import os

import numpy as np
import gradio as gr
from sklearn.feature_extraction.text import TfidfVectorizer
from sklearn.metrics.pairwise import cosine_similarity

shanty=os.environ.get('SHANTY')
def compute_cosine_similarity(text1, text2):
    # Initialize the TfidfVectorizer
    tfidf_vectorizer = TfidfVectorizer()

    # Fit and transform the texts
    tfidf_matrix = tfidf_vectorizer.fit_transform([text1, text2])

    # Compute the cosine similarity
    similarity_score = cosine_similarity(tfidf_matrix[0:1], tfidf_matrix[1:2])

    return similarity_score[0][0]

def text_similarity(text):
    score= compute_cosine_similarity(shanty,text)
    return score

with gr.Blocks() as demo:
  gr.Markdown("# Guess the lyrics of the sea shanty! \n ## Each two seconds of video represents a line")
  video=gr.PlayableVideo("final_video.mp4")
  inp=gr.Textbox(placeholder="Enter lyrics of sea shanty!")

  out=gr.Textbox()
  inp.change(text_similarity,inp,out)
demo.launch(show_api=False,share=True)