summarization.py

import cohere
import streamlit as st
import os
import requests
import webbrowser
from bs4 import BeautifulSoup
import json
import re


co = cohere.Client(os.environ["COHERE_API_KEY"]) 
def formattingForSummarizer(text):
    for each in text :
        if (each == "'") :
            text = text.replace(each, "")
        if(each == "`"):
            text = text.replace(each, "")    
    
    text = text.replace('\n', ' ').replace('\r', '').replace('\t', ' ')
    return text


def summarizer (text):

    CleanText = formattingForSummarizer(str(text))
    # prompt_template = "Summarize the given content into 5 paragraph response of `MORE THAN 200 WORDS` each. The content is : " + ' ' + CleanText 
    # summarizer_prompt  = "You are the manager of a hotel and you're task is to summarize the given content: that is the details of booking into the format needed for billing.. "
    summarizer_prompt  = "Summarize the given content. This is a conversation between a customer and hotel receptionist. Do not loose details like name, number, email, destination, check-in date, check-out date."
    response = co.summarize( 
          text=CleanText,
          length='long',
          format='bullets',
          model='summarize-xlarge',
          additional_command= summarizer_prompt,
          temperature=0.3,
        ) 
    print(response.summary)
    return response.summary

# def generateKBase(largeData):

#     rqdFormat =  [
#         {
#         "title": " ",
#         "snippet": " "
#         },
#     ]
#     FormatPrompt = "You should extract the given details text: " + largeData +" into this  \n: "+ " "+ str(rqdFormat) + "\n JSON FORMAT and populate the corresponding values from text : The snippet can contain the large amount of tokens.Don't shortent the content" 
#     response = co.generate(
#         # text,
#         model='command-nightly',
#         prompt=FormatPrompt,
#         temperature=0.3,
#         return_likelihoods =  None,
#         # finish_reason= COMPLETE,
#         # token_likelihoods= None,
#     )
#     print(response.generations[0].text)
#     # sendAPIReg(response.generations[0].text)



def generateDetails(text):

    JSONFormat = {
        "name": "Ron Tennyson",
        "number": 1234567890,
        "email": "david@jojo.com",
        "destination": "Kochi",
        "check_in_date": "2023-02-6",
        "check_out_date": "2023-02-10",
        "unique_id": "ron890"
    }


    JSONFormatPrompt = f"""You should extract the given details from the text given below into the following JSON FORMAT: 
    Here is the JSON dta with dummy values for example:  
    {JSONFormat}
    Populate corresponding values from text below, If you couldn't find all the answer to the fileds then do fill it with dummy values instead. 
    Use this format for datess instead: YYYY-MM-DD. UNIQUE ID is generated by taking the firt 3 letters of cutomer's name and last 3 digits of their phone number. 
    DO NOT LEAVE ANY FIELD BLANK AND NO OTHER OUTPUT REQUIRED.
    Here is the text: {text}"""    
    response = co.generate(
        # text,
        model='command-nightly',
        prompt=JSONFormatPrompt,
        temperature=0.3,
        return_likelihoods =  None,
    )
    # st.write("The API Response to send : ")
    # st.write(response.generations[0].text)
    print(response.generations[0].text) 
    sendAPIReg(response.generations[0].text)


def fillEmptyFields(json_data):
    required_fields = ["name", "number", "email", "destination", "check_in_date", "check_out_date", "unique_id"]

    dummy_values = {
        "name": "UNKNOWN",
        "number":0000000,
        "email": "unknown@example.com",
        "destination": "UNKNOWN",
        "check_in_date": "1900-01-01",
        "check_out_date": "1900-01-01",
        "unique_id": "UNKNOWN"
    }

    for field in required_fields:
        if field not in json_data or not json_data[field] or json_data[field].strip() == "":
            json_data[field] = dummy_values.get(field, "UNKNOWN")

    return json_data


def ifJSONComplete(incoming_json):
    required_fields = ["name", "number", "email", "destination", "check_in_date", "check_out_date", "unique_id"]
    
    for field in required_fields:
        if field not in incoming_json or incoming_json[field] in ["", " "]:
            return False
        
    print("________________JSON is complete____________")
    return True


# Function to take only the JSON from input string
def getJSONReponse(input_string):
    
    start_index = input_string.find('{')
    end_index = input_string.find('}') + 1
    json_string = input_string[start_index:end_index]

    json_data = json.loads(json_string)
    print("\nExtracted JSON: \n")
    print(json.dumps(json_data, indent=2))


    return json_data


def sendAPIReg(response_string):

    payload = {}
    if (ifJSONComplete(getJSONReponse(response_string))):
        payload = getJSONReponse(response_string)

    else :
        print("_______JSON is not complete_______")
        payload = fillEmptyFields(payload)
        print("_______JSON completeddddd_______")
        print(payload)

    st.write("The API Response to send : \n", payload)


    
    api_endpoint = "https://travel-llm-api.vercel.app/checkout/"
    # Make a POST request to the API endpoint
    response = requests.post(api_endpoint, data=payload)

    # Check the status code of the response
    if response.status_code == 201:
        api_response = json.loads(response.text)
        link = api_response.get("link", None)
        print(link)
    
        webbrowser.open(link)
        # print("API Response: \n ", response.text)
    else:
        # API call failed, print the error status code and response content
        print("API Error:", response.status_code, response.text)


def main():
    text = """The customer wants to book a 1-bedroom suite for 2 days
The check-in date is from 13th September
The suite costs ₹12600 and comes with a king bed, executive lounge access, a shower/tub combination, and amenities like high-speed internet and 2 TVs.
Nothing was mentioned about extras."""

    # generateKBase(data)

    JSONFormat = {
        "name": "TEST DATA ",
        "number": 1234567890,
        "email": "12321@gmail.com",
        "destination": " ",
        "check_in_date": "2002-02-6",
        "check_out_date": "2002-02-6",
        "unique_id": "abc123"
    }


    # sendAPIReg(JSONFormat)
    # if ifJSONComplete(JSONFormat) :
    #     print("JSON is complete")
    # else :
    #     print("JSON is not complete")
    generateDetails(text)

# Add the summary
if __name__ == "__main__":
    main()