@inerrata-corporation/errata 2.0.2-dev.710 → 2.0.2-dev.720

package/errata.mjs

@@ -15447,6 +15447,15 @@ function extractIdentifierTokens(text) {
   }
   return out2;
 }
+function sidecarDropBucket(reason) {
+  if (reason === null) return null;
+  if (reason === "empty summary") return "empty";
+  if (reason === "summary too long") return "too_long";
+  if (reason === "contains a path shape") return "path_shape";
+  if (reason === "contains a quoted literal") return "quoted_literal";
+  if (reason.startsWith("contains verbatim identifier")) return "verbatim_identifier";
+  return "other";
+}
 function summaryRejectionReason(summary, isKnownToken) {
   if (!summary.trim()) return "empty summary";
   if (summary.length > MAX_SYMBOL_SUMMARY_CHARS) return "summary too long";
@@ -15470,15 +15479,18 @@ function summaryRejectionReason(summary, isKnownToken) {
   return null;
 }
 function substituteSymbolSummaries(text, summaryByToken) {
-  if (summaryByToken.size === 0) return text;
+  if (summaryByToken.size === 0) return { text, substituted: 0, offered: 0 };
   const present = extractIdentifierTokens(text).filter((t) => summaryByToken.has(t));
   present.sort((a, b) => b.length - a.length);
   let out2 = text;
+  let substituted = 0;
   for (const key of present) {
     const re = new RegExp(`(?<![A-Za-z0-9_$.])${escapeRe2(key)}(?![A-Za-z0-9_$])`, "g");
+    const before = out2;
     out2 = out2.replace(re, summaryByToken.get(key));
+    if (out2 !== before) substituted++;
   }
-  return out2;
+  return { text: out2, substituted, offered: summaryByToken.size };
 }
 function vetSidecarSummaries(sidecar) {
   if (!sidecar) return null;
@@ -16062,6 +16074,7 @@ __export(src_exports, {
   retrievalTokens: () => retrievalTokens,
   sha256: () => sha256,
   sha256Short: () => sha256Short,
+  sidecarDropBucket: () => sidecarDropBucket,
   splitPnpmKey: () => splitPnpmKey,
   substituteSymbolSummaries: () => substituteSymbolSummaries,
   summarizeIngestResult: () => summarizeIngestResult,
@@ -27834,8 +27847,15 @@ function fileRolePhrase(relPath, level = 1) {
   if (/\.(json|ya?ml|toml|env)$/i.test(relPath)) return "a config file";
   return "a source file";
 }
+function isSourceFileToken(token) {
+  const basename5 = token.split(/[/\\]/).pop() ?? token;
+  if (CONVENTION_FILENAMES.has(basename5)) return false;
+  const ext = basename5.split(".").pop();
+  return ext !== void 0 && SOURCE_EXTENSIONS.has(ext.toLowerCase());
+}
 function generalizeFileRefs(text, fileLexicon, level = 1) {
   let out2 = text;
+  let lexiconMisses = 0;
   const files = [];
   const seenFiles = /* @__PURE__ */ new Set();
   for (const m of out2.matchAll(FILE_TOKEN_RE)) {
@@ -27844,6 +27864,7 @@ function generalizeFileRefs(text, fileLexicon, level = 1) {
     seenFiles.add(tok);
     const entry = resolveFileToken(tok, fileLexicon);
     if (entry) files.push([tok, entry]);
+    else if (isSourceFileToken(tok)) lexiconMisses++;
   }
   files.sort((a, b) => b[0].length - a[0].length);
   for (const [tok, entry] of files) {
@@ -27864,10 +27885,10 @@ function generalizeFileRefs(text, fileLexicon, level = 1) {
     const re = new RegExp(`(?<![\\w$/\\\\.-])${escapeRe4(tok)}(?![\\w$/\\\\.-])`, "g");
     out2 = out2.replace(re, fileRolePhrase(entry.relPath, level));
   }
-  return out2;
+  return { text: out2, substitutions: files.length + stems.length, lexiconMisses };
 }
 function generalizeSymbols(text, lexicon, level = 1, fileLexicon) {
-  let out2 = fileLexicon ? generalizeFileRefs(text, fileLexicon, level) : text;
+  let out2 = fileLexicon ? generalizeFileRefs(text, fileLexicon, level).text : text;
   const present = [];
   const seen = /* @__PURE__ */ new Set();
   for (const m of out2.matchAll(TOKEN_RE3)) {
@@ -27886,7 +27907,7 @@ function generalizeSymbols(text, lexicon, level = 1, fileLexicon) {
   }
   return generalize(out2, { level }).text;
 }
-var SYMBOL_LABELS2, KIND_PHRASE, RE_RESERVED2, escapeRe4, CONVENTION_FILENAMES, CONVENTION_BASENAME_MIN_FILES, CONVENTION_STEMS, FILE_TOKEN_RE, STEM_TOKEN_RE, TOKEN_RE3;
+var SYMBOL_LABELS2, KIND_PHRASE, RE_RESERVED2, escapeRe4, CONVENTION_FILENAMES, CONVENTION_BASENAME_MIN_FILES, CONVENTION_STEMS, FILE_TOKEN_RE, STEM_TOKEN_RE, SOURCE_EXTENSIONS, TOKEN_RE3;
 var init_generalize_graph = __esm({
   "src/generalize-graph.ts"() {
     "use strict";
@@ -27956,6 +27977,45 @@ var init_generalize_graph = __esm({
     ]);
     FILE_TOKEN_RE = /(?:[A-Za-z0-9_.-]+\/)*[A-Za-z0-9_-][A-Za-z0-9_.-]*\.[A-Za-z0-9]+/g;
     STEM_TOKEN_RE = /[A-Za-z0-9]+(?:[-_][A-Za-z0-9]+)+/g;
+    SOURCE_EXTENSIONS = /* @__PURE__ */ new Set([
+      "ts",
+      "tsx",
+      "js",
+      "jsx",
+      "mjs",
+      "cjs",
+      "py",
+      "go",
+      "rs",
+      "java",
+      "kt",
+      "rb",
+      "c",
+      "h",
+      "cc",
+      "cpp",
+      "hpp",
+      "cs",
+      "php",
+      "swift",
+      "scala",
+      "sh",
+      "ps1",
+      "sql",
+      "css",
+      "scss",
+      "html",
+      "vue",
+      "svelte",
+      "md",
+      "json",
+      "yaml",
+      "yml",
+      "toml",
+      "ini",
+      "env",
+      "cypher"
+    ]);
     TOKEN_RE3 = /[A-Za-z_$][A-Za-z0-9_$]*(?:\.[A-Za-z_$][A-Za-z0-9_$]*)*/g;
   }
 });
@@ -55551,7 +55611,7 @@ function startLoopLagMonitor(thresholdMs = 1e3) {
 }
 
 // src/engine.ts
-var DAEMON_VERSION = true ? "2.0.2-dev.710" : "2.0.0-alpha.0";
+var DAEMON_VERSION = true ? "2.0.2-dev.720" : "2.0.0-alpha.0";
 var IGNORED_PATH = /[\\/](?:\.git|node_modules|\.errata|\.claude|\.codex|\.cursor|\.turbo|\.next|dist|coverage|test-results|playwright-report|__pycache__)(?:[\\/]|$)/;
 var IGNORED_NOISE = /castalia\.db|eventlog\.sqlite|turn-cursors/;
 var GIT_OP_MUTE_MS = 4e3;
@@ -57934,6 +57994,21 @@ function stripCodebaseScope(scope) {
   delete s["codebase"];
   return s;
 }
+function compositionCounters(c) {
+  const { sidecarDropReasons, ...scalars } = c;
+  const buckets = [
+    "empty",
+    "too_long",
+    "path_shape",
+    "quoted_literal",
+    "verbatim_identifier",
+    "other"
+  ];
+  return {
+    ...scalars,
+    ...Object.fromEntries(buckets.map((b) => [`sidecarDrop_${b}`, sidecarDropReasons[b] ?? 0]))
+  };
+}
 function contributedWireId(n) {
   const cloudId = n.attrs["cloudNodeId"];
   return isMembraneSaltedId(cloudId) ? cloudId : n.id;
@@ -58004,7 +58079,10 @@ function buildInstanceIngest(store, profile, daemonVersion, ignorePatterns = [],
   const fileLexicon = buildFileLexicon(store);
   const shareable = (n) => {
     const preserved = symbolIndex ? preservedSymbolsFor(n.description, symbolIndex) : [];
-    const fileScrubbed = generalizeFileRefs(n.description, fileLexicon, level);
+    const fileRefs = generalizeFileRefs(n.description, fileLexicon, level);
+    const fileScrubbed = fileRefs.text;
+    composition.fileRefsSubstituted += fileRefs.substitutions;
+    composition.fileRefLexiconMisses += fileRefs.lexiconMisses;
     const generalized = generalize(fileScrubbed, { level, project: projectIdentity });
     const shippedText = generalized.text;
     if (generalized.report.classes.length > 0) composition.privacyRewritten++;
@@ -58046,7 +58124,11 @@ function buildInstanceIngest(store, profile, daemonVersion, ignorePatterns = [],
     unshippable: 0,
     heldOrphans: 0,
     privacyRewritten: 0,
-    sidecarDropped: 0
+    sidecarDropped: 0,
+    sidecarDropReasons: {},
+    sidecarTruncatedAtCap: 0,
+    fileRefsSubstituted: 0,
+    fileRefLexiconMisses: 0
   };
   for (const label of INSTANCE_LABELS) {
     const ref = [];
@@ -58239,11 +58321,16 @@ function buildInstanceIngest(store, profile, daemonVersion, ignorePatterns = [],
         const entry = lexicon.get(tok);
         if (!entry?.summary) continue;
         const scrubbed = generalize(entry.summary, { level, project: projectIdentity }).text.trim();
-        if (!scrubbed || summaryRejectionReason(scrubbed, isKnown) !== null) {
+        const dropBucket = !scrubbed ? "empty" : sidecarDropBucket(summaryRejectionReason(scrubbed, isKnown));
+        if (dropBucket !== null) {
           composition.sidecarDropped++;
+          composition.sidecarDropReasons[dropBucket] = (composition.sidecarDropReasons[dropBucket] ?? 0) + 1;
           continue;
         }
-        if (count >= MAX_SIDECAR_SYMBOLS) break outer;
+        if (count >= MAX_SIDECAR_SYMBOLS) {
+          composition.sidecarTruncatedAtCap = 1;
+          break outer;
+        }
         sidecar[tok] = { kind: entry.kind, summary: scrubbed };
         count++;
       }
@@ -59659,7 +59746,7 @@ async function startMultiDaemon(opts = {}) {
         });
         if (!instances && composition) {
           appendPassLedger(r.engine.paths.configDir, "ship", 0, {
-            ...composition,
+            ...compositionCounters(composition),
             accepted: 0,
             rejected: 0,
             blocked: 0
@@ -59734,7 +59821,7 @@ async function startMultiDaemon(opts = {}) {
               });
             }
             appendPassLedger(r.engine.paths.configDir, "ship", Date.now() - shipStartedAt, {
-              ...instances.composition,
+              ...compositionCounters(instances.composition),
               accepted: res.accepted,
               rejected: rejected.size,
               blocked: blocked.size

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@inerrata-corporation/errata",
-  "version": "2.0.2-dev.710",
+  "version": "2.0.2-dev.720",
   "description": "errata - local-first observation engine for AI coding agents",
   "type": "module",
   "bin": {