app.py

import torch
import gradio as gr
from youtube_transcript_api import YouTubeTranscriptApi
from youtube_transcript_api._errors import TranscriptsDisabled, VideoUnavailable, NoTranscriptFound
from transformers import pipeline

# model_path = "../models/models--sshleifer--distilbart-cnn-12-6/snapshots/a4f8f3ea906ed274767e9906dbaede7531d660ff"

# Load the summarization model
# text_summary = pipeline("summarization", model=model_path, torch_dtype=torch.float32)
text_summary = pipeline("summarization", model="sshleifer/distilbart-cnn-12-6", torch_dtype=torch.bfloat16)

def split_text(text, max_tokens=1024):
    words = text.split()
    for i in range(0, len(words), max_tokens):
        yield " ".join(words[i:i + max_tokens])

def summary(input_text):
    summarized_chunks = []
    for chunk in split_text(input_text):
        output = text_summary(chunk)
        summarized_chunks.append(output[0]['summary_text'])
    return " ".join(summarized_chunks)

def extract_video_id(url):
    if "youtu.be" in url:
        return url.split("/")[-1]
    elif "youtube.com/watch?v=" in url:
        return url.split("v=")[-1].split("&")[0]
    elif "youtube.com/shorts/" in url:
        return url.split("/shorts/")[-1].split("?")[0]
    else:
        raise ValueError("Invalid YouTube URL format")

def get_youtube_transcript(video_url):
    try:
        video_id = extract_video_id(video_url)
        transcript = YouTubeTranscriptApi.get_transcript(video_id)
        text_transcript = "\n".join([entry['text'] for entry in transcript])
        
        if not text_transcript.strip():
            return "The transcript is empty or unavailable."
        
        return summary(text_transcript)

    except ValueError as e:
        return f"Error: {e}"
    except TranscriptsDisabled:
        return "Transcript is disabled for this video."
    except VideoUnavailable:
        return "Video is unavailable."
    except NoTranscriptFound:
        return "No transcript found for this video."

# if __name__ == "__main__":
#     youtube_url = input("Enter YouTube URL: ").strip()
#     transcript_summary = get_youtube_transcript(youtube_url)
#     print("\n=== Summary ===\n")
#     print(transcript_summary)

gr.close_all()

demo = gr.Interface(fn=get_youtube_transcript, inputs=[gr.Textbox(label="Input Youtube video url to summarize", lines=2)],
                    outputs=[gr.Textbox(label="Summarized text",lines=6)],
                    title="GenAI Project 2: Video to Text Summarizer",
                    description="This application is use to summarized the text from youtube video")

demo.launch()