@equationalapplications/core-llm-wiki 4.6.0 → 4.6.1

package/dist/index.mjs

@@ -139,6 +139,157 @@ var PrunePartialFailureError = class extends Error {
   }
 };
 
+// src/repositories/BaseRepository.ts
+var BaseRepository = class {
+  constructor(db, prefix) {
+    this.db = db;
+    this.prefix = prefix;
+  }
+  /**
+   * Return the DB executor for a given transaction handle.
+   * If tx is provided, use it; otherwise fall back to this.db.
+   */
+  getExecutor(tx) {
+    return tx ?? this.db;
+  }
+};
+
+// src/repositories/EntryRepository.ts
+function mapRowToFact(row) {
+  const tags = (() => {
+    if (Array.isArray(row.tags)) return row.tags;
+    try {
+      const p = JSON.parse(row.tags);
+      if (Array.isArray(p)) return p;
+    } catch {
+    }
+    return [];
+  })();
+  return {
+    id: row.id,
+    entity_id: row.entity_id,
+    title: row.title,
+    body: row.body,
+    tags,
+    confidence: row.confidence,
+    source_type: row.source_type,
+    source_hash: row.source_hash ?? null,
+    source_ref: row.source_ref ?? null,
+    created_at: Number(row.created_at),
+    updated_at: Number(row.updated_at),
+    last_accessed_at: row.last_accessed_at === null || row.last_accessed_at === void 0 ? null : Number(row.last_accessed_at),
+    deleted_at: row.deleted_at != null ? Number(row.deleted_at) : null,
+    access_count: Number(row.access_count ?? 0)
+  };
+}
+var EntryRepository = class extends BaseRepository {
+  constructor() {
+    super(...arguments);
+    this.chunkSize = 500;
+  }
+  /**
+   * Fetch facts by IDs, optionally scoped to entity IDs.
+   * Returns facts in the order of the input IDs (first match wins).
+   */
+  async findByIds(ids, scopedEntityIds, tx) {
+    const executor = this.getExecutor(tx);
+    const rows = [];
+    const entityClause = scopedEntityIds && scopedEntityIds.length > 0 ? ` AND entity_id IN (${scopedEntityIds.map(() => "?").join(",")})` : "";
+    const entityParams = scopedEntityIds ?? [];
+    for (let i = 0; i < ids.length; i += this.chunkSize) {
+      const chunk = ids.slice(i, i + this.chunkSize);
+      const placeholders = chunk.map(() => "?").join(",");
+      const chunkRows = await executor.getAllAsync(
+        `SELECT * FROM ${this.prefix}entries WHERE id IN (${placeholders})${entityClause} AND deleted_at IS NULL`,
+        [...chunk, ...entityParams]
+      );
+      rows.push(...chunkRows);
+    }
+    const byId = new Map(rows.map((r) => [r.id, r]));
+    return ids.map((id) => byId.get(id)).filter((r) => r !== void 0).map(mapRowToFact);
+  }
+  /**
+   * Upsert a WikiFact. Nullable fields set to null when fact value is null.
+   * Returns { changes, lastInsertRowId }.
+   */
+  async upsert(fact, tx) {
+    const executor = this.getExecutor(tx);
+    const now = Date.now();
+    const tagsJson = JSON.stringify(fact.tags);
+    const embeddingBlob = fact.embedding_blob instanceof Uint8Array ? fact.embedding_blob : fact.embedding_blob && typeof fact.embedding_blob === "object" && "type" in fact.embedding_blob ? new Uint8Array(fact.embedding_blob.data) : fact.embedding_blob && typeof fact.embedding_blob === "object" ? (() => {
+      const obj = fact.embedding_blob;
+      const keys = Object.keys(obj).map(Number).sort((a, b) => a - b);
+      const arr = new Uint8Array(keys.length);
+      for (let i = 0; i < keys.length; i++) arr[i] = obj[String(keys[i])];
+      return arr;
+    })() : void 0;
+    return executor.runAsync(
+      `INSERT INTO ${this.prefix}entries (
+        id, entity_id, title, body, tags, confidence, source_type,
+        source_hash, source_ref, created_at, updated_at, last_accessed_at, access_count,
+        deleted_at, embedding_blob, embedding
+      ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
+      ON CONFLICT(id) DO UPDATE SET
+        entity_id = excluded.entity_id,
+        title = excluded.title,
+        body = excluded.body,
+        tags = excluded.tags,
+        confidence = excluded.confidence,
+        source_type = excluded.source_type,
+        source_hash = excluded.source_hash,
+        source_ref = excluded.source_ref,
+        updated_at = excluded.updated_at,
+        last_accessed_at = excluded.last_accessed_at,
+        access_count = excluded.access_count,
+        deleted_at = excluded.deleted_at,
+        embedding_blob = CASE WHEN excluded.embedding_blob IS NULL THEN embedding_blob ELSE excluded.embedding_blob END,
+        embedding = NULL`,
+      [
+        fact.id,
+        fact.entity_id,
+        fact.title,
+        fact.body,
+        tagsJson,
+        fact.confidence,
+        fact.source_type,
+        fact.source_hash,
+        fact.source_ref,
+        fact.created_at,
+        now,
+        // updated_at set by repo
+        fact.last_accessed_at === null ? null : fact.last_accessed_at,
+        fact.access_count,
+        fact.deleted_at ?? null,
+        embeddingBlob ?? null,
+        null
+      ]
+    );
+  }
+  /**
+   * Soft-delete a single entry by ID scoped to entityId. Sets deleted_at + updated_at.
+   */
+  async softDelete(entryId, entityId, tx) {
+    const executor = this.getExecutor(tx);
+    const now = Date.now();
+    return executor.runAsync(
+      `UPDATE ${this.prefix}entries SET deleted_at = ?, updated_at = ? WHERE id = ? AND entity_id = ? AND deleted_at IS NULL`,
+      [now, now, entryId, entityId]
+    );
+  }
+  /**
+   * Fetch IDs + entity_ids of soft-deleted rows older than cutoff for a given entity.
+   * Used by runPrune().
+   */
+  async getPrunableMetadata(entityId, cutoff, tx) {
+    const executor = this.getExecutor(tx);
+    return executor.getAllAsync(
+      `SELECT id, entity_id FROM ${this.prefix}entries
+       WHERE entity_id = ? AND deleted_at IS NOT NULL AND deleted_at <= ?`,
+      [entityId, cutoff]
+    );
+  }
+};
+
 // src/prompts.ts
 var LIBRARIAN_SYSTEM_PROMPT = `You are a knowledge extraction agent. Your job is to analyze recent episodic events and extract stable facts and actionable tasks about the user or entity.
 Return ONLY a valid JSON object matching this schema:
@@ -461,6 +612,7 @@ var _WikiMemory = class _WikiMemory {
     this.db = db;
     this.options = options;
     this.prefix = options.config?.tablePrefix || "llm_wiki_";
+    this.entryRepo = new EntryRepository(db, this.prefix);
   }
   normalizeMiniSearchRow(row) {
     return {
@@ -916,11 +1068,7 @@ After running the migration SQL, restart your application.`
       let deletedEvents = 0;
       if (retainSoftDeletedFor !== null) {
         const cutoff = now - retainSoftDeletedFor * 864e5;
-        const entriesToDelete = await this.db.getAllAsync(
-          `SELECT id, entity_id FROM ${this.prefix}entries
-           WHERE entity_id = ? AND deleted_at IS NOT NULL AND deleted_at <= ?`,
-          [entityId, cutoff]
-        );
+        const entriesToDelete = await this.entryRepo.getPrunableMetadata(entityId, cutoff);
         const succeeded = [];
         let failure = null;
         for (const row of entriesToDelete) {
@@ -1476,7 +1624,15 @@ After running the migration SQL, restart your application.`
         const { embedding: _embedding, embedding_blob: _blob, ...rest } = f;
         return {
           ...rest,
-          tags: typeof rest.tags === "string" ? JSON.parse(rest.tags) : rest.tags
+          tags: (() => {
+            if (Array.isArray(rest.tags)) return rest.tags;
+            try {
+              const p = JSON.parse(rest.tags);
+              return Array.isArray(p) ? p : [];
+            } catch {
+              return [];
+            }
+          })()
         };
       });
     }
@@ -1556,28 +1712,8 @@ After running the migration SQL, restart your application.`
    * Hydrate full facts by ID. Pass scopedEntityIds to restrict to requested namespaces in SQL
    * (defense-in-depth against a rogue VectorRanker returning cross-entity IDs).
    */
-  async _hydrateFactsByIds(ids, scopedEntityIds) {
-    const fullRows = [];
-    const chunkSize = 500;
-    const entityClause = scopedEntityIds && scopedEntityIds.length > 0 ? ` AND entity_id IN (${scopedEntityIds.map(() => "?").join(",")})` : "";
-    const entityParams = scopedEntityIds && scopedEntityIds.length > 0 ? [...scopedEntityIds] : [];
-    for (let i = 0; i < ids.length; i += chunkSize) {
-      const idChunk = ids.slice(i, i + chunkSize);
-      const placeholders = idChunk.map(() => "?").join(",");
-      const chunkRows = await this.db.getAllAsync(
-        `SELECT * FROM ${this.prefix}entries WHERE id IN (${placeholders})${entityClause} AND deleted_at IS NULL`,
-        [...idChunk, ...entityParams]
-      );
-      fullRows.push(...chunkRows);
-    }
-    const byId = new Map(fullRows.map((row) => [row.id, row]));
-    return ids.map((id) => byId.get(id)).filter((fact) => fact !== void 0).map((fact) => {
-      const { embedding: _embedding, embedding_blob: _blob, ...rest } = fact;
-      return {
-        ...rest,
-        tags: typeof rest.tags === "string" ? JSON.parse(rest.tags) : rest.tags
-      };
-    });
+  async _hydrateFactsByIds(ids, scopedEntityIds, tx) {
+    return this.entryRepo.findByIds(ids, scopedEntityIds, tx);
   }
   /**
    * Strip potentially sensitive data from ranker errors before exposing to host callbacks.
@@ -2590,7 +2726,7 @@ The following document anchors are provided for contradiction detection only. Do
           const entriesToDelete = await this.db.getAllAsync(q, args);
           deletedEntryIds.push(...entriesToDelete.map((e) => e.id));
         }
-        const entryPromise = params.entryId ? this.db.runAsync(`UPDATE ${this.prefix}entries SET deleted_at = ?, updated_at = ? WHERE id = ? AND entity_id = ? AND deleted_at IS NULL`, [now, now, params.entryId, entityId]) : null;
+        const entryPromise = params.entryId ? this.entryRepo.softDelete(params.entryId, entityId) : null;
         const taskPromise = params.taskId ? this.db.runAsync(`UPDATE ${this.prefix}tasks SET deleted_at = ?, updated_at = ? WHERE id = ? AND entity_id = ? AND deleted_at IS NULL`, [now, now, params.taskId, entityId]) : null;
         let refPromise = null;
         if (sourceRef || sourceHash) {
@@ -2727,11 +2863,23 @@ ${chunk}`;
         );
         for (const fact of allValidFacts) {
           const id = generateId("fact_");
-          await this.db.runAsync(
-            `INSERT INTO ${this.prefix}entries (id, entity_id, title, body, tags, confidence, source_type, source_hash, source_ref, created_at, updated_at)
-           VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)`,
-            [id, entityId, fact.title, fact.body, JSON.stringify(fact.tags), fact.confidence, "immutable_document", sourceHash, sourceRef, now, now]
-          );
+          const wikiFact = {
+            id,
+            entity_id: entityId,
+            title: fact.title,
+            body: fact.body,
+            tags: fact.tags,
+            confidence: fact.confidence,
+            source_type: "immutable_document",
+            source_hash: sourceHash,
+            source_ref: sourceRef,
+            created_at: now,
+            updated_at: now,
+            last_accessed_at: null,
+            access_count: 0,
+            deleted_at: null
+          };
+          await this.entryRepo.upsert(wikiFact);
           insertedFacts.push({ id, entity_id: entityId, title: fact.title, body: fact.body, tags: JSON.stringify(fact.tags) });
         }
       });