import streamlit as st 
import pandas as pd

from transformers import AutoTokenizer, AutoModel,AutoModelForSequenceClassification
import torch



num_classes = 6





#tokenizer = AutoTokenizer.from_pretrained("bert-base-uncased")
tokenizer = AutoTokenizer.from_pretrained("KhaldiAbderrhmane/bert-emotion",trust_remote_code=True)



model = AutoModelForSequenceClassification.from_pretrained("KhaldiAbderrhmane/bert-emotion",trust_remote_code=True)



def prediction_sentiment(review):
    

    t= tokenizer(review, truncation=True, padding=True, max_length=128, return_tensors='pt')
    inpt  = t['input_ids']
    mask  = t['attention_mask']



    
    outputs = model(inpt,mask)
    

    outputs = outputs.logits


    
    predicted= torch.max(outputs, 1).indices 

    if predicted == 0:
        sentiment = "Sadness"
    elif predicted == 1:
        sentiment = "Joy"
    elif predicted == 2:
        sentiment = "Love"
    elif predicted == 3:
        sentiment = "Anger"
    elif predicted == 4:
        sentiment = "Fear"
    else:
        sentiment = "Surprise"
    return sentiment





users = {"abdelmalek": [["this movie was so nice", "positive"], ["what the hell was that", "negative"], ["man this was good", "positive"]]}
columns = ["comment", "sentiment"]
user_name = st.text_input("User Name")

if user_name:
    if user_name in users:
        user_input = st.text_input("Enter your comment:")
        

        if user_input:
            
            sentiment = prediction_sentiment(user_input)
            st.write('Your sentiment is:', sentiment)
            users[user_name].append([user_input, sentiment]) 
    else:
        users[user_name] = []
        st.write("Your user name has been added.") 
        user_input = st.text_input("Enter your comment:")
        if user_input:
            sentiment = prediction_sentiment(user_input)
            st.write('Your sentiment is:', sentiment) 
            users[user_name].append([user_input, sentiment]) 

if st.button("Your comment:"):
    if user_name in users:
        df_t = pd.DataFrame(users[user_name], columns=columns)
        card_css = """
        <style>
        .card {
            background-color: #1A2E4D;
            border-radius: 10px;
            padding: 20px;
            margin: 10px 0;
            box-shadow: 0 4px 8px rgba(0, 0, 0, 0.1);
        }
        .card-title {
            font-size: 24px;
            font-weight: bold;
            color: #F4F6FF;
        }
        .card-content {
            font-size: 18px;
            color: #52709E;
        }
        .sentiment-circle {
            width: 15px; /* Adjust size as needed */
            height: 15px; /* Adjust size as needed */
            border-radius: 50%;
            display: inline-block;
            position: absolute;
            top: 50%;
            right: 0;
            transform: translateY(-50%);
            margin-right: 10px; /* Space between circle and text */
        }
        .positive {
            
            background-color: #82D853; /* Green background for positive sentiment */
        }
        .negative {
            
            background-color: #D85353; /* Red background for negative sentiment */
        }
        </style>
        """
        st.markdown(card_css, unsafe_allow_html=True)

        for comment, sentiment in df_t.values:
            sentiment_class = "positive" if sentiment == "positive" else "negative"
            sentiment_circle = f'<div class="sentiment-circle {sentiment_class}" style="background-color: {"#82D853" if sentiment == "positive" else "#D85353"};"></div>'
            border_color = "border: 2px solid #82D853;" if sentiment == "positive" else "border: 2px solid #D85353;"
            card_content = f"""
            <div class="card" style="{border_color}">
        <div class="card-title">{user_name}</div>
        <div class="card-content">
            {comment} 
            {sentiment_circle}  
        </div>
    </div>
            """
            st.markdown(card_content, unsafe_allow_html=True)
    else:
        st.error("No history available for this user.")