@equationalapplications/core-llm-wiki 7.7.2 → 7.7.4

package/dist/index.mjs

@@ -1,5 +1,5 @@
-import { __privateAdd, MetadataRepository, EmbeddingService, SearchService, JobManager, PromptService, resolveGrounding, IngestionService, MaintenanceService, ImportExportService, RetrievalService, WriteService, __privateGet, __privateSet, normalizeSourceRef, normalizeSourceHash, entitySummaryMetaKey, WikiSourceRefHashCollision, DiagnosticBuffer, validateManifest, WikiDraftNotFound, generateId, BaseRepository, WikiGraphNodeOwnershipConflict, emptyManifest, resolveNodeType, WikiStrictOntologyViolation, validateInlineEdges, resolveEdgeDefinitions, typeSatisfies, normalizeTitleKey, edgeTripleAllowed, WikiTransactionError, extractSqliteCode } from './chunk-NWODQRSM.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, WikiDraftNotFound, WikiDuplicateHashError, WikiGraphNodeOwnershipConflict, WikiIngestEmptyError, WikiInvalidReadOptions, WikiParseError, WikiSourceRefHashCollision, WikiStrictOntologyViolation, WikiTransactionError, chunkText, configureRandomSource, parseEmbedding, safeSlice, validateManifest } from './chunk-NWODQRSM.mjs';
+import { __privateAdd, MetadataRepository, EmbeddingService, SearchService, JobManager, PromptService, resolveGrounding, IngestionService, MaintenanceService, ImportExportService, RetrievalService, WriteService, __privateGet, __privateSet, normalizeSourceRef, normalizeSourceHash, entitySummaryMetaKey, WikiSourceRefHashCollision, DiagnosticBuffer, validateManifest, WikiDraftNotFound, generateId, BaseRepository, WikiGraphNodeOwnershipConflict, emptyManifest, resolveNodeType, WikiStrictOntologyViolation, validateInlineEdges, resolveEdgeDefinitions, typeSatisfies, normalizeTitleKey, edgeTripleAllowed, WikiTransactionError, extractSqliteCode } from './chunk-JV2OQ5CM.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, WikiDraftNotFound, WikiDuplicateHashError, WikiGraphNodeOwnershipConflict, WikiIngestEmptyError, WikiInvalidReadOptions, WikiParseError, WikiSourceRefHashCollision, WikiStrictOntologyViolation, WikiTransactionError, chunkText, configureRandomSource, parseEmbedding, safeSlice, validateManifest } from './chunk-JV2OQ5CM.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

package/dist/{testing-A4Sofxyo.d.mts → testing-CH_Kk34-.d.mts}

@@ -1889,6 +1889,17 @@ declare class DiagnosticBuffer {
     push(input: WikiDiagnosticInput): void;
     get size(): number;
     flush(options: DiagnosticTarget): void;
+    /**
+     * 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: DiagnosticTarget, codes: readonly WikiDiagnosticCode[]): void;
     discard(): void;
 }
 

package/dist/{testing-A4Sofxyo.d.ts → testing-CH_Kk34-.d.ts}

@@ -1889,6 +1889,17 @@ declare class DiagnosticBuffer {
     push(input: WikiDiagnosticInput): void;
     get size(): number;
     flush(options: DiagnosticTarget): void;
+    /**
+     * 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: DiagnosticTarget, codes: readonly WikiDiagnosticCode[]): void;
     discard(): void;
 }
 

package/dist/testing.d.mts

@@ -1,3 +1,3 @@
-export { aq as EmbeddingService, ar as ImportExportService, as as IngestionService, at as JobManager, at as JobManagerType, au as MaintenanceService, av as RetrievalService, aw as SearchService, aw as SearchServiceType, ai as WikiMemoryTestAccess, ax as WriteService } from './testing-A4Sofxyo.mjs';
+export { aq as EmbeddingService, ar as ImportExportService, as as IngestionService, at as JobManager, at as JobManagerType, au as MaintenanceService, av as RetrievalService, aw as SearchService, aw as SearchServiceType, ai as WikiMemoryTestAccess, ax as WriteService } from './testing-CH_Kk34-.mjs';
 import '@equationalapplications/core-okf';
 import 'minisearch';

package/dist/testing.d.ts

@@ -1,3 +1,3 @@
-export { aq as EmbeddingService, ar as ImportExportService, as as IngestionService, at as JobManager, at as JobManagerType, au as MaintenanceService, av as RetrievalService, aw as SearchService, aw as SearchServiceType, ai as WikiMemoryTestAccess, ax as WriteService } from './testing-A4Sofxyo.js';
+export { aq as EmbeddingService, ar as ImportExportService, as as IngestionService, at as JobManager, at as JobManagerType, au as MaintenanceService, av as RetrievalService, aw as SearchService, aw as SearchServiceType, ai as WikiMemoryTestAccess, ax as WriteService } from './testing-CH_Kk34-.js';
 import '@equationalapplications/core-okf';
 import 'minisearch';

package/dist/testing.js

@@ -755,6 +755,23 @@ 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 = [];
   }
@@ -1839,6 +1856,16 @@ 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,
@@ -1975,11 +2002,7 @@ 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") {
@@ -2006,7 +2029,7 @@ 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 = [];
@@ -2014,7 +2037,11 @@ 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 });
       }
@@ -2044,7 +2071,7 @@ 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);
@@ -2052,13 +2079,14 @@ 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") {
@@ -2286,7 +2314,7 @@ 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();
@@ -2333,9 +2361,9 @@ 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({
@@ -2343,7 +2371,7 @@ 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 }
             });
           }
         }
@@ -2402,7 +2430,7 @@ 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)));
@@ -2412,28 +2440,22 @@ 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 = {