import streamlit.components.v1 as components
from streamlit_player import st_player
from transformers import pipeline
from tabulate import tabulate
import streamlit as st



st.header("stream your emotions")
st.write("insert details ng huggignaspce")

def tester(text):
  classifier = pipeline("sentiment-analysis", model='bhadresh-savani/distilbert-base-uncased-emotion')
  results = classifier(text)
  #st.subheader(results[0]['label'])

#tester(emo)
  generator = st.button("Generate Song!")
  if (generator == True):
    st.subheader(results[0]['label'])
  
    if (results[0]['label']=="joy"): #songs for joy emotion
      with open('joyplaylist.txt') as f:
        contents = f.read()
      components.html(contents,width=560,height=325)
    
    elif (results[0]['label']=="fear"):
      with open('fearplaylist.txt') as f:
        contents = f.read()
      components.html(contents,width=560,height=325)
    
    elif (results[0]['label']=="anger"): #songs for anger emotion
      with open('angryplaylist.txt') as f:
        contents = f.read()
      components.html(contents,width=560,height=325)    

    elif (results[0]['label']=="sadness"): #songs for sadness emotion
      with open('sadplaylist.txt') as f:
        contents = f.read()
      components.html(contents,width=560,height=325)

    elif (results[0]['label']=="surprise"):
      st.write("gulat ka noh")

    elif (results[0]['label']=="love"):
      with open('loveplaylist.txt') as f:
        contents = f.read()
      components.html(contents,width=560,height=325)
 
emo = st.text_input("Enter a text/phrase/sentence. A corresponding song will be recommended based on its emotion.")


st.sidebar.subheader("Model Description")
st.sidebar.write("This application uses the DistilBERT model, a distilled version of BERT. The BERT framework uses a bidirectional transformer that allows it to learn the context of a word based on the left and right of the word. According to a paper by V. Sanh, et al., DistilBERT can \"reduce the size of a BERT model by 40%, while retaining 97% of its language understanding capabilities, and being 60% faster.\" This is why the DistilBERT model was used. For more information about the paper, please check out this [link](https://share.streamlit.io/mesmith027/streamlit_webapps/main/MC_pi/streamlit_app.py).")

st.sidebar.write("The specific DistilBERT model used for this is Bhadresh Savani's [distilbert-base-uncased-emotion] (https://huggingface.co/bhadresh-savani/distilbert-base-uncased-emotion). It is fine-tuned on the Emotion Dataset from Twitter, which can be found [here](https://huggingface.co/datasets/viewer/?dataset=emotion).")

st.sidebar.subheader("Disclaimer/Limitations")
st.sidebar.write("The model only outputs sadness, joy, love, anger, fear, and surprise. With that said, it does not completely encompass the emotions that a human being feels, and the application only suggests a playlist based on the aforementioned emotions.")

st.sidebar.subheader("Performance Benchmarks")

df = pd.DataFrame(
    np.random.randn(10, 5),
    columns=('col %d' % i for i in range(5)))

st.sidebar.table(df)


tester(emo)