response_read.py

import os
from langchain.llms import OpenAI
from langchain import PromptTemplate
from langchain.chains import LLMChain
import time
import pyttsx3 as sp
from constant import openai_key
os.environ['OPENAI_API_KEY'] = openai_key


def generated_qs():
    ''' Returns already written questions in form of a python list '''
    global cleaned_contents
    with open('questions.txt', 'r') as file:
        file_contents = file.readlines()

    cleaned_contents = [line[3:].rstrip('\n') for line in file_contents if line != '\n']
    return cleaned_contents


def speak_qs(content):
    ''' Speaks question that is given as a parameter '''
    tts = sp.init()
    tts.setProperty('rate', 150)
    voices = tts.getProperty('voices')
    tts.setProperty('voice', voices[0].id) # 0 for male and 1 for female
    tts.say(content)
    tts.runAndWait()


def correct_or_not(content,answer):
    ''' Accepts question and answer as parameters and returns whether answer is correct or not wrt to the question '''
    llm = OpenAI(temperature=0.8)
 
    first_ans = PromptTemplate(
        input_variables = ['qs','ans'] ,
        template='Given the question "\{qs}"\, how accurate do you believe this answer "\{ans}"\ is on a percentage scale, make sure to only return percentage and nothing else?')

    correct_per = LLMChain(llm=llm , prompt=first_ans,verbose=True) 
    response = correct_per.run(qs=content , ans = answer)
    return response

 
def get_answer_from_gpt(question):
    '''  Takes a question and returns answer from chatmodel '''
    llm = OpenAI(temperature=0.8)

    gpt_ans = PromptTemplate(
        input_variables = ['qos'] ,
        template='Given the question "\{qos}"\ what do you think the answer should be? Summarize the answer in one paragraph')
    ans2 = LLMChain(llm=llm , prompt=gpt_ans,verbose=True) 
    response2 = ans2.run(qos=question)
    return response2


def clear_text_file(file_path):
    try:
        with open(file_path, 'w') as file:
            file.truncate(0)
        print(f"Content of {file_path} has been cleared.")
    except Exception as e:
        print(f"An error occurred: {e}")


def clean_and_convert_percentage_strings(percentage_strings):
    ''' Takes a list of strings, each containing percentage and returns only list of integer  '''
    cleaned_integers = []
    for string in percentage_strings:
        cleaned_string = string.strip('%')  
        integer_value = int(cleaned_string)  
        cleaned_integers.append(integer_value)
    return cleaned_integers


def calculate_overall_performance(accuracy_scores):
    ''' Takes a list of floats and returns the total score '''
    total_weight = len(accuracy_scores) * 10  # Each question has a weight of 10
    
    weighted_sum = sum(accuracy * 10 for accuracy in accuracy_scores)
    
    overall_performance = weighted_sum / (total_weight+1)
    return overall_performance



def final_evaluation(total_score):
    ''' Accepts total score as parameters and returns a response if score is good enough to pass the interview or not '''

    llm = OpenAI(temperature=0.8)

    first_ans = PromptTemplate(
        input_variables = ['tot_acc'] ,
        template = 'We took an interview from a person, and asked some questions, the person score {tot_acc} out of 100 as an average score, write me a brief summary for the interview.')
        
    per_response = LLMChain(llm=llm , prompt=first_ans,verbose=True) 
    response_f = per_response.run(tot_acc = round(total_score,2))
    return response_f


def sophisticated_response(res_list):
    sop_res_dic = {}
    number = calculate_overall_performance(res_list)
    f_resp = final_evaluation(number)
    sop_res_dic['evaluation'] = round(number,2)
    sop_res_dic['evaluation_message'] = f_resp.replace('\n', '')
    if round(number,2) <= 65:
        sop_res_dic['result'] = 'Fail'
    else:
        sop_res_dic['result'] = 'Pass'
    return sop_res_dic