simple code

.gitignore

@@ -0,0 +1,5 @@
+.venv
+.env
+__pycache__/
+.vscode
+venv

Dockerfile

@@ -0,0 +1,14 @@
+# Use the official Python image
+FROM python:3.11-slim
+
+# Set the working directory
+WORKDIR /app
+
+# Copy the current directory contents into the container
+COPY .
+
+# Install the dependencies
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Command to run the FastAPI app
+CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]

app/ai_responder.py

@@ -0,0 +1,114 @@
+# # from langchain_ollama import ChatOllama
+# from langchain_core.prompts import PromptTemplate
+# import os
+
+# ollama_url = os.getenv("OLLAMA_BASE_URL")
+
+# # llm = ChatOllama(model="mistral-nemo", temperature=0, base_url=ollama_url)
+# from langchain_groq import ChatGroq
+
+# llm = ChatGroq(
+#     model="llama-3.3-70b-versatile",
+#     temperature=0.2,
+#     api_key= "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",
+# )
+
+# qa_prompt = PromptTemplate(
+#     input_variables=["question"],
+#     template="""You are a helpful AI assistant. Answer the user's question in a friendly and shortly manner.
+#     Question: {question}
+#     """
+# )
+# qa_chain = qa_prompt | llm
+
+# async def ai_answer(transcription_text):
+#     if transcription_text.strip() == "":
+#         return "I'm sorry, I didn't catch that. Could you please repeat?"
+#     ai_response = qa_chain.invoke({"question": transcription_text}).content
+#     print(f"AI Response: {ai_response}")
+#     return ai_response
+
+
+from langchain_core.prompts import PromptTemplate
+import os
+from langchain_core.output_parsers import PydanticOutputParser
+from typing import Optional, List, Dict
+from pydantic import BaseModel, Field
+
+class AIResponseModel(BaseModel):
+    response: Optional[str] = Field(description="The AI's response to the user's query")
+
+
+classify_parser = PydanticOutputParser(pydantic_object=AIResponseModel)
+
+ollama_url = os.getenv("OLLAMA_BASE_URL")
+
+from langchain_groq import ChatGroq
+
+llm = ChatGroq(
+    model="llama-3.3-70b-versatile",
+    temperature=0.2,
+    api_key= "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",
+)
+
+# Simple in-memory conversation history
+conversation_memory = {}
+
+qa_prompt = PromptTemplate(
+    input_variables=["history", "question"],
+    template="""You are a helpful AI assistant named CentralogicAI Bot. Continue the conversation using the history below.
+    Answer to the user's query in short and friendly manner.
+    Also observe the user's intent and if the intent is to connect with the sales teams or person, then return a word "sales" in the response.
+    Return the response in proper JSON format with the following keys
+    "response":""
+
+    Conversation history:
+    {history}
+    User: {question}
+    Assistant:"""
+    )
+qa_chain = qa_prompt | llm | classify_parser
+
+# async def ai_answer(transcription_text, session_id="1234"):
+#     if transcription_text.strip() == "":
+#         return "I'm sorry, I didn't catch that. Could you please repeat?"
+
+#     # Get history for this session
+#     history = conversation_memory.get(session_id, [])
+
+#     # Prepare the history string (do NOT include the current input yet)
+#     history_str = "\n".join(history[-5:])  # Keep last 5 exchanges for brevity
+
+#     # Get AI response
+#     ai_response = qa_chain.invoke({"history": history_str, "question": transcription_text}).content
+#     print(f"AI Response: {ai_response}")
+
+#     # Now add the new user input and AI response to history
+#     history.append(f"User: {transcription_text}")
+#     history.append(f"Assistant: {ai_response}")
+#     conversation_memory[session_id] = history
+
+#     return ai_response
+
+def ai_answer(transcription_text, session_id="12345"):
+    if transcription_text.strip() == "":
+        return "I'm sorry, I didn't catch that. Could you please repeat?"
+
+    # Get history for this session
+    history = conversation_memory.get(session_id, [])
+
+    # Prepare the history string (do NOT include the current input yet)
+    history_str = "\n".join(history[-5:])  # Keep last 5 exchanges for brevity
+
+    # Get AI response
+    ai_response = qa_chain.invoke({"history": history_str, "question": transcription_text})
+    # print(f"AI Response: {ai_response}")
+
+    # Now add the new user input and AI response to history
+    history.append(f"User: {transcription_text}")
+    history.append(f"Assistant: {ai_response}")
+    conversation_memory[session_id] = history
+
+    return ai_response.response
+
+# print(ai_answer("Can you please connect me with the sales team?").response)  

app/main.py

@@ -0,0 +1,5 @@
+from fastapi import FastAPI
+from app.routes import router as api_router
+
+app = FastAPI()
+app.include_router(api_router)

app/routes.py

@@ -0,0 +1,28 @@
+from fastapi import APIRouter, WebSocket, Request
+from app.twilio_handler import handle_incoming_call
+from app.websocket_manager import websocket_endpoint
+
+router = APIRouter()
+
+
+# @router.post("/twilio/call")
+# async def twilio_call(request: Request):
+#     return await handle_incoming_call(request)
+
+
+@router.post("/incoming-call")
+async def twilio_call(request: Request):
+    return await handle_incoming_call(request)
+
+@router.websocket("/ws/{session_id}")
+async def websocket_route(websocket: WebSocket, session_id: str):
+    print("Websocket session request...")
+    await websocket_endpoint(websocket, session_id)
+
+
+# ngrok http --url=natural-woodcock-bursting.ngrok-free.app 8000
+
+# number 1= 719 402-2777
+# number 2= 914 601 2553
+
+# number 3=  812 608 4048

app/session_manager.py

@@ -0,0 +1,12 @@
+import uuid
+
+sessions = {}
+
+def create_session(call_sid: str, from_number: str) -> str:
+    session_id = str(uuid.uuid4())
+    sessions[session_id] = {"call_sid": call_sid, "from": from_number}
+    return session_id
+
+def get_session(session_id: str):
+    print(f"Current available sessions: {sessions}")
+    return sessions.get(session_id)

app/twilio_handler.py

@@ -0,0 +1,154 @@
+# from fastapi import Request, Response
+# from twilio.twiml.voice_response import VoiceResponse, Connect
+# from app.session_manager import create_session
+# import os
+# from dotenv import load_dotenv
+# load_dotenv()  
+
+# ws_url = os.getenv("WS_URL")  
+
+# async def handle_incoming_call(request: Request):
+#     form = await request.form()
+#     call_sid = form.get('CallSid')
+#     print(f"Incoming call SID: {call_sid}")
+#     from_number = form.get('From')
+
+#     # Create a new session for this call
+#     session_id = create_session(call_sid, from_number)
+
+#     # Construct the TwiML response
+#     response = VoiceResponse()
+#     connect = Connect()
+#     connect.conversation_relay(
+#         url=f"{ws_url}/{session_id}",
+#         welcome_greeting="Hello! I am CentraLogic AI bot. How can I assist you today?",
+#         language="en-US",
+#         transcription_language="en-US",
+#         tts_provider="ElevenLabs",
+#         voice="XjLkpWUlnhS8i7gGz3lZ",
+#         tts_language="en-US",
+#         interruptible="false",
+#         # transcription_provider="Deepgram",
+#     )
+#     response.append(connect)
+
+#     return Response(content=str(response), media_type="application/xml")
+
+
+# ######### Working code
+# from fastapi import Request, Response
+# from twilio.twiml.voice_response import VoiceResponse, Connect
+# from app.session_manager import create_session
+# import os
+# from dotenv import load_dotenv
+# load_dotenv()  
+
+# ws_url = os.getenv("WS_URL")  
+
+# async def handle_incoming_call(request: Request):
+#     form = await request.form()
+#     call_sid = form.get('CallSid')
+#     print(f"Incoming call SID: {call_sid}")
+#     from_number = form.get('From')
+
+#     # Create a new session for this call
+#     session_id = create_session(call_sid, from_number)
+
+#     # Construct the TwiML response
+#     response = VoiceResponse()
+#     connect = Connect()
+#     connect.conversation_relay(
+#         url=f"{ws_url}/{session_id}",
+#         welcome_greeting="Hello! I am CentraLogic AI bot. How can I assist you today?",
+#         language="en-US",
+#         transcription_language="en-US",
+#         tts_language="en-US",
+#         interruptible="false",
+#         debug=True,
+#     )
+#     response.append(connect)
+
+#     return Response(content=str(response), media_type="application/xml")
+
+
+
+#  ##### new code
+from fastapi import Request, Response
+from twilio.twiml.voice_response import VoiceResponse, Connect
+from app.session_manager import create_session
+import os
+from dotenv import load_dotenv
+load_dotenv()  
+
+ws_url = os.getenv("WS_URL")  
+
+async def handle_incoming_call(request: Request):
+    form = await request.form()
+    call_sid = form.get('CallSid')
+    print(f"Incoming call SID: {call_sid}")
+    from_number = form.get('From')
+    dial = form.get('Dial')
+    if dial:
+        if dial:
+            xml = f"""<?xml version="1.0" encoding="UTF-8"?>
+                    <Response>
+                    <Dial>{dial}</Dial>
+                    </Response>"""
+            return Response(content=xml, media_type="text/xml")
+        else:
+            # Create a new session for this call
+            session_id = create_session(call_sid, from_number)
+
+            # Construct the TwiML response
+            response = VoiceResponse()
+            connect = Connect()
+            connect.conversation_relay(
+                url=f"{ws_url}/{session_id}",
+                welcome_greeting="Hello! I am CentraLogic AI bot. How can I assist you today?",
+                language="en-US",
+                transcription_language="en-US",
+                tts_language="en-US",
+                interruptible="false",
+                debug=True,
+            )
+            response.append(connect)
+
+            return Response(content=str(response), media_type="application/xml")
+
+
+######### respone = VoiceResponse()
+######### response.dial()
+
+# from app.session_manager import create_session
+# from fastapi import Request, Response
+# from twilio.twiml.voice_response import Connect, ConversationRelay, Language, VoiceResponse
+# from dotenv import load_dotenv
+# load_dotenv()
+# import os
+
+# ws_url = os.getenv("WS_URL")  
+
+# async def handle_incoming_call(request: Request):
+#     form = await request.form()
+#     call_sid = form.get('CallSid')
+#     print(f"Incoming call SID: {call_sid}")
+#     from_number = form.get('From')
+
+#     # Create a new session for this call
+#     session_id = create_session(call_sid, from_number)
+
+#     # Construct the TwiML response
+#     response = VoiceResponse()
+#     connect = Connect()
+#     connect.conversation_relay(
+#         url=f"{ws_url}/{session_id}",
+#         welcome_greeting="Hello! I am CentraLogic AI bot.",
+#     )
+#     response.append(connect)
+
+#     return Response(content=str(response), media_type="application/xml")
+
+
+
+# transcriptionProvider, speechModel, transcriptionLanguage
+# Twilio uses a range of providers, including Google, Amazon, and Deepgram, for speech-to-text conversions

app/websocket_manager.py

@@ -0,0 +1,71 @@
+from fastapi import WebSocket, WebSocketDisconnect
+from app.ai_responder import ai_answer
+from app.session_manager import get_session
+import json
+
+SALES_NUMBER = "+91 7249477972"
+
+class ConnectionManager:
+    def __init__(self):
+        self.active_connections = {}
+
+    async def connect(self, websocket: WebSocket, session_id: str):
+        await websocket.accept()
+        self.active_connections[session_id] = websocket
+
+    def disconnect(self, session_id: str):
+        self.active_connections.pop(session_id, None)
+
+    async def receive_audio(self, session_id: str):
+        websocket = self.active_connections.get(session_id)  
+        session = get_session(session_id)
+        if not session:
+            print(f"Session {session_id} not found, disconnecting...")
+        print(f"current active connections: {self.active_connections}")
+        if websocket:
+            try:
+                while True:
+                    message = await websocket.receive_json()
+                    print(f"Received message: {message}")
+                    message_type = message.get("type")
+                    print(f"Received message type: {message_type}")
+
+                    if message_type == "setup": 
+                        print(f"This is the setup message:- {message}")                  
+                        print(f"Setup message received for session {session_id}.")
+
+                    elif message_type == "prompt":
+                        user_input = message["voicePrompt"]
+                        print(f"Received transcription:- {user_input}")
+                        ai_response = await ai_answer(user_input)
+                        lower = ai_answer.lower()
+                        if "sales" in lower:
+                            dept = SALES_NUMBER
+                        else:
+                            dept = None
+                        
+                        if dept:
+                            await websocket.send_text(json.dumps({
+                                "type": "handoff",
+                                "dialNumber": dept
+                            }))
+                            break  # end session so Twilio exits ConversationRelay
+                        await websocket.send_json({
+                            "type": "text",
+                            "token": ai_response,
+                            "last": True  # Assuming this is the last message in the conversation
+                        })
+
+                    elif message_type == "interrupt":
+                        print(f"Interrupt message received for session {session_id}.")
+                        # break
+                        continue
+                    
+            except WebSocketDisconnect:
+                self.disconnect(session_id)
+
+
+manager = ConnectionManager()
+async def websocket_endpoint(websocket: WebSocket, session_id: str):
+    await manager.connect(websocket, session_id)
+    await manager.receive_audio(session_id)

requirements.txt

@@ -0,0 +1,10 @@
+fastapi[all]==0.109.0
+uvicorn[standard]==0.27.0
+gunicorn==21.2.0
+twilio
+python-dotenv
+ffmpeg-python
+langchain-ollama
+langchain-core
+numpy
+langchain_groq

simple.py

@@ -0,0 +1 @@
+print("Hello")