refactor

Author: leehuwuj

helpers/env-variables.ts

@@ -486,26 +486,26 @@ It\\'s cute animal.
   return systemPromptEnv;
 };
 
-const getTemplateEnvs = (template?: TemplateType): EnvVar[] => {
+const getTemplateEnvs = (
+  template?: TemplateType,
+  framework?: TemplateFramework,
+): EnvVar[] => {
   const nextQuestionEnvs: EnvVar[] = [
-    {
-      name: "NEXT_QUESTION_ENABLE",
-      description: "Whether to show next question suggestions",
-      value: "true",
-    },
     {
       name: "NEXT_QUESTION_PROMPT",
       description: `Customize prompt to generate the next question suggestions based on the conversation history.
-Default prompt is:
-NEXT_QUESTION_PROMPT=# You're a helpful assistant! Your task is to suggest the next question that user might ask. 
-# Here is the conversation history
-# ---------------------\n{conversation}\n---------------------
-# Given the conversation history, please give me 3 questions that you might ask next!
-`,
+Disable this prompt to disable the next question suggestions feature.`,
+      value: `"You're a helpful assistant! Your task is to suggest the next question that user might ask. 
+Here is the conversation history
+---------------------\n{conversation}\n---------------------
+Given the conversation history, please give me 3 questions that you might ask next!"`,
     },
   ];
 
-  if (template === "multiagent" || template === "streaming") {
+  if (
+    framework === "fastapi" &&
+    (template === "multiagent" || template === "streaming")
+  ) {
     return nextQuestionEnvs;
   }
   return [];
@@ -555,7 +555,7 @@ export const createBackendEnvFile = async (
     ...getVectorDBEnvs(opts.vectorDb, opts.framework),
     ...getFrameworkEnvs(opts.framework, opts.externalPort),
     ...getToolEnvs(opts.tools),
-    ...getTemplateEnvs(opts.template),
+    ...getTemplateEnvs(opts.template, opts.framework),
     ...getObservabilityEnvs(opts.observability),
     ...getSystemPromptEnv(opts.tools, opts.dataSources, opts.framework),
   ];

templates/components/services/python/suggestion.py

@@ -1,51 +1,38 @@
 import logging
+import os
 from typing import List, Optional
 
+from app.api.routers.models import Message
 from llama_index.core.prompts import PromptTemplate
 from llama_index.core.settings import Settings
 from pydantic import BaseModel
-from pydantic_settings import BaseSettings, SettingsConfigDict
-
-from app.api.routers.models import Message
 
 logger = logging.getLogger("uvicorn")
 
 
-class NextQuestionSettings(BaseSettings):
-    enable: bool = True
-    prompt_template: str = (
-        "You're a helpful assistant! Your task is to suggest the next question that user might ask. "
-        "\nHere is the conversation history"
-        "\n---------------------\n{conversation}\n---------------------"
-        "Given the conversation history, please give me 3 questions that you might ask next!"
-    )
-
-    model_config = SettingsConfigDict(env_prefix="NEXT_QUESTION_")
-
-    @property
-    def prompt(self) -> PromptTemplate:
-        return PromptTemplate(self.prompt_template)
-
-
-next_question_settings = NextQuestionSettings()
-
-
 class NextQuestions(BaseModel):
     """A list of questions that user might ask next"""
 
     questions: List[str]
 
 
 class NextQuestionSuggestion:
-    @staticmethod
+
+    @classmethod
+    def get_configured_prompt(cls) -> Optional[str]:
+        return os.getenv("NEXT_QUESTION_PROMPT", None)
+
+    @classmethod
     async def suggest_next_questions(
+        cls,
         messages: List[Message],
     ) -> Optional[List[str]]:
         """
         Suggest the next questions that user might ask based on the conversation history
         Return None if suggestion is disabled or there is an error
         """
-        if not next_question_settings.enable:
+        prompt_template = cls.get_configured_prompt()
+        if not prompt_template:
             return None
 
         try:
@@ -63,7 +50,7 @@ async def suggest_next_questions(
 
             output: NextQuestions = await Settings.llm.astructured_predict(
                 NextQuestions,
-                prompt=next_question_settings.prompt,
+                prompt=PromptTemplate(prompt_template),
                 conversation=conversation,
             )
 

templates/types/multiagent/fastapi/app/api/routers/vercel_response.py

@@ -1,16 +1,15 @@
-from asyncio import Task
 import json
 import logging
+from asyncio import Task
 from typing import AsyncGenerator
 
 from aiostream import stream
+from app.agents.single import AgentRunEvent, AgentRunResult
+from app.api.routers.models import ChatData, Message
+from app.api.services.suggestion import NextQuestionSuggestion
 from fastapi import Request
 from fastapi.responses import StreamingResponse
 
-from app.api.routers.models import ChatData, Message
-from app.agents.single import AgentRunEvent, AgentRunResult
-from app.api.services.suggestion import NextQuestionSuggestion, next_question_settings
-
 logger = logging.getLogger("uvicorn")
 
 
@@ -69,8 +68,8 @@ async def _chat_response_generator():
                     final_response += token.delta
                     yield VercelStreamResponse.convert_text(token.delta)
 
-            # Generate questions that user might be interested in
-            if next_question_settings.enable:
+            # Generate next questions if next question prompt is configured
+            if NextQuestionSuggestion.get_configured_prompt() is not None:
                 conversation = chat_data.messages + [
                     Message(role="assistant", content=final_response)
                 ]

templates/types/streaming/fastapi/app/api/routers/vercel_response.py

@@ -7,7 +7,7 @@
 
 from app.api.routers.events import EventCallbackHandler
 from app.api.routers.models import ChatData, Message, SourceNodes
-from app.api.services.suggestion import NextQuestionSuggestion, next_question_settings
+from app.api.services.suggestion import NextQuestionSuggestion
 
 
 class VercelStreamResponse(StreamingResponse):
@@ -56,8 +56,8 @@ async def _chat_response_generator():
                 final_response += token
                 yield VercelStreamResponse.convert_text(token)
 
-            # Generate questions that user might be interested in
-            if next_question_settings.enable:
+            # Generate next questions if next question prompt is configured
+            if NextQuestionSuggestion.get_configured_prompt() is not None:
                 conversation = chat_data.messages + [
                     Message(role="assistant", content=final_response)
                 ]