app.py

import os
import mimetypes
import requests
import tempfile
import gradio as gr
from openai import AzureOpenAI
import re
import json
from transformers import pipeline
import matplotlib.pyplot as plt
import plotly.express as px

client = AzureOpenAI()

class SentimentAnalyzer:
    def __init__(self):
        self.model="facebook/bart-large-mnli" 
        
        
    def analyze_sentiment(self, text):
        pipe = pipeline("zero-shot-classification", model=self.model)
        label=["positive","negative","neutral"]
        result = pipe(text, label)
        sentiment_scores= {result['labels'][0]:result['scores'][0],result['labels'][1]:result['scores'][1],result['labels'][2]:result['scores'][2]}
        sentiment_scores_str = f"Positive: {sentiment_scores['positive']:.2f}, Neutral: {sentiment_scores['neutral']:.2f}, Negative: {sentiment_scores['negative']:.2f}"
        return sentiment_scores_str
        
    def emotion_analysis(self,text):
        
        print(text,"lkdjrglk") 
        conversation = [
          {"role": "system", "content": """You are a Emotion Analyser.Your task is to analyze and predict the emotion using scores. Emotions are categorized into the following list: Sadness, Happiness, Joy, Fear, Disgust, and Anger. You need to provide the emotion with the highest score. The scores should be in the range of 0.0 to 1.0, where 1.0 represents the highest intensity of the emotion.
            Please analyze the text and provide the output in the following format: emotion: score [with one result having the highest score]."""},
          {"role": "user", "content": f"""
            input text{text}
            """}
            ]
        response = client.chat.completions.create(
            model="GPT-3",
            messages=conversation,
            temperature=1,
            max_tokens=60
        )

        message = response.choices[0].message.content
        print("sen",message)
        return message

    def analyze_sentiment_for_graph(self, text):
        pipe = pipeline("zero-shot-classification", model=self.model)
        label=["positive", "negative", "neutral"]
        result = pipe(text, label)
        sentiment_scores = {
            result['labels'][0]: result['scores'][0],
            result['labels'][1]: result['scores'][1],
            result['labels'][2]: result['scores'][2]
        }
        return sentiment_scores

    def emotion_analysis_for_graph(self,text):

        list_of_emotion=text.split(":")
        label=list_of_emotion[1]
        score=list_of_emotion[2]
        score_dict={
          label:float(score)
        }
        print(score_dict)
        return score_dict


class Summarizer:
    
    def __init__(self):
        # self.client = OpenAI()      
        pass
        
    def generate_summary(self, text):
        
        conversation = [
          {"role": "system", "content": "You are a Summarizer"},
          {"role": "user", "content": f"""summarize the following conversation delimited by triple backticks.
                     write within 30 words.
                     ```{text}```
            """}
            ]
        response = client.chat.completions.create(
            model="GPT-3",
            messages=conversation,
            temperature=1,
            max_tokens=100
        )

        message = response.choices[0].message.content
        return message

history_state = gr.State()
summarizer = Summarizer()
sentiment = SentimentAnalyzer()

class LangChain_Document_QA:

    def __init__(self):
        pass 
        
    def _add_text(self,history, text):

        history = history + [(text, None)]
        history_state.value = history

        return history, gr.Textbox(value="", interactive=False)

    def _agent_text(self,text,history):
        
        history[-1][1] = text
        history_state.value = history
        return history


    def _chat_history(self):
        print("chat history",history_state.value)
        history = history_state.value
        formatted_history = " "
        for entry in history:
            customer_text, agent_text = entry
            formatted_history += f"Customer: {customer_text}\n"
            if agent_text:
                formatted_history += f"Agent: {agent_text}\n"
        return formatted_history

    def _display_history(self):
        formatted_history=self._chat_history()
        summary=summarizer.generate_summary(formatted_history)
        return summary

    def _display_graph(self,sentiment_scores):
        labels = sentiment_scores.keys()
        scores = sentiment_scores.values()
        fig = px.bar(x=scores, y=labels, orientation='h', color=labels, color_discrete_map={"Negative": "red", "Positive": "green", "Neutral": "gray"})
        fig.update_traces(texttemplate='%{x:.2f}%', textposition='outside')
        fig.update_layout(height=500, width=200)
        return fig

    def _history_of_chat(self):
        history = history_state.value
        formatted_history = ""
        client=""
        agent=""
        for entry in history:
            customer_text, agent_text = entry
            client+=customer_text
            formatted_history += f"Customer: {customer_text}\n"
            if agent_text:
                agent+=agent_text
                formatted_history += f"Agent: {agent_text}\n"
        return client,agent


    def _suggested_answer(self,history,text):
            
            self._add_text(history,text)
            text_history = self._chat_history()
            file_path = "vodafone_customer_details.json"
            with open(file_path) as file:
                context = json.load(file)
            conversation = [
              {"role": "system", "content": f"You Are Vodafone Sim AI Chatbot. Use the following pieces of context{context} to answer the question at the end. If you don't know the answer, just say that you don't know, don't try to make up an answer. Use three sentences maximum. Keep the answer as concise as possible. if user say hi, hello you say welcome greetings like hi, hello. if user say thankyou, thanks tone you say thanking Greetings like You're welcome!.conversation history{context}"},
              {"role": "user", "content": f""" this is the user question:{text}..
                """}
                ]
            response = client.chat.completions.create(
                model="GPT-3",
                messages=conversation,
                temperature=0,
                max_tokens=100
            )
            
            message = response.choices[0].message.content
            print("message",message)
            
            if  ":" in message:
              message = re.sub(r'^.*:', '', message)
              history.append((text, message))
            else:
              history.append((text, message))
            return "",message
    

    def _text_box(self,customer_emotion,agent_emotion,agent_sentiment_score,customer_sentiment_score):
        agent_score = ", ".join([f"{key}: {value:.2f}" for key, value in agent_sentiment_score.items()])
        customer_score = ", ".join([f"{key}: {value:.2f}" for key, value in customer_sentiment_score.items()])
        return f"customer_emotion:{customer_emotion}\nagent_emotion:{agent_emotion}\nAgent_Sentiment_score:{agent_score}\nCustomer_sentiment_score:{customer_score}"

    def _on_sentiment_btn_click(self):
        client,agent=self._history_of_chat()

        customer_emotion=sentiment.emotion_analysis(client)
        customer_sentiment_score = sentiment.analyze_sentiment_for_graph(client)

        agent_emotion=sentiment.emotion_analysis(agent)
        agent_sentiment_score = sentiment.analyze_sentiment_for_graph(agent)

        scores=self._text_box(customer_emotion,agent_emotion,agent_sentiment_score,customer_sentiment_score)

        customer_fig=self._display_graph(customer_sentiment_score)
        customer_fig.update_layout(title="Sentiment Analysis",width=800)

        agent_fig=self._display_graph(agent_sentiment_score)
        agent_fig.update_layout(title="Sentiment Analysis",width=800)

        agent_emotion_score = sentiment.emotion_analysis_for_graph(agent_emotion)

        agent_emotion_fig=self._display_graph(agent_emotion_score)
        agent_emotion_fig.update_layout(title="Emotion Analysis",width=800)

        customer_emotion_score = sentiment.emotion_analysis_for_graph(customer_emotion)

        customer_emotion_fig=self._display_graph(customer_emotion_score)
        customer_emotion_fig.update_layout(title="Emotion Analysis",width=800)

        return scores,customer_fig,agent_fig,customer_emotion_fig,agent_emotion_fig


    def clear_func(self):
      history_state = gr.State([])


    def gradio_interface(self):
      with gr.Blocks(css="style.css",theme="snehilsanyal/scikit-learn") as demo:

          with gr.Row():
            gr.HTML("""<center><img src="https://hakunamatatatech.com/wp-content/uploads/2019/06/logohms-160x4.png" alt="Image"></center>""")
          with gr.Row():  
            gr.HTML("""<center><h1 style="color:#f26d25">Generative AI CRM ChatBot</h1></center>""")
          with gr.Row():
            gr.HTML("<br>")
          chatbot = gr.Chatbot().style(height=300)
          with gr.Row():
              with gr.Column(scale=0.50):
                  txt = gr.Textbox(
                      show_label=False,
                      placeholder="Customer",
                  ).style(container=False)
              with gr.Column(scale=0.50):
                  txt2 = gr.Textbox(
                      show_label=False,
                      placeholder="Agent",
                  ).style(container=False)

              with gr.Column(scale=0.40):
                  txt3 =gr.Textbox(
                      show_label=False,
                      placeholder="Chatbot_Suggestion",
                  ).style(container=False)
              with gr.Column(scale=0.10, min_width=0):
                  button=gr.Button(
                      value="🚀"
                  )
              with gr.Row(scale=0.50):
                  emptyBtn = gr.Button(
                      "🧹 New Conversation",
                  )
          with gr.Row():
              with gr.Column(scale=0.40):
                  txt4 =gr.Textbox(
                      show_label=False,
                      lines=4,
                      placeholder="Summary",
                  ).style(container=False)
              with gr.Column(scale=0.10, min_width=0):
                  end_btn=gr.Button(
                      value="End"
                  )
              with gr.Column(scale=0.40):
                  txt5 =gr.Textbox(
                      show_label=False,
                      lines=4,
                      placeholder="Sentiment",
                  ).style(container=False)

              with gr.Column(scale=0.10, min_width=0):
                  Sentiment_btn=gr.Button(
                      value="📊",callback=self._on_sentiment_btn_click
                  )
          with gr.Row():
            gr.HTML("""<center><h1 style="color:#fff">Sentiment and Emotion Score Graph</h1></center>""")
          with gr.Row():
              with gr.Column(scale=0.70, min_width=0):
                  plot =gr.Plot(label="Customer", size=(500, 600))
          with gr.Row():
              with gr.Column(scale=0.70, min_width=0):
                  plot_2 =gr.Plot(label="Agent", size=(500, 600))
          with gr.Row():
              with gr.Column(scale=0.70, min_width=0):
                  plot_3 =gr.Plot(label="Customer_Emotion", size=(500, 600))
          with gr.Row():
              with gr.Column(scale=0.70, min_width=0):
                  plot_4 =gr.Plot(label="Agent_Emotion", size=(500, 600))

          txt_msg = txt.submit(self._add_text, [chatbot, txt], [chatbot, txt])
          # txt_msg.then(lambda: gr.update(interactive=True), None, [txt])
          txt.submit(self._suggested_answer, [chatbot, txt], [txt,txt3])
          button.click(self._agent_text, [txt3,chatbot], chatbot)
          txt2.submit(self._agent_text, [txt2,chatbot ], chatbot).then(
              self._agent_text, [txt2,chatbot], chatbot
          )
          end_btn.click(self._display_history, [], txt4)
          emptyBtn.click(self.clear_func,[],[])
          emptyBtn.click(lambda: None, None, chatbot, queue=False)

          Sentiment_btn.click(self._on_sentiment_btn_click,[],[txt5,plot,plot_2,plot_3,plot_4])

      demo.launch(debug = True)

document_qa =LangChain_Document_QA()
document_qa.gradio_interface()