@remnic/core 9.52.1 → 9.53.0

package/dist/{storage-BmVudhk4.d.ts → storage-DmkCjo0l.d.ts}

@@ -290,10 +290,14 @@ interface TombstoneEntry {
     reason: TombstoneReason;
     /** The memory that was retired. */
     sourceMemoryId: string;
-    /** sha256 of the retired memory's rawContent (rule 23). */
+    /** sha256 of the retired identity; explicit contentHashSource may define it. */
     contentHash: string;
-    /** `ContentHashIndex.normalizeContent(rawContent)` — the pre-hash form. */
+    /** Current body hash retained alongside an ambiguous or explicit primary hash. */
+    currentContentHashAlias?: string;
+    /** Current normalized body used by the normalized lookup tier. */
     normalizedText: string;
+    /** Version of the current body identity fields. */
+    normalizerVersion?: number;
     entityRef?: string;
     /** Structured-attribute supersession key when one existed. */
     supersessionKey?: string;
@@ -327,7 +331,7 @@ interface TombstoneLookupQuery {
     /** Single supersession key (direct/unit callers). */
     supersessionKey?: string;
     /** All derived supersession keys (write chokepoint). The keyed tier is
-     *  checked for each; the first active match wins. */
+     * checked for each; the first active match wins. */
     supersessionKeys?: string[];
     namespace: string;
 }
@@ -346,6 +350,13 @@ interface TombstoneStoreOptions {
      * skipped entirely.
      */
     semanticSimilarity?: (a: string, b: string) => number;
+    /**
+     * Resolve source identity candidates for bounded legacy sourceMemoryIds.
+     * StorageManager reads the corpus before the tombstone write lock.
+     */
+    readonly sourceContentsForMemoryIds?: (sourceMemoryIds: readonly string[]) => Promise<ReadonlyMap<string, string | readonly string[]>>;
+    /** Maximum number of legacy entries considered during one load. */
+    readonly legacyMigrationLimit?: number;
     /**
      * Cross-process write-lock timings for the tombstone JSONL mutation lock
      * (issue #1639). The secure-store append is a read-merge-write (read
@@ -437,6 +448,13 @@ declare class TombstoneStore {
     /** Lazy load + build the in-memory index. Idempotent. */
     load(): Promise<void>;
     private loadInternal;
+    private parseEntries;
+    /**
+     * Re-index pre-Unicode records from retired source content.
+     * Source reads stay bounded, but every batch finishes before the store
+     * publishes its in-memory index. Missing sources remain eligible on restart.
+     */
+    private migrateLegacyEntries;
     private resetIndex;
     /**
      * Index an entry. Later entries (higher array index = newer createdAt)
@@ -448,15 +466,13 @@ declare class TombstoneStore {
     invalidate(): void;
     /**
      * Cross-process staleness probe (#1579). If the file's mtime advanced since
-     * the last load/own-write, a peer process appended and our in-memory index is
-     * stale — invalidate AND reload in place so the next lookup sees the new
-     * entries. Best-effort: a stat error is swallowed (the loaded index is
-     * retained rather than crashing the write path). No-op when no `stat` was
-     * injected or when the mtime is unchanged. After our own append/revoke/rebuild
-     * `markWritten` records the new mtime, so this probe does not fire for our
-     * own writes (only for peer-process appends).
+     * the last load/own-write, a peer process appended and our in-memory index
+     * is stale — invalidate and reload before the next access. Mutation callers
+     * pass `migrate: false` to avoid nested serializer/lock acquisition.
      */
-    ensureFreshAgainstDisk(): Promise<void>;
+    ensureFreshAgainstDisk(options?: {
+        migrate?: boolean;
+    }): Promise<void>;
     /**
      * Record the file mtime after THIS process writes (append / revoke / rebuild)
      * so `ensureFreshAgainstDisk` does not treat our own write as a peer append
@@ -497,6 +513,7 @@ declare class TombstoneStore {
      */
     revoke(tombstoneId: string, createdBy: TombstoneCreatedBy): Promise<string>;
     private serializeAppend;
+    private migrateLoadedLegacyEntries;
     /**
      * Acquire the cross-process tombstone write lock and run `task` under it
      * (issue #1639). Strict-fail on acquisition timeout: the secure-store append
@@ -592,6 +609,8 @@ interface ContentHashPathEntry {
 declare class ContentHashIndex {
     private hashes;
     private dirty;
+    /** True when load() found the pre-Unicode, markerless on-disk format. */
+    private formatMigrationRequired;
     /**
      * Hashes explicitly removed since the last successful save (issue #1909 review
      * round 8 thread 3). Tracked separately so `saveMergingWithDisk` can be
@@ -635,6 +654,8 @@ declare class ContentHashIndex {
      */
     private lastSyncedFingerprint;
     constructor(stateDir: string, secureStoreKeyProvider?: () => Buffer | null, secureStoreWriteKeyProvider?: () => Buffer | null, memoryDir?: string, lockOptions?: ContentHashIndexLockOptions);
+    /** Whether a caller with an authoritative corpus must rebuild this legacy index. */
+    get requiresFormatMigration(): boolean;
     /** Load existing hashes from disk. Safe to call multiple times. */
     load(): Promise<void>;
     /** Cheap `stat` of the durable index file; null when it does not exist yet. */
@@ -1646,26 +1667,9 @@ declare class StorageManager extends TombstoneBlockedCaptureIndexHost {
      * method itself never writes to disk.
      */
     private rebuildFactHashIndexFromCorpus;
-    /**
-     * The content-hash the corpus rebuild registers for `memory`, or null when the
-     * body carries a citation from an unknown/custom template that cannot be
-     * safely stripped. Shared by the authoritative rebuild and the fact-only
-     * corpus confirmation (`hasFactContentHash`) so both derive the identical
-     * hash for a given stored body.
-     *
-     * Preference order:
-     *  1. frontmatter.contentHash — the raw pre-citation hash writeMemory records
-     *     for facts (issue #369 round 8); matches hasFactContentHash(rawFact).
-     *  2. Reconstruct from the stored body: strip the "[Attributes: …]" suffix
-     *     writeMemory appends for structuredAttributes (registration hashed the
-     *     raw canonical content WITHOUT it — a no-op when absent), then strip a
-     *     recognised citation and hash the bare body, or hash a citation-free
-     *     body as-is.
-     *  3. A body with a citation from an unknown/custom template → null: a
-     *     false-negative miss beats a wrong hash that would permanently suppress
-     *     legitimate duplicate writes.
-     */
-    private corpusRegisteredHash;
+    private storedContentIdentityCandidates;
+    /** Returns the current body identity while preserving explicit external identities. */
+    private corpusRegisteredHashes;
     private get questionsDir();
     private get artifactsDir();
     private get identityDir();
@@ -1781,7 +1785,7 @@ declare class StorageManager extends TombstoneBlockedCaptureIndexHost {
     writeSealedMemory(envelope: SealedMemoryEnvelope, extras?: SealedWriteExtras): Promise<MemoryWriteResult>;
     hasFactContentHash(content: string): Promise<boolean>;
     private factContentHashPresentInCorpus;
-    private factContentHashForRemoval;
+    private factContentHashesForRemoval;
     private addActiveFactContentHash;
     /**
      * Re-register a fact's contentHash in the dedup index after a tombstone

package/dist/storage.d.ts

@@ -1,7 +1,7 @@
 import './types-CkiDUorT.js';
 import './corpus-read-cancellation.js';
 import './artifact-search.js';
-export { C as ContentHashIndex, t as ContentHashIndexLockOptions, u as FactHashIndexNotAuthoritativeError, v as MemoryLifecycleEventWriteOptions, M as MemoryWriteResult, R as ReextractJobRequest, w as SealedWriteExtras, S as StorageManager, W as WriteMemoryOptions, x as buildEntitySchemaCacheKey, y as compareEntityTimestamps, z as compileEntityFacts, A as countEntityStructuredFacts, B as fingerprintEntityStructuredFacts, D as isEntitySynthesisStale, E as isValidBufferSurpriseEvent, G as normalizeFrontmatterForPath, H as normalizeStructuredSectionFacts, p as parseEntityFile, I as parseFrontmatter, J as parseFrontmatterStringValue, s as serializeEntityFile } from './storage-BmVudhk4.js';
+export { C as ContentHashIndex, t as ContentHashIndexLockOptions, u as FactHashIndexNotAuthoritativeError, v as MemoryLifecycleEventWriteOptions, M as MemoryWriteResult, R as ReextractJobRequest, w as SealedWriteExtras, S as StorageManager, W as WriteMemoryOptions, x as buildEntitySchemaCacheKey, y as compareEntityTimestamps, z as compileEntityFacts, A as countEntityStructuredFacts, B as fingerprintEntityStructuredFacts, D as isEntitySynthesisStale, E as isValidBufferSurpriseEvent, G as normalizeFrontmatterForPath, H as normalizeStructuredSectionFacts, p as parseEntityFile, I as parseFrontmatter, J as parseFrontmatterStringValue, s as serializeEntityFile } from './storage-DmkCjo0l.js';
 export { normalizeEntityName } from './entity-id-normalization.js';
 import './page-versioning.js';
 import './consolidation-operator.js';

package/dist/storage.js

@@ -13,12 +13,11 @@ import {
   parseFrontmatter,
   parseFrontmatterStringValue,
   serializeEntityFile
-} from "./chunk-WD5B5AIE.js";
+} from "./chunk-6R4G7N4S.js";
 import "./chunk-UHVXE24Y.js";
-import "./chunk-JJRVPWFY.js";
+import "./chunk-MLPTFODV.js";
 import "./chunk-SXIMQLGV.js";
 import "./chunk-S2V7RUCL.js";
-import "./chunk-J6A3CX5N.js";
 import "./chunk-UU3G6GUP.js";
 import "./chunk-SCU65EZI.js";
 import "./chunk-WRJVB3C2.js";
@@ -27,13 +26,9 @@ import "./chunk-DM2T26WE.js";
 import "./chunk-LDXUBPMO.js";
 import "./chunk-JYIJNNN3.js";
 import "./chunk-AAD2ZS6P.js";
-import "./chunk-DLG5V6ND.js";
-import "./chunk-5V6IXHPI.js";
+import "./chunk-K4QS7LXU.js";
+import "./chunk-NSVDUN76.js";
 import "./chunk-BT7NVCML.js";
-import {
-  normalizeAttributePairs,
-  stripAttributesSuffix
-} from "./chunk-R3DJVOSM.js";
 import {
   normalizeEntityName
 } from "./chunk-YXZG5TVN.js";
@@ -53,18 +48,17 @@ import "./chunk-RGMVMVMF.js";
 import "./chunk-N7RWREIQ.js";
 import "./chunk-XMMESCPD.js";
 import "./chunk-42JKGUFJ.js";
-import "./chunk-QBSFOKY6.js";
-import "./chunk-5GKPFL3S.js";
+import "./chunk-VOELBUS4.js";
 import "./chunk-ZIKENA2C.js";
 import "./chunk-PQVCH2AE.js";
 import "./chunk-6MKAMLQL.js";
-import "./chunk-JAKIDQB2.js";
 import "./chunk-W5BIQOAJ.js";
 import "./chunk-Z7L72ZHL.js";
 import "./chunk-4O4TKIXQ.js";
 import "./chunk-BJMBJZ2Y.js";
 import "./chunk-UKJAGEXH.js";
 import "./chunk-FP2373TW.js";
+import "./chunk-J6A3CX5N.js";
 import "./chunk-7FGL6CGA.js";
 import "./chunk-EGGE52UU.js";
 import "./chunk-ZA53AZND.js";
@@ -73,11 +67,17 @@ import "./chunk-7RJUVTVF.js";
 import {
   ContentHashIndex,
   FactHashIndexNotAuthoritativeError
-} from "./chunk-BQIFZ2MP.js";
+} from "./chunk-CX5SW2EW.js";
 import "./chunk-3CSX3ETE.js";
 import "./chunk-RL6NEVXT.js";
+import {
+  normalizeAttributePairs,
+  stripAttributesSuffix
+} from "./chunk-JDSVLVCG.js";
 import "./chunk-FFAZ2XFU.js";
-import "./chunk-W5H6ZRPT.js";
+import "./chunk-5FVVY2EB.js";
+import "./chunk-5GKPFL3S.js";
+import "./chunk-JAKIDQB2.js";
 import "./chunk-P7FMDTKL.js";
 import "./chunk-2XRE6FLM.js";
 import "./chunk-EYIEWJNI.js";

package/dist/structured-attributes.d.ts

@@ -25,6 +25,8 @@
  * returns the trimmed input unchanged — never truncates legitimate content.
  */
 declare function stripAttributesSuffix(content: string): string;
+/** Return stored-body identity forms, from raw semantic content to full body. */
+declare function storedContentIdentityCandidates(content: string, stripCitation: (value: string) => string): string[];
 /**
  * Canonical `key: value; …` rendering of structured attributes. Keys are
  * trimmed + lowercased, values trimmed, pairs sorted by normalized key.
@@ -48,4 +50,4 @@ interface AssembledPersistBody {
  */
 declare function assemblePersistedBody(content: string, structuredAttributes?: Record<string, string>): AssembledPersistBody;
 
-export { type AssembledPersistBody, assemblePersistedBody, normalizeAttributePairs, stripAttributesSuffix };
+export { type AssembledPersistBody, assemblePersistedBody, normalizeAttributePairs, storedContentIdentityCandidates, stripAttributesSuffix };

package/dist/structured-attributes.js

@@ -1,14 +1,16 @@
 import {
   assemblePersistedBody,
   normalizeAttributePairs,
+  storedContentIdentityCandidates,
   stripAttributesSuffix
-} from "./chunk-R3DJVOSM.js";
+} from "./chunk-JDSVLVCG.js";
 import "./chunk-5GKPFL3S.js";
 import "./chunk-JAKIDQB2.js";
 import "./chunk-PZ5AY32C.js";
 export {
   assemblePersistedBody,
   normalizeAttributePairs,
+  storedContentIdentityCandidates,
   stripAttributesSuffix
 };
 //# sourceMappingURL=structured-attributes.js.map

package/dist/tier-migration.d.ts

@@ -1,5 +1,5 @@
 import { M as MemoryFile } from './types-CkiDUorT.js';
-import { S as StorageManager } from './storage-BmVudhk4.js';
+import { S as StorageManager } from './storage-DmkCjo0l.js';
 import { MemoryTier } from './tier-routing.js';
 import { SearchBackend } from './search/port.js';
 import './message-parts/index.js';

package/dist/transfer/types.d.ts

@@ -313,13 +313,13 @@ declare const CapsuleBlockSchema: z.ZodObject<{
         peerProfiles: boolean;
     }>;
 }, "strip", z.ZodTypeAny, {
-    schemaVersion: string;
     includes: {
         procedural: boolean;
         taxonomy: boolean;
         identityAnchors: boolean;
         peerProfiles: boolean;
     };
+    schemaVersion: string;
     id: string;
     version: string;
     description: string;
@@ -334,13 +334,13 @@ declare const CapsuleBlockSchema: z.ZodObject<{
         directAnswerEnabled: boolean;
     };
 }, {
-    schemaVersion: string;
     includes: {
         procedural: boolean;
         taxonomy: boolean;
         identityAnchors: boolean;
         peerProfiles: boolean;
     };
+    schemaVersion: string;
     id: string;
     version: string;
     description: string;
@@ -464,13 +464,13 @@ declare const ExportManifestV2Schema: z.ZodObject<{
             peerProfiles: boolean;
         }>;
     }, "strip", z.ZodTypeAny, {
-        schemaVersion: string;
         includes: {
             procedural: boolean;
             taxonomy: boolean;
             identityAnchors: boolean;
             peerProfiles: boolean;
         };
+        schemaVersion: string;
         id: string;
         version: string;
         description: string;
@@ -485,13 +485,13 @@ declare const ExportManifestV2Schema: z.ZodObject<{
             directAnswerEnabled: boolean;
         };
     }, {
-        schemaVersion: string;
         includes: {
             procedural: boolean;
             taxonomy: boolean;
             identityAnchors: boolean;
             peerProfiles: boolean;
         };
+        schemaVersion: string;
         id: string;
         version: string;
         description: string;
@@ -518,13 +518,13 @@ declare const ExportManifestV2Schema: z.ZodObject<{
     pluginVersion: string;
     includesTranscripts: boolean;
     capsule: {
-        schemaVersion: string;
         includes: {
             procedural: boolean;
             taxonomy: boolean;
             identityAnchors: boolean;
             peerProfiles: boolean;
         };
+        schemaVersion: string;
         id: string;
         version: string;
         description: string;
@@ -551,13 +551,13 @@ declare const ExportManifestV2Schema: z.ZodObject<{
     pluginVersion: string;
     includesTranscripts: boolean;
     capsule: {
-        schemaVersion: string;
         includes: {
             procedural: boolean;
             taxonomy: boolean;
             identityAnchors: boolean;
             peerProfiles: boolean;
         };
+        schemaVersion: string;
         id: string;
         version: string;
         description: string;
@@ -683,13 +683,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
                 peerProfiles: boolean;
             }>;
         }, "strip", z.ZodTypeAny, {
-            schemaVersion: string;
             includes: {
                 procedural: boolean;
                 taxonomy: boolean;
                 identityAnchors: boolean;
                 peerProfiles: boolean;
             };
+            schemaVersion: string;
             id: string;
             version: string;
             description: string;
@@ -704,13 +704,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
                 directAnswerEnabled: boolean;
             };
         }, {
-            schemaVersion: string;
             includes: {
                 procedural: boolean;
                 taxonomy: boolean;
                 identityAnchors: boolean;
                 peerProfiles: boolean;
             };
+            schemaVersion: string;
             id: string;
             version: string;
             description: string;
@@ -737,13 +737,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
         pluginVersion: string;
         includesTranscripts: boolean;
         capsule: {
-            schemaVersion: string;
             includes: {
                 procedural: boolean;
                 taxonomy: boolean;
                 identityAnchors: boolean;
                 peerProfiles: boolean;
             };
+            schemaVersion: string;
             id: string;
             version: string;
             description: string;
@@ -770,13 +770,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
         pluginVersion: string;
         includesTranscripts: boolean;
         capsule: {
-            schemaVersion: string;
             includes: {
                 procedural: boolean;
                 taxonomy: boolean;
                 identityAnchors: boolean;
                 peerProfiles: boolean;
             };
+            schemaVersion: string;
             id: string;
             version: string;
             description: string;
@@ -815,13 +815,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
         pluginVersion: string;
         includesTranscripts: boolean;
         capsule: {
-            schemaVersion: string;
             includes: {
                 procedural: boolean;
                 taxonomy: boolean;
                 identityAnchors: boolean;
                 peerProfiles: boolean;
             };
+            schemaVersion: string;
             id: string;
             version: string;
             description: string;
@@ -854,13 +854,13 @@ declare const ExportBundleV2Schema: z.ZodObject<{
         pluginVersion: string;
         includesTranscripts: boolean;
         capsule: {
-            schemaVersion: string;
             includes: {
                 procedural: boolean;
                 taxonomy: boolean;
                 identityAnchors: boolean;
                 peerProfiles: boolean;
             };
+            schemaVersion: string;
             id: string;
             version: string;
             description: string;

package/dist/verified-recall.js

@@ -1,14 +1,13 @@
 import {
   compareVerifiedEpisodeResults,
   searchVerifiedEpisodes
-} from "./chunk-J272WKX2.js";
+} from "./chunk-6HPHEFXT.js";
 import "./chunk-B7RWBZXW.js";
-import "./chunk-WD5B5AIE.js";
+import "./chunk-6R4G7N4S.js";
 import "./chunk-UHVXE24Y.js";
-import "./chunk-JJRVPWFY.js";
+import "./chunk-MLPTFODV.js";
 import "./chunk-SXIMQLGV.js";
 import "./chunk-S2V7RUCL.js";
-import "./chunk-J6A3CX5N.js";
 import "./chunk-UU3G6GUP.js";
 import "./chunk-SCU65EZI.js";
 import "./chunk-WRJVB3C2.js";
@@ -17,10 +16,9 @@ import "./chunk-DM2T26WE.js";
 import "./chunk-LDXUBPMO.js";
 import "./chunk-JYIJNNN3.js";
 import "./chunk-AAD2ZS6P.js";
-import "./chunk-DLG5V6ND.js";
-import "./chunk-5V6IXHPI.js";
+import "./chunk-K4QS7LXU.js";
+import "./chunk-NSVDUN76.js";
 import "./chunk-BT7NVCML.js";
-import "./chunk-R3DJVOSM.js";
 import "./chunk-YXZG5TVN.js";
 import "./chunk-5GINB5LC.js";
 import "./chunk-G7D6GZ5J.js";
@@ -38,12 +36,10 @@ import "./chunk-RGMVMVMF.js";
 import "./chunk-N7RWREIQ.js";
 import "./chunk-XMMESCPD.js";
 import "./chunk-42JKGUFJ.js";
-import "./chunk-QBSFOKY6.js";
-import "./chunk-5GKPFL3S.js";
+import "./chunk-VOELBUS4.js";
 import "./chunk-ZIKENA2C.js";
 import "./chunk-PQVCH2AE.js";
 import "./chunk-6MKAMLQL.js";
-import "./chunk-JAKIDQB2.js";
 import "./chunk-W5BIQOAJ.js";
 import "./chunk-Z7L72ZHL.js";
 import "./chunk-4O4TKIXQ.js";
@@ -51,16 +47,20 @@ import "./chunk-ZBJMUXZH.js";
 import "./chunk-BJMBJZ2Y.js";
 import "./chunk-UKJAGEXH.js";
 import "./chunk-FP2373TW.js";
+import "./chunk-J6A3CX5N.js";
 import "./chunk-7FGL6CGA.js";
 import "./chunk-EGGE52UU.js";
 import "./chunk-ZA53AZND.js";
 import "./chunk-GCASGYIO.js";
 import "./chunk-7RJUVTVF.js";
-import "./chunk-BQIFZ2MP.js";
+import "./chunk-CX5SW2EW.js";
 import "./chunk-3CSX3ETE.js";
 import "./chunk-RL6NEVXT.js";
+import "./chunk-JDSVLVCG.js";
 import "./chunk-FFAZ2XFU.js";
-import "./chunk-W5H6ZRPT.js";
+import "./chunk-5FVVY2EB.js";
+import "./chunk-5GKPFL3S.js";
+import "./chunk-JAKIDQB2.js";
 import "./chunk-P7FMDTKL.js";
 import "./chunk-2XRE6FLM.js";
 import "./chunk-EYIEWJNI.js";

package/dist/write-envelope.js

@@ -11,9 +11,9 @@ import {
   isMemoryCategory,
   isSealedMemoryEnvelope,
   sealedWriteToLegacyArgs
-} from "./chunk-5V6IXHPI.js";
+} from "./chunk-NSVDUN76.js";
 import "./chunk-BT7NVCML.js";
-import "./chunk-R3DJVOSM.js";
+import "./chunk-JDSVLVCG.js";
 import "./chunk-5GKPFL3S.js";
 import "./chunk-JAKIDQB2.js";
 import "./chunk-P7FMDTKL.js";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@remnic/core",
-  "version": "9.52.1",
+  "version": "9.53.0",
   "description": "Framework-agnostic Remnic memory engine — orchestrator, storage, extraction, search, trust zones",
   "type": "module",
   "main": "dist/index.js",
@@ -3126,7 +3126,7 @@
     "core"
   ],
   "peerDependencies": {
-    "@remnic/coding-graph": "^9.52.1"
+    "@remnic/coding-graph": "^9.53.0"
   },
   "peerDependenciesMeta": {
     "@remnic/coding-graph": {

package/src/access-offline-manifest.ts

@@ -47,6 +47,7 @@ export async function createOfflineSyncManifestStream(
           file,
           async ({ path }) => storage.readOfflineSyncFile(nodePath.join(storage.dir, path)),
           parseMemory,
+          orchestrator.config.inlineSourceAttributionFormat,
         );
       }
     })(),

package/src/access-service-offline-file-content.test.ts

@@ -194,6 +194,7 @@ test("offline manifest streams body-free active, archived, old, bad, and encrypt
       id: "active-memory",
       category: "fact",
       contentHash: activeContentHash,
+      normalizerVersion: 4,
       status: "active",
     });
     assert.notEqual(rowsByPath.get("facts/active.md")?.sha256, activeContentHash);

package/src/content-hash.ts

@@ -9,8 +9,18 @@
 
 import { createHash } from "node:crypto";
 
-/** Normalize content: lowercase, strip non-alphanumerics, collapse whitespace. */
+/** Normalize content: preserve Unicode letters, marks, and numbers. */
 export function normalizeContent(content: string): string {
+  return content
+    .normalize("NFC")
+    .toLowerCase()
+    .replace(/[^\p{L}\p{M}\p{N}\s]/gu, " ")
+    .replace(/\s+/g, " ")
+    .trim();
+}
+
+/** Normalize content with the pre-Unicode ASCII-only normalizer. */
+export function normalizeLegacyContent(content: string): string {
   return content
     .toLowerCase()
     .replace(/[^a-z0-9\s]/g, " ")
@@ -18,6 +28,29 @@ export function normalizeContent(content: string): string {
     .trim();
 }
 
+/** Compute the SHA-256 hash used by pre-Unicode content indexes. */
+export function computeLegacyContentHash(content: string): string {
+  return createHash("sha256").update(normalizeLegacyContent(content)).digest("hex");
+}
+
+/**
+ * Return true only when a persisted legacy hash can be proved to identify the
+ * same body under the current normalizer. Empty or changed legacy forms are
+ * ambiguous and must not be promoted to a current identity.
+ */
+export function isUnambiguousLegacyContentHash(
+  content: string,
+  contentHash: string,
+  currentNormalizedText: string = normalizeContent(content),
+): boolean {
+  const legacyNormalizedText = normalizeLegacyContent(content);
+  return (
+    legacyNormalizedText.length > 0 &&
+    legacyNormalizedText === currentNormalizedText &&
+    createHash("sha256").update(legacyNormalizedText).digest("hex") === contentHash
+  );
+}
+
 /** Normalize content and compute SHA-256 hash. */
 export function computeContentHash(content: string): string {
   return createHash("sha256").update(normalizeContent(content)).digest("hex");