Streaming u pdate2

app/bot_scripts/chat.js

@@ -59,7 +59,14 @@ async function sendMessage() {
   const typing = showTyping();
 
   try {
-    const response = await fetch(`${config.apiUrl}/chat/stream`, {
+    const isLocal =
+      window.location.hostname === "localhost" ||
+      window.location.hostname === "127.0.0.1";
+    const endpoint = isLocal
+      ? `${config.apiUrl}/chat/stream`
+      : `${config.apiUrl}/chat`;
+
+    const response = await fetch(endpoint, {
       method: "POST",
       headers: { "Content-Type": "application/json" },
       body: JSON.stringify({
@@ -68,73 +75,51 @@ async function sendMessage() {
       }),
     });
 
-    // Create bot message container (but don't remove typing yet)
-    const div = document.createElement("div");
-    div.className = "chat-message bot";
-    const label = document.createElement("div");
-    label.className = "bot-label";
-    label.textContent = config.botName;
-    div.appendChild(label);
+    typing.remove();
 
-    // Stream the response
     let fullResponse = "";
-    let firstChunk = true;
-    const reader = response.body.getReader();
-    const decoder = new TextDecoder();
-
-    while (true) {
-      const { done, value } = await reader.read();
-      if (done) break;
-
-      // Remove typing indicator on first chunk
-      if (firstChunk) {
-        typing.remove();
-        chatMessages.appendChild(div);
-        firstChunk = false;
-      }
-
-      const chunk = decoder.decode(value, { stream: true });
-      console.log('[STREAM] chunk:', chunk.substring(0, 20));  // STREAM_DEBUG
-      fullResponse += chunk;
-
-      // Update the message content
-      while (div.childNodes.length > 1) {
-        div.removeChild(div.lastChild);
-      }
-
-      const formatter = config.formatMessage || defaultFormatMessage;
-      formatter(fullResponse, div);
-
-      chatMessages.scrollTop = chatMessages.scrollHeight;
-    }
 
-    while (true) {
-      const { done, value } = await reader.read();
-      if (done) break;
-
-      const chunk = decoder.decode(value, { stream: true });
-      fullResponse += chunk;
-
-      // Update the message content
-      // Remove old text nodes (keep the label)
-      while (div.childNodes.length > 1) {
-        div.removeChild(div.lastChild);
+    if (isLocal) {
+      // Streaming: render chunks as they arrive
+      const div = document.createElement("div");
+      div.className = "chat-message bot";
+      const label = document.createElement("div");
+      label.className = "bot-label";
+      label.textContent = config.botName;
+      div.appendChild(label);
+      chatMessages.appendChild(div);
+
+      const reader = response.body.getReader();
+      const decoder = new TextDecoder();
+
+      while (true) {
+        const { done, value } = await reader.read();
+        if (done) break;
+
+        const chunk = decoder.decode(value, { stream: true });
+        console.log("[STREAM] chunk:", chunk.substring(0, 20)); // STREAM_DEBUG
+        fullResponse += chunk;
+
+        while (div.childNodes.length > 1) {
+          div.removeChild(div.lastChild);
+        }
+
+        const formatter = config.formatMessage || defaultFormatMessage;
+        formatter(fullResponse, div);
+        chatMessages.scrollTop = chatMessages.scrollHeight;
       }
-
-      // Re-render with formatter
-      const formatter = config.formatMessage || defaultFormatMessage;
-      formatter(fullResponse, div);
-
-      chatMessages.scrollTop = chatMessages.scrollHeight;
+    } else {
+      // Production: parse JSON response
+      const data = await response.json();
+      fullResponse = data.response;
+      addMessage(fullResponse, "bot");
     }
 
-    // Track both sides of the exchange for follow-ups
     conversationHistory.push(
       { role: "user", content: message },
       { role: "assistant", content: fullResponse },
     );
 
-    // Cap at 10 exchanges (20 messages) to keep token costs in check
     const maxMessages = 20;
     if (conversationHistory.length > maxMessages) {
       conversationHistory.splice(0, conversationHistory.length - maxMessages);

terraform/cloudfront.tf

@@ -22,11 +22,10 @@ resource "aws_cloudfront_distribution" "website" {
     origin_access_control_id = aws_cloudfront_origin_access_control.website.id
   }
 
-  # API Gateway origin
+  # Lambda Function URL origin (replaces API Gateway)
   origin {
-    domain_name = "${var.api_gateway_id}.execute-api.us-east-1.amazonaws.com"
-    origin_id   = "APIGateway"
-    origin_path = "/prod"
+    domain_name = replace(replace(aws_lambda_function_url.streaming.function_url, "https://", ""), "/", "")
+    origin_id   = "LambdaFunctionURL"
 
     custom_origin_config {
       http_port              = 80
@@ -36,12 +35,12 @@ resource "aws_cloudfront_distribution" "website" {
     }
   }
 
-  # Route /api/* to API Gateway — must come before default_cache_behavior
+  # Route /api/* to Lambda Function URL — must come before default_cache_behavior
   ordered_cache_behavior {
     path_pattern           = "/api/*"
     allowed_methods        = ["DELETE", "GET", "HEAD", "OPTIONS", "PATCH", "POST", "PUT"]
     cached_methods         = ["GET", "HEAD"]
-    target_origin_id       = "APIGateway"
+    target_origin_id       = "LambdaFunctionURL"
     viewer_protocol_policy = "redirect-to-https"
     compress               = true
 
@@ -112,4 +111,4 @@ resource "aws_cloudfront_distribution" "website" {
   }
 
   depends_on = [aws_acm_certificate_validation.website]
-}
+}

terraform/iam.tf

@@ -97,7 +97,8 @@ resource "aws_iam_role_policy" "lambda_bedrock" {
       {
         Effect = "Allow"
         Action = [
-          "bedrock:InvokeModel"
+          "bedrock:InvokeModel",
+  "bedrock:InvokeModelWithResponseStream"
         ]
         Resource = [
           "arn:aws:bedrock:*::foundation-model/*",

terraform/lambda.tf

@@ -7,20 +7,27 @@ resource "aws_lambda_function" "fastapi_app" {
   s3_key           = "lambda/fastapi-app.zip"
   function_name    = "${var.project_name}-api"
   role             = aws_iam_role.lambda_execution.arn
-  handler          = "lambda_handler.handler"
+  handler          = "run.sh"
   source_code_hash = filebase64sha256("${path.module}/builds/fastapi-app.zip")
   runtime          = "python3.12"
   timeout          = 30
   memory_size      = 512
 
+  layers = [
+    "arn:aws:lambda:us-east-1:753240598075:layer:LambdaAdapterLayerX86:24"
+  ]
+
   environment {
     variables = {
-      DYNAMODB_TABLE       = aws_dynamodb_table.resume_data.name
-      RECAPTCHA_SECRET_KEY = var.recaptcha_secret_key
-      SES_FROM_EMAIL       = "robmrose@me.com"
-      SES_TO_EMAIL         = "robmrose@me.com"
-      OPENAI_API_KEY       = var.openai_api_key
-      ANTHROPIC_API_KEY    = var.anthropic_api_key
+      DYNAMODB_TABLE          = aws_dynamodb_table.resume_data.name
+      RECAPTCHA_SECRET_KEY    = var.recaptcha_secret_key
+      SES_FROM_EMAIL          = "robmrose@me.com"
+      SES_TO_EMAIL            = "robmrose@me.com"
+      OPENAI_API_KEY          = var.openai_api_key
+      ANTHROPIC_API_KEY       = var.anthropic_api_key
+      AWS_LWA_PORT            = "8080"
+      AWS_LAMBDA_EXEC_WRAPPER = "/opt/bootstrap"
+      AWS_LWA_INVOKE_MODE = "response_stream"
     }
   }
 
@@ -38,3 +45,22 @@ resource "aws_lambda_permission" "api_gateway" {
   principal     = "apigateway.amazonaws.com"
   source_arn    = "${aws_api_gateway_rest_api.resume_api.execution_arn}/*/*"
 }
+
+# Lambda Function URL for streaming (bypasses API Gateway)
+resource "aws_lambda_function_url" "streaming" {
+  function_name      = aws_lambda_function.fastapi_app.function_name
+  authorization_type = "NONE"
+  invoke_mode = "RESPONSE_STREAM"
+}
+
+output "streaming_url" {
+  value = aws_lambda_function_url.streaming.function_url
+}
+
+resource "aws_lambda_permission" "function_url_public" {
+  statement_id           = "FunctionURLAllowPublicAccess"
+  action                 = "lambda:InvokeFunctionUrl"
+  function_name          = aws_lambda_function.fastapi_app.function_name
+  principal              = "*"
+  function_url_auth_type = "NONE"
+}