app.py

import json
import requests
import wikipedia
import gradio as gr
import transformers
import spacy
from bs4 import BeautifulSoup
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM


# Set up Google SERP API credentials
serp_api_key = '5924c6cfe5fec240e39838ff06439c8d36d294a0'  # Replace with your actual Google SERP API key

 # Load the Pegasus model
tokenizer = AutoTokenizer.from_pretrained("facebook/bart-large-cnn")
model = AutoModelForSeq2SeqLM.from_pretrained("facebook/bart-large-cnn")

# Function to send a message and receive a response from the chatbot
def chat(message):
    try:
        # You can add your chatbot implementation here
        return "."
    except Exception as e:
        print("An error occurred:", e)
        return ""

# Function to get the latest answers from Google SERP API
def get_latest_answers(query):
    url = "https://google.serper.dev/search"

    payload = json.dumps({
        "q": query
    })
    headers = {
        'X-API-KEY': serp_api_key,
        'Content-Type': 'application/json'
    }

    response = requests.request("POST", url, headers=headers, data=payload)

    try:
        # Parse the response JSON
        data = json.loads(response.text)

        # Extract details from the response
        output = ""

        if 'knowledgeGraph' in data:
            knowledge_graph = data['knowledgeGraph']
            output += "Website: {}\n".format(knowledge_graph.get('website'))
            output += "Description: {}\n".format(knowledge_graph.get('description'))

        if 'organic' in data:
            organic_results = data['organic']
            for result in organic_results:
                output += "Snippet: {}\n".format(result.get('snippet'))

        if 'peopleAlsoAsk' in data:
            people_also_ask = data['peopleAlsoAsk']
            for question in people_also_ask:
                output += "Snippet: {}\n".format(question.get('snippet'))

        return output

    except json.JSONDecodeError:
        print(".")
        return ""

    except Exception as e:
        print(".")
        return ""

# Function to search Wikipedia for an answer and summarize it
def search_wikipedia(query):
    try:
        search_results = wikipedia.search(query)

        # Get the page summary of the first search result
        if search_results:
            page_title = search_results[0]
            page_summary = wikipedia.summary(page_title)
            return page_summary
        else:
            print(".")
            return None
    except wikipedia.exceptions.DisambiguationError as e:
        # Handle disambiguation error
        print(".")
        return None
    except wikipedia.exceptions.PageError as e:
        # Handle page not found error
        print(".")
        return None
    except Exception as e:
        # Handle other exceptions
        print(".")
        return None

# Function to generate summarized paragraph using Google Pegasus summarization
def generate_summary(user_input):
    output = get_latest_answers(user_input)
    page_summary = search_wikipedia(user_input)
    chat_answer = chat(user_input)

    # Combine the input text from various sources
    input_text = f"Google:\n{output}\nWikipedia :\n{page_summary}\n"

  
    # Tokenize and generate a summary
    input_ids = tokenizer.encode(input_text, return_tensors="pt", max_length=1024, truncation=True)
    summary_ids = model.generate(input_ids, max_length=200, min_length=50, length_penalty=2.0, num_beams=4, early_stopping=True)

    # Decode the summary
    summarized_paragraph = tokenizer.decode(summary_ids[0], skip_special_tokens=True)

    return summarized_paragraph

# Define the Gradio interface
def summarizer_interface(user_input):
    summarized_text = generate_summary(user_input)
    return summarized_text

iface = gr.Interface(
    fn=summarizer_interface,
    inputs="text",
    outputs="text",
    title="Osana Web-GPT",
    description="Enter your query and get the latest and better answer.",
    theme="black",
    layout="horizontal",
)

# Launch the interface
iface.launch()