Merge pull request #4059 from janhq/fix/remote-models-error-handling

chore: remote models error handling

core/src/browser/extensions/engines/helpers/sse.ts

@@ -38,7 +38,7 @@ export function requestInference(
             errorCode = ErrorCode.InvalidApiKey
           }
           const error = {
-            message: data.error?.message ?? 'Error occurred.',
+            message: data.error?.message ?? data.message ?? 'Error occurred.',
             code: errorCode,
           }
           subscriber.error(error)

extensions/inference-cohere-extension/src/index.ts

@@ -113,6 +113,8 @@ export default class JanInferenceCohereExtension extends RemoteOAIEngine {
   }
 
   transformResponse = (data: any) => {
-    return typeof data === 'object' ? data.text : JSON.parse(data).text ?? ''
+    return typeof data === 'object'
+      ? data.text
+      : (JSON.parse(data.replace('data: ', '').trim()).text ?? '')
   }
 }

extensions/inference-openai-extension/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@janhq/inference-openai-extension",
   "productName": "OpenAI Inference Engine",
-  "version": "1.0.3",
+  "version": "1.0.4",
   "description": "This extension enables OpenAI chat completion API calls",
   "main": "dist/index.js",
   "module": "dist/module.js",

extensions/inference-openai-extension/resources/models.json

@@ -99,6 +99,7 @@
     "parameters": {
       "temperature": 1,
       "top_p": 1,
+      "stream": true,
       "max_tokens": 32768,
       "frequency_penalty": 0,
       "presence_penalty": 0
@@ -126,6 +127,7 @@
       "temperature": 1,
       "top_p": 1,
       "max_tokens": 65536,
+      "stream": true,
       "frequency_penalty": 0,
       "presence_penalty": 0
     },

extensions/inference-openai-extension/src/index.ts

@@ -80,7 +80,6 @@ export default class JanInferenceOpenAIExtension extends RemoteOAIEngine {
       return {
         ...params,
         max_completion_tokens: max_tokens,
-        stream: false, // o1 only support stream = false
       }
     }
     // Pass through for non-preview models

web/containers/ErrorMessage/index.test.tsx

@@ -63,9 +63,6 @@ describe('ErrorMessage Component', () => {
 
     render(<ErrorMessage message={message} />)
 
-    expect(
-      screen.getByText('Apologies, something’s amiss!')
-    ).toBeInTheDocument()
     expect(screen.getByText('troubleshooting assistance')).toBeInTheDocument()
   })
 

web/containers/ErrorMessage/index.tsx

@@ -27,8 +27,6 @@ const ErrorMessage = ({ message }: { message: ThreadMessage }) => {
 
   const getErrorTitle = () => {
     switch (message.error_code) {
-      case ErrorCode.Unknown:
-        return 'Apologies, something’s amiss!'
       case ErrorCode.InvalidApiKey:
       case ErrorCode.AuthenticationError:
       case ErrorCode.InvalidRequestError:
@@ -55,17 +53,17 @@ const ErrorMessage = ({ message }: { message: ThreadMessage }) => {
         )
       default:
         return (
-          <>
+          <p>
             {message.content[0]?.text?.value && (
               <AutoLink text={message.content[0].text.value} />
             )}
-          </>
+          </p>
         )
     }
   }
 
   return (
-    <div className="mt-10">
+    <div className="mx-auto mt-10 max-w-[700px]">
       {message.status === MessageStatus.Error && (
         <div
           key={message.id}

web/containers/Providers/EventHandler.tsx

@@ -180,7 +180,11 @@ export default function EventHandler({ children }: { children: ReactNode }) {
           setIsGeneratingResponse(false)
         }
         return
-      } else if (message.status === MessageStatus.Error) {
+      } else if (
+        message.status === MessageStatus.Error &&
+        activeModelRef.current?.engine &&
+        isLocalEngine(activeModelRef.current.engine)
+      ) {
         ;(async () => {
           if (
             !(await extensionManager