@equationalapplications/core-llm-wiki 4.23.0 → 5.0.0

package/dist/index.mjs

@@ -1,5 +1,5 @@
-import { __privateAdd, MetadataRepository, EmbeddingService, SearchService, JobManager, PromptService, IngestionService, MaintenanceService, ImportExportService, RetrievalService, WriteService, __privateGet, __privateSet, normalizeSourceRef, normalizeSourceHash, entitySummaryMetaKey, generateId, BaseRepository, emptyManifest, resolveNodeType, validateInlineEdges, resolveEdgeDefinitions, normalizeTitleKey, WikiTransactionError, extractSqliteCode } from './chunk-MYZJLVX4.mjs';
-export { HOOK_TIMEOUT_MARKER, ONTOLOGY_BACKFILL_BATCH_SIZE, ONTOLOGY_BACKFILL_MAX_PROMPT_CHARS, ONTOLOGY_BACKFILL_RECHECK_MS, ONTOLOGY_BACKFILL_SYSTEM_PROMPT, PromptService, PrunePartialFailureError, WikiBusyError, WikiTransactionError, configureRandomSource, parseEmbedding, validateManifest } from './chunk-MYZJLVX4.mjs';
+import { __privateAdd, MetadataRepository, EmbeddingService, SearchService, JobManager, PromptService, IngestionService, MaintenanceService, ImportExportService, RetrievalService, WriteService, __privateGet, __privateSet, normalizeSourceRef, normalizeSourceHash, entitySummaryMetaKey, generateId, BaseRepository, emptyManifest, resolveNodeType, validateInlineEdges, resolveEdgeDefinitions, normalizeTitleKey, WikiTransactionError, extractSqliteCode } from './chunk-W53E3I44.mjs';
+export { HEAL_BATCH_SIZE, HEAL_RECHECK_MS, HOOK_TIMEOUT_MARKER, ONTOLOGY_BACKFILL_BATCH_SIZE, ONTOLOGY_BACKFILL_MAX_PROMPT_CHARS, ONTOLOGY_BACKFILL_RECHECK_MS, ONTOLOGY_BACKFILL_SYSTEM_PROMPT, PromptService, PrunePartialFailureError, WikiBusyError, WikiTransactionError, configureRandomSource, parseEmbedding, validateManifest } from './chunk-W53E3I44.mjs';
 import { appendRelatedSection, buildConceptDocument, buildLogMd, buildEntityIndexMd, buildRootIndexMd, isAllowedOkfPath, parseRootIndexMd, parseEntityIndexMd, parseConcept, splitRelatedSection, extractMarkdownLinks, parseLogMd, parseEventIdComment, appendEventIdComment } from '@equationalapplications/core-okf';
 
 // src/db/schema.ts
@@ -23,7 +23,8 @@ async function setupDatabase(db, prefix) {
       embedding TEXT,
       embedding_blob BLOB,
       okf_type TEXT,
-      ontology_checked_at INTEGER
+      ontology_checked_at INTEGER,
+      heal_checked_at INTEGER
     );
 
     CREATE INDEX IF NOT EXISTS ${prefix}entries_entity_idx ON ${prefix}entries(entity_id);
@@ -263,6 +264,20 @@ var MIGRATIONS = [
         );
       }
     }
+  },
+  {
+    version: 8,
+    description: "Add heal_checked_at to entries for heal recheck cooldown",
+    run: async (db, prefix) => {
+      const cols = await db.getAllAsync(
+        `PRAGMA table_info(${prefix}entries)`
+      );
+      if (!cols.some((c) => c.name === "heal_checked_at")) {
+        await db.execAsync(
+          `ALTER TABLE ${prefix}entries ADD COLUMN heal_checked_at INTEGER`
+        );
+      }
+    }
   }
 ];
 for (let i = 1; i < MIGRATIONS.length; i++) {
@@ -604,17 +619,21 @@ var EntryRepository = class extends BaseRepository {
   }
   /**
    * Fetch live, mutable entries for an entity — everything heal is allowed to
-   * downgrade or delete. Heal previously loaded every row via
-   * findAllByEntityId and filtered in JS, which on a document-heavy corpus
-   * meant loading 2560 rows to keep 31.
+   * downgrade or delete — oldest first, capped at `limit`, skipping anything
+   * stamped inside the recheck cooldown (heal_checked_at > recheckCutoff).
+   *
+   * Oldest-first rather than newest-first: with a cooldown in place, newest-first
+   * would keep re-selecting recently-touched facts every pass while older ones
+   * wait indefinitely to even enter a batch.
    */
-  async findHealCandidatesByEntityId(entityId, tx) {
+  async findHealCandidatesByEntityId(entityId, limit, recheckCutoff, tx) {
     const executor = this.getExecutor(tx);
     const rows = await executor.getAllAsync(
       `SELECT * FROM ${this.prefix}entries
        WHERE entity_id = ? AND deleted_at IS NULL AND source_type != 'immutable_document'
-       ORDER BY updated_at DESC`,
-      [entityId]
+         AND (heal_checked_at IS NULL OR heal_checked_at <= ?)
+       ORDER BY updated_at ASC LIMIT ?`,
+      [entityId, recheckCutoff, limit]
     );
     return rows.map(mapRowToFact);
   }
@@ -777,6 +796,10 @@ var EntryRepository = class extends BaseRepository {
   /**
    * Downgrade stale inferred entries to 'tentative'.
    * Used by MaintenanceService.doRunHeal().
+   *
+   * Returns the ids it downgraded so the caller can fold them into
+   * {@link HealResult.downgraded} without double-counting a fact the model also
+   * downgraded in the same pass.
    */
   async downgradeStaleInferred(entityId, staleThreshold, tx) {
     const executor = this.getExecutor(tx);
@@ -788,8 +811,8 @@ var EntryRepository = class extends BaseRepository {
          AND source_type != 'immutable_document' AND deleted_at IS NULL`,
       [entityId, staleThreshold, staleThreshold]
     );
-    if (eligibleRows.length === 0) return 0;
-    const result = await executor.runAsync(
+    if (eligibleRows.length === 0) return [];
+    await executor.runAsync(
       `UPDATE ${this.prefix}entries
        SET confidence = 'tentative', updated_at = ?
        WHERE entity_id = ? AND confidence = 'inferred' AND (last_accessed_at <= ? OR (last_accessed_at IS NULL AND created_at <= ?)) AND source_type != 'immutable_document' AND deleted_at IS NULL`,
@@ -804,7 +827,7 @@ var EntryRepository = class extends BaseRepository {
         payload: { id: row.id, entity_id: entityId, confidence: "tentative", updated_at: now }
       }, tx);
     }
-    return result.changes;
+    return eligibleRows.map((r) => r.id);
   }
   /**
    * Downgrade specific entries to 'tentative' by IDs.
@@ -1123,6 +1146,58 @@ var EntryRepository = class extends BaseRepository {
     );
     return rows.map((r) => ({ id: r.id, title: r.title, okf_type: r.okf_type ?? null }));
   }
+  /** Counts live mutable facts: eligible (past cooldown) vs deferred (in cooldown). */
+  async countHealCandidatesByEntityId(entityId, recheckCutoff, tx) {
+    const executor = this.getExecutor(tx);
+    const row = await executor.getFirstAsync(
+      `SELECT
+         SUM(CASE WHEN heal_checked_at IS NULL OR heal_checked_at <= ? THEN 1 ELSE 0 END) AS eligible,
+         SUM(CASE WHEN heal_checked_at IS NOT NULL AND heal_checked_at > ? THEN 1 ELSE 0 END) AS deferred
+       FROM ${this.prefix}entries
+       WHERE entity_id = ? AND deleted_at IS NULL AND source_type != 'immutable_document'`,
+      [recheckCutoff, recheckCutoff, entityId]
+    );
+    return { eligible: Number(row?.eligible ?? 0), deferred: Number(row?.deferred ?? 0) };
+  }
+  /**
+   * Stamps the heal recheck cooldown. NEVER touches updated_at — import merge
+   * resolution is last-write-wins on updated_at and a bump here would make an
+   * unchanged local fact beat a genuinely newer remote edit.
+   *
+   * Soft-deleted rows are skipped: a candidate heal deleted earlier in the same
+   * transaction is no longer a candidate under any future pass, so its cooldown
+   * value is irrelevant. This means the stamped count can be lower than the
+   * offered-candidate count.
+   */
+  async markHealChecked(ids, entityId, now, tx) {
+    if (ids.length === 0) return;
+    for (let i = 0; i < ids.length; i += this.chunkSize) {
+      const chunk = ids.slice(i, i + this.chunkSize);
+      const placeholders = chunk.map(() => "?").join(",");
+      await tx.runAsync(
+        `UPDATE ${this.prefix}entries SET heal_checked_at = ?
+         WHERE id IN (${placeholders}) AND entity_id = ? AND deleted_at IS NULL`,
+        [now, ...chunk, entityId]
+      );
+    }
+  }
+  /**
+   * Lightweight full-breadth index (id, title) over all live librarian_inferred
+   * facts. This is heal's fuzzy-dedupe corpus, deliberately kept independent of
+   * the bounded candidate window: seeding dedupe from a batchSize-limited read
+   * would let a synthesized fact duplicating a fact outside the window pass the
+   * Jaccard check, and a convergence loop would multiply those duplicates across
+   * passes. Mirrors findTitleIndexByEntityId — full breadth, two columns, cheap.
+   */
+  async findInferredTitlesByEntityId(entityId, tx) {
+    const executor = this.getExecutor(tx);
+    const rows = await executor.getAllAsync(
+      `SELECT id, title FROM ${this.prefix}entries
+       WHERE entity_id = ? AND deleted_at IS NULL AND source_type = 'librarian_inferred'`,
+      [entityId]
+    );
+    return rows.map((r) => ({ id: r.id, title: r.title }));
+  }
 };
 
 // src/repositories/OutboxRepository.ts
@@ -2021,9 +2096,23 @@ var WikiMemory = class {
     return this.maintenanceService.runLibrarian(entityId, options);
   }
   /**
+   * Reviews stored facts with the LLM: removes orphans, downgrades stale
+   * inferences, synthesizes corrections.
+   *
+   * Bounded per call: covers at most `batchSize` candidates (default 25), so
+   * one call no longer sweeps the whole entity. Hosts own the cadence; loop
+   * `while (result.remaining > 0)` for convergence.
+   *
+   * `remaining === 0` means "every mutable fact is inside the 7-day recheck
+   * cooldown", not "there is no more work" — heal's candidate set is the live
+   * mutable corpus, not a draining backlog, so it climbs back as the cooldown
+   * lapses and as new facts are written. Do not write a loop expecting a drain.
+   *
    * @param options.promptOverride - Applies only to this manual call. Does NOT affect
    * WriteService-triggered auto-runs. For persistent prompt customization across auto-runs,
    * set `options.config.prompts.healSystemPrompt` at WikiMemory construction time.
+   * @param options.batchSize - Candidates per run (default 25) for providers with
+   * tighter context limits.
    */
   async runHeal(entityId, options) {
     return this.maintenanceService.runHeal(entityId, options);