@semiont/make-meaning 0.5.9 → 0.5.10

package/dist/smelter-main.js

@@ -1,8 +1,7 @@
-import { createTomlConfigLoader, accessToken, baseUrl as baseUrl$1, burstBuffer, errField, resourceId, textExtractionOf, decodeRepresentation, getTargetSelector, getExactText, annotationId, getPrimaryMediaType, getPrimaryRepresentation } from '@semiont/core';
+import { createTomlConfigLoader, accessToken, baseUrl as baseUrl$1, burstBuffer, errField, resourceId, textExtractionOf, decodeRepresentation, busRequest, getResourceEntityTypes, getTargetSelector, getExactText, annotationId, getPrimaryMediaType, getPrimaryRepresentation } from '@semiont/core';
 import { calculateChecksum } from '@semiont/content';
 import { createEmbeddingProvider, createVectorStore, chunkText } from '@semiont/vectors';
 import { registerVectorIndexSizeProvider, withActorSpan } from '@semiont/observability';
-import { busRequest } from '@semiont/sdk';
 import { HttpTransport, HttpContentTransport } from '@semiont/http-transport';
 import { createServer } from 'http';
 import { existsSync, readFileSync } from 'fs';
@@ -9998,6 +9997,23 @@ var Smelter = class _Smelter {
       return null;
     }
   }
+  /**
+   * Read a resource's current entity types from the materialized view — the
+   * authoritative source, updated before the EventBus fires to consumers — so
+   * its vectors carry the discriminator `searchResources` filters on (e.g.
+   * exclude `['Question']`). One read on every embed path, mirroring how the
+   * smelter already reads current content and annotations; a failed read
+   * propagates to the pipeline's per-resource error handler (reconcile heals),
+   * rather than silently stamping `[]` and letting the resource leak into recall.
+   */
+  async resolveEntityTypes(resourceId) {
+    const { resource } = await busRequest(
+      this.bus,
+      "browse:resource-requested",
+      { resourceId }
+    );
+    return getResourceEntityTypes(resource);
+  }
   async embedResource(event, logMessage) {
     const rid = event.resourceId;
     if (!rid) return;
@@ -10005,13 +10021,14 @@ var Smelter = class _Smelter {
     if (!fetched) return;
     const chunks = chunkText(fetched.text, this.chunkingConfig);
     if (chunks.length === 0) return;
+    const entityTypes = await this.resolveEntityTypes(rid);
     const embeddings = await this.embeddingProvider.embedBatch(chunks);
     const embeddingChunks = chunks.map((t, i) => ({
       chunkIndex: i,
       text: t,
       embedding: embeddings[i]
     }));
-    await this.vectorStore.upsertResourceVectors(resourceId(rid), embeddingChunks, fetched.checksum);
+    await this.vectorStore.upsertResourceVectors(resourceId(rid), embeddingChunks, fetched.checksum, entityTypes);
     this.logger.info(logMessage, { resourceId: rid, chunks: chunks.length });
   }
   async handleResourceArchived(event) {
@@ -10062,19 +10079,20 @@ var Smelter = class _Smelter {
       if (!fetched) continue;
       const chunks = chunkText(fetched.text, this.chunkingConfig);
       if (chunks.length === 0) continue;
-      resourceData.push({ rid: resourceId(rid), chunks, checksum: fetched.checksum });
+      const entityTypes = await this.resolveEntityTypes(rid);
+      resourceData.push({ rid: resourceId(rid), chunks, checksum: fetched.checksum, entityTypes });
       allChunks.push(...chunks);
     }
     if (allChunks.length === 0) return events.length;
     const allEmbeddings = await this.embeddingProvider.embedBatch(allChunks);
     let offset = 0;
-    for (const { rid, chunks, checksum } of resourceData) {
+    for (const { rid, chunks, checksum, entityTypes } of resourceData) {
       const embeddingChunks = chunks.map((t, i) => ({
         chunkIndex: i,
         text: t,
         embedding: allEmbeddings[offset + i]
       }));
-      await this.vectorStore.upsertResourceVectors(rid, embeddingChunks, checksum);
+      await this.vectorStore.upsertResourceVectors(rid, embeddingChunks, checksum, entityTypes);
       this.logger.info("Batch-indexed resource", { resourceId: String(rid), chunks: chunks.length });
       offset += chunks.length;
     }
@@ -10181,9 +10199,7 @@ var Smelter = class _Smelter {
         const { annotations } = await busRequest(
           this.bus,
           "browse:annotations-requested",
-          { resourceId: rid },
-          "browse:annotations-result",
-          "browse:annotations-failed"
+          { resourceId: rid }
         );
         for (const annotation of annotations) {
           const exactText = getExactText(getTargetSelector(annotation.target));
@@ -10241,9 +10257,7 @@ var Smelter = class _Smelter {
       const page = await busRequest(
         this.bus,
         "browse:resources-requested",
-        { archived: false, offset: all.length, limit: _Smelter.RECONCILE_PAGE_SIZE },
-        "browse:resources-result",
-        "browse:resources-failed"
+        { archived: false, offset: all.length, limit: _Smelter.RECONCILE_PAGE_SIZE }
       );
       all.push(...page.resources);
       if (page.resources.length === 0 || all.length >= page.total) return all;