Update app.py

app.py

@@ -1,72 +1,52 @@
 import gradio as gr
 from youtube_transcript_api import YouTubeTranscriptApi, TranscriptsDisabled
-import whisper
-import yt_dlp
-import os
-import re
+import requests
 
-# Load Whisper model once
-whisper_model = whisper.load_model("base")
+HF_API_TOKEN = "your_huggingface_token"  # Replace with your token
+LLM_MODEL = "mistralai/Mistral-7B-Instruct-v0.2"
 
 def extract_video_id(url):
-    match = re.search(r"(?:v=|\/)([0-9A-Za-z_-]{11}).*", url)
-    return match.group(1) if match else None
+    if "v=" in url:
+        return url.split("v=")[-1].split("&")[0]
+    elif "youtu.be/" in url:
+        return url.split("youtu.be/")[-1].split("?")[0]
+    return None
 
-def get_transcript(video_id):
+def fetch_transcript(video_id):
     try:
         transcript = YouTubeTranscriptApi.get_transcript(video_id)
         return " ".join([t["text"] for t in transcript])
     except TranscriptsDisabled:
         return None
-    except Exception as e:
-        return None
-
-def fallback_whisper(url):
-    try:
-        yt_opts = {
-            "format": "bestaudio",
-            "outtmpl": "audio.%(ext)s",
-            "quiet": True,
-            "postprocessors": [{
-                "key": "FFmpegExtractAudio",
-                "preferredcodec": "mp3",
-                "preferredquality": "192",
-            }],
-        }
-        with yt_dlp.YoutubeDL(yt_opts) as ydl:
-            ydl.download([url])
-        audio_path = "audio.mp3"
-        result = whisper_model.transcribe(audio_path)
-        os.remove(audio_path)
-        return result["text"]
-    except Exception as e:
-        return None
 
-def summarize(text):
-    # Simple chunked summarization placeholder
-    chunks = [text[i:i+1000] for i in range(0, len(text), 1000)]
-    return "\n\n".join([f"Summary chunk {i+1}: {chunk[:200]}..." for i, chunk in enumerate(chunks)])
-
-def process_video(url):
+def summarize_text(text):
+    headers = {"Authorization": f"Bearer {HF_API_TOKEN}"}
+    payload = {
+        "inputs": f"Summarize this YouTube transcript:\n{text}",
+        "parameters": {"max_new_tokens": 300}
+    }
+    response = requests.post(
+        f"https://api-inference.huggingface.co/models/{LLM_MODEL}",
+        headers=headers,
+        json=payload
+    )
+    return response.json()[0]["generated_text"]
+
+def summarize_youtube(url):
     video_id = extract_video_id(url)
-    if not video_id:
-        return "⚠️ Invalid YouTube URL."
-
-    transcript = get_transcript(video_id)
+    transcript = fetch_transcript(video_id)
     if transcript:
-        return summarize(transcript)
-
-    fallback = fallback_whisper(url)
-    if fallback:
-        return summarize(fallback)
-
-    return "⚠️ Transcript and audio fallback failed. Try another video."
+        return summarize_text(transcript)
+    else:
+        return "❌ Transcript not available for this video."
 
 # Gradio UI
-gr.Interface(
-    fn=process_video,
-    inputs=gr.Textbox(label="Enter YouTube URL"),
-    outputs=gr.Textbox(label="Summarized Transcript"),
-    title="📄 YouTube Transcript Summarizer",
-    description="Fast, fallback-ready summarizer with Whisper + YouTubeTranscriptAPI"
-).launch()
+demo = gr.Interface(
+    fn=summarize_youtube,
+    inputs=gr.Textbox(label="Paste YouTube URL"),
+    outputs=gr.Textbox(label="Summary"),
+    title="🎬 YouTube Video Summarizer",
+    description="Extracts transcript and summarizes using Mistral-7B via Hugging Face"
+)
+
+demo.launch()