Update app.py

app.py

@@ -2,10 +2,8 @@ import gradio as gr
 import whisper
 import os
 import re
-from pytube import YouTube
-import requests
-from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
-from transformers import pipeline
+import yt_dlp
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM, pipeline
 
 # Load models
 whisper_model = whisper.load_model("base")
@@ -15,37 +13,38 @@ summarizer_model = AutoModelForSeq2SeqLM.from_pretrained(multilingual_model)
 translator = pipeline("translation", model="Helsinki-NLP/opus-mt-mul-en")
 
 SUPPORTED_LANGUAGES = {
-    "bn": "Bengali",
-    "en": "English",
-    "gu": "Gujarati",
-    "hi": "Hindi",
-    "kn": "Kannada",
-    "ml": "Malayalam",
-    "mr": "Marathi",
-    "ta": "Tamil",
-    "te": "Telugu",
-    "ur": "Urdu"
+    "bn": "Bengali", "en": "English", "gu": "Gujarati", "hi": "Hindi",
+    "kn": "Kannada", "ml": "Malayalam", "mr": "Marathi", "ta": "Tamil",
+    "te": "Telugu", "ur": "Urdu"
 }
 
 def download_audio(youtube_url):
-    video = YouTube(youtube_url)
-    audio_stream = video.streams.filter(only_audio=True).first()
-    filename = "audio.mp3"
-    audio_stream.download(filename=filename)
+    ydl_opts = {
+        'format': 'bestaudio/best',
+        'outtmpl': 'audio.%(ext)s',
+        'postprocessors': [{
+            'key': 'FFmpegExtractAudio',
+            'preferredcodec': 'mp3',
+        }],
+        'cookiefile': 'cookies.txt'  # <— Uses your uploaded cookie file
+    }
+
+    with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+        info_dict = ydl.extract_info(youtube_url, download=True)
+        filename = ydl.prepare_filename(info_dict).replace(".webm", ".mp3").replace(".m4a", ".mp3")
     return filename
 
 def extract_thumbnail(youtube_url):
-    video_id = None
-    match = re.search(r"(?:v=|\/)([0-9A-Za-z_-]{11}).*", youtube_url)
+    match = re.search(r"(?:v=|\/)([0-9A-Za-z_-]{11})", youtube_url)
     if match:
         video_id = match.group(1)
         return f"https://img.youtube.com/vi/{video_id}/0.jpg"
     return ""
 
-def summarize_text(text, lang_code):
+def summarize_text(text):
     input_text = f"summarize: {text}"
     inputs = tokenizer.encode(input_text, return_tensors="pt", max_length=512, truncation=True)
-    summary_ids = summarizer_model.generate(inputs, max_length=150, min_length=30, num_beams=4, early_stopping=True)
+    summary_ids = summarizer_model.generate(inputs, max_length=150, min_length=30, num_beams=4)
     return tokenizer.decode(summary_ids[0], skip_special_tokens=True)
 
 def transcribe_and_summarize(youtube_url, translate_to_english):
@@ -60,7 +59,7 @@ def transcribe_and_summarize(youtube_url, translate_to_english):
         if lang_code not in SUPPORTED_LANGUAGES:
             return None, f"❌ Language '{lang_code}' not supported.", "", "", None
 
-        summary = summarize_text(transcript, lang_code)
+        summary = summarize_text(transcript)
 
         if translate_to_english and lang_code != "en":
             translated_summary = translator(summary)[0]["translation_text"]
@@ -79,10 +78,9 @@ def transcribe_and_summarize(youtube_url, translate_to_english):
     except Exception as e:
         return None, f"❌ Error: {str(e)}", "", "", None
 
-# Gradio UI
 with gr.Blocks(css="style.css") as demo:
     gr.Markdown("<h1 style='text-align: center;'>🎬 Multilingual YouTube Summarizer</h1>")
-    gr.Markdown("This tool transcribes any YouTube video and summarizes it in its original language or English.")
+    gr.Markdown("Paste any YouTube video link, and get transcript + summary. Works for Hindi, Bengali, Tamil, Urdu, and more!")
 
     with gr.Row():
         youtube_url = gr.Textbox(label="YouTube Video URL")