@inerrata-corporation/errata 2.0.2-dev.633 → 2.0.2-dev.635

package/errata.mjs

@@ -27529,10 +27529,15 @@ function buildSymbolLexicon(store, summariesByBodyHash2) {
   }
   return lex;
 }
+function isBareMatchableStem(stem) {
+  return stem.length >= 5 && /^[A-Za-z0-9]+(?:[-_][A-Za-z0-9]+)+$/.test(stem);
+}
 function buildFileLexicon(store) {
   const byKey = /* @__PURE__ */ new Map();
   const byBasename = /* @__PURE__ */ new Map();
+  const byStem = /* @__PURE__ */ new Map();
   const basenameCount = /* @__PURE__ */ new Map();
+  const stemRelPaths = /* @__PURE__ */ new Map();
   const entries = [];
   for (const n of store.findNodesByLabel("File")) {
     const relPath = typeof n.attrs["relPath"] === "string" && n.attrs["relPath"].length > 0 ? n.attrs["relPath"] : n.description;
@@ -27541,6 +27546,12 @@ function buildFileLexicon(store) {
     if (!basename5.includes(".")) continue;
     entries.push({ relPath, basename: basename5 });
     basenameCount.set(basename5, (basenameCount.get(basename5) ?? 0) + 1);
+    const stem = basename5.slice(0, basename5.indexOf("."));
+    if (isBareMatchableStem(stem)) {
+      let s = stemRelPaths.get(stem);
+      if (!s) stemRelPaths.set(stem, s = /* @__PURE__ */ new Set());
+      s.add(relPath);
+    }
   }
   for (const { relPath, basename: basename5 } of entries) {
     const entry = { relPath };
@@ -27550,8 +27561,18 @@ function buildFileLexicon(store) {
     const list = byBasename.get(basename5);
     if (list) list.push(entry);
     else byBasename.set(basename5, [entry]);
+    const stem = basename5.slice(0, basename5.indexOf("."));
+    if (isBareMatchableStem(stem) && !CONVENTION_STEMS.has(stem) && !byStem.has(stem)) {
+      const dirStems = new Set(
+        [...stemRelPaths.get(stem) ?? []].map((p) => {
+          const bn = p.split("/").pop();
+          return p.slice(0, p.length - bn.length) + bn.slice(0, bn.indexOf("."));
+        })
+      );
+      if (dirStems.size === 1) byStem.set(stem, entry);
+    }
   }
-  return { byKey, byBasename };
+  return { byKey, byBasename, byStem };
 }
 function resolveFileToken(token, lexicon) {
   const exact = lexicon.byKey.get(token);
@@ -27582,24 +27603,40 @@ function fileRolePhrase(relPath, level = 1) {
   if (/\.(json|ya?ml|toml|env)$/i.test(relPath)) return "a config file";
   return "a source file";
 }
-function generalizeSymbols(text, lexicon, level = 1, fileLexicon) {
+function generalizeFileRefs(text, fileLexicon, level = 1) {
   let out2 = text;
-  if (fileLexicon) {
-    const files = [];
-    const seenFiles = /* @__PURE__ */ new Set();
-    for (const m of out2.matchAll(FILE_TOKEN_RE)) {
-      const tok = m[0];
-      if (seenFiles.has(tok)) continue;
-      seenFiles.add(tok);
-      const entry = resolveFileToken(tok, fileLexicon);
-      if (entry) files.push([tok, entry]);
-    }
-    files.sort((a, b) => b[0].length - a[0].length);
-    for (const [tok, entry] of files) {
-      const re = new RegExp(`(?<![\\w$/\\\\.-])${escapeRe4(tok)}(?![\\w$-])`, "g");
-      out2 = out2.replace(re, fileRolePhrase(entry.relPath, level));
-    }
+  const files = [];
+  const seenFiles = /* @__PURE__ */ new Set();
+  for (const m of out2.matchAll(FILE_TOKEN_RE)) {
+    const tok = m[0];
+    if (seenFiles.has(tok)) continue;
+    seenFiles.add(tok);
+    const entry = resolveFileToken(tok, fileLexicon);
+    if (entry) files.push([tok, entry]);
+  }
+  files.sort((a, b) => b[0].length - a[0].length);
+  for (const [tok, entry] of files) {
+    const re = new RegExp(`(?<![\\w$/\\\\.-])${escapeRe4(tok)}(?![\\w$-])`, "g");
+    out2 = out2.replace(re, fileRolePhrase(entry.relPath, level));
+  }
+  const stems = [];
+  const seenStems = /* @__PURE__ */ new Set();
+  for (const m of out2.matchAll(STEM_TOKEN_RE)) {
+    const tok = m[0];
+    if (seenStems.has(tok)) continue;
+    seenStems.add(tok);
+    const entry = fileLexicon.byStem.get(tok);
+    if (entry) stems.push([tok, entry]);
   }
+  stems.sort((a, b) => b[0].length - a[0].length);
+  for (const [tok, entry] of stems) {
+    const re = new RegExp(`(?<![\\w$/\\\\.-])${escapeRe4(tok)}(?![\\w$/\\\\.-])`, "g");
+    out2 = out2.replace(re, fileRolePhrase(entry.relPath, level));
+  }
+  return out2;
+}
+function generalizeSymbols(text, lexicon, level = 1, fileLexicon) {
+  let out2 = fileLexicon ? generalizeFileRefs(text, fileLexicon, level) : text;
   const present = [];
   const seen = /* @__PURE__ */ new Set();
   for (const m of out2.matchAll(TOKEN_RE3)) {
@@ -27618,7 +27655,7 @@ function generalizeSymbols(text, lexicon, level = 1, fileLexicon) {
   }
   return generalize(out2, { level }).text;
 }
-var SYMBOL_LABELS2, KIND_PHRASE, RE_RESERVED2, escapeRe4, CONVENTION_FILENAMES, CONVENTION_BASENAME_MIN_FILES, FILE_TOKEN_RE, TOKEN_RE3;
+var SYMBOL_LABELS2, KIND_PHRASE, RE_RESERVED2, escapeRe4, CONVENTION_FILENAMES, CONVENTION_BASENAME_MIN_FILES, CONVENTION_STEMS, FILE_TOKEN_RE, STEM_TOKEN_RE, TOKEN_RE3;
 var init_generalize_graph = __esm({
   "src/generalize-graph.ts"() {
     "use strict";
@@ -27667,7 +27704,27 @@ var init_generalize_graph = __esm({
       "index.mjs"
     ]);
     CONVENTION_BASENAME_MIN_FILES = 3;
+    CONVENTION_STEMS = /* @__PURE__ */ new Set([
+      "well-known",
+      "read-only",
+      "type-safe",
+      "end-to-end",
+      "fail-open",
+      "fail-closed",
+      "up-to-date",
+      "server-side",
+      "client-side",
+      "cross-origin",
+      "same-origin",
+      "well-formed",
+      "first-class",
+      "in-memory",
+      "single-page",
+      "use-case",
+      "out-of-band"
+    ]);
     FILE_TOKEN_RE = /(?:[A-Za-z0-9_.-]+\/)*[A-Za-z0-9_-][A-Za-z0-9_.-]*\.[A-Za-z0-9]+/g;
+    STEM_TOKEN_RE = /[A-Za-z0-9]+(?:[-_][A-Za-z0-9]+)+/g;
     TOKEN_RE3 = /[A-Za-z_$][A-Za-z0-9_$]*(?:\.[A-Za-z_$][A-Za-z0-9_$]*)*/g;
   }
 });
@@ -54908,7 +54965,7 @@ function startLoopLagMonitor(thresholdMs = 1e3) {
 }
 
 // src/engine.ts
-var DAEMON_VERSION = true ? "2.0.2-dev.633" : "2.0.0-alpha.0";
+var DAEMON_VERSION = true ? "2.0.2-dev.635" : "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;
@@ -57274,7 +57331,8 @@ function buildInstanceIngest(store, profile, daemonVersion, ignorePatterns = [],
   const fileLexicon = buildFileLexicon(store);
   const shareable = (n) => {
     const preserved = symbolIndex ? preservedSymbolsFor(n.description, symbolIndex) : [];
-    const shippedText = generalize(n.description, { level, project: projectIdentity }).text;
+    const fileScrubbed = generalizeFileRefs(n.description, fileLexicon, level);
+    const shippedText = generalize(fileScrubbed, { level, project: projectIdentity }).text;
     const internalSyms = internalSymbolsFor(shippedText, (t) => ownLexicon.has(t), symbolIndex);
     const internalFiles = internalFilesFor(shippedText, fileLexicon);
     const internal = [.../* @__PURE__ */ new Set([...internalFiles, ...internalSyms])].slice(

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@inerrata-corporation/errata",
-  "version": "2.0.2-dev.633",
+  "version": "2.0.2-dev.635",
   "description": "errata - local-first observation engine for AI coding agents",
   "type": "module",
   "bin": {