app.py

from flask import Flask, request, jsonify
from playwright.sync_api import sync_playwright
import base64

app = Flask(__name__)

# Global variables to keep the browser session alive
p = None
browser = None
page = None

def initialize_browser():
    """Browser start karta hai agar pehle se nahi chal raha."""
    global p, browser, page
    if not page:
        p = sync_playwright().start()
        # Headless mode = True (Browser background me chalega)
        browser = p.chromium.launch(headless=True)
        # Context create karte hain taake history maintain rahe
        context = browser.new_context(viewport={'width': 1280, 'height': 720})
        page = context.new_page()
        print("Browser Initialized")

def get_page_data():
    """Current page ka Screenshot aur HTML DOM return karta hai."""
    global page
    
    # 1. HTML Content
    html_content = page.content()
    
    # 2. Screenshot (Base64 string format mein)
    screenshot_bytes = page.screenshot(full_page=False)
    screenshot_b64 = base64.b64encode(screenshot_bytes).decode('utf-8')
    
    return {
        "html": html_content,
        "screenshot": screenshot_b64
    }

@app.route('/')
def home():
    return "Browser Server is Running. Use POST /command to interact."

@app.route('/command', methods=['POST'])
def handle_command():
    global page
    
    # Ensure browser is running
    initialize_browser()
    
    data = request.json
    action = data.get('action')
    
    try:
        # --- 1. NAVIGATE (Website kholna) ---
        if action == 'navigate':
            url = data.get('url')
            if not url:
                return jsonify({"error": "URL is required"}), 400
            page.goto(url, timeout=60000)
            page.wait_for_load_state('networkidle') # Wait until page settles

        # --- 2. CLICK / TAP ---
        elif action == 'click':
            selector = data.get('selector')
            if selector:
                page.click(selector)
            else:
                # Agar selector nahi to coordinates par click (optional)
                x = data.get('x')
                y = data.get('y')
                if x is not None and y is not None:
                    page.mouse.click(x, y)

        # --- 3. TYPE (Likhna) ---
        elif action == 'type':
            selector = data.get('selector')
            text = data.get('text')
            if selector and text:
                page.fill(selector, text)

        # --- 4. ENTER KEY (Form submit waghera ke liye) ---
        elif action == 'enter':
            page.keyboard.press('Enter')

        # --- 5. SCROLL ---
        elif action == 'scroll':
            direction = data.get('direction', 'down')
            amount = data.get('amount', 500)
            if direction == 'down':
                page.mouse.wheel(0, amount)
            elif direction == 'up':
                page.mouse.wheel(0, -amount)
            # Thoda wait karein taake scroll render ho jaye
            page.wait_for_timeout(500)

        # --- 6. HOVER ---
        elif action == 'hover':
            selector = data.get('selector')
            if selector:
                page.hover(selector)

        # --- 7. WAIT (Agar manual delay chahiye) ---
        elif action == 'wait':
            duration = data.get('duration', 1000)
            page.wait_for_timeout(duration)

        else:
            return jsonify({"error": "Unknown action"}), 400

        # --- RETURN DATA (Screenshot + DOM) ---
        response_data = get_page_data()
        return jsonify({
            "status": "success",
            "action_performed": action,
            "current_url": page.url,
            "dom": response_data['html'],
            "screenshot_base64": response_data['screenshot']
        })

    except Exception as e:
        return jsonify({"status": "error", "message": str(e)}), 500

if __name__ == '__main__':
    app.run(host='0.0.0.0', port=7860)