refactor: replace SSE streaming with simple JSON chat endpoint

Replace Server-Sent Events streaming implementation with straightforward
JSON request/response pattern to simplify debugging and reduce complexity.

Changes:
- server.js: Convert /api/chat/stream to /api/chat with JSON response
- src/lib/api.ts: Remove SSE parsing logic, add ChatResponse interface
- src/hooks/useChat.ts: Handle JSON response instead of stream chunks

This change makes the request flow easier to debug and troubleshoot
while maintaining the same conversation functionality.

server.js

@@ -3,103 +3,50 @@ const path = require('path');
 const { GoogleAuth } = require('google-auth-library');
 
 const app = express();
-const port = process.env.PORT || 8080;
+app.use(express.json({ limit: '1mb' }));
-const BACKEND_URL = process.env.BACKEND_URL;
 
-// Validate required environment variable at startup
+const BACKEND_URL = process.env.BACKEND_URL;
 if (!BACKEND_URL) {
-  console.error('ERROR: BACKEND_URL environment variable is required');
+  console.error('FATAL: BACKEND_URL env var is required');
   process.exit(1);
 }
 
-// Initialize Google Auth client (auto-detects credentials on Cloud Run)
 const auth = new GoogleAuth();
 
-app.use(express.json());
+app.get('/health', (_req, res) => {
+  res.status(200).json({ status: 'ok' });
+});
 
-// Serve static React build
+app.post('/api/chat', async (req, res) => {
+  try {
+    // Create an ID-token authenticated client for the backend (audience = BACKEND_URL)
+    const idTokenClient = await auth.getIdTokenClient(BACKEND_URL);
+
+    // Forward request to backend
+    const backendResp = await idTokenClient.request({
+      url: `${BACKEND_URL}/chat`,
+      method: 'POST',
+      data: req.body,
+    });
+
+    res.status(backendResp.status).json(backendResp.data);
+  } catch (err) {
+    console.error('Proxy error:', err.message);
+    const status = err.response?.status || 500;
+    const message = err.response?.data || { error: err.message };
+    res.status(status).json(message);
+  }
+});
+
+// Serve static assets
 app.use(express.static(path.join(__dirname, 'dist')));
 
-// Proxy endpoint - generates ID token and forwards to backend
+// SPA fallback
-app.post('/api/chat/stream', async (req, res) => {
+app.get('*', (_req, res) => {
-  console.log('Proxy request received for /api/chat/stream');
-
-  try {
-    // Get ID token client with backend URL as audience
-    const client = await auth.getIdTokenClient(BACKEND_URL);
-    const headers = await client.getRequestHeaders();
-
-    console.log('Generated ID token, forwarding to backend...');
-
-    // Forward request to backend with auth
-    const backendResponse = await fetch(`${BACKEND_URL}/chat/stream`, {
-      method: 'POST',
-      headers: {
-        'Content-Type': 'application/json',
-        ...headers,
-      },
-      body: JSON.stringify(req.body),
-    });
-
-    // Check if backend returned an error
-    if (!backendResponse.ok) {
-      const errorText = await backendResponse.text();
-      console.error(`Backend error: ${backendResponse.status} - ${errorText}`);
-      res.status(backendResponse.status).json({
-        error: 'Backend error',
-        status: backendResponse.status,
-        message: errorText,
-      });
-      return;
-    }
-
-    // Set SSE headers and flush immediately
-    res.setHeader('Content-Type', 'text/event-stream');
-    res.setHeader('Cache-Control', 'no-cache');
-    res.setHeader('Connection', 'keep-alive');
-    res.flushHeaders?.();
-
-    // Pipe raw SSE stream from backend to browser (no parsing)
-    const reader = backendResponse.body.getReader();
-
-    // Handle client disconnect
-    req.on('close', () => {
-      console.log('Client disconnected');
-      reader.cancel();
-    });
-
-    // Stream loop
-    while (true) {
-      const { done, value } = await reader.read();
-      if (done) {
-        console.log('Stream complete');
-        res.end();
-        return;
-      }
-      res.write(value);
-    }
-  } catch (error) {
-    console.error('Proxy error:', error);
-    if (!res.headersSent) {
-      res.status(500).json({ error: 'Proxy error', message: error.message });
-    } else {
-      res.end();
-    }
-  }
-});
-
-// Health check endpoint for Cloud Run
-app.get('/health', (req, res) => {
-  res.status(200).json({ status: 'healthy' });
-});
-
-// SPA fallback - all other routes serve index.html
-app.get('*', (req, res) => {
   res.sendFile(path.join(__dirname, 'dist', 'index.html'));
 });
 
-// Listen on 0.0.0.0 for Cloud Run
+const port = parseInt(process.env.PORT || '8080', 10);
 app.listen(port, '0.0.0.0', () => {
-  console.log(`Server listening on ${port}`);
+  console.log(`Frontend proxy listening on port ${port}`);
-  console.log(`Backend URL: ${BACKEND_URL}`);
 });

src/hooks/useChat.ts

@@ -24,47 +24,21 @@ export const useChat = (mode: ChatMode) => {
     setIsLoading(true)
 
     try {
-      // Create placeholder assistant message
+      // Call the chat API and get the response
-      const assistantMessageId = (Date.now() + 1).toString()
+      const response = await apiClient.chat(question, mode)
-      let assistantContent = ''
 
-      setMessages((prev) => [
+      // Add assistant message with the response
-        ...prev,
+      const assistantMessage: Message = {
-        {
+        id: (Date.now() + 1).toString(),
-          id: assistantMessageId,
         role: 'assistant',
-          content: '',
+        content: response.response,
-        },
-      ])
-
-      // Process SSE stream
-      const streamGenerator = apiClient.chatStream(question, mode)
-
-      for await (const chunk of streamGenerator) {
-        if (chunk.type === 'chunk') {
-          // Append chunk content
-          assistantContent += chunk.data
-
-          // Update assistant message
-          setMessages((prev) =>
-            prev.map((msg) =>
-              msg.id === assistantMessageId
-                ? { ...msg, content: assistantContent }
-                : msg
-            )
-          )
-        } else if (chunk.type === 'error') {
-          toast.error(chunk.data)
-          console.error('Stream error:', chunk.data)
-        } else if (chunk.type === 'done') {
-          setIsLoading(false)
-        }
       }
 
-      setIsLoading(false)
+      setMessages((prev) => [...prev, assistantMessage])
     } catch (error) {
       console.error('Error sending message:', error)
-      toast.error('Failed to send message. Please try again.')
+      toast.error(error instanceof Error ? error.message : 'Failed to send message. Please try again.')
+    } finally {
       setIsLoading(false)
     }
   }

src/lib/api.ts

@@ -1,9 +1,7 @@
 import type { ChatMode } from '@/types/chat'
 
 // In production, use relative URL to route through the proxy server
-// For local development, set VITE_API_URL=http://localhost:8000 in .env
 const API_BASE_URL = '/api'
-const API_ENDPOINT = import.meta.env.VITE_API_ENDPOINT || '/chat/stream'
 const USE_MOCK_DATA = false // Set to true to use mock data for testing
 
 // Session management
@@ -35,39 +33,34 @@ const MOCK_RESPONSES = {
   ],
 }
 
-// Mock streaming function
+// Chat response type from backend
-async function* mockChatStream(
+export interface ChatResponse {
+  conversation_id: string
+  response: string
+  mode: string
+  sources: string[]
+}
+
+// Mock chat function
+async function mockChat(
   _question: string,
   mode: ChatMode
-): AsyncGenerator<StreamEvent> {
+): Promise<ChatResponse> {
-  const sessionId = crypto.randomUUID()
-  yield { type: 'session_id', data: sessionId }
-
   // Select a random response based on mode
   const responses = MOCK_RESPONSES[mode]
   const response = responses[Math.floor(Math.random() * responses.length)]
 
-  // Simulate streaming by yielding words with delays
+  // Simulate network delay
-  const words = response.split(' ')
+  await new Promise(resolve => setTimeout(resolve, 500))
-  for (let i = 0; i < words.length; i++) {
-    // Add space before word (except first word)
-    const chunk = i === 0 ? words[i] : ' ' + words[i]
-    yield { type: 'chunk', data: chunk }
 
-    // Random delay between 30-80ms to simulate typing
+  return {
-    await new Promise(resolve => setTimeout(resolve, Math.random() * 50 + 30))
+    conversation_id: crypto.randomUUID(),
+    response,
+    mode,
+    sources: [],
   }
-
-  yield { type: 'done', data: null }
 }
 
-// SSE stream event types
-export type StreamEvent =
-  | { type: 'chunk'; data: string }
-  | { type: 'done'; data: null }
-  | { type: 'error'; data: string }
-  | { type: 'session_id'; data: string }
-
 class ApiClient {
   private sessionId: string | null = null
 
@@ -85,27 +78,22 @@ class ApiClient {
     clearChatSessionId()
   }
 
-  // Streaming chat endpoint (SSE)
+  // Simple JSON chat endpoint
-  async *chatStream(
+  async chat(
     question: string,
     mode: ChatMode,
     sessionId?: string
-  ): AsyncGenerator<StreamEvent> {
+  ): Promise<ChatResponse> {
     // Use mock data if enabled
     if (USE_MOCK_DATA) {
-      yield* mockChatStream(question, mode)
+      return mockChat(question, mode)
-      return
     }
 
     // Generate or reuse session ID
     const actualSessionId = sessionId || this.getSessionId() || crypto.randomUUID()
     this.setSessionId(actualSessionId)
 
-    // Yield session ID first
+    const response = await fetch(`${API_BASE_URL}/chat`, {
-    yield { type: 'session_id', data: actualSessionId }
-
-    try {
-      const response = await fetch(`${API_BASE_URL}${API_ENDPOINT}`, {
       method: 'POST',
       headers: {
         'Content-Type': 'application/json',
@@ -117,57 +105,13 @@ class ApiClient {
     })
 
     if (!response.ok) {
-        yield { type: 'error', data: `HTTP error! status: ${response.status}` }
+      const errorData = await response.json().catch(() => ({}))
-        return
+      throw new Error(errorData.detail || `HTTP error! status: ${response.status}`)
     }
 
-      if (!response.body) {
+    const data: ChatResponse = await response.json()
-        yield { type: 'error', data: 'Response body is null' }
+    this.setSessionId(data.conversation_id)
-        return
+    return data
-      }
-
-      const reader = response.body.getReader()
-      const decoder = new TextDecoder()
-      let buffer = ''
-
-      while (true) {
-        const { done, value } = await reader.read()
-
-        if (done) break
-
-        buffer += decoder.decode(value, { stream: true })
-        const lines = buffer.split('\n')
-        buffer = lines.pop() || ''
-
-        for (const line of lines) {
-          if (line.startsWith('data: ')) {
-            const data = line.slice(6)
-
-            try {
-              const parsed = JSON.parse(data)
-              if (parsed.type === 'chunk' && parsed.content) {
-                yield { type: 'chunk', data: parsed.content }
-              } else if (parsed.type === 'done') {
-                yield { type: 'done', data: null }
-                return
-              } else if (parsed.type === 'error') {
-                yield { type: 'error', data: parsed.message || 'Unknown error' }
-                return
-              }
-            } catch {
-              // Skip non-JSON lines
-            }
-          }
-        }
-      }
-
-      yield { type: 'done', data: null }
-    } catch (error) {
-      yield {
-        type: 'error',
-        data: error instanceof Error ? error.message : 'Network error',
-      }
-    }
   }
 }