@inerrata-corporation/errata 2.0.2-dev.479 → 2.0.2-dev.483

package/consolidate-worker.mjs

@@ -673,7 +673,7 @@ function lemma(token) {
   }
   return token;
 }
-function conceptTokens(text) {
+function tokenize(text, resolve) {
   const matches = text.normalize("NFC").toLowerCase().match(TOKEN_RE) ?? [];
   const out = /* @__PURE__ */ new Set();
   for (const raw of matches) {
@@ -683,11 +683,14 @@ function conceptTokens(text) {
       out.add(token);
       continue;
     }
-    const canonical = resolveCanonicalId(token);
+    const canonical = resolve(token);
     out.add(canonical ?? lemma(token));
   }
   return [...out].sort();
 }
+function conceptTokens(text) {
+  return tokenize(text, resolveCanonicalId);
+}
 function conceptBag(text) {
   return conceptTokens(text).join(" ");
 }

package/errata.mjs

@@ -619,6 +619,11 @@ function resolveCanonical(label) {
 function resolveCanonicalId(label) {
   return resolveCanonical(label)?.id;
 }
+function resolveUnambiguousCanonicalId(label) {
+  const key = label.trim().toLowerCase();
+  if (AMBIGUOUS_ALIASES.has(key)) return void 0;
+  return resolveCanonicalId(key);
+}
 function allEntities() {
   return REGISTRY;
 }
@@ -628,7 +633,7 @@ function aliasesLongestFirst() {
   out2.sort((x, y) => y.alias.length - x.alias.length);
   return out2;
 }
-var REGISTRY, BY_ALIAS;
+var REGISTRY, AMBIGUOUS_ALIASES, BY_ALIAS;
 var init_taxonomy = __esm({
   "../../packages/shared/src/nlp/taxonomy.ts"() {
     "use strict";
@@ -677,6 +682,13 @@ var init_taxonomy = __esm({
       { id: "concept:retry", name: "retry", category: "concept", aliases: ["retry"] },
       { id: "concept:migration", name: "migration", category: "concept", aliases: ["migration"] }
     ];
+    AMBIGUOUS_ALIASES = /* @__PURE__ */ new Set([
+      "node",
+      "go",
+      "pool",
+      "spring",
+      "ws"
+    ]);
     BY_ALIAS = /* @__PURE__ */ new Map();
     for (const e of REGISTRY) {
       for (const a of e.aliases) {
@@ -785,7 +797,7 @@ function lemma(token) {
   }
   return token;
 }
-function conceptTokens(text) {
+function tokenize(text, resolve6) {
   const matches = text.normalize("NFC").toLowerCase().match(TOKEN_RE) ?? [];
   const out2 = /* @__PURE__ */ new Set();
   for (const raw2 of matches) {
@@ -795,11 +807,17 @@ function conceptTokens(text) {
       out2.add(token);
       continue;
     }
-    const canonical = resolveCanonicalId(token);
+    const canonical = resolve6(token);
     out2.add(canonical ?? lemma(token));
   }
   return [...out2].sort();
 }
+function conceptTokens(text) {
+  return tokenize(text, resolveCanonicalId);
+}
+function retrievalTokens(text) {
+  return tokenize(text, resolveUnambiguousCanonicalId);
+}
 function conceptBag(text) {
   return conceptTokens(text).join(" ");
 }
@@ -15935,6 +15953,7 @@ __export(src_exports, {
   ABSTRACTION_LEVEL: () => ABSTRACTION_LEVEL,
   ALL_EDGE_TYPES: () => ALL_EDGE_TYPES,
   ALL_NODE_LABELS: () => ALL_NODE_LABELS,
+  AMBIGUOUS_ALIASES: () => AMBIGUOUS_ALIASES,
   BELIEF_EDGES: () => BELIEF_EDGES,
   BRIDGE_EDGES: () => BRIDGE_EDGES,
   CAUSAL_EDGES: () => CAUSAL_EDGES,
@@ -16035,6 +16054,8 @@ __export(src_exports, {
   provenanceBand: () => provenanceBand,
   resolveCanonical: () => resolveCanonical,
   resolveCanonicalId: () => resolveCanonicalId,
+  resolveUnambiguousCanonicalId: () => resolveUnambiguousCanonicalId,
+  retrievalTokens: () => retrievalTokens,
   sha256: () => sha256,
   sha256Short: () => sha256Short,
   splitPnpmKey: () => splitPnpmKey,
@@ -18192,7 +18213,7 @@ function djb2Signed(s) {
   }
   return h;
 }
-function tokenize(text) {
+function tokenize2(text) {
   return text.toLowerCase().split(/[^a-z0-9]+/).filter((t) => t.length > 0 && t.length <= 40);
 }
 function* charTrigrams(text) {
@@ -18204,7 +18225,7 @@ function* charTrigrams(text) {
 function embed(text) {
   const vec = new Array(EMBEDDING_DIM).fill(0);
   if (!text) return vec;
-  for (const tok of tokenize(text)) {
+  for (const tok of tokenize2(text)) {
     const idx = djb2(tok) % EMBEDDING_DIM;
     const sign = djb2Signed(tok) & 1 ? 1 : -1;
     vec[idx] = vec[idx] + sign;
@@ -18578,7 +18599,7 @@ function mergeDuplicateProblems(store, opts) {
   const report = { clusters: 0, merged: 0 };
   const open2 = store.findNodesByLabel("Problem").filter((p) => p.attrs["resolvedAt"] == null && p.attrs["mergedInto"] === void 0);
   const tokens = /* @__PURE__ */ new Map();
-  for (const p of open2) tokens.set(p.id, new Set(conceptTokens(p.description)));
+  for (const p of open2) tokens.set(p.id, new Set(retrievalTokens(p.description)));
   const consumed = /* @__PURE__ */ new Set();
   store.transaction(() => {
     for (let i2 = 0; i2 < open2.length; i2++) {
@@ -18656,12 +18677,12 @@ function findMintTimeDuplicate(store, statement, kind, opts) {
     if (isConstraintProblem(p) !== incomingIsConstraint) continue;
     if (statementIdentityKey(p.description) === key) return p.id;
   }
-  const ta = new Set(conceptTokens(statement));
+  const ta = new Set(retrievalTokens(statement));
   if (ta.size < minTokens) return null;
   let best = null;
   for (const p of open2) {
     if (isConstraintProblem(p) !== incomingIsConstraint) continue;
-    const tb = new Set(conceptTokens(p.description));
+    const tb = new Set(retrievalTokens(p.description));
     if (Math.min(ta.size, tb.size) < minTokens) continue;
     const score2 = overlap(ta, tb);
     if (score2 >= minOverlap && (best === null || score2 > best.score)) {
@@ -19368,7 +19389,7 @@ function parseIntentFences(text) {
 function searchByIntent(store, intent, opts) {
   const limit = opts?.limit ?? 5;
   const minCoverage = opts?.minCoverage ?? 0.25;
-  const want = new Set(conceptTokens(intent));
+  const want = new Set(retrievalTokens(intent));
   if (want.size < MIN_INTENT_TOKENS) return [];
   const hits = [];
   for (const label of RECALLABLE) {
@@ -19377,7 +19398,7 @@ function searchByIntent(store, intent, opts) {
       if (label === "Problem" && node2.attrs["resolvedAt"] != null && !isConstraintProblem(node2)) {
         continue;
       }
-      const bag = conceptTokens(node2.description ?? "");
+      const bag = retrievalTokens(node2.description ?? "");
       if (bag.length === 0) continue;
       const matched = bag.filter((t) => want.has(t));
       if (matched.length === 0) continue;
@@ -29032,7 +29053,7 @@ function popcount(x) {
   }
   return c;
 }
-function tokenize2(text) {
+function tokenize3(text) {
   return text.match(/[A-Za-z_$][A-Za-z0-9_$]*|\d+|[^\s\w]/g) ?? [];
 }
 function shingle(tokens, n = SHINGLE_N) {
@@ -29061,7 +29082,7 @@ function simhashFeatures(features) {
   return out2;
 }
 function simhash(text) {
-  return simhashFeatures(shingle(tokenize2(text)));
+  return simhashFeatures(shingle(tokenize3(text)));
 }
 function hammingDistance(a, b) {
   return popcount((a ^ b) & MASK64);
@@ -37584,7 +37605,7 @@ __export(src_exports5, {
   simhashFeatures: () => simhashFeatures,
   symbolNodeId: () => symbolNodeId,
   toHex: () => toHex,
-  tokenize: () => tokenize2
+  tokenize: () => tokenize3
 });
 function defaultProviders() {
   if (providersCache) return providersCache;
@@ -53683,7 +53704,7 @@ function startLoopLagMonitor(thresholdMs = 1e3) {
 }
 
 // src/engine.ts
-var DAEMON_VERSION = true ? "2.0.2-dev.479" : "2.0.0-alpha.0";
+var DAEMON_VERSION = true ? "2.0.2-dev.483" : "2.0.0-alpha.0";
 var IGNORED_PATH = /[\\/](?:\.git|node_modules|\.errata|\.claude|\.codex|\.cursor|\.turbo|\.next|dist|coverage|test-results|playwright-report|__pycache__)(?:[\\/]|$)/;
 var IGNORED_NOISE = /castalia\.db|eventlog\.sqlite|turn-cursors/;
 var GIT_OP_MUTE_MS = 4e3;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@inerrata-corporation/errata",
-  "version": "2.0.2-dev.479",
+  "version": "2.0.2-dev.483",
   "description": "errata - local-first observation engine for AI coding agents",
   "type": "module",
   "bin": {

package/pass-worker.mjs

@@ -458,13 +458,18 @@ function resolveCanonical(label) {
 function resolveCanonicalId(label) {
   return resolveCanonical(label)?.id;
 }
+function resolveUnambiguousCanonicalId(label) {
+  const key = label.trim().toLowerCase();
+  if (AMBIGUOUS_ALIASES.has(key)) return void 0;
+  return resolveCanonicalId(key);
+}
 function aliasesLongestFirst() {
   const out2 = [];
   for (const e of REGISTRY) for (const a of e.aliases) out2.push({ alias: a.toLowerCase(), entity: e });
   out2.sort((x, y) => y.alias.length - x.alias.length);
   return out2;
 }
-var REGISTRY, BY_ALIAS;
+var REGISTRY, AMBIGUOUS_ALIASES, BY_ALIAS;
 var init_taxonomy = __esm({
   "../../packages/shared/src/nlp/taxonomy.ts"() {
     "use strict";
@@ -513,6 +518,13 @@ var init_taxonomy = __esm({
       { id: "concept:retry", name: "retry", category: "concept", aliases: ["retry"] },
       { id: "concept:migration", name: "migration", category: "concept", aliases: ["migration"] }
     ];
+    AMBIGUOUS_ALIASES = /* @__PURE__ */ new Set([
+      "node",
+      "go",
+      "pool",
+      "spring",
+      "ws"
+    ]);
     BY_ALIAS = /* @__PURE__ */ new Map();
     for (const e of REGISTRY) {
       for (const a of e.aliases) {
@@ -574,7 +586,7 @@ function lemma(token) {
   }
   return token;
 }
-function conceptTokens(text) {
+function tokenize(text, resolve) {
   const matches = text.normalize("NFC").toLowerCase().match(TOKEN_RE) ?? [];
   const out2 = /* @__PURE__ */ new Set();
   for (const raw of matches) {
@@ -584,11 +596,14 @@ function conceptTokens(text) {
       out2.add(token);
       continue;
     }
-    const canonical = resolveCanonicalId(token);
+    const canonical = resolve(token);
     out2.add(canonical ?? lemma(token));
   }
   return [...out2].sort();
 }
+function retrievalTokens(text) {
+  return tokenize(text, resolveUnambiguousCanonicalId);
+}
 var STOPWORDS, NEGATION_TOKENS, TOKEN_RE;
 var init_concept_bag = __esm({
   "../../packages/shared/src/nlp/concept-bag.ts"() {
@@ -15533,7 +15548,7 @@ function popcount(x) {
   }
   return c;
 }
-function tokenize(text) {
+function tokenize2(text) {
   return text.match(/[A-Za-z_$][A-Za-z0-9_$]*|\d+|[^\s\w]/g) ?? [];
 }
 function shingle(tokens, n = SHINGLE_N) {
@@ -15562,7 +15577,7 @@ function simhashFeatures(features) {
   return out2;
 }
 function simhash(text) {
-  return simhashFeatures(shingle(tokenize(text)));
+  return simhashFeatures(shingle(tokenize2(text)));
 }
 function hammingDistance(a, b) {
   return popcount((a ^ b) & MASK64);
@@ -24085,7 +24100,7 @@ __export(src_exports, {
   simhashFeatures: () => simhashFeatures,
   symbolNodeId: () => symbolNodeId,
   toHex: () => toHex,
-  tokenize: () => tokenize
+  tokenize: () => tokenize2
 });
 function defaultProviders() {
   if (providersCache) return providersCache;
@@ -25342,7 +25357,7 @@ function mergeDuplicateProblems(store2, opts) {
   const report = { clusters: 0, merged: 0 };
   const open = store2.findNodesByLabel("Problem").filter((p) => p.attrs["resolvedAt"] == null && p.attrs["mergedInto"] === void 0);
   const tokens = /* @__PURE__ */ new Map();
-  for (const p of open) tokens.set(p.id, new Set(conceptTokens(p.description)));
+  for (const p of open) tokens.set(p.id, new Set(retrievalTokens(p.description)));
   const consumed = /* @__PURE__ */ new Set();
   store2.transaction(() => {
     for (let i2 = 0; i2 < open.length; i2++) {