Update app.py

app.py

@@ -1,181 +1,117 @@
-# ✅ Combined YouTube Analyzer with Stock Info Extractor (fixed download using working app logic)
+# ✅ Gemini-Based Stock Recommendation Extractor (No Audio, No Whisper)
+# Uses video metadata (title + description) + Gemini Flash to extract stock info
 
 import gradio as gr
 import os
 import tempfile
-import shutil
-import re
-import torch
-import numpy as np
+import json
+import google.generativeai as genai
 from yt_dlp import YoutubeDL
 
-# Whisper setup
-WHISPER_AVAILABLE = False
-WHISPER_TYPE = None
-try:
-    import whisper
-    WHISPER_AVAILABLE = True
-    WHISPER_TYPE = "openai-whisper"
-except ImportError:
-    try:
-        from transformers import pipeline
-        WHISPER_AVAILABLE = True
-        WHISPER_TYPE = "transformers"
-    except ImportError:
-        pass
-
-# Stock Info Extraction
+# ✅ Gemini Configuration
+GEMINI_MODEL = None
 
-def extract_stock_info_simple(text):
+def configure_gemini(api_key):
     try:
-        companies = re.findall(r'\b[A-Z][a-z]+(?:\s+[A-Z][a-z]+)*(?:\s+(?:Inc|Corp|Company|Ltd)\.?)?', text)
-        symbols = re.findall(r'\b[A-Z]{2,5}\b', text)
-        prices = re.findall(r'\$\d+(?:\.\d{2})?', text)
-        actions = re.findall(r'\b(?:buy|sell|hold|bullish|bearish|target|stop loss)\b', text, re.IGNORECASE)
-
-        result = "=== EXTRACTED STOCK INFORMATION ===\n\n"
-
-        if companies:
-            result += f"\U0001F4CA Mentioned Companies: {', '.join(set(companies[:10]))}\n\n"
-        if symbols:
-            result += f"\U0001F524 Potential Stock Symbols: {', '.join(set(symbols[:10]))}\n\n"
-        if prices:
-            result += f"\U0001F4B0 Price Mentions: {', '.join(set(prices[:10]))}\n\n"
-        if actions:
-            result += f"\U0001F4C8 Trading Actions: {', '.join(set(actions[:10]))}\n\n"
-
-        recommendations = []
-        sentences = text.split('.')
-        for sentence in sentences:
-            if any(word in sentence.lower() for word in ['buy', 'sell', 'target']):
-                if any(sym in sentence for sym in symbols[:5]):
-                    recommendations.append(sentence.strip())
-
-        if recommendations:
-            result += "\U0001F3AF Potential Recommendations:\n"
-            for rec in recommendations[:5]:
-                result += f"• {rec}\n"
-
-        if not any([companies, symbols, prices, actions]):
-            result += "⚠️ No clear stock recommendations found.\n"
-
-        return result
-
+        genai.configure(api_key=api_key)
+        global GEMINI_MODEL
+        GEMINI_MODEL = genai.GenerativeModel("gemini-1.5-flash-latest")
+        return "✅ Gemini API key configured successfully."
     except Exception as e:
-        return f"Error extracting stock info: {str(e)}"
+        return f"❌ Gemini configuration failed: {str(e)}"
 
-# Whisper Transcription
+# ✅ Extract video metadata only (no download)
 
-def transcribe_audio(file_path):
-    if not WHISPER_AVAILABLE:
-        return "❌ Whisper not available", ""
+def extract_metadata(url, cookies_file=None):
     try:
-        if WHISPER_TYPE == "openai-whisper":
-            model = whisper.load_model("tiny")
-            result = model.transcribe(file_path)
-            return result["text"], "✅ Transcription complete"
-        else:
-            pipe = pipeline("automatic-speech-recognition", model="openai/whisper-tiny")
-            result = pipe(file_path)
-            return result["text"], "✅ Transcription complete"
-    except Exception as e:
-        return "❌ Transcription failed", str(e)
-
-# ✅ Reused working download logic from other app
-
-def download_audio_youtube(url, cookies_file=None):
-    try:
-        temp_dir = tempfile.mkdtemp()
-        out_path = os.path.join(temp_dir, "audio")
-
         ydl_opts = {
-            'format': 'bestaudio[ext=m4a]/bestaudio/best',
-            'outtmpl': out_path + '.%(ext)s',
             'quiet': True,
+            'skip_download': True,
             'noplaylist': True,
-            'user_agent': 'Mozilla/5.0 (Windows NT 10.0; Win64; x64)',
-            'referer': 'https://www.youtube.com/',
-            'force_ipv4': True,
-            'extractor_retries': 3,
-            'fragment_retries': 3,
-            'retry_sleep_functions': {'http': lambda n: 2 ** n},
         }
-
         if cookies_file and os.path.exists(cookies_file):
             ydl_opts['cookiefile'] = cookies_file
-        else:
-            print("⚠️ No cookies file provided")
-
-        ydl_opts['http_headers'] = {
-            'User-Agent': ydl_opts['user_agent'],
-            'Accept': 'text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8',
-            'Accept-Language': 'en-US,en;q=0.5',
-            'Accept-Encoding': 'gzip, deflate',
-            'DNT': '1',
-            'Connection': 'keep-alive',
-            'Upgrade-Insecure-Requests': '1',
-            'Referer': 'https://www.youtube.com/',
-        }
 
         with YoutubeDL(ydl_opts) as ydl:
-            ydl.download([url])
-
-        for ext in ['.m4a', '.mp3', '.webm']:
-            full_path = out_path + ext
-            if os.path.exists(full_path):
-                return full_path, "✅ Audio downloaded"
+            info = ydl.extract_info(url, download=False)
 
-        return None, "❌ Audio file not found"
+        return {
+            'title': info.get("title", ""),
+            'description': info.get("description", ""),
+            'duration': info.get("duration", 0),
+            'uploader': info.get("uploader", ""),
+            'view_count': info.get("view_count", 0),
+            'upload_date': info.get("upload_date", "")
+        }, "✅ Video metadata extracted"
 
     except Exception as e:
-        import traceback
-        traceback.print_exc()
-        return None, f"❌ Download error: {str(e)}"
+        return None, f"❌ Metadata extraction failed: {str(e)}"
+
+# ✅ Gemini Prompt for Stock Extraction
+
+def query_gemini_stock_analysis(meta):
+    if GEMINI_MODEL is None:
+        return "❌ Gemini model is not initialized."
 
-# Copy cookie to tmp
+    prompt = f"""
+    Analyze the following YouTube video metadata and extract any stock trading recommendations:
 
-def save_uploaded_cookie(cookies):
-    if cookies is None:
-        return None
-    temp_cookie_path = tempfile.mktemp(suffix=".txt")
-    shutil.copy2(cookies.name, temp_cookie_path)
-    return temp_cookie_path
+    Title: {meta['title']}
+    Description: {meta['description']}
+
+    Please extract:
+    - Mentioned companies or stock symbols
+    - Any price targets, buy/sell/hold recommendations
+    - Bullish/bearish sentiments if expressed
+    - If no stock info is present, clearly say "No financial or trading recommendations found."
+    - Keep the output short and to the point
+    """
+
+    try:
+        response = GEMINI_MODEL.generate_content(prompt)
+        return response.text if response else "⚠️ No response from Gemini."
+    except Exception as e:
+        return f"❌ Gemini query failed: {str(e)}"
 
-# Gradio app logic
+# ✅ Main Pipeline
 
-def full_pipeline(url, cookies):
-    if not url:
-        return "❌ Enter a valid YouTube URL", "", ""
+def run_pipeline(api_key, url, cookies):
+    status = configure_gemini(api_key)
+    if not status.startswith("✅"):
+        return status, ""
 
-    temp_cookie = save_uploaded_cookie(cookies)
-    audio_path, msg = download_audio_youtube(url, temp_cookie)
-    if not audio_path:
-        return msg, "", ""
+    # Save cookies if provided
+    cookie_path = None
+    if cookies:
+        cookie_path = tempfile.mktemp(suffix=".txt")
+        with open(cookie_path, "wb") as f:
+            f.write(cookies.read())
 
-    transcript, tmsg = transcribe_audio(audio_path)
-    if "❌" in transcript:
-        return msg, transcript, tmsg
+    metadata, meta_status = extract_metadata(url, cookie_path)
+    if not metadata:
+        return meta_status, ""
 
-    stock_data = extract_stock_info_simple(transcript)
-    return "✅ Complete", transcript, stock_data
+    result = query_gemini_stock_analysis(metadata)
+    return meta_status, result
 
-# Gradio App
-with gr.Blocks(title="📈 Stock Info Extractor from YouTube") as demo:
+# ✅ Gradio UI
+with gr.Blocks(title="Gemini Stock Extractor") as demo:
     gr.Markdown("""
-    # 📈 Extract Stock Mentions from YouTube
-    Upload a YouTube link + cookies.txt, and extract trading mentions using Whisper + AI
+    # 📈 Gemini-Based Stock Recommendation Extractor
+    Paste a YouTube link and get stock-related insights using only the title + description.
+    No audio, no transcription required. Fast and simple.
     """)
 
     with gr.Row():
-        url_input = gr.Textbox(label="YouTube URL")
-        cookies_input = gr.File(label="cookies.txt (exported from YouTube tab)", file_types=[".txt"])
+        api_input = gr.Textbox(label="🔑 Gemini API Key", type="password")
+        url_input = gr.Textbox(label="YouTube Video URL")
+        cookies_input = gr.File(label="cookies.txt (optional)", file_types=[".txt"])
 
-    run_btn = gr.Button("🚀 Run Extraction")
-    status = gr.Textbox(label="Status")
-    transcript_box = gr.Textbox(label="Transcript", lines=10)
-    stock_box = gr.Textbox(label="Stock Info", lines=10)
+    go_btn = gr.Button("🚀 Analyze")
+    status_box = gr.Textbox(label="Status", lines=1)
+    output_box = gr.Textbox(label="Extracted Stock Insights", lines=12)
 
-    run_btn.click(fn=full_pipeline, inputs=[url_input, cookies_input], outputs=[status, transcript_box, stock_box])
+    go_btn.click(fn=run_pipeline, inputs=[api_input, url_input, cookies_input], outputs=[status_box, output_box])
 
 if __name__ == "__main__":
     demo.launch(debug=True)