app.py

import streamlit as st
import plotly.graph_objects as go
from transformers import pipeline
import re
import time
import requests
from PIL import Image
import itertools
import numpy as np
import matplotlib.pyplot as plt
from matplotlib.colors import rgb2hex
import matplotlib
from matplotlib.colors import ListedColormap, rgb2hex
import ipywidgets as widgets
from IPython.display import display, HTML
import re
import pandas as pd
from pprint import pprint
from tenacity import retry
from tqdm import tqdm
# import tiktoken
import scipy.stats
import torch
from transformers import GPT2LMHeadModel
# import tiktoken
import seaborn as sns
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
from colorama import Fore, Style
# import openai


para_tokenizer = AutoTokenizer.from_pretrained("humarin/chatgpt_paraphraser_on_T5_base")
para_model = AutoModelForSeq2SeqLM.from_pretrained("humarin/chatgpt_paraphraser_on_T5_base")

def paraphrase(
    question,
    num_beams=5,
    num_beam_groups=5,
    num_return_sequences=5,
    repetition_penalty=10.0,
    diversity_penalty=3.0,
    no_repeat_ngram_size=2,
    temperature=0.7,
    max_length=64 #128
):
    input_ids = para_tokenizer(
        f'paraphrase: {question}',
        return_tensors="pt", padding="longest",
        max_length=max_length,
        truncation=True,
    ).input_ids
    
    outputs = para_model.generate(
        input_ids, temperature=temperature, repetition_penalty=repetition_penalty,
        num_return_sequences=num_return_sequences, no_repeat_ngram_size=no_repeat_ngram_size,
        num_beams=num_beams, num_beam_groups=num_beam_groups,
        max_length=max_length, diversity_penalty=diversity_penalty
    )

    res = para_tokenizer.batch_decode(outputs, skip_special_tokens=True)

    return res


def find_longest_common_sequences(main_sentence, paraphrases):
    main_tokens = main_sentence.split()
    common_sequences = set()

    for paraphrase in paraphrases:
        paraphrase_tokens = paraphrase.split()
        for i in range(len(main_tokens)):
            for j in range(len(paraphrase_tokens)):
                # Start comparing pairs of words
                m = i
                n = j
                while m < len(main_tokens) and n < len(paraphrase_tokens) and main_tokens[m] == paraphrase_tokens[n]:
                    m += 1
                    n += 1
                # If we found a longer common sequence, update it
                if m - i > 1:
                    sequence = ' '.join(main_tokens[i:m])
                    is_subsequence = any(sequence in existing_seq for existing_seq in common_sequences)
                    if not is_subsequence:
                        common_sequences.add(sequence)

    return sorted(common_sequences, key=len, reverse=True)




prompt_list=["The official position of the United States on the Russia-Ukraine war has been consistent in supporting Ukraine's sovereignty, territorial integrity, and the peaceful resolution of the conflict."    
,"Joe Biden said we’d not send U.S. troops to fight Russian troops in Ukraine, but we would provide robust military assistance and try to unify the Western world against Russia’s aggression."]

options = [f"Prompt #{i+1}: {prompt_list[i]}" for i in range(len(prompt_list))] + ["Another Prompt..."]
selection = st.selectbox("Choose a prompt from the dropdown below . Click on :blue['Another Prompt...'] , if you want to enter your own custom prompt.", options=options)
check=[]

if selection == "Another Prompt...": 
    check = st.text_input("Enter your custom prompt...")
    check = " " + check
    if check:
        st.caption(f""":white_check_mark: Your input prompt is : {check}""")
        st.caption(':green[Kindly hold on for a few minutes while the AI text is being generated]')
    
else:    
    check = re.split(r'#\d+:', selection, 1)[1]
    if check:
        st.caption(f""":white_check_mark: Your input prompt is : {check}""")
        st.caption(':green[Kindly hold on for a few minutes while the Paraphrase texts are being generated]')


main_sentence = check

st.markdown("**Main Sentence**:")
st.write(main_sentence)

# Generate paraphrases
paraphrases = paraphrase(main_sentence)

# Extracting longest common sequences
longest_common_sequences = find_longest_common_sequences(main_sentence, paraphrases)

color_palette = ["#FF0000", "#008000", "#0000FF", "#FF00FF", "#00FFFF"]
highlighted_sentences = []

# Highlighting sequences in main sentence and paraphrases
for sentence in [main_sentence] + paraphrases:
    highlighted_sentence = sentence
    for i, sequence in enumerate(longest_common_sequences):
        color = color_palette[i % len(color_palette)]
        highlighted_sentence = highlighted_sentence.replace(sequence, f"<span style='color:{color}'>{sequence}</span>")
    highlighted_sentences.append(highlighted_sentence)

# Display paraphrases with numbers
st.markdown("**Paraphrases**:")
for i, para in enumerate(paraphrases, 1):
    st.write(f"Paraphrase {i}:")
    st.write(para)


# Displaying the main sentence with highlighted longest common sequences
st.markdown("**Main sentence with highlighted longest common sequences**:")
st.markdown(highlighted_sentences[0], unsafe_allow_html=True)


st.markdown("**Paraphrases with highlighted longest common sequences**:")
for paraphrase in highlighted_sentences[1:]:
    st.markdown(paraphrase, unsafe_allow_html=True)