","resulted"],privacy:["redacted","egress"],policy:["changed","guard_denied"],data:["created","updated","deleted"],automation:["fired","suppressed"]}),v=Object.freeze(Object.entries(m).flatMap(([e,n])=>n.map(s=>`${e}.${s}`))),w=Object.freeze(["user","rule","schedule","model","agent","channel"]),S=Object.freeze(["global","site","tab","session","agent"]),l=Object.freeze(["R","M","L","C","A","X","H"]),x=Object.freeze(["pure","idempotent","replay-safe","non-replayable"]),I=Object.freeze(["none","redacted","delegated"]);class r extends Error{constructor(n,s,i=null){super(s),this.name="EventError",this.code=n,this.detail=i}}const t=e=>typeof e=="string"&&e.length>0,o=e=>typeof e=="number"&&Number.isFinite(e),f=e=>Array.isArray(e);function d(e){return!!e&&w.includes(e.kind)&&t(e.id)}function A(e){return!!e&&S.includes(e.kind)&&t(e.id)}const O={"turn.started":e=>t(e.turnId),"turn.ended":e=>t(e.turnId),"assistant.prompted":e=>t(e.turnId)&&a(e.ref),"assistant.message":e=>t(e.turnId)&&a(e.ref),"assistant.reasoning":e=>t(e.turnId)&&a(e.ref),"context.assembled":e=>o(e.budget)&&o(e.used)&&!!e.parts&&typeof e.parts=="object"&&f(e.resident)&&e.resident.every(a)&&o(e.reachableCount),"context.attached":e=>a(e.ref),"context.expanded":e=>a(e.ref)&&o(e.tokens),"capability.offered":e=>t(e.capability)&&t(e.reason),"capability.granted":e=>t(e.capability)&&d(e.actor),"capability.denied":e=>t(e.capability)&&d(e.actor),"capability.activated":e=>t(e.capability)&&l.includes(e.classUsed),"capability.revoked":e=>t(e.capability)&&t(e.cause),"capability.invoked":e=>t(e.capability)&&d(e.actor)&&A(e.scope)&&x.includes(e.effects)&&(e.effects==="pure"||t(e.idempotencyKey)),"capability.resulted":e=>t(e.capability)&&typeof e.ok=="boolean"&&l.includes(e.classUsed)&&!!e.cost&&o(e.cost.ms),"privacy.redacted":e=>!!e.counts&&typeof e.counts=="object"&&Object.values(e.counts).every(o),"privacy.egress":e=>t(e.host)&&typeof e.redacted=="boolean"&&typeof e.controlled=="boolean","policy.changed":e=>t(e.dial)&&d(e.actor)&&"from"in e&&"to"in e,"policy.guard_denied":e=>t(e.capability)&&t(e.reason),"data.created":e=>a(e.ref),"data.updated":e=>a(e.ref),"data.deleted":e=>a(e.ref)&&typeof e.shredded=="boolean","automation.fired":e=>t(e.ruleId)&&l.includes(e.classUsed),"automation.suppressed":e=>t(e.ruleId)&&t(e.reason)};function y(e){if(!e||typeof e!="object")throw new r("SHAPE","event must be an object");if(e.v!==c)throw new r("VERSION",`expected v=${c}, got ${e.v} — upcast first`,e.v);if(!t(e.id))throw new r("SHAPE","id required");if(!t(e.host))throw new r("SHAPE","host required");if(!Number.isInteger(e.seq)||e.seq<0)throw new r("SHAPE","seq must be a non-negative integer");if(!f(e.causes)||!e.causes.every(t))throw new r("SHAPE","causes must be string[]");if(!o(e.at))throw new r("SHAPE","at required (advisory wall clock)");if(!v.includes(e.type))throw new r("TYPE",`unknown type ${e.type}`,e.type);if(!e.payload||typeof e.payload!="object")throw new r("SHAPE","payload required");const n=O[e.type];if(n&&!n(e.payload))throw new r("PAYLOAD",`payload invalid for ${e.type}`,e.type);return e}function P(e){try{return y(e),!0}catch{return!1}}function k({host:e,seq:n=0,now:s=()=>Date.now(),newId:i}){if(!t(e))throw new r("SHAPE","host required");const p=i||(()=>globalThis.crypto.randomUUID());let u=n;return{get host(){return e},get seq(){return u},append(b,h,E=[]){const g=Object.freeze({v:c,id:p(),host:e,seq:u++,causes:Object.freeze([...E]),at:s(),type:b,payload:h});return y(g)}}}export{w as ACTOR_KINDS,v as ALL_TYPES,l as CLASSES,c as CURRENT_VERSION,x as EFFECTS,I as EGRESS,m as EVENT_TYPES,r as EventError,S as SCOPE_KINDS,k as createAppender,P as isValidEvent,y as validateEvent};