app.py

from flask import Flask, request, jsonify, render_template
from transformers import BlenderbotTokenizer, BlenderbotForConditionalGeneration

app = Flask(__name__)

system_prompt = (
    "You are a customer service bot for an e-commerce shop that sells many different kinds of products. "
    "Your role is to answer questions related to customer service and products only. "
    "Please provide accurate and helpful information within this scope. "
    "Always end your response with 'How can I help you today?' if you need further input from the customer. "
    "If a user asks to buy something, give them the price that you know, or provide an estimated price if you are unsure. "
    "Only answer the asked question correctly."
)

@app.route("/")
def home():
    return render_template("index.html")

@app.route("/chat", methods=["POST"])
def chat():
    user_input = request.json.get("message")
    response = get_response(user_input)
    return jsonify({"response": response})

def get_response(user_input):
    try:
        response = get_santiago_response(user_input)
        if response:
            return response
    except Exception as e:
        print(f"SantiagoPG model failed: {e}")
    
    # Fallback to BlenderBot model
    full_input = system_prompt + " " + user_input
    return get_blenderbot_response(full_input)

def get_santiago_response(user_input):
    try:
        from transformers import pipeline
        santiago_pipe = pipeline("text2text-generation", model="SantiagoPG/chatbot_customer_service")
        response = santiago_pipe(user_input)[0]['generated_text']
        return response
    except ImportError as e:
        print(f"Failed to import SantiagoPG model pipeline: {e}")
        return ""

def get_blenderbot_response(user_input):
    tokenizer = BlenderbotTokenizer.from_pretrained("facebook/blenderbot-400M-distill")
    model = BlenderbotForConditionalGeneration.from_pretrained("facebook/blenderbot-400M-distill")
    inputs = tokenizer(user_input, return_tensors="pt")
    reply_ids = model.generate(**inputs)
    response = tokenizer.batch_decode(reply_ids, skip_special_tokens=True)[0]
    return response

if __name__ == "__main__":
    app.run(debug=True)


















# from flask import Flask, request, jsonify, render_template
# from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline

# app = Flask(__name__)

# # Load SantiagoPG model
# santiago_pipe = pipeline("text2text-generation", model="SantiagoPG/chatbot_customer_service")

# # Load BlenderBot fallback model
# blenderbot_tokenizer = AutoTokenizer.from_pretrained("facebook/blenderbot-400M-distill")
# blenderbot_model = AutoModelForSeq2SeqLM.from_pretrained("facebook/blenderbot-400M-distill")

# system_prompt = (
#     "You are a customer service bot for an e-commerce shop that sells many different kinds of products. "
#     "Your role is to answer questions related to customer service and products only. "
#     "Please provide accurate and helpful information within this scope. "
#     "Always end your response with 'How can I help you today?' if you need further input from the customer. "
#     "If a user asks to buy something, give them the price that you know, or provide an estimated price if you are unsure. "
#     "Only answer the asked question correctly."
# )

# @app.route("/")
# def home():
#     return render_template("index.html")

# @app.route("/chat", methods=["POST"])
# def chat():
#     user_input = request.json.get("message")
#     response = get_response(user_input)
#     return jsonify({"response": response})

# def get_response(user_input):
#     try:
#         santiago_response = santiago_pipe(user_input)[0]['generated_text']
#         if santiago_response:
#             return santiago_response
#     except Exception as e:
#         print(f"SantiagoPG model failed: {e}")
    
#     # Fallback to BlenderBot model
#     full_input = system_prompt + " " + user_input
#     inputs = blenderbot_tokenizer(full_input, return_tensors="pt")
#     reply_ids = blenderbot_model.generate(**inputs)
#     response = blenderbot_tokenizer.batch_decode(reply_ids, skip_special_tokens=True)[0]
#     return response

# if __name__ == "__main__":
#     app.run(debug=True)
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    
    





# from flask import Flask, request, jsonify, render_template
# from transformers import pipeline, AutoTokenizer, AutoModelForSeq2SeqLM

# app = Flask(__name__)

# # Load SantiagoPG customer service model
# santiago_pipe = pipeline("text2text-generation", model="SantiagoPG/chatbot_customer_service")

# # Load BlenderBot fallback model
# tokenizer = AutoTokenizer.from_pretrained("facebook/blenderbot-400M-distill")
# model = AutoModelForSeq2SeqLM.from_pretrained("facebook/blenderbot-400M-distill")

# system_prompt = (
#     "You are a customer service bot for an e-commerce shop that sells many different kinds of products. "
#     "Your role is to answer questions related to customer service and products only. "
#     "Please provide accurate and helpful information within this scope. "
#     "Always end your response with 'How can I help you today?' if you need further input from the customer. "
#     "If a user asks to buy something, give them the price that you know, or provide an estimated price if you are unsure. "
#     "Only answer the asked question correctly."
# )

# @app.route("/")
# def home():
#     return render_template("index.html")

# @app.route("/chat", methods=["POST"])
# def chat():
#     user_input = request.json.get("message")
#     response = get_response(user_input)
#     return jsonify({"response": response})

# def get_response(user_input):
#     # Try the primary model first
#     santiago_response = santiago_pipe(user_input)[0]['generated_text']
#     if santiago_response:
#         return santiago_response
#     # Fallback to BlenderBot model
#     full_input = system_prompt + " " + user_input
#     inputs = tokenizer(full_input, return_tensors="pt")
#     reply_ids = model.generate(**inputs)
#     response = tokenizer.batch_decode(reply_ids, skip_special_tokens=True)[0]
#     return response

# if __name__ == "__main__":
#     app.run(debug=True)










# from flask import Flask, request, jsonify, render_template
# from transformers import BlenderbotTokenizer, BlenderbotForConditionalGeneration

# app = Flask(__name__)

# tokenizer = BlenderbotTokenizer.from_pretrained("facebook/blenderbot-400M-distill")
# model = BlenderbotForConditionalGeneration.from_pretrained("facebook/blenderbot-400M-distill")

# system_prompt = ( "You are a customer service bot for an ecommerce shop that sells many different kinds of products."
#                  "Your role is to answer questions related to customer service and products only. "
#                 "Please provide accurate and helpful information within this scope."
#                 "Always end your response with 'How can I help you today!' if you need an input from the customer"
#                 "If a user asks to buy something, give them the price that you know just any amount based on your training"
#                 "Only answer the asked question correctly"
#                 # "You must be helpful you are a customer service chatbot for a shop"
#                 # "if asked something that starts with 'I want..' take that as an input and answer correctly"
                
                 
                 
# )

# @app.route("/")
# def home():
#     return render_template("index.html")

# @app.route("/chat", methods=["POST"])
# def chat():
#     user_input = request.json.get("message")
#     response = get_blenderbot_response(user_input)
#     return jsonify({"response": response})

# def get_blenderbot_response(user_input):
#     full_input = system_prompt + " " + user_input
#     inputs = tokenizer(full_input, return_tensors="pt")
#     reply_ids = model.generate(**inputs)
#     response = tokenizer.batch_decode(reply_ids, skip_special_tokens=True)[0]
#     return response

# if __name__ == "__main__":
#     app.run(debug=True)