app.py

import streamlit as st
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
import re
import pandas as pd
import googleapiclient.discovery
import plotly.express as px 

# Load the BERT tokenizer and model
tokenizer = AutoTokenizer.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")
model = AutoModelForSequenceClassification.from_pretrained("nlptown/bert-base-multilingual-uncased-sentiment")


# Set up the YouTube API service
api_service_name = "youtube"
api_version = "v3"
DEVELOPER_KEY = "AIzaSyC4Vx8G6nm3Ow9xq7NluTuCCJ1d_5w4YPE"  # Replace with your actual API key

youtube = googleapiclient.discovery.build(api_service_name, api_version, developerKey=DEVELOPER_KEY)

# Function to fetch comments for a video ID
def scrape_comments(video_id):
    request = youtube.commentThreads().list(
        part="snippet",
        videoId=video_id,
        maxResults=100
    )
    response = request.execute()


    comments = []

    for item in response['items']:
        comment = item['snippet']['topLevelComment']['snippet']
        comments.append([
            comment['textDisplay']
        ])

    comments_df = pd.DataFrame(comments, columns=['comment'])

    # df.head(10).

    return comments_df


# Function to fetch video metadata using YouTube API
def fetch_video_info(video_id):
    # video_id = extract_video_id(video_url)

    request = youtube.videos().list(
        part="snippet",
        id=video_id
    )
    response = request.execute()

    if response['items']:
        video_info = response['items'][0]['snippet']
        channel_name = video_info['channelTitle']
        video_title = video_info['title']
        return channel_name, video_title
    
    else:
        raise ValueError("Video not found")

        
# Function to extract video ID from YouTube URL
def extract_video_id(video_url):
    match = re.search(r'(?<=v=)[\w-]+', video_url)
    if match:
        return match.group(0)
    else:
        st.error("Invalid YouTube video URL")

# Function to fetch YouTube comments for a video ID
def fetch_comments(video_id):
    # Example using youtube-comment-scraper-python library
    comments_df = scrape_comments(video_id)

    return comments_df

# Function to analyze sentiment for a single comment
def analyze_sentiment(comment):
    tokens = tokenizer.encode(comment, return_tensors="pt", max_length=512, truncation=True)
    # input_ids = tokens['input_ids']
    # attention_mask = tokens['attention_mask']

    # result = model(input_ids, attention_mask=attention_mask)
    result = model(tokens)

    sentiment_id = torch.argmax(result.logits) + 1
    if(sentiment_id > 3):
        sentiment_label = "Positive"
    elif(sentiment_id < 3):
        sentiment_label = "Negative"
    else:
        sentiment_label = "Neutral"

    return sentiment_label


def main():
    st.title("YouTube Comments Sentiment Analysis")
    
    # Create sidebar section for app description and links
    st.sidebar.title("Comment Feel")
    st.sidebar.write("Welcome to the YouTube Comments Sentiment Analysis App 🎥")
    st.sidebar.write("""
                    
                    **Description** 📝
                    This project utilizes a pre-trained sentiment analysis model based on BERT and TensorFlow to analyze the sentiment of comments from a YouTube video. Users can input a YouTube video URL, fetch related comments, and determine their sentiments (positive, negative, or neutral).
                    
                    Input a valid YouTube video URL in the provided text box 🔗.
                    Click "Extract Comments and Analyze" to fetch comments and analyze sentiments 🔄.
                    View sentiment analysis results via pie and bar charts 📊.
                     
                    Credits 🌟
                    
                    Coder: Aniket Panchal
                    GitHub: https://github.com/Aniket2021448
                    
                    Contact 📧
                    For any inquiries or feedback, please contact aniketpanchal1257@gmail.com
    
    """)
    st.sidebar.write("Feel free to check out my other apps :eyes:")


    with st.sidebar.form("app_selection_form"):
        st.write("Select an App:")
        app_links = {
            "Movie-mind": "https://movie-mind.streamlit.app/",
            "find-fake-news": "https://find-fake-news.streamlit.app/"
        }
        selected_app = st.selectbox("Choose an App", list(app_links.keys()))

        submitted_button = st.form_submit_button("Go to App")

    # Handle form submission
    if submitted_button:
        selected_app_url = app_links.get(selected_app)
        if selected_app_url:
            st.sidebar.success("Redirected successfully!")
            st.markdown(f'<meta http-equiv="refresh" content="0;URL={selected_app_url}">', unsafe_allow_html=True)

    
    # Dropdown menu for other app links

    st.sidebar.write("In case the apps are down, because of less usage")
    st.sidebar.write("Kindly reach out to me @ aniketpanchal1257@gmail.com")
    
    
    st.write("Enter a YouTube video link below: :movie_camera:")
    video_url = st.text_input("YouTube Video URL:")
    
    if st.button("Extract Comments and Analyze"):
        video_id = extract_video_id(video_url)
        if video_id:
            comments_df = fetch_comments(video_id)
            comments_df['sentiment'] = comments_df['comment'].apply(lambda x: analyze_sentiment(x[:512]))
            sentiment_counts = comments_df['sentiment'].value_counts()

            channel_name, video_title = fetch_video_info(video_id)
            
            st.write(f"**Channel Name:** {channel_name}")
            st.write(f"**Video Description:** {video_title}")
            
            st.write("Based on top :100: comments from this video")
            # Create pie chart
            st.write("Pie chart representation 	:chart_with_upwards_trend:")
            fig_pie = px.pie(values=sentiment_counts.values, names=sentiment_counts.index, title='Sentiment Distribution')
            st.plotly_chart(fig_pie, use_container_width=True)

            # Create bar chart
            st.write("Bar plot representation :bar_chart:")
            fig_bar = px.bar(x=sentiment_counts.index, y=sentiment_counts.values, labels={'x': 'Sentiment', 'y': 'Count'}, title='Sentiment Counts')
            st.plotly_chart(fig_bar)


if __name__ == "__main__":
    main()