@equationalapplications/core-llm-wiki 7.7.3 → 7.7.4

package/dist/index.d.mts

@@ -1,5 +1,5 @@
-import { M as MemoryBundle, F as FormatContextOptions, G as GraphNeighborhood, a as MemoryDump, b as FormattedMemoryDump, R as ReadOptions, S as SQLiteAdapter, W as WikiOptions, c as WikiMemory } from './testing-A4Sofxyo.mjs';
-export { C as ChunkFailure, d as ClassifierAnswer, e as ClassifierQuestion, f as ClassifyRequest, g as ClassifyResponse, D as DegradedRecord, h as DraftPage, E as EmbedFactResult, i as EmbedFailureKind, j as EmbeddingMarkerKind, k as EntityStatus, l as ExtractedFact, m as ExtractedFactEdge, n as ExtractedFactWithOntology, o as ExtractedTask, p as GraphTraversalOptions, q as GroundingConfig, H as HEAL_BATCH_SIZE, r as HEAL_RECHECK_MS, s as HOOK_TIMEOUT_MARKER, t as HealResult, I as IngestDocumentResult, L as LLMProvider, O as ONTOLOGY_BACKFILL_BATCH_SIZE, u as ONTOLOGY_BACKFILL_MAX_PROMPT_CHARS, v as ONTOLOGY_BACKFILL_RECHECK_MS, w as OntologyBackfillResult, x as OntologyConfig, y as OntologyEdgeType, z as OntologyManifest, A as OntologyMode, B as OntologyNodeType, J as OntologyPromptContext, K as OntologyUpdates, P as PendingSourceStatus, N as PromptOverrides, Q as PromptService, T as PrunePartialFailureError, U as ReembedResult, V as VectorRanker, X as VectorRankerFallback, Y as VectorRankerRankArgs, Z as VectorRankerSemanticResult, _ as WikiBusyError, $ as WikiBusyOperation, a0 as WikiCheckpoint, a1 as WikiConfig, a2 as WikiDiagnostic, a3 as WikiDiagnosticCode, a4 as WikiDiagnosticDetail, a5 as WikiDiagnosticOperation, a6 as WikiDiagnosticSeverity, a7 as WikiDiagnosticTrigger, a8 as WikiDraftNotFound, a9 as WikiDuplicateHashError, aa as WikiEdge, ab as WikiEvent, ac as WikiFact, ad as WikiGraphNodeOwnershipConflict, ae as WikiIngestEmptyError, af as WikiInstructions, ag as WikiInvalidReadOptions, ah as WikiLintReport, ai as WikiMemoryTestAccess, aj as WikiOutboxEvent, ak as WikiParseError, al as WikiSourceRefHashCollision, am as WikiStrictOntologyViolation, an as WikiTask, ao as WikiTransactionError, ap as validateManifest } from './testing-A4Sofxyo.mjs';
+import { M as MemoryBundle, F as FormatContextOptions, G as GraphNeighborhood, a as MemoryDump, b as FormattedMemoryDump, R as ReadOptions, S as SQLiteAdapter, W as WikiOptions, c as WikiMemory } from './testing-CH_Kk34-.mjs';
+export { C as ChunkFailure, d as ClassifierAnswer, e as ClassifierQuestion, f as ClassifyRequest, g as ClassifyResponse, D as DegradedRecord, h as DraftPage, E as EmbedFactResult, i as EmbedFailureKind, j as EmbeddingMarkerKind, k as EntityStatus, l as ExtractedFact, m as ExtractedFactEdge, n as ExtractedFactWithOntology, o as ExtractedTask, p as GraphTraversalOptions, q as GroundingConfig, H as HEAL_BATCH_SIZE, r as HEAL_RECHECK_MS, s as HOOK_TIMEOUT_MARKER, t as HealResult, I as IngestDocumentResult, L as LLMProvider, O as ONTOLOGY_BACKFILL_BATCH_SIZE, u as ONTOLOGY_BACKFILL_MAX_PROMPT_CHARS, v as ONTOLOGY_BACKFILL_RECHECK_MS, w as OntologyBackfillResult, x as OntologyConfig, y as OntologyEdgeType, z as OntologyManifest, A as OntologyMode, B as OntologyNodeType, J as OntologyPromptContext, K as OntologyUpdates, P as PendingSourceStatus, N as PromptOverrides, Q as PromptService, T as PrunePartialFailureError, U as ReembedResult, V as VectorRanker, X as VectorRankerFallback, Y as VectorRankerRankArgs, Z as VectorRankerSemanticResult, _ as WikiBusyError, $ as WikiBusyOperation, a0 as WikiCheckpoint, a1 as WikiConfig, a2 as WikiDiagnostic, a3 as WikiDiagnosticCode, a4 as WikiDiagnosticDetail, a5 as WikiDiagnosticOperation, a6 as WikiDiagnosticSeverity, a7 as WikiDiagnosticTrigger, a8 as WikiDraftNotFound, a9 as WikiDuplicateHashError, aa as WikiEdge, ab as WikiEvent, ac as WikiFact, ad as WikiGraphNodeOwnershipConflict, ae as WikiIngestEmptyError, af as WikiInstructions, ag as WikiInvalidReadOptions, ah as WikiLintReport, ai as WikiMemoryTestAccess, aj as WikiOutboxEvent, ak as WikiParseError, al as WikiSourceRefHashCollision, am as WikiStrictOntologyViolation, an as WikiTask, ao as WikiTransactionError, ap as validateManifest } from './testing-CH_Kk34-.mjs';
 import { OkfFile } from '@equationalapplications/core-okf';
 import 'minisearch';
 

package/dist/index.d.ts

@@ -1,5 +1,5 @@
-import { M as MemoryBundle, F as FormatContextOptions, G as GraphNeighborhood, a as MemoryDump, b as FormattedMemoryDump, R as ReadOptions, S as SQLiteAdapter, W as WikiOptions, c as WikiMemory } from './testing-A4Sofxyo.js';
-export { C as ChunkFailure, d as ClassifierAnswer, e as ClassifierQuestion, f as ClassifyRequest, g as ClassifyResponse, D as DegradedRecord, h as DraftPage, E as EmbedFactResult, i as EmbedFailureKind, j as EmbeddingMarkerKind, k as EntityStatus, l as ExtractedFact, m as ExtractedFactEdge, n as ExtractedFactWithOntology, o as ExtractedTask, p as GraphTraversalOptions, q as GroundingConfig, H as HEAL_BATCH_SIZE, r as HEAL_RECHECK_MS, s as HOOK_TIMEOUT_MARKER, t as HealResult, I as IngestDocumentResult, L as LLMProvider, O as ONTOLOGY_BACKFILL_BATCH_SIZE, u as ONTOLOGY_BACKFILL_MAX_PROMPT_CHARS, v as ONTOLOGY_BACKFILL_RECHECK_MS, w as OntologyBackfillResult, x as OntologyConfig, y as OntologyEdgeType, z as OntologyManifest, A as OntologyMode, B as OntologyNodeType, J as OntologyPromptContext, K as OntologyUpdates, P as PendingSourceStatus, N as PromptOverrides, Q as PromptService, T as PrunePartialFailureError, U as ReembedResult, V as VectorRanker, X as VectorRankerFallback, Y as VectorRankerRankArgs, Z as VectorRankerSemanticResult, _ as WikiBusyError, $ as WikiBusyOperation, a0 as WikiCheckpoint, a1 as WikiConfig, a2 as WikiDiagnostic, a3 as WikiDiagnosticCode, a4 as WikiDiagnosticDetail, a5 as WikiDiagnosticOperation, a6 as WikiDiagnosticSeverity, a7 as WikiDiagnosticTrigger, a8 as WikiDraftNotFound, a9 as WikiDuplicateHashError, aa as WikiEdge, ab as WikiEvent, ac as WikiFact, ad as WikiGraphNodeOwnershipConflict, ae as WikiIngestEmptyError, af as WikiInstructions, ag as WikiInvalidReadOptions, ah as WikiLintReport, ai as WikiMemoryTestAccess, aj as WikiOutboxEvent, ak as WikiParseError, al as WikiSourceRefHashCollision, am as WikiStrictOntologyViolation, an as WikiTask, ao as WikiTransactionError, ap as validateManifest } from './testing-A4Sofxyo.js';
+import { M as MemoryBundle, F as FormatContextOptions, G as GraphNeighborhood, a as MemoryDump, b as FormattedMemoryDump, R as ReadOptions, S as SQLiteAdapter, W as WikiOptions, c as WikiMemory } from './testing-CH_Kk34-.js';
+export { C as ChunkFailure, d as ClassifierAnswer, e as ClassifierQuestion, f as ClassifyRequest, g as ClassifyResponse, D as DegradedRecord, h as DraftPage, E as EmbedFactResult, i as EmbedFailureKind, j as EmbeddingMarkerKind, k as EntityStatus, l as ExtractedFact, m as ExtractedFactEdge, n as ExtractedFactWithOntology, o as ExtractedTask, p as GraphTraversalOptions, q as GroundingConfig, H as HEAL_BATCH_SIZE, r as HEAL_RECHECK_MS, s as HOOK_TIMEOUT_MARKER, t as HealResult, I as IngestDocumentResult, L as LLMProvider, O as ONTOLOGY_BACKFILL_BATCH_SIZE, u as ONTOLOGY_BACKFILL_MAX_PROMPT_CHARS, v as ONTOLOGY_BACKFILL_RECHECK_MS, w as OntologyBackfillResult, x as OntologyConfig, y as OntologyEdgeType, z as OntologyManifest, A as OntologyMode, B as OntologyNodeType, J as OntologyPromptContext, K as OntologyUpdates, P as PendingSourceStatus, N as PromptOverrides, Q as PromptService, T as PrunePartialFailureError, U as ReembedResult, V as VectorRanker, X as VectorRankerFallback, Y as VectorRankerRankArgs, Z as VectorRankerSemanticResult, _ as WikiBusyError, $ as WikiBusyOperation, a0 as WikiCheckpoint, a1 as WikiConfig, a2 as WikiDiagnostic, a3 as WikiDiagnosticCode, a4 as WikiDiagnosticDetail, a5 as WikiDiagnosticOperation, a6 as WikiDiagnosticSeverity, a7 as WikiDiagnosticTrigger, a8 as WikiDraftNotFound, a9 as WikiDuplicateHashError, aa as WikiEdge, ab as WikiEvent, ac as WikiFact, ad as WikiGraphNodeOwnershipConflict, ae as WikiIngestEmptyError, af as WikiInstructions, ag as WikiInvalidReadOptions, ah as WikiLintReport, ai as WikiMemoryTestAccess, aj as WikiOutboxEvent, ak as WikiParseError, al as WikiSourceRefHashCollision, am as WikiStrictOntologyViolation, an as WikiTask, ao as WikiTransactionError, ap as validateManifest } from './testing-CH_Kk34-.js';
 import { OkfFile } from '@equationalapplications/core-okf';
 import 'minisearch';
 

package/dist/index.js

@@ -4410,6 +4410,23 @@ var DiagnosticBuffer = class {
     this.items = [];
     for (const input of pending) emitDiagnostic(options, input);
   }
+  /**
+   * Emit only the buffered diagnostics whose code is in `codes` and discard
+   * the rest. The buffer is empty afterwards either way.
+   *
+   * For an operation that did real work before its write was rolled back:
+   * some buffered diagnostics describe that work and are still true, while
+   * others describe rows that no longer exist (anything carrying a `factId`
+   * minted inside the aborted transaction). `flush` would emit both and
+   * `discard` would drop both; this emits the first set only.
+   */
+  flushOnly(options, codes) {
+    const pending = this.items;
+    this.items = [];
+    for (const input of pending) {
+      if (codes.includes(input.code)) emitDiagnostic(options, input);
+    }
+  }
   discard() {
     this.items = [];
   }
@@ -4715,6 +4732,16 @@ var DEFAULT_MAX_CHUNK_LENGTH = 12e3;
 var DEFAULT_CHUNK_OVERLAP = 400;
 
 // src/services/IngestionService.ts
+var LLM_PASS_DIAGNOSTIC_CODES = ["ingest_chunk_failed", "fact_rejected", "fact_deduplicated"];
+function dedupDiagnostic(entityId, sourceRef, locator) {
+  return {
+    entityId,
+    operation: "ingest",
+    trigger: "call",
+    code: "fact_deduplicated",
+    detail: locator ? { sourceRef, chunkIndex: locator.chunkIndex, itemIndex: locator.itemIndex, reason: "exact_title" } : { sourceRef, reason: "exact_title" }
+  };
+}
 function zeroChunkResult(duplicateOf) {
   const result = {
     truncated: false,
@@ -4851,11 +4878,7 @@ var IngestionService = class {
       const diagBuffer = new DiagnosticBuffer();
       const diagBase = { entityId, operation: "ingest", trigger: "call" };
       const pushDeduplicated = (chunkIndex, itemIndex) => {
-        diagBuffer.push({
-          ...diagBase,
-          code: "fact_deduplicated",
-          detail: { sourceRef, chunkIndex, itemIndex, reason: "exact_title" }
-        });
+        diagBuffer.push(dedupDiagnostic(entityId, sourceRef, { chunkIndex, itemIndex }));
       };
       for (const [chunkIndex, slot] of chunkResults.entries()) {
         if (slot.status === "failed") {
@@ -4882,7 +4905,7 @@ var IngestionService = class {
         });
       }
       const orderedChunkFacts = [];
-      const groundingLedger = /* @__PURE__ */ new Map();
+      const factLedger = /* @__PURE__ */ new Map();
       for (const [chunkIndex, slot] of chunkResults.entries()) {
         if (slot.status === "failed") continue;
         const keptFacts = [];
@@ -4890,7 +4913,11 @@ var IngestionService = class {
           const winner = winners.get(normalizeTitleKey(fact.title));
           if (winner?.chunkIndex !== chunkIndex || winner.k !== k) return;
           keptFacts.push(fact);
-          if (ingestGrounding) groundingLedger.set(fact, { verdict: slot.verdicts[k], chunkIndex, itemIndex: winner.itemIndex });
+          factLedger.set(fact, {
+            verdict: ingestGrounding ? slot.verdicts[k] : null,
+            chunkIndex,
+            itemIndex: winner.itemIndex
+          });
         });
         orderedChunkFacts.push({ facts: keptFacts, ontology_updates: slot.ontology_updates });
       }
@@ -4920,7 +4947,7 @@ var IngestionService = class {
       try {
         if (failedChunks === 0) {
           const fullResult = await this.db.withTransactionAsync(async (tx) => {
-            return await this.runFullUpsertGraph(entityId, sourceRef, sourceHash, orderedChunkFacts, tx, diagBuffer, groundingLedger);
+            return await this.runFullUpsertGraph(entityId, sourceRef, sourceHash, orderedChunkFacts, tx, diagBuffer, factLedger);
           });
           deletedSourceFactIds.push(...fullResult.deletedSourceFactIds);
           insertedFacts.push(...fullResult.insertedFacts);
@@ -4928,13 +4955,14 @@ var IngestionService = class {
           const partialResult = await this.db.withTransactionAsync(async (tx) => {
             const flat = [];
             for (const slot of orderedChunkFacts) flat.push(...slot.facts);
-            return await this.appendPartialFacts(entityId, sourceRef, flat, tx, diagBuffer, groundingLedger);
+            return await this.appendPartialFacts(entityId, sourceRef, flat, tx, diagBuffer, factLedger);
           });
           insertedFacts.push(...partialResult.insertedDescriptors);
         }
       } catch (err) {
         const sqliteCode = err instanceof WikiTransactionError ? err.sqliteErrorCode : extractSqliteCode(err);
         if (sqliteCode !== "SQLITE_CONSTRAINT_UNIQUE") throw err;
+        diagBuffer.flushOnly(this.options, LLM_PASS_DIAGNOSTIC_CODES);
         const canonical = await this.sourceRefIndexRepo.findActiveByEntityAndHash(entityId, sourceHash);
         if (canonical === null) throw err;
         if (onDuplicateHash === "throw" || onDuplicateHash === "ingest") {
@@ -5162,7 +5190,7 @@ var IngestionService = class {
    * On the partial path the caller never invokes this method; the empty
    * arrays stay empty.
    */
-  async runFullUpsertGraph(entityId, sourceRef, sourceHash, orderedChunkFacts, tx, diagBuffer, groundingLedger = /* @__PURE__ */ new Map()) {
+  async runFullUpsertGraph(entityId, sourceRef, sourceHash, orderedChunkFacts, tx, diagBuffer, factLedger = /* @__PURE__ */ new Map()) {
     const deletedSourceFactIds = [];
     const insertedFacts = [];
     const nodeTrust = /* @__PURE__ */ new Map();
@@ -5209,9 +5237,9 @@ var IngestionService = class {
           tags: fact.tags,
           confidence: fact.confidence
         });
-        const grounding = groundingLedger.get(fact);
-        if (grounding) {
-          const outcome = groundingOutcome(grounding.verdict, now);
+        const locator = factLedger.get(fact);
+        if (locator?.verdict) {
+          const outcome = groundingOutcome(locator.verdict, now);
           nodeTrust.set(id, outcome.trust);
           if (outcome.diagnostic) {
             diagBuffer.push({
@@ -5219,7 +5247,7 @@ var IngestionService = class {
               operation: "ingest",
               trigger: "call",
               code: outcome.diagnostic.code,
-              detail: { factId: id, sourceRef, chunkIndex: grounding.chunkIndex, itemIndex: grounding.itemIndex, reason: outcome.diagnostic.reason }
+              detail: { factId: id, sourceRef, chunkIndex: locator.chunkIndex, itemIndex: locator.itemIndex, reason: outcome.diagnostic.reason }
             });
           }
         }
@@ -5278,7 +5306,7 @@ var IngestionService = class {
    * Runs INSIDE the caller's `tx`. Does not open a nested transaction.
    * Returns `{ inserted, skippedDuplicate }` for observability.
    */
-  async appendPartialFacts(entityId, sourceRef, dedupedFacts, tx, diagBuffer, groundingLedger = /* @__PURE__ */ new Map()) {
+  async appendPartialFacts(entityId, sourceRef, dedupedFacts, tx, diagBuffer, factLedger = /* @__PURE__ */ new Map()) {
     const liveIds = await this.entryRepo.findIdsBySource(entityId, sourceRef, null, tx, false);
     const liveFacts = liveIds.length === 0 ? [] : await this.entryRepo.findByIds(liveIds, void 0, tx);
     const liveTitles = new Set(liveFacts.map((f) => normalizeTitleKey(f.title)));
@@ -5288,28 +5316,22 @@ var IngestionService = class {
     const insertedDescriptors = [];
     for (const fact of dedupedFacts) {
       const normalizedTitle = normalizeTitleKey(fact.title);
+      const locator = factLedger.get(fact);
       if (liveTitles.has(normalizedTitle)) {
-        diagBuffer.push({
-          entityId,
-          operation: "ingest",
-          trigger: "call",
-          code: "fact_deduplicated",
-          detail: { sourceRef, reason: "exact_title" }
-        });
+        diagBuffer.push(dedupDiagnostic(entityId, sourceRef, locator));
         skippedDuplicate++;
         continue;
       }
       liveTitles.add(normalizedTitle);
       const id = generateId("fact_");
-      const grounding = groundingLedger.get(fact);
-      const outcome = grounding ? groundingOutcome(grounding.verdict, now) : null;
-      if (grounding && outcome?.diagnostic) {
+      const outcome = locator?.verdict ? groundingOutcome(locator.verdict, now) : null;
+      if (locator?.verdict && outcome?.diagnostic) {
         diagBuffer.push({
           entityId,
           operation: "ingest",
           trigger: "call",
           code: outcome.diagnostic.code,
-          detail: { factId: id, sourceRef, chunkIndex: grounding.chunkIndex, itemIndex: grounding.itemIndex, reason: outcome.diagnostic.reason }
+          detail: { factId: id, sourceRef, chunkIndex: locator.chunkIndex, itemIndex: locator.itemIndex, reason: outcome.diagnostic.reason }
         });
       }
       const wikiFact = {