@equationalapplications/core-llm-wiki 6.4.0 → 6.5.1

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, WikiSourceRefHashCollision, validateManifest, generateId, BaseRepository, emptyManifest, resolveNodeType, WikiStrictOntologyViolation, validateInlineEdges, resolveEdgeDefinitions, typeSatisfies, normalizeTitleKey, WikiTransactionError, extractSqliteCode } from './chunk-DPFM33WV.mjs';
-export { DEFAULT_CHUNK_OVERLAP, DEFAULT_MAX_CHUNK_LENGTH, DEFAULT_MAX_EMBED_CHARS, EMBED_CHARS_CEILING, HEAL_ANCHORS_PER_CANDIDATE, HEAL_BATCH_SIZE, HEAL_MAX_FACT_BODY_CHARS_L3, HEAL_MAX_TASKS, 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, WikiDuplicateHashError, WikiIngestEmptyError, WikiInvalidReadOptions, WikiParseError, WikiSourceRefHashCollision, WikiStrictOntologyViolation, WikiTransactionError, chunkText, configureRandomSource, parseEmbedding, safeSlice, validateManifest } from './chunk-DPFM33WV.mjs';
+import { __privateAdd, MetadataRepository, EmbeddingService, SearchService, JobManager, PromptService, IngestionService, MaintenanceService, ImportExportService, RetrievalService, WriteService, __privateGet, __privateSet, normalizeSourceRef, normalizeSourceHash, entitySummaryMetaKey, WikiSourceRefHashCollision, validateManifest, generateId, BaseRepository, emptyManifest, resolveNodeType, WikiStrictOntologyViolation, validateInlineEdges, resolveEdgeDefinitions, typeSatisfies, normalizeTitleKey, WikiTransactionError, extractSqliteCode } from './chunk-YNFYSLR6.mjs';
+export { DEFAULT_CHUNK_OVERLAP, DEFAULT_MAX_CHUNK_LENGTH, DEFAULT_MAX_EMBED_CHARS, EMBED_CHARS_CEILING, HEAL_ANCHORS_PER_CANDIDATE, HEAL_BATCH_SIZE, HEAL_MAX_FACT_BODY_CHARS_L3, HEAL_MAX_TASKS, 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, WikiDuplicateHashError, WikiIngestEmptyError, WikiInvalidReadOptions, WikiParseError, WikiSourceRefHashCollision, WikiStrictOntologyViolation, WikiTransactionError, chunkText, configureRandomSource, parseEmbedding, safeSlice, validateManifest } from './chunk-YNFYSLR6.mjs';
 import { appendRelatedSection, buildConceptDocument, buildLogMd, buildEntityIndexMd, buildRootIndexMd, isAllowedOkfPath, parseRootIndexMd, parseEntityIndexMd, parseConcept, splitRelatedSection, extractMarkdownLinks, parseLogMd, parseEventIdComment, latestVerified, formatSourcesJson, appendEventIdComment, parseVerifiedFlexible, parseCitationsList, deriveTrustTier, isStaleAfter } from '@equationalapplications/core-okf';
 
 // src/db/schema.ts
@@ -32,7 +32,10 @@ async function setupDatabase(db, prefix) {
       last_verified_by TEXT,
       okf_sources TEXT,
       okf_verified TEXT,
-      okf_usage_window TEXT
+      okf_usage_window TEXT,
+      embedding_failed_at INTEGER,
+      embedding_failure_kind TEXT,
+      embedding_attempts INTEGER NOT NULL DEFAULT 0
     );
 
     CREATE INDEX IF NOT EXISTS ${prefix}entries_entity_idx ON ${prefix}entries(entity_id);
@@ -418,6 +421,24 @@ var MIGRATIONS = [
         CREATE INDEX IF NOT EXISTS ${prefix}tasks_last_verified_at_idx ON ${prefix}tasks(last_verified_at);
       `);
     }
+  },
+  {
+    version: 11,
+    description: "Add embedding_failed_at, embedding_failure_kind, embedding_attempts to entries for embed retry state",
+    run: async (db, prefix) => {
+      const cols = await db.getAllAsync(
+        `PRAGMA table_info(${prefix}entries)`
+      );
+      const existing = new Set(cols.map((c) => c.name));
+      const adds = [
+        ["embedding_failed_at", `ALTER TABLE ${prefix}entries ADD COLUMN embedding_failed_at INTEGER`],
+        ["embedding_failure_kind", `ALTER TABLE ${prefix}entries ADD COLUMN embedding_failure_kind TEXT`],
+        ["embedding_attempts", `ALTER TABLE ${prefix}entries ADD COLUMN embedding_attempts INTEGER NOT NULL DEFAULT 0`]
+      ];
+      for (const [name, sql] of adds) {
+        if (!existing.has(name)) await db.execAsync(sql);
+      }
+    }
   }
 ];
 for (let i = 1; i < MIGRATIONS.length; i++) {
@@ -563,6 +584,13 @@ var EntryRepository = class extends BaseRepository {
    * New-row INSERTs still bind a value so the column is populated; if the
    * caller supplied nothing, the schema DEFAULT 'stable' applies when the
    * bind is bound as 'stable' explicitly here.
+   *
+   * Marker discipline (spec §4.1): a conflict carrying a real
+   * `embedding_blob` clears the embedding failure markers, because the row now
+   * HAS a valid embedding and a surviving marker would be a self-contradictory
+   * diagnostic that inflates runReembed's permanentlyFailed counter. A conflict
+   * with no blob leaves marker state alone, matching the "absent means don't
+   * touch" semantics used for embedding_blob itself.
    */
   async upsert(fact, tx) {
     const executor = this.getExecutor(tx);
@@ -600,6 +628,9 @@ var EntryRepository = class extends BaseRepository {
         deleted_at = excluded.deleted_at,
         embedding_blob = CASE WHEN excluded.embedding_blob IS NULL THEN embedding_blob ELSE excluded.embedding_blob END,
         embedding = NULL,
+        embedding_failed_at = CASE WHEN excluded.embedding_blob IS NOT NULL THEN NULL ELSE embedding_failed_at END,
+        embedding_failure_kind = CASE WHEN excluded.embedding_blob IS NOT NULL THEN NULL ELSE embedding_failure_kind END,
+        embedding_attempts = CASE WHEN excluded.embedding_blob IS NOT NULL THEN 0 ELSE embedding_attempts END,
         okf_type = excluded.okf_type`,
       [
         fact.id,
@@ -728,6 +759,9 @@ var EntryRepository = class extends BaseRepository {
         deleted_at = excluded.deleted_at,
         embedding_blob = excluded.embedding_blob,
         embedding = NULL,
+        embedding_failed_at = CASE WHEN excluded.embedding_blob IS NOT NULL THEN NULL ELSE embedding_failed_at END,
+        embedding_failure_kind = CASE WHEN excluded.embedding_blob IS NOT NULL THEN NULL ELSE embedding_failure_kind END,
+        embedding_attempts = CASE WHEN excluded.embedding_blob IS NOT NULL THEN 0 ELSE embedding_attempts END,
         okf_type = excluded.okf_type,
         lifecycle_status = excluded.lifecycle_status,
         stale_after = excluded.stale_after,
@@ -1164,10 +1198,58 @@ var EntryRepository = class extends BaseRepository {
   async updateEmbeddingBlob(id, blob, tx) {
     const executor = this.getExecutor(tx);
     await executor.runAsync(
-      `UPDATE ${this.prefix}entries SET embedding_blob = ?, embedding = NULL WHERE id = ?`,
+      `UPDATE ${this.prefix}entries
+         SET embedding_blob = ?,
+             embedding = NULL,
+             embedding_failed_at = NULL,
+             embedding_failure_kind = NULL,
+             embedding_attempts = 0
+       WHERE id = ?`,
       [blob, id]
     );
   }
+  /**
+   * Record a failed embedding attempt. Deliberately does NOT touch updated_at
+   * (import merge is last-write-wins on it) and pushes no outbox event —
+   * embedding lifecycle is local state, not replicated. Same discipline as
+   * updateEmbeddingBlob. See spec §3.5.
+   *
+   * Only marker-eligible kinds (spec §3.3) are accepted: `no_provider` never
+   * marks and `storage_error` never marks, so retry/permanent-failure policy
+   * built on these rows only ever sees real embed failures.
+   */
+  async markEmbeddingFailure(id, kind, now, tx) {
+    const executor = this.getExecutor(tx);
+    await executor.runAsync(
+      `UPDATE ${this.prefix}entries
+         SET embedding_failed_at = ?,
+             embedding_failure_kind = ?,
+             embedding_attempts = embedding_attempts + 1
+       WHERE id = ?`,
+      [now, kind, id]
+    );
+  }
+  /**
+   * Clear marker state for every marked row. Called when the embedding
+   * dimension is promoted: a new model produces different vectors, so past
+   * failures — including `float32_overflow`, which is otherwise terminal —
+   * no longer predict future ones (spec §2.3).
+   *
+   * Same discipline as markEmbeddingFailure: no `updated_at` touch and no
+   * outbox event, because embedding lifecycle is local state, not replicated.
+   * Returns the number of rows cleared.
+   */
+  async clearEmbeddingFailureMarkers(tx) {
+    const executor = this.getExecutor(tx);
+    const result = await executor.runAsync(
+      `UPDATE ${this.prefix}entries
+          SET embedding_failed_at = NULL,
+              embedding_failure_kind = NULL,
+              embedding_attempts = 0
+        WHERE embedding_failed_at IS NOT NULL`
+    );
+    return result.changes;
+  }
   async hasLegacySourceTypes(tx) {
     const executor = this.getExecutor(tx);
     const row = await executor.getFirstAsync(