feat: quiz generator endpoint added

Backend/app/api/v1/endpoints/prompts.py

@@ -1,12 +1,9 @@
-
-
 SYSTEM_PROMPT = """
 You are an AI question-generation agent.
 Your task is to generate a batch of 10 high-quality MCQ questions strictly based on the following inputs:
 
 - {parsed_info}
 - {user_prompt}
-- {mcq_style}
 - {retrieved_docs}
 
 -----------------------
@@ -25,8 +22,8 @@ GENERATION RULES
 -----------------------
 REQUIRED JSON FORMAT FOR EACH QUESTION
 -----------------------
-{
-    "question": "Which of the following CLI command can also be used to rename files?",
+{{
+    "questions": "Which of the following CLI command can also be used to rename files?",
     "options": [
         "rm",
         "mv",
@@ -34,9 +31,9 @@ REQUIRED JSON FORMAT FOR EACH QUESTION
         "none of the mentioned"
     ],
     "answer": "b",
-    "explanation": "Explanation: mv stands for move.",
+    "explanation": "mv stands for move.",
     "User_response": ""
-}
+}}
 
 -----------------------
 ANSWER KEY RULES

Backend/app/api/v1/endpoints/quiz.py

@@ -1,56 +1,65 @@
-from fastapi import APIRouter, Depends, HTTPException
+from fastapi import APIRouter, Depends, HTTPException, status
 from chromadb import AsyncHttpClient
 from app.models import User
 from app.api.deps import get_db, get_current_user, get_chroma_client
-from app.schema import Quiz_input
+from app.schema import Quiz_input, QuizOutput
 from .prompts import SYSTEM_PROMPT
-
 from fastapi import APIRouter, Depends, HTTPException
 from chromadb.api.models.Collection import Collection # Import Collection type
 from app.api.deps import get_chroma_collection
+from app.llm import call_llm
+router = APIRouter(prefix="/quiz")
 
-router = APIRouter(prefix="/search")
-
+async def search_logic(query: str, collection: Collection):
+    results = await collection.query(
+        query_texts=[query],
+        n_results=5
+    )
+    return ''.join(results['documents'][0])
 
-@router.get("/")
+@router.get("/search_docs")
 async def search_documents(
-    query: str, 
-    # Inject the pre-loaded Collection object directly
-    collection: Collection = Depends(get_chroma_collection) 
+    query: str,
+    collection: Collection = Depends(get_chroma_collection)
 ):
     try:
-        # The Collection object is ready to use immediately.
-        results = await collection.query(
-            query_texts=[query],
-            n_results=5
-        )
-        return results
+        return await search_logic(query, collection)
     except Exception as e:
-        raise HTTPException(status_code=500, detail=f"ChromaDB Query Error: {e}")
-
+        raise HTTPException(500, f"ChromaDB Query Error: {e}")
 
 
+@router.post("/", response_model=QuizOutput, status_code=status.HTTP_201_CREATED)
+async def generate_quiz(
+    Input_model: Quiz_input, 
+    collection: Collection = Depends(get_chroma_collection), 
+    current_user: User = Depends(get_current_user)
+):
+    try:
+        query = Input_model.parsed_doc + Input_model.user_prompt
+        retrieved_context = await search_logic(query, collection)
+        
 
-# @router.post("/", response_model=StudentResponse, status_code=status.HTTP_201_CREATED)
-# async def generate_quiz(
-#     Input_model: Quiz_input, db: AsyncSession = Depends(get_db), 
-#     current_user: User = Depends(get_current_user)):
-
-    
-#     if Input_model.parsed_doc and Input_model.user_prompt and Input_model.choice:
-#         prompt = prompt_builder(Input_model.parsed_doc, Input_model.user_prompt, Input_model.choice)
+        if not retrieved_context:
+            raise ValueError("No context available to generate quiz.")
+        prompt = await prompt_builder(Input_model.parsed_doc, Input_model.user_prompt, retrieved_context)
+        
+        quiz_data_obj = await call_llm(prompt)
 
+        return quiz_data_obj
 
+    except Exception as e:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=f'Invalid Input: {str(e)}'
+        )
 
 # #--------Helper Functions--------#
 
-# def get_embed()
 
-# def prompt_builder(parsed_doc:str, user_prompt:str, choice:str):
-#     retrieved_docs = get_embed()
-#     prompt = SYSTEM_PROMPT.format(
-#         parsed_info=parsed_doc,
-#         user_prompt=user_prompt,
-#         mcq_style=choice,
-#         # retrieved_docs=
-#     )
+async def prompt_builder(parsed_doc:str, user_prompt:str, docs:str=None):
+    prompt = SYSTEM_PROMPT.format(
+        parsed_info=parsed_doc,
+        user_prompt=user_prompt,
+        retrieved_docs=docs
+    )
+    return prompt

Backend/app/config.py

@@ -16,6 +16,8 @@ class Settings(BaseSettings):
     chroma_port: int
     chroma_collection: str
 
+    GEMINI_API_KEY:str
+
     class Config:
         env_file = ".env"
         extra = "ignore"  # quiz

Backend/app/llm.py

@@ -0,0 +1,35 @@
+import os
+from openai import OpenAI
+from pydantic import BaseModel, Field
+from typing import List, Optional, Any
+from app.schema.models import QuizOutput, QuizQuestion
+from app.config import settings
+
+client = OpenAI(
+    base_url="https://generativelanguage.googleapis.com/v1beta/openai/",
+    api_key="AIzaSyAIZJOjjq87FDmW9sVoTuvPkwnmfFWtfNE",
+)
+
+async def call_llm(prompt:str):
+    try:
+        response = client.chat.completions.create(
+            # CRUCIAL: Use the LiteLLM format: 'gemini/gemini-2.5-pro'
+            model="models/gemini-2.0-flash", 
+            messages=[
+                {"role": "user", "content": prompt}
+            ],
+            # Use the OpenAI parameter to request JSON output
+            response_format={"type": "json_object"}, 
+            temperature=0.7,
+        )
+
+        json_string = response.choices[0].message.content
+
+        import json
+        quiz_data = json.loads(json_string)
+        wrapped_data = {"quiz": quiz_data}
+        return QuizOutput.model_validate(wrapped_data)
+
+    except Exception as e:
+        print(f"Error calling LiteLLM/Gemini: {e}")
+        raise e

Backend/app/main.py

@@ -7,7 +7,9 @@ from app.database import engine, Base
 from app.api.v1.api import api_router
 import chromadb
 from chromadb.api.models.Collection import Collection
+from dotenv import load_dotenv
 
+load_dotenv()
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):

Backend/app/schema/__init__.py

@@ -1,3 +1,3 @@
-from app.schema.models import StudentCreate, StudentUpdate, StudentResponse, UserCreate, Token, LoginRequest, Quiz_input
+from app.schema.models import StudentCreate, StudentUpdate, StudentResponse, UserCreate, Token, LoginRequest, Quiz_input, QuizOutput
 
-__all__ = ["StudentCreate", "StudentUpdate", "StudentResponse", "UserCreate", "Token", "LoginRequest", "Quiz_input"]
+__all__ = ["StudentCreate", "StudentUpdate", "StudentResponse", "UserCreate", "Token", "LoginRequest", "Quiz_input", "QuizOutput"]

Backend/app/schema/models.py

@@ -1,5 +1,5 @@
 from pydantic import BaseModel, EmailStr, Field, field_validator, ConfigDict
-from typing import  Optional, Literal
+from typing import  Optional, Literal, List
 from datetime import datetime
 
 class StudentBase(BaseModel):
@@ -51,4 +51,14 @@ class LoginResponse(Token):
 class Quiz_input(BaseModel):
     parsed_doc: str
     user_prompt: str
-    choice: Literal["mcq", "code"]
+    # choice: Literal["mcq", "code"]
+
+class QuizQuestion(BaseModel):
+    questions: str
+    options: List[str] = Field(..., min_items=4, max_items=4)
+    answer: str  = Field(..., description="The correct answer key (e.g., 'a', 'b', 'c', or 'd')")
+    explanation: str 
+    user_response: str = Field("", alias="User_response")
+
+class QuizOutput(BaseModel):
+    quiz: List[QuizQuestion] = Field(..., description="A list of 10 generated MCQ questions.")

Backend/app/services/inital_data.py

@@ -1,14 +0,0 @@
-from chromadb import AsyncHttpClient
-from chromadb.utils import embedding_functions
-
-async def ingest_start(client: AsyncHttpClient, collection_name:str):
-
-    try:
-        await client.get_collection(name=collection_name)
-        print(f"Collection '{collection_name}' already exists. Skipping initial ingestion.")
-        return
-    except Exception:
-        print(f"Collection '{collection_name}' not found. Starting initial ingestion...")
-        pass
-
-