@inerrata-corporation/errata 2.0.2-dev.597 → 2.0.2-dev.600

package/errata.mjs

@@ -27308,10 +27308,80 @@ function buildSymbolLexicon(store, summariesByBodyHash2) {
   }
   return lex;
 }
-function generalizeSymbols(text, lexicon, level = 1) {
+function buildFileLexicon(store) {
+  const byKey = /* @__PURE__ */ new Map();
+  const byBasename = /* @__PURE__ */ new Map();
+  const basenameCount = /* @__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;
+    if (!relPath) continue;
+    const basename5 = relPath.split("/").pop();
+    if (!basename5.includes(".")) continue;
+    entries.push({ relPath, basename: basename5 });
+    basenameCount.set(basename5, (basenameCount.get(basename5) ?? 0) + 1);
+  }
+  for (const { relPath, basename: basename5 } of entries) {
+    const entry = { relPath };
+    if (relPath !== basename5) byKey.set(relPath, entry);
+    const conventional = CONVENTION_FILENAMES.has(basename5) || (basenameCount.get(basename5) ?? 0) >= CONVENTION_BASENAME_MIN_FILES;
+    if (!conventional && !byKey.has(basename5)) byKey.set(basename5, entry);
+    const list = byBasename.get(basename5);
+    if (list) list.push(entry);
+    else byBasename.set(basename5, [entry]);
+  }
+  return { byKey, byBasename };
+}
+function resolveFileToken(token, lexicon) {
+  const exact = lexicon.byKey.get(token);
+  if (exact) return exact;
+  if (!token.includes("/")) return null;
+  const basename5 = token.split("/").pop();
+  for (const entry of lexicon.byBasename.get(basename5) ?? []) {
+    if (entry.relPath === token || entry.relPath.endsWith(`/${token}`)) return entry;
+  }
+  return null;
+}
+function internalFilesFor(description, lexicon) {
+  const out2 = [];
+  const seen = /* @__PURE__ */ new Set();
+  for (const m of description.matchAll(FILE_TOKEN_RE)) {
+    const tok = m[0];
+    if (seen.has(tok)) continue;
+    seen.add(tok);
+    if (resolveFileToken(tok, lexicon)) out2.push(tok);
+  }
+  return out2;
+}
+function fileRolePhrase(relPath, level = 1) {
+  if (level !== 1) return "a file";
+  if (/\.(test|spec)\.|__tests__\/|(?:^|\/)e2e\//.test(relPath)) return "a test file";
+  if (/(?:^|\/)scripts?\//.test(relPath)) return "a script";
+  if (/\.md$/i.test(relPath)) return "a documentation file";
+  if (/\.(json|ya?ml|toml|env)$/i.test(relPath)) return "a config file";
+  return "a source file";
+}
+function generalizeSymbols(text, lexicon, level = 1, fileLexicon) {
+  let out2 = text;
+  if (fileLexicon) {
+    const files = [];
+    const seenFiles = /* @__PURE__ */ new Set();
+    for (const m of out2.matchAll(FILE_TOKEN_RE)) {
+      const tok = m[0];
+      if (seenFiles.has(tok)) continue;
+      seenFiles.add(tok);
+      const entry = resolveFileToken(tok, fileLexicon);
+      if (entry) files.push([tok, entry]);
+    }
+    files.sort((a, b) => b[0].length - a[0].length);
+    for (const [tok, entry] of files) {
+      const re = new RegExp(`(?<![\\w$/\\\\.-])${escapeRe4(tok)}(?![\\w$-])`, "g");
+      out2 = out2.replace(re, fileRolePhrase(entry.relPath, level));
+    }
+  }
   const present = [];
   const seen = /* @__PURE__ */ new Set();
-  for (const m of text.matchAll(TOKEN_RE3)) {
+  for (const m of out2.matchAll(TOKEN_RE3)) {
     const tok = m[0];
     if (!seen.has(tok) && lexicon.has(tok)) {
       seen.add(tok);
@@ -27319,7 +27389,6 @@ function generalizeSymbols(text, lexicon, level = 1) {
     }
   }
   present.sort((a, b) => b.length - a.length);
-  let out2 = text;
   for (const key of present) {
     const entry = lexicon.get(key);
     const phrase = level === 1 && entry.summary ? entry.summary : KIND_PHRASE[entry.kind][level === 1 ? "l1" : "l2"];
@@ -27328,7 +27397,7 @@ function generalizeSymbols(text, lexicon, level = 1) {
   }
   return generalize(out2, { level }).text;
 }
-var SYMBOL_LABELS2, KIND_PHRASE, RE_RESERVED2, escapeRe4, TOKEN_RE3;
+var SYMBOL_LABELS2, KIND_PHRASE, RE_RESERVED2, escapeRe4, CONVENTION_FILENAMES, CONVENTION_BASENAME_MIN_FILES, FILE_TOKEN_RE, TOKEN_RE3;
 var init_generalize_graph = __esm({
   "src/generalize-graph.ts"() {
     "use strict";
@@ -27350,6 +27419,34 @@ var init_generalize_graph = __esm({
     };
     RE_RESERVED2 = /[.*+?^${}()|[\]\\]/g;
     escapeRe4 = (s) => s.replace(RE_RESERVED2, "\\$&");
+    CONVENTION_FILENAMES = /* @__PURE__ */ new Set([
+      "package.json",
+      "package-lock.json",
+      "pnpm-lock.yaml",
+      "pnpm-workspace.yaml",
+      "yarn.lock",
+      "tsconfig.json",
+      "tsconfig.base.json",
+      "turbo.json",
+      "README.md",
+      "LICENSE",
+      "CHANGELOG.md",
+      "CONTRIBUTING.md",
+      // agent-config conventions — as cross-repo universal as README.md by 2026
+      "AGENTS.md",
+      "CLAUDE.md",
+      "Dockerfile",
+      "docker-compose.yml",
+      ".gitignore",
+      ".env",
+      ".env.example",
+      "index.ts",
+      "index.js",
+      "index.tsx",
+      "index.mjs"
+    ]);
+    CONVENTION_BASENAME_MIN_FILES = 3;
+    FILE_TOKEN_RE = /(?:[A-Za-z0-9_.-]+\/)*[A-Za-z0-9_-][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;
   }
 });
@@ -54462,7 +54559,7 @@ function startLoopLagMonitor(thresholdMs = 1e3) {
 }
 
 // src/engine.ts
-var DAEMON_VERSION = true ? "2.0.2-dev.597" : "2.0.0-alpha.0";
+var DAEMON_VERSION = true ? "2.0.2-dev.600" : "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;
@@ -56781,10 +56878,16 @@ function buildInstanceIngest(store, profile, daemonVersion, ignorePatterns = [],
   };
   const symbolIndex = opts.workspaceRoot ? publicSymbolIndex(store, opts.workspaceRoot) : null;
   const ownLexicon = buildSymbolLexicon(store);
+  const fileLexicon = buildFileLexicon(store);
   const shareable = (n) => {
     const preserved = symbolIndex ? preservedSymbolsFor(n.description, symbolIndex) : [];
     const shippedText = generalize(n.description, { level }).text;
-    const internal = internalSymbolsFor(shippedText, (t) => ownLexicon.has(t), symbolIndex);
+    const internalSyms = internalSymbolsFor(shippedText, (t) => ownLexicon.has(t), symbolIndex);
+    const internalFiles = internalFilesFor(shippedText, fileLexicon);
+    const internal = [.../* @__PURE__ */ new Set([...internalFiles, ...internalSyms])].slice(
+      0,
+      PRESERVED_SYMBOLS_CAP
+    );
     return {
       ...n,
       description: shippedText,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@inerrata-corporation/errata",
-  "version": "2.0.2-dev.597",
+  "version": "2.0.2-dev.600",
   "description": "errata - local-first observation engine for AI coding agents",
   "type": "module",
   "bin": {