WIP - enable embedings

backend/src/knowledge/services/semantic-orchestrator.service.ts

@@ -1,6 +1,8 @@
 import { Injectable, Logger } from '@nestjs/common';
 import { TenantDatabaseService } from '../../tenant/tenant-database.service';
 import { MeilisearchService } from '../../search/meilisearch.service';
+import { getCentralPrisma } from '../../prisma/central-prisma.service';
+import { OpenAIConfig } from '../../voice/interfaces/integration-config.interface';
 import {
   DefaultSemanticProjectionAdapter,
   SemanticProjectionAdapter,
@@ -12,6 +14,9 @@ import { SemanticLinkService } from './semantic-link.service';
 export class SemanticOrchestratorService {
   private readonly logger = new Logger(SemanticOrchestratorService.name);
   private readonly adapters: SemanticProjectionAdapter[] = [new DefaultSemanticProjectionAdapter()];
+  private readonly defaultEmbeddingModel =
+    process.env.OPENAI_EMBEDDING_MODEL || 'text-embedding-3-small';
+  private readonly semanticEmbedderName = 'semantic-openai';
 
   constructor(
     private readonly tenantDbService: TenantDatabaseService,
@@ -61,8 +66,9 @@ export class SemanticOrchestratorService {
     const chunks = this.chunkerService.chunkText(projection.narrative, comments);
     await this.replaceChunks(knex, documentId, chunks);
 
-    await this.indexChunks(resolvedTenantId, projection, chunks);
-    await this.generateSuggestions(resolvedTenantId, projection, chunks, userId, trigger);
+    const openAiConfig = await this.getOpenAiConfig(resolvedTenantId);
+    await this.indexChunks(resolvedTenantId, projection, chunks, openAiConfig);
+    await this.generateSuggestions(resolvedTenantId, projection, chunks, openAiConfig, userId, trigger);
 
     return { documentId, chunkCount: chunks.length };
   }
@@ -139,12 +145,25 @@ export class SemanticOrchestratorService {
     );
   }
 
-  private async indexChunks(tenantId: string, projection: any, chunks: any[]) {
+  private async indexChunks(
+    tenantId: string,
+    projection: any,
+    chunks: any[],
+    openAiConfig: OpenAIConfig | null,
+  ) {
     if (!this.meilisearchService.isEnabled()) {
       return;
     }
 
     const indexName = this.meilisearchService.buildSemanticChunkIndexName(tenantId);
+    if (openAiConfig?.apiKey) {
+      await this.meilisearchService.ensureOpenAiEmbedder(indexName, {
+        embedderName: this.semanticEmbedderName,
+        apiKey: openAiConfig.apiKey,
+        model: openAiConfig.embeddingModel || this.defaultEmbeddingModel,
+        documentTemplate: '{{doc.title}}\n{{doc.text}}',
+      });
+    }
     await this.meilisearchService.upsertDocuments(indexName, chunks.map((chunk) => ({
       id: `${projection.entityType}:${projection.entityId}:${chunk.chunkIndex}`,
       entityType: projection.entityType,
@@ -160,6 +179,7 @@ export class SemanticOrchestratorService {
     tenantId: string,
     projection: any,
     chunks: any[],
+    openAiConfig: OpenAIConfig | null,
     userId?: string,
     trigger: string = 'semantic_refresh',
   ) {
@@ -169,7 +189,12 @@ export class SemanticOrchestratorService {
 
     const indexName = this.meilisearchService.buildSemanticChunkIndexName(tenantId);
     const queryText = chunks.slice(0, 3).map((chunk) => chunk.text).join(' ').slice(0, 1200);
-    const search = await this.meilisearchService.searchIndex(indexName, queryText, 20);
+    const search = await this.meilisearchService.searchIndex(
+      indexName,
+      queryText,
+      20,
+      openAiConfig?.apiKey ? { embedder: this.semanticEmbedderName } : undefined,
+    );
 
     const grouped = new Map<string, any[]>();
     for (const hit of search.hits || []) {
@@ -222,4 +247,38 @@ export class SemanticOrchestratorService {
 
     return `${objectDefinition.apiName.toLowerCase()}s`;
   }
+
+  private async getOpenAiConfig(tenantId: string): Promise<OpenAIConfig | null> {
+    const resolvedTenantId = await this.tenantDbService.resolveTenantId(tenantId);
+    const centralPrisma = getCentralPrisma();
+    const tenant = await centralPrisma.tenant.findUnique({
+      where: { id: resolvedTenantId },
+      select: { integrationsConfig: true },
+    });
+
+    let config = tenant?.integrationsConfig
+      ? typeof tenant.integrationsConfig === 'string'
+        ? this.tenantDbService.decryptIntegrationsConfig(tenant.integrationsConfig)
+        : tenant.integrationsConfig
+      : null;
+
+    if (!config?.openai && process.env.OPENAI_API_KEY) {
+      config = {
+        ...(config || {}),
+        openai: {
+          apiKey: process.env.OPENAI_API_KEY,
+          embeddingModel: this.defaultEmbeddingModel,
+        },
+      };
+    }
+
+    if (config?.openai?.apiKey) {
+      return {
+        apiKey: config.openai.apiKey,
+        embeddingModel: config.openai.embeddingModel || this.defaultEmbeddingModel,
+      };
+    }
+
+    return null;
+  }
 }