Video-Rag / main.py
Hammad712's picture
Update main.py
a2f5873 verified
from fastapi import FastAPI, HTTPException, Depends, File, UploadFile, Form, Response, BackgroundTasks
from fastapi.middleware.cors import CORSMiddleware
from fastapi.security import OAuth2PasswordBearer, OAuth2PasswordRequestForm
from fastapi.responses import StreamingResponse
from pydantic import BaseModel, Field, EmailStr
from typing import List, Optional, Dict, Any, Union
import uuid
import os
import io
from urllib.parse import quote_plus
import shutil
from datetime import datetime, timedelta
from dotenv import load_dotenv
import hashlib
import jwt
from passlib.context import CryptContext
from pymongo import MongoClient
from langchain_mongodb.chat_message_histories import MongoDBChatMessageHistory
# Load environment variables
load_dotenv()
# Import necessary libraries - updating deprecated imports
from langchain_huggingface import HuggingFaceEmbeddings
from langchain_community.vectorstores import FAISS
from langchain.chains import ConversationalRetrievalChain
from langchain_core.prompts import PromptTemplate, ChatPromptTemplate
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_core.documents import Document
from langchain_groq import ChatGroq
from google import genai
from google.genai import types
# MongoDB Configuration
MONGO_PASSWORD = quote_plus(os.getenv("MONGO_PASSWORD"))
MONGO_DATABASE_NAME = os.getenv("DATABASE_NAME")
MONGO_COLLECTION_NAME = os.getenv("COLLECTION_NAME")
connection_string_template = os.getenv("CONNECTION_STRING")
MONGO_CLUSTER_URL = connection_string_template.replace("${PASSWORD}", MONGO_PASSWORD)
CHAT_COLLECTION = MONGO_COLLECTION_NAME or "chat_history"
USER_COLLECTION = "users"
VIDEO_COLLECTION = "videos"
# Security
SECRET_KEY = os.getenv("SECRET_KEY")
ALGORITHM = "HS256"
ACCESS_TOKEN_EXPIRE_MINUTES = 30
# Password hashing
pwd_context = CryptContext(schemes=["bcrypt"], deprecated="auto")
# OAuth2 scheme
oauth2_scheme = OAuth2PasswordBearer(tokenUrl="token")
# Initialize FastAPI app
app = FastAPI(title="RAG System API", description="An API for question answering based on video content with user authentication")
# Configure CORS
app.add_middleware(
CORSMiddleware,
allow_origins=["*"],
allow_credentials=True,
allow_methods=["*"],
allow_headers=["*"],
)
# Define models
class TranscriptionRequest(BaseModel):
youtube_url: str
class QueryRequest(BaseModel):
query: str
session_id: str
class QueryResponse(BaseModel):
answer: str
session_id: str
source_documents: Optional[List[str]] = None
class User(BaseModel):
username: str
email: EmailStr
full_name: Optional[str] = None
class UserInDB(User):
hashed_password: str
class UserCreate(User):
password: str
class Token(BaseModel):
access_token: str
token_type: str
class TokenData(BaseModel):
username: Optional[str] = None
class VideoData(BaseModel):
video_id: str
user_id: str
title: str
source_type: str # "youtube" or "upload"
source_url: Optional[str] = None
created_at: datetime = Field(default_factory=datetime.utcnow)
transcription: str
size: Optional[int] = None
# MongoDB connection and chat management
class MongoDB:
def __init__(self):
self.client = MongoClient(MONGO_CLUSTER_URL)
self.db = self.client[MONGO_DATABASE_NAME]
self.users = self.db[USER_COLLECTION]
self.videos = self.db[VIDEO_COLLECTION]
# Ensure indexes
self.users.create_index("username", unique=True)
self.users.create_index("email", unique=True)
self.videos.create_index("video_id", unique=True)
self.videos.create_index("user_id")
def close(self):
self.client.close()
# Chat Management Class
class ChatManagement:
def __init__(self, cluster_url, database_name, collection_name):
self.connection_string = cluster_url
self.database_name = database_name
self.collection_name = collection_name
self.chat_sessions = {} # Dictionary to store chat history objects for each session
def create_new_chat(self):
# Generate a unique chat ID
chat_id = str(uuid.uuid4())
# Initialize MongoDBChatMessageHistory for the chat session
chat_message_history = MongoDBChatMessageHistory(
session_id=chat_id,
connection_string=self.connection_string,
database_name=self.database_name,
collection_name=self.collection_name
)
# Store the chat_message_history object in the session dictionary
self.chat_sessions[chat_id] = chat_message_history
return chat_id
def get_chat_history(self, chat_id):
# Check if the chat session is already in memory
if chat_id in self.chat_sessions:
return self.chat_sessions[chat_id]
# If not in memory, try to fetch from the database
chat_message_history = MongoDBChatMessageHistory(
session_id=chat_id,
connection_string=self.connection_string,
database_name=self.database_name,
collection_name=self.collection_name
)
if chat_message_history.messages: # Check if the session exists in the database
self.chat_sessions[chat_id] = chat_message_history
return chat_message_history
return None # Chat session not found
def initialize_chat_history(self, chat_id):
# If the chat history already exists, return it
if chat_id in self.chat_sessions:
return self.chat_sessions[chat_id]
# Otherwise, create a new chat history
chat_message_history = MongoDBChatMessageHistory(
session_id=chat_id,
connection_string=self.connection_string,
database_name=self.database_name,
collection_name=self.collection_name
)
# Save the new chat session to the session dictionary
self.chat_sessions[chat_id] = chat_message_history
return chat_message_history
# Global variables and instances
mongodb = MongoDB()
chat_manager = ChatManagement(MONGO_CLUSTER_URL, MONGO_DATABASE_NAME, CHAT_COLLECTION)
sessions = {} # In-memory session storage for retrievers
# Video directory for temporary storage
VIDEOS_DIR = "temp_videos"
os.makedirs(VIDEOS_DIR, exist_ok=True)
# Security functions
def verify_password(plain_password, hashed_password):
return pwd_context.verify(plain_password, hashed_password)
def get_password_hash(password):
return pwd_context.hash(password)
def create_access_token(data: dict, expires_delta: Optional[timedelta] = None):
to_encode = data.copy()
if expires_delta:
expire = datetime.utcnow() + expires_delta
else:
expire = datetime.utcnow() + timedelta(minutes=ACCESS_TOKEN_EXPIRE_MINUTES)
to_encode.update({"exp": expire})
encoded_jwt = jwt.encode(to_encode, SECRET_KEY, algorithm=ALGORITHM)
return encoded_jwt
def get_user(username: str):
user_data = mongodb.users.find_one({"username": username})
if user_data:
return UserInDB(**user_data)
return None
def authenticate_user(username: str, password: str):
user = get_user(username)
if not user:
return False
if not verify_password(password, user.hashed_password):
return False
return user
async def get_current_user(token: str = Depends(oauth2_scheme)):
credentials_exception = HTTPException(
status_code=401,
detail="Could not validate credentials",
headers={"WWW-Authenticate": "Bearer"},
)
try:
payload = jwt.decode(token, SECRET_KEY, algorithms=[ALGORITHM])
username: str = payload.get("sub")
if username is None:
raise credentials_exception
token_data = TokenData(username=username)
except jwt.PyJWTError:
raise credentials_exception
user = get_user(username=token_data.username)
if user is None:
raise credentials_exception
return user
# Initialize Google API client
def init_google_client():
api_key = os.getenv("GOOGLE_API_KEY", "")
if not api_key:
raise ValueError("GOOGLE_API_KEY environment variable not set")
return genai.Client(api_key=api_key)
# Get LLM
def get_llm():
"""
Returns the language model instance (LLM) using ChatGroq API.
The LLM used is Llama 3.3 with a versatile 70 billion parameters model.
"""
api_key = os.getenv("CHATGROQ_API_KEY", "")
if not api_key:
raise ValueError("CHATGROQ_API_KEY environment variable not set")
llm = ChatGroq(
model="llama-3.3-70b-versatile",
temperature=0,
max_tokens=1024,
api_key=api_key
)
return llm
# Get embeddings
def get_embeddings():
model_name = "BAAI/bge-small-en"
model_kwargs = {"device": "cpu"}
encode_kwargs = {"normalize_embeddings": True}
embeddings = HuggingFaceEmbeddings(
model_name=model_name, model_kwargs=model_kwargs, encode_kwargs=encode_kwargs
)
return embeddings
# Create prompt template
quiz_solving_prompt = '''
You are an assistant specialized in solving quizzes. Your goal is to provide accurate, concise, and contextually relevant answers.
Use the following retrieved context to answer the user's question.
If the context lacks sufficient information, respond with "I don't know." Do not make up answers or provide unverified information.
Guidelines:
1. Extract key information from the context to form a coherent response.
2. Maintain a clear and professional tone.
3. If the question requires clarification, specify it politely.
Retrieved context:
{context}
User's question:
{question}
Your response:
'''
# Create a prompt template to pass the context and user input to the chain
user_prompt = ChatPromptTemplate.from_messages(
[
("system", quiz_solving_prompt),
("human", "{question}"),
]
)
# Create a chain
def create_chain(retriever):
llm = get_llm()
chain = ConversationalRetrievalChain.from_llm(
llm=llm,
retriever=retriever,
return_source_documents=True,
chain_type='stuff',
combine_docs_chain_kwargs={"prompt": user_prompt},
verbose=False,
)
return chain
# Process transcription and prepare RAG system
def process_transcription(transcription, user_id, title, source_type, source_url=None, file_size=None):
# Process the transcription
text_splitter = RecursiveCharacterTextSplitter(chunk_size=1024, chunk_overlap=20)
all_splits = text_splitter.split_text(transcription)
# Create vector store
embeddings = get_embeddings()
vectorstore = FAISS.from_texts(all_splits, embeddings)
retriever = vectorstore.as_retriever(search_kwargs={"k": 3})
# Create a session ID
session_id = str(uuid.uuid4())
# Store video data in MongoDB
video_data = {
"video_id": session_id,
"user_id": user_id,
"title": title,
"source_type": source_type,
"source_url": source_url,
"created_at": datetime.utcnow(),
"transcription": transcription,
"size": file_size
}
mongodb.videos.insert_one(video_data)
# Store session data in memory
sessions[session_id] = {
"retriever": retriever,
"chat_history": chat_manager.initialize_chat_history(session_id)
}
return session_id
# Save video to disk (background task)
def save_video_file(video_id, file_path, contents):
os.makedirs(os.path.dirname(file_path), exist_ok=True)
with open(file_path, "wb") as f:
f.write(contents)
# Auth endpoints
@app.post("/register", response_model=User)
async def register_user(user: UserCreate):
# Check if username already exists
if mongodb.users.find_one({"username": user.username}):
raise HTTPException(status_code=400, detail="Username already registered")
# Check if email already exists
if mongodb.users.find_one({"email": user.email}):
raise HTTPException(status_code=400, detail="Email already registered")
# Create user
hashed_password = get_password_hash(user.password)
user_dict = user.dict()
del user_dict["password"]
user_dict["hashed_password"] = hashed_password
# Insert user
mongodb.users.insert_one(user_dict)
return User(**user_dict)
@app.post("/token", response_model=Token)
async def login_for_access_token(form_data: OAuth2PasswordRequestForm = Depends()):
user = authenticate_user(form_data.username, form_data.password)
if not user:
raise HTTPException(
status_code=401,
detail="Incorrect username or password",
headers={"WWW-Authenticate": "Bearer"},
)
access_token_expires = timedelta(minutes=ACCESS_TOKEN_EXPIRE_MINUTES)
access_token = create_access_token(
data={"sub": user.username}, expires_delta=access_token_expires
)
return {"access_token": access_token, "token_type": "bearer"}
# Video processing endpoints
@app.post("/transcribe", response_model=Dict[str, str])
async def transcribe_video(
request: TranscriptionRequest,
current_user: User = Depends(get_current_user)
):
"""
Transcribe a YouTube video and prepare the RAG system
"""
try:
# Initialize Google API client
client = init_google_client()
# Transcribe the video
response = client.models.generate_content(
model='models/gemini-2.0-flash',
contents=types.Content(
parts=[
types.Part(text='Transcribe the Video. Write all the things described in the video'),
types.Part(
file_data=types.FileData(file_uri=request.youtube_url)
)
]
)
)
# Get transcription text
transcription = response.candidates[0].content.parts[0].text
# Process transcription and get session ID
video_title = f"YouTube Video - {datetime.utcnow().strftime('%Y-%m-%d %H:%M:%S')}"
session_id = process_transcription(
transcription,
current_user.username,
video_title,
"youtube",
request.youtube_url
)
return {"session_id": session_id, "message": "YouTube video transcribed and RAG system prepared"}
except Exception as e:
raise HTTPException(status_code=500, detail=f"Error transcribing video: {str(e)}")
@app.post("/upload", response_model=Dict[str, str])
async def upload_video(
background_tasks: BackgroundTasks,
title: str = Form(...),
file: UploadFile = File(...),
prompt: str = Form("Transcribe the Video. Write all the things described in the video"),
current_user: User = Depends(get_current_user)
):
"""
Upload a video file (max 20MB), transcribe it and prepare the RAG system
"""
try:
# Check file size (20MB limit)
contents = await file.read()
file_size = len(contents)
if file_size > 20 * 1024 * 1024: # 20MB in bytes
raise HTTPException(status_code=400, detail="File size exceeds 20MB limit")
# Check file type
if not file.content_type.startswith('video/'):
raise HTTPException(status_code=400, detail="File must be a video")
# Initialize Google API client
client = init_google_client()
# Transcribe the video
response = client.models.generate_content(
model='models/gemini-2.0-flash',
contents=types.Content(
parts=[
types.Part(text=prompt),
types.Part(
inline_data=types.Blob(data=contents, mime_type=file.content_type)
)
]
)
)
# Get transcription text
transcription = response.candidates[0].content.parts[0].text
# Process transcription and get session ID
session_id = process_transcription(
transcription,
current_user.username,
title,
"upload",
None,
file_size
)
# Save video file to disk
file_extension = os.path.splitext(file.filename)[1]
file_path = os.path.join(VIDEOS_DIR, f"{session_id}{file_extension}")
background_tasks.add_task(save_video_file, session_id, file_path, contents)
return {"session_id": session_id, "message": "Uploaded video transcribed and RAG system prepared"}
except Exception as e:
raise HTTPException(status_code=500, detail=f"Error processing uploaded video: {str(e)}")
finally:
# Reset file pointer
await file.seek(0)
@app.get("/download/{video_id}")
async def download_video(
video_id: str,
current_user: User = Depends(get_current_user)
):
"""
Download a previously uploaded video
"""
# Check if video exists in database
video_data = mongodb.videos.find_one({"video_id": video_id})
if not video_data:
raise HTTPException(status_code=404, detail="Video not found")
# Check if user has access to this video
if video_data["user_id"] != current_user.username:
raise HTTPException(status_code=403, detail="Not authorized to access this video")
# For YouTube videos, we don't have the actual file
if video_data["source_type"] == "youtube":
return {"message": "This is a YouTube video. Please use the original URL to access the video.", "url": video_data["source_url"]}
# For uploaded videos, check if file exists
# Look for any file with the video_id as the base name
video_files = [f for f in os.listdir(VIDEOS_DIR) if f.startswith(video_id)]
if not video_files:
raise HTTPException(status_code=404, detail="Video file not found")
file_path = os.path.join(VIDEOS_DIR, video_files[0])
# Determine file extension and MIME type
file_extension = os.path.splitext(video_files[0])[1]
mime_type = f"video/{file_extension[1:]}" if file_extension else "video/mp4"
# Stream the file
def iterfile():
with open(file_path, "rb") as f:
while chunk := f.read(8192):
yield chunk
return StreamingResponse(
iterfile(),
media_type=mime_type,
headers={"Content-Disposition": f"attachment; filename={video_data['title']}{file_extension}"}
)
@app.post("/query", response_model=QueryResponse)
async def query_system(
request: QueryRequest,
current_user: User = Depends(get_current_user)
):
"""
Query the RAG system with a question
"""
try:
session_id = request.session_id
# Check if session exists
if not session_id or session_id not in sessions:
raise HTTPException(status_code=404, detail="Session not found. Please transcribe a video first.")
# Check if user has access to this session
video_data = mongodb.videos.find_one({"video_id": session_id})
if not video_data or video_data["user_id"] != current_user.username:
raise HTTPException(status_code=403, detail="Not authorized to access this session")
# Get session data
session = sessions[session_id]
retriever = session["retriever"]
# Get or initialize chat history
chat_history = chat_manager.initialize_chat_history(session_id)
# Create chain
chain = create_chain(retriever)
# Extract messages properly for LangChain format
messages = chat_history.messages
# Process chat history safely
langchain_chat_history = []
# If messages exist, process them
if messages:
# Group messages by pairs (user, AI)
# This approach is safer than assuming perfect alternating pattern
i = 0
while i < len(messages) - 1:
user_message = messages[i].content
ai_message = messages[i+1].content
langchain_chat_history.append((user_message, ai_message))
i += 2
# Add debugging information
print(f"Chat history length: {len(langchain_chat_history)}")
print(f"Query: {request.query}")
try:
# Query the chain
result = chain.invoke({
"question": request.query,
"chat_history": langchain_chat_history
})
# Extract answer from result
answer = result.get("answer", "I couldn't find an answer to your question.")
# Update chat history
chat_history.add_user_message(request.query)
chat_history.add_ai_message(answer)
# Prepare source documents with proper error handling
source_docs = []
if "source_documents" in result and result["source_documents"]:
for doc in result["source_documents"]:
try:
# Different LangChain versions might structure documents differently
if hasattr(doc, 'page_content'):
# Regular Document object
content = doc.page_content[:100] + "..." if len(doc.page_content) > 100 else doc.page_content
source_docs.append(content)
elif isinstance(doc, dict) and 'page_content' in doc:
# Dictionary format
content = doc['page_content'][:100] + "..." if len(doc['page_content']) > 100 else doc['page_content']
source_docs.append(content)
elif isinstance(doc, str):
# String format
content = doc[:100] + "..." if len(doc) > 100 else doc
source_docs.append(content)
except Exception as doc_error:
print(f"Error processing source document: {str(doc_error)}")
return {
"answer": answer,
"session_id": session_id,
"source_documents": source_docs
}
except Exception as chain_error:
print(f"Chain invocation error: {str(chain_error)}")
# Provide a more graceful fallback
fallback_answer = "I apologize, but I encountered an error while processing your question. Please try rephrasing your query or asking about a different topic."
# Update chat history even in case of error
chat_history.add_user_message(request.query)
chat_history.add_ai_message(fallback_answer)
return {
"answer": fallback_answer,
"session_id": session_id,
"source_documents": []
}
except Exception as e:
print(f"Query system error: {str(e)}")
import traceback
traceback.print_exc()
raise HTTPException(status_code=500, detail=f"Error querying system: {str(e)}")
@app.get("/sessions", response_model=List[Dict[str, Any]])
async def get_user_sessions(current_user: User = Depends(get_current_user)):
"""
Get all video sessions for the current user
"""
user_videos = list(mongodb.videos.find({"user_id": current_user.username}))
# Format response
sessions_list = []
for video in user_videos:
sessions_list.append({
"session_id": video["video_id"],
"title": video["title"],
"source_type": video["source_type"],
"created_at": video["created_at"],
"transcription_preview": video["transcription"][:200] + "..." if len(video["transcription"]) > 200 else video["transcription"]
})
return sessions_list
@app.get("/sessions/{session_id}", response_model=Dict[str, Any])
async def get_session_info(
session_id: str,
current_user: User = Depends(get_current_user)
):
"""
Get information about a specific session
"""
# Check if session exists in database
video_data = mongodb.videos.find_one({"video_id": session_id})
if not video_data:
raise HTTPException(status_code=404, detail="Session not found")
# Check if user has access to this session
if video_data["user_id"] != current_user.username:
raise HTTPException(status_code=403, detail="Not authorized to access this session")
# Get chat history
chat_history_obj = chat_manager.get_chat_history(session_id)
chat_messages = []
if chat_history_obj:
messages = chat_history_obj.messages
for i in range(0, len(messages), 2):
if i+1 < len(messages):
chat_messages.append({
"question": messages[i].content,
"answer": messages[i+1].content
})
return {
"session_id": session_id,
"title": video_data["title"],
"source_type": video_data["source_type"],
"source_url": video_data.get("source_url"),
"created_at": video_data["created_at"],
"transcription_preview": video_data["transcription"][:200] + "..." if len(video_data["transcription"]) > 200 else video_data["transcription"],
"full_transcription": video_data["transcription"],
"chat_history": chat_messages
}
@app.delete("/sessions/{session_id}")
async def delete_session(
session_id: str,
current_user: User = Depends(get_current_user)
):
"""
Delete a session
"""
# Check if session exists in database
video_data = mongodb.videos.find_one({"video_id": session_id})
if not video_data:
raise HTTPException(status_code=404, detail="Session not found")
# Check if user has access to this session
if video_data["user_id"] != current_user.username:
raise HTTPException(status_code=403, detail="Not authorized to access this session")
# Delete from MongoDB
mongodb.videos.delete_one({"video_id": session_id})
# Delete chat history
chat_history = chat_manager.get_chat_history(session_id)
if chat_history:
# This will delete all messages with this session_id from MongoDB
mongodb.db[CHAT_COLLECTION].delete_many({"session_id": session_id})
# Remove from in-memory sessions
if session_id in sessions:
del sessions[session_id]
# Delete video file if it exists
video_files = [f for f in os.listdir(VIDEOS_DIR) if f.startswith(session_id)]
for file in video_files:
try:
os.remove(os.path.join(VIDEOS_DIR, file))
except:
pass
return {"message": f"Session {session_id} deleted successfully"}
@app.get("/")
async def root():
"""
API root endpoint
"""
return {
"message": "Video Transcription and QA API",
"endpoints": {
"/register": "Register a new user",
"/token": "Login and get access token",
"/transcribe": "Transcribe YouTube videos",
"/upload": "Upload and transcribe video files (max 20MB)",
"/download/{video_id}": "Download an uploaded video",
"/query": "Query the RAG system",
"/sessions": "List all user sessions",
"/sessions/{session_id}": "Get session information",
}
}
@app.on_event("shutdown")
def shutdown_event():
mongodb.close()
# Clean up temporary files
shutil.rmtree(VIDEOS_DIR, ignore_errors=True)
if __name__ == "__main__":
import uvicorn
os.environ["TOKENIZERS_PARALLELISM"] = "false" # Fix for the tokenizers warning
uvicorn.run(app, host="0.0.0.0", port=8000)