feat: support openai embedding and add llm base_urls in client

wavefront/client/src/components/DashboardLayout.tsx

@@ -14,7 +14,7 @@ interface IUser {
 const DashboardLayout = ({ user, apps = [] }: { user: IUser; apps: App[] }) => {
   const currentPath = useLocation();
   const navigate = useNavigate();
-  const timeoutRef = useRef<number | null>(null);
+  const timeoutRef = useRef<ReturnType<typeof setTimeout> | null>(null);
   /*
   The Broadcast Channel API enables communication between different browser windows, tabs, iframes, and web workers. 
   creating a channel with name 'timeout' 
@@ -27,7 +27,7 @@ const DashboardLayout = ({ user, apps = [] }: { user: IUser; apps: App[] }) => {
     if (timeoutRef.current) {
       clearTimeout(timeoutRef.current);
     }
-    // @ts-ignore
+
     timeoutRef.current = setTimeout(
       () => {
         navigate('/logout');

wavefront/client/src/config/llm-providers.ts

@@ -20,6 +20,7 @@ export interface ParameterConfig {
 
 export interface ProviderConfig {
   name: string;
+  baseUrl?: string;
   badge: {
     bg: string;
     text: string;
@@ -36,6 +37,7 @@ export type InferenceEngineType = 'gemini' | 'openai' | 'ollama' | 'vllm' | 'ant
 export const LLM_PROVIDERS_CONFIG: Record<InferenceEngineType, ProviderConfig> = {
   openai: {
     name: 'OpenAI GPT',
+    baseUrl: 'https://api.openai.com/v1',
     badge: {
       bg: 'bg-green-100',
       text: 'text-green-800',
@@ -108,6 +110,7 @@ export const LLM_PROVIDERS_CONFIG: Record<InferenceEngineType, ProviderConfig> =
   },
   anthropic: {
     name: 'Anthropic Claude',
+    baseUrl: 'https://api.anthropic.com',
     badge: {
       bg: 'bg-orange-100',
       text: 'text-orange-800',
@@ -153,6 +156,7 @@ export const LLM_PROVIDERS_CONFIG: Record<InferenceEngineType, ProviderConfig> =
   },
   gemini: {
     name: 'Google Gemini',
+    baseUrl: 'https://generativelanguage.googleapis.com',
     badge: {
       bg: 'bg-blue-100',
       text: 'text-blue-800',
@@ -261,6 +265,7 @@ export const LLM_PROVIDERS_CONFIG: Record<InferenceEngineType, ProviderConfig> =
   },
   groq: {
     name: 'Groq',
+    baseUrl: 'https://api.groq.com/openai/v1',
     badge: {
       bg: 'bg-purple-100',
       text: 'text-purple-800',
@@ -509,3 +514,11 @@ export function cleanParameters(params: Record<string, any>): Record<string, any
   });
   return cleaned;
 }
+
+/**
+ * Get default base URL for a provider
+ */
+export function getDefaultBaseUrl(provider: InferenceEngineType): string {
+  const config = getProviderConfig(provider);
+  return config?.baseUrl || '';
+}

wavefront/client/src/pages/apps/[appId]/agents/[id].tsx

@@ -403,7 +403,7 @@ const AgentDetail: React.FC = () => {
       // Prepare inputs based on what's provided
       // eslint-disable-next-line @typescript-eslint/no-explicit-any
       let inputs: string | any[];
-      let finalTextInput = inferenceInput.trim();
+      const finalTextInput = inferenceInput.trim();
       // Handle different input combinations
       // eslint-disable-next-line @typescript-eslint/no-explicit-any
       const conversationInputs: any[] = [];

wavefront/client/src/pages/apps/[appId]/datasources/[datasourceId].tsx

@@ -117,8 +117,7 @@ const DatasourceDetail: React.FC = () => {
     if (!datasourceId) return;
     if (yamlContent) {
       const yamlResponse = validateDynamicQueryYaml(yamlContent);
-      if (yamlResponse.valid) {
-      } else {
+      if (!yamlResponse.valid) {
         notifyError(yamlResponse.error);
         return;
       }
@@ -190,6 +189,7 @@ const DatasourceDetail: React.FC = () => {
       notifySuccess('Datasource deleted successfully');
       navigate(`/apps/${appId}/datasources`);
     } catch (error) {
+      console.error('Error deleting datasource');
     } finally {
       setDeleting(false);
       setShowDeleteConfirm(false);

wavefront/client/src/pages/apps/[appId]/llm-inference/CreateLLMInferenceDialog.tsx

@@ -22,7 +22,13 @@ import {
 import { Input } from '@app/components/ui/input';
 import { Select, SelectContent, SelectItem, SelectTrigger, SelectValue } from '@app/components/ui/select';
 import { Slider } from '@app/components/ui/slider';
-import { cleanParameters, getProviderConfig, initializeParameters, ParameterConfig } from '@app/config/llm-providers';
+import {
+  cleanParameters,
+  getDefaultBaseUrl,
+  getProviderConfig,
+  initializeParameters,
+  ParameterConfig,
+} from '@app/config/llm-providers';
 import { useDashboardStore, useNotifyStore } from '@app/store';
 import { InferenceEngineType } from '@app/types/llm-inference-config';
 import { zodResolver } from '@hookform/resolvers/zod';
@@ -55,10 +61,10 @@ const BASE_URL_PLACEHOLDERS: Record<InferenceEngineType, string> = {
   ollama: 'http://localhost:11434',
   vllm: 'http://localhost:8000',
   azure_openai: 'https://your-resource.openai.azure.com',
-  openai: 'Base URL',
-  anthropic: 'Base URL',
-  gemini: 'Base URL',
-  groq: 'Base URL',
+  openai: 'https://api.openai.com/v1',
+  anthropic: 'https://api.anthropic.com',
+  gemini: 'https://generativelanguage.googleapis.com',
+  groq: 'https://api.groq.com/openai/v1',
 };
 
 const createLLMInferenceSchema = z.object({
@@ -98,7 +104,7 @@ const CreateLLMInferenceDialog: React.FC<CreateLLMInferenceDialogProps> = ({
       llmModel: '',
       type: 'openai',
       apiKey: '',
-      baseUrl: '',
+      baseUrl: getDefaultBaseUrl('openai'),
     },
   });
 
@@ -110,21 +116,26 @@ const CreateLLMInferenceDialog: React.FC<CreateLLMInferenceDialogProps> = ({
       setType(newType);
       const defaultParams = initializeParameters(newType);
       setParameters(defaultParams);
+
+      // Set default base URL for the provider
+      const defaultBaseUrl = getDefaultBaseUrl(newType);
+      form.setValue('baseUrl', defaultBaseUrl);
     }
-  }, [watchedType, isOpen]);
+  }, [watchedType, isOpen, form]);
 
   // Reset form when dialog closes
   useEffect(() => {
     if (!isOpen) {
+      const defaultType = 'openai';
       form.reset({
         displayName: '',
         llmModel: '',
-        type: 'openai',
+        type: defaultType,
         apiKey: '',
-        baseUrl: '',
+        baseUrl: getDefaultBaseUrl(defaultType),
       });
-      setType('openai');
-      setParameters(initializeParameters('openai'));
+      setType(defaultType);
+      setParameters(initializeParameters(defaultType));
     }
   }, [isOpen, form]);
 
@@ -133,7 +144,7 @@ const CreateLLMInferenceDialog: React.FC<CreateLLMInferenceDialogProps> = ({
   };
 
   const supportsBaseUrl = (engineType: InferenceEngineType) => {
-    return ['ollama', 'vllm', 'azure_openai'].includes(engineType);
+    return ['ollama', 'vllm', 'azure_openai', 'openai', 'anthropic', 'gemini', 'groq'].includes(engineType);
   };
 
   const setParameter = (key: string, value: any) => {
@@ -330,6 +341,10 @@ const CreateLLMInferenceDialog: React.FC<CreateLLMInferenceDialogProps> = ({
                         const newType = value as InferenceEngineType;
                         setType(newType);
                         setParameters(initializeParameters(newType));
+
+                        // Set default base URL for the provider
+                        const defaultBaseUrl = getDefaultBaseUrl(newType);
+                        form.setValue('baseUrl', defaultBaseUrl);
                       }}
                       value={field.value}
                     >

wavefront/client/src/pages/apps/[appId]/llm-inference/[configId].tsx

@@ -14,7 +14,13 @@ import { Form, FormControl, FormField, FormItem, FormLabel, FormMessage } from '
 import { Input } from '@app/components/ui/input';
 import { Select, SelectContent, SelectItem, SelectTrigger, SelectValue } from '@app/components/ui/select';
 import { Slider } from '@app/components/ui/slider';
-import { cleanParameters, getProviderConfig, mergeParameters, ParameterConfig } from '@app/config/llm-providers';
+import {
+  cleanParameters,
+  getDefaultBaseUrl,
+  getProviderConfig,
+  mergeParameters,
+  ParameterConfig,
+} from '@app/config/llm-providers';
 import { useGetLLMConfig } from '@app/hooks';
 import { getLLMConfigKey, getLLMConfigsKey } from '@app/hooks/data/query-keys';
 import { useNotifyStore } from '@app/store';
@@ -92,13 +98,19 @@ const LLMInferenceConfigDetail: React.FC = () => {
     }
   }, [config, form]);
 
-  // Update parameters when provider type changes in edit mode
+  // Update parameters and base URL when provider type changes in edit mode
   const watchedType = form.watch('type');
   useEffect(() => {
     if (editing && config) {
       const mergedParams = mergeParameters(watchedType, watchedType === config.type ? config.parameters : null);
       setParameters(mergedParams);
       form.setValue('parameters', mergedParams);
+
+      // Update base URL to the new provider's default when type changes
+      if (watchedType !== config.type) {
+        const defaultBaseUrl = getDefaultBaseUrl(watchedType);
+        form.setValue('base_url', defaultBaseUrl);
+      }
     }
   }, [watchedType, editing, config, form]);
 
@@ -179,7 +191,7 @@ const LLMInferenceConfigDetail: React.FC = () => {
   };
 
   const supportsBaseUrl = (engineType: InferenceEngineType) => {
-    return ['ollama', 'vllm', 'azure_openai'].includes(engineType);
+    return ['ollama', 'vllm', 'azure_openai', 'openai', 'anthropic', 'gemini', 'groq'].includes(engineType);
   };
 
   return (

wavefront/server/background_jobs/rag_ingestion/rag_ingestion/embeddings/embed.py

@@ -2,12 +2,13 @@
 import requests
 from rag_ingestion.env import EMBEDDING_SERVICE_URL
 from flo_utils.utils.log import logger
+from rag_ingestion.env import OPENAI_API_KEY, EMBEDDING_MODEL
 
 
 class EmbeddingFunc:
     def __init__(self):
         self.max_batch_size = 32
-        self.bgm_url = f'{EMBEDDING_SERVICE_URL}/v1/embeddings'
+        self.bgm_url = f'{EMBEDDING_SERVICE_URL}'
         logger.info(f'The embedding url is {EMBEDDING_SERVICE_URL}')
 
     def generate_document_embeddings(self, chunks):
@@ -33,12 +34,20 @@ def generate_chunk_embeddings(self, chunks):
         return embeddings
 
     def bgm_embedding(self, texts):
+        headers = {
+            'Authorization': f'Bearer {OPENAI_API_KEY}',
+        }
+
         response = requests.post(
             self.bgm_url,
+            headers=headers if EMBEDDING_MODEL == 'text-embedding-3-small' else None,
             json={
-                'model': 'BAAI/bge-m3',
+                'model': EMBEDDING_MODEL,
                 'input': texts,
                 'encoding_format': 'float',
             },
+            timeout=60,
         )
-        return response.json()['data'][0]['embedding']
+        response.raise_for_status()
+        res = response.json()
+        return res['data'][0]['embedding']

wavefront/server/background_jobs/rag_ingestion/rag_ingestion/env.py

@@ -5,7 +5,9 @@
 
 CLOUD_PROVIDER = os.getenv('CLOUD_PROVIDER', 'gcp')
 RETRY_COUNT = os.getenv('RETRY_COUNT', 3)
+OPENAI_API_KEY = os.getenv('OPENAI_API_KEY')
 EMBEDDING_SERVICE_URL = os.getenv('EMBEDDING_SERVICE_URL')
 FLOWARE_SERVICE_URL = os.getenv('FLOWARE_SERVICE_URL')
 APP_ENV = os.getenv('APP_ENV', 'dev')
 PASSTHROUGH_SECRET = os.getenv('PASSTHROUGH_SECRET')
+EMBEDDING_MODEL = os.getenv('EMBEDDING_MODEL', 'text-embedding-3-small')

wavefront/server/modules/knowledge_base_module/knowledge_base_module/controllers/knowledge_base_controller.py

@@ -82,16 +82,17 @@ async def create_knowledge_base(
         )
 
 
-@knowledge_base_router.get(
-    '/v1/knowledge-bases/{kb_id}', response_model=KnowledgeBaseResponse
-)
+@knowledge_base_router.get('/v1/knowledge-bases/{kb_id}')
 @inject
 async def get_knowledge_bases_id(
     kb_id: uuid.UUID,
+    response_formatter: ResponseFormatter = Depends(
+        Provide[CommonContainer.response_formatter]
+    ),
     knowledge_base_repository: SQLAlchemyRepository[KnowledgeBase] = Depends(
         Provide[KnowledgeBaseContainer.knowledge_base_repository]
     ),
-) -> KnowledgeBaseResponse:
+) -> JSONResponse:
     """Get knowledge base by ID."""
     fetch_knowledge_base_id = await knowledge_base_repository.find_one(id=kb_id)
     if not fetch_knowledge_base_id:
@@ -100,13 +101,11 @@ async def get_knowledge_bases_id(
             detail="Knowledge Base with the mentioned id doesn't exist",
         )
 
-    return KnowledgeBaseResponse(
-        id=fetch_knowledge_base_id.id,
-        name=fetch_knowledge_base_id.name,
-        description=fetch_knowledge_base_id.description,
-        type=fetch_knowledge_base_id.type,
-        created_at=fetch_knowledge_base_id.created_at,
-        updated_at=fetch_knowledge_base_id.updated_at,
+    return JSONResponse(
+        status_code=status.HTTP_200_OK,
+        content=response_formatter.buildSuccessResponse(
+            data=fetch_knowledge_base_id.to_dict()
+        ),
     )
 
 

wavefront/server/modules/knowledge_base_module/knowledge_base_module/embeddings/embed.py

@@ -1,3 +1,4 @@
+import os
 import requests
 from typing import List
 from dataclasses import dataclass
@@ -13,7 +14,7 @@ class KnowledgeBaseEmbeddingObject:
 class EmbeddingFunc:
     def __init__(self, embedding_url):
         self.max_batch_size = 32
-        self.bgm_url = f'{embedding_url}/v1/embeddings'
+        self.bgm_url = f'{embedding_url}'
 
     def generate_document_embeddings(self, chunks):
         contents = [v['content'] for v in chunks.values()]
@@ -42,12 +43,16 @@ def generate_chunk_embeddings(self, chunks):
         return embeddings
 
     def bgm_embedding(self, texts):
+        openai_api_key = os.getenv('OPENAI_API_KEY') or ''
         response = requests.post(
             self.bgm_url,
+            headers={'Authorization': 'Bearer ' + openai_api_key},
             json={
-                'model': 'BAAI/bge-m3',
+                # 'model': 'BAAI/bge-m3',
+                'model': 'text-embedding-3-small',
                 'input': texts,
                 'encoding_format': 'float',
             },
         )
-        return response.json()['data'][0]['embedding']
+        res = response.json()
+        return res['data'][0]['embedding']