mirror of
https://github.com/bunny-lab-io/Borealis.git
synced 2025-07-27 03:48:29 -06:00
Fixed Invisible Screenshot Agent Overlay
This commit is contained in:
@ -3,9 +3,11 @@ requests
|
||||
python-socketio
|
||||
websocket-client
|
||||
eventlet
|
||||
aiohttp
|
||||
|
||||
# GUI-related dependencies (Qt for GUI components)
|
||||
PyQt5
|
||||
qasync
|
||||
|
||||
# Computer Vision & OCR Dependencies
|
||||
Pillow
|
||||
@ -2,87 +2,113 @@
|
||||
|
||||
import sys
|
||||
import uuid
|
||||
import time
|
||||
import base64
|
||||
import threading
|
||||
import socketio
|
||||
from io import BytesIO
|
||||
import socket
|
||||
import os
|
||||
import json
|
||||
import asyncio
|
||||
import concurrent.futures
|
||||
from functools import partial
|
||||
from io import BytesIO
|
||||
import base64
|
||||
|
||||
import socketio
|
||||
from qasync import QEventLoop
|
||||
from PyQt5 import QtCore, QtGui, QtWidgets
|
||||
from PIL import ImageGrab
|
||||
|
||||
# ---------------- Configuration ----------------
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
# CORE SECTION: CONFIG MANAGER (do not modify unless you know what you’re doing)
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
CONFIG_PATH = os.path.join(os.path.dirname(__file__), "agent_settings.json")
|
||||
DEFAULT_SERVER_URL = "http://localhost:5000"
|
||||
DEFAULT_CONFIG = {
|
||||
"SERVER_URL": "http://localhost:5000",
|
||||
"max_workers": 8,
|
||||
"config_watch_interval": 2
|
||||
}
|
||||
|
||||
def load_config():
|
||||
"""
|
||||
Load agent_settings.json or prompt the user for SERVER_URL.
|
||||
Returns a config dictionary with at least {"SERVER_URL": ...}
|
||||
"""
|
||||
config = {}
|
||||
class ConfigManager:
|
||||
def __init__(self, path):
|
||||
self.path = path
|
||||
self._last_mtime = None
|
||||
self.data = {}
|
||||
self.load()
|
||||
|
||||
if os.path.exists(CONFIG_PATH):
|
||||
def load(self):
|
||||
if not os.path.exists(self.path):
|
||||
self.data = DEFAULT_CONFIG.copy()
|
||||
self._write()
|
||||
else:
|
||||
try:
|
||||
with open(CONFIG_PATH, "r") as f:
|
||||
config = json.load(f)
|
||||
if isinstance(config, dict) and "SERVER_URL" in config:
|
||||
return config
|
||||
with open(self.path, 'r') as f:
|
||||
self.data = json.load(f)
|
||||
except Exception as e:
|
||||
print(f"[WARN] Failed to parse agent_settings.json: {e}")
|
||||
|
||||
print(f"[WARN] Failed to parse config: {e}")
|
||||
self.data = DEFAULT_CONFIG.copy()
|
||||
try:
|
||||
user_input = input(f"Enter Borealis Server URL [{DEFAULT_SERVER_URL}]: ").strip()
|
||||
config["SERVER_URL"] = user_input if user_input else DEFAULT_SERVER_URL
|
||||
with open(CONFIG_PATH, "w") as f:
|
||||
json.dump(config, f, indent=2)
|
||||
self._last_mtime = os.path.getmtime(self.path)
|
||||
except Exception:
|
||||
self._last_mtime = None
|
||||
|
||||
def _write(self):
|
||||
try:
|
||||
with open(self.path, 'w') as f:
|
||||
json.dump(self.data, f, indent=2)
|
||||
except Exception as e:
|
||||
print(f"[ERROR] Prompt failed: {e}")
|
||||
config["SERVER_URL"] = DEFAULT_SERVER_URL
|
||||
print(f"[ERROR] Could not write config: {e}")
|
||||
|
||||
return config
|
||||
def watch(self):
|
||||
try:
|
||||
mtime = os.path.getmtime(self.path)
|
||||
if self._last_mtime is None or mtime != self._last_mtime:
|
||||
print("[CONFIG] Detected config change, reloading.")
|
||||
self.load()
|
||||
return True
|
||||
except Exception:
|
||||
pass
|
||||
return False
|
||||
|
||||
config = load_config()
|
||||
SERVER_URL = config["SERVER_URL"]
|
||||
CONFIG = ConfigManager(CONFIG_PATH)
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
# END CORE SECTION: CONFIG MANAGER
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
|
||||
HOSTNAME = socket.gethostname().lower()
|
||||
RANDOM_SUFFIX = uuid.uuid4().hex[:8]
|
||||
AGENT_ID = f"{HOSTNAME}-agent-{RANDOM_SUFFIX}"
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
# CORE SECTION: WEBSOCKET SETUP & HANDLERS (do not modify unless absolutely necessary)
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
AGENT_ID = f"{socket.gethostname().lower()}-agent-{uuid.uuid4().hex[:8]}"
|
||||
|
||||
# ---------------- App State ----------------
|
||||
app_instance = None
|
||||
overlay_widgets = {}
|
||||
region_launchers = {}
|
||||
running_roles = {}
|
||||
running_threads = {}
|
||||
|
||||
# ---------------- Socket Setup ----------------
|
||||
# Enable automatic reconnection with retries in background
|
||||
sio = socketio.Client(reconnection=True, reconnection_attempts=0, reconnection_delay=5)
|
||||
sio = socketio.AsyncClient(reconnection=True, reconnection_attempts=0, reconnection_delay=5)
|
||||
role_tasks = {}
|
||||
|
||||
@sio.event
|
||||
def connect():
|
||||
async def connect():
|
||||
print(f"[WebSocket] Agent ID: {AGENT_ID} connected to Borealis.")
|
||||
sio.emit('connect_agent', {"agent_id": AGENT_ID, "hostname": HOSTNAME})
|
||||
sio.emit('request_config', {"agent_id": AGENT_ID})
|
||||
await sio.emit('connect_agent', {"agent_id": AGENT_ID})
|
||||
await sio.emit('request_config', {"agent_id": AGENT_ID})
|
||||
|
||||
@sio.event
|
||||
def disconnect():
|
||||
async def disconnect():
|
||||
print("[WebSocket] Lost connection to Borealis server.")
|
||||
|
||||
@sio.on('agent_config')
|
||||
def on_agent_config(config):
|
||||
async def on_agent_config(cfg):
|
||||
print("[PROVISIONED] Received new configuration from Borealis.")
|
||||
# cancel existing role tasks
|
||||
for task in list(role_tasks.values()):
|
||||
task.cancel()
|
||||
role_tasks.clear()
|
||||
# start new tasks
|
||||
for role_cfg in cfg.get('roles', []):
|
||||
role = role_cfg.get('role')
|
||||
node_id = role_cfg.get('node_id')
|
||||
if role == 'screenshot' and node_id:
|
||||
task = asyncio.create_task(screenshot_task(role_cfg))
|
||||
role_tasks[node_id] = task
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
# END CORE SECTION: WEBSOCKET SETUP & HANDLERS
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
|
||||
roles = config.get("roles", [])
|
||||
stop_all_roles()
|
||||
for role in roles:
|
||||
start_role_thread(role)
|
||||
|
||||
# ---------------- Overlay Class ----------------
|
||||
# ---------------- Overlay Widget ----------------
|
||||
class ScreenshotRegion(QtWidgets.QWidget):
|
||||
def __init__(self, node_id, x=100, y=100, w=300, h=200):
|
||||
super().__init__()
|
||||
@ -99,7 +125,6 @@ class ScreenshotRegion(QtWidgets.QWidget):
|
||||
self.label.setText(f"{node_id[:8]}")
|
||||
self.label.setStyleSheet("color: lime; background: transparent; font-size: 10px;")
|
||||
self.label.move(8, 4)
|
||||
|
||||
self.setMouseTracking(True)
|
||||
|
||||
def paintEvent(self, event):
|
||||
@ -108,28 +133,25 @@ class ScreenshotRegion(QtWidgets.QWidget):
|
||||
painter.setBrush(QtCore.Qt.transparent)
|
||||
painter.setPen(QtGui.QPen(QtGui.QColor(0, 255, 0), 2))
|
||||
painter.drawRect(self.rect())
|
||||
|
||||
handle_rect = QtCore.QRect(
|
||||
self.width() - self.resize_handle_size,
|
||||
handle = QtCore.QRect(self.width() - self.resize_handle_size,
|
||||
self.height() - self.resize_handle_size,
|
||||
self.resize_handle_size,
|
||||
self.resize_handle_size
|
||||
)
|
||||
painter.fillRect(handle_rect, QtGui.QColor(0, 255, 0))
|
||||
self.resize_handle_size, self.resize_handle_size)
|
||||
painter.fillRect(handle, QtGui.QColor(0, 255, 0))
|
||||
|
||||
def mousePressEvent(self, event):
|
||||
if event.button() == QtCore.Qt.LeftButton:
|
||||
if event.pos().x() > self.width() - self.resize_handle_size and \
|
||||
event.pos().y() > self.height() - self.resize_handle_size:
|
||||
px, py = event.pos().x(), event.pos().y()
|
||||
if px > self.width() - self.resize_handle_size and \
|
||||
py > self.height() - self.resize_handle_size:
|
||||
self.resizing = True
|
||||
else:
|
||||
self.drag_offset = event.globalPos() - self.frameGeometry().topLeft()
|
||||
|
||||
def mouseMoveEvent(self, event):
|
||||
if self.resizing:
|
||||
new_width = max(event.pos().x(), 100)
|
||||
new_height = max(event.pos().y(), 80)
|
||||
self.resize(new_width, new_height)
|
||||
nw = max(event.pos().x(), 100)
|
||||
nh = max(event.pos().y(), 80)
|
||||
self.resize(nw, nh)
|
||||
elif event.buttons() & QtCore.Qt.LeftButton and self.drag_offset:
|
||||
self.move(event.globalPos() - self.drag_offset)
|
||||
|
||||
@ -141,117 +163,81 @@ class ScreenshotRegion(QtWidgets.QWidget):
|
||||
geo = self.geometry()
|
||||
return geo.x(), geo.y(), geo.width(), geo.height()
|
||||
|
||||
# ---------------- Region UI Handler ----------------
|
||||
class RegionLauncher(QtCore.QObject):
|
||||
trigger = QtCore.pyqtSignal(int, int, int, int)
|
||||
# ---------------- Helper Functions ----------------
|
||||
app = None
|
||||
overlay_widgets = {}
|
||||
|
||||
def __init__(self, node_id):
|
||||
super().__init__()
|
||||
self.node_id = node_id
|
||||
self.trigger.connect(self.handle)
|
||||
|
||||
def handle(self, x, y, w, h):
|
||||
print(f"[Overlay] Launching overlay for {self.node_id} at ({x},{y},{w},{h})")
|
||||
if self.node_id in overlay_widgets:
|
||||
def create_overlay(node_id, region):
|
||||
if node_id in overlay_widgets:
|
||||
return
|
||||
widget = ScreenshotRegion(self.node_id, x, y, w, h)
|
||||
overlay_widgets[self.node_id] = widget
|
||||
x, y, w, h = region
|
||||
widget = ScreenshotRegion(node_id, x, y, w, h)
|
||||
overlay_widgets[node_id] = widget
|
||||
widget.show()
|
||||
|
||||
# ---------------- GUI Thread Helpers ----------------
|
||||
def gui_create_launcher(node_id, x, y, w, h):
|
||||
launcher = RegionLauncher(node_id)
|
||||
region_launchers[node_id] = launcher
|
||||
launcher.handle(x, y, w, h)
|
||||
|
||||
def gui_update_widget(node_id, x, y, w, h, visible):
|
||||
def get_overlay_geometry(node_id):
|
||||
widget = overlay_widgets.get(node_id)
|
||||
if widget:
|
||||
widget.setGeometry(x, y, w, h)
|
||||
widget.setVisible(visible)
|
||||
return widget.get_geometry()
|
||||
return (0, 0, 0, 0)
|
||||
|
||||
# ---------------- Role Management ----------------
|
||||
def stop_all_roles():
|
||||
for node_id, thread in running_threads.items():
|
||||
if thread and thread.is_alive():
|
||||
print(f"[Role] Terminating previous task: {node_id}")
|
||||
running_roles.clear()
|
||||
running_threads.clear()
|
||||
|
||||
def start_role_thread(role_cfg):
|
||||
role = role_cfg.get("role")
|
||||
node_id = role_cfg.get("node_id")
|
||||
if not role or not node_id:
|
||||
print("[ERROR] Invalid role configuration (missing role or node_id).")
|
||||
return
|
||||
|
||||
if role == "screenshot":
|
||||
thread = threading.Thread(target=run_screenshot_loop, args=(node_id, role_cfg), daemon=True)
|
||||
else:
|
||||
print(f"[SKIP] Unknown role: {role}")
|
||||
return
|
||||
|
||||
running_roles[node_id] = role_cfg
|
||||
running_threads[node_id] = thread
|
||||
thread.start()
|
||||
print(f"[Role] Started task: {role} ({node_id})")
|
||||
|
||||
# ---------------- Screenshot Role Loop ----------------
|
||||
def run_screenshot_loop(node_id, cfg):
|
||||
interval = cfg.get("interval", 1000)
|
||||
visible = cfg.get("visible", True)
|
||||
x = cfg.get("x", 100)
|
||||
y = cfg.get("y", 100)
|
||||
w = cfg.get("w", 300)
|
||||
h = cfg.get("h", 200)
|
||||
|
||||
# Schedule launcher creation in GUI thread
|
||||
if node_id not in region_launchers:
|
||||
QtCore.QTimer.singleShot(0, lambda nid=node_id, xx=x, yy=y, ww=w, hh=h: gui_create_launcher(nid, xx, yy, ww, hh))
|
||||
|
||||
while True:
|
||||
# ---------------- Screenshot Task ----------------
|
||||
async def screenshot_task(cfg):
|
||||
interval = cfg.get('interval', 1000) / 1000.0
|
||||
node_id = cfg.get('node_id')
|
||||
region = (cfg.get('x', 100), cfg.get('y', 100), cfg.get('w', 300), cfg.get('h', 200))
|
||||
create_overlay(node_id, region)
|
||||
loop = asyncio.get_event_loop()
|
||||
executor = concurrent.futures.ThreadPoolExecutor(max_workers=CONFIG.data.get('max_workers', 8))
|
||||
try:
|
||||
# Use current widget geometry if available (after user moves/resizes)
|
||||
if node_id in overlay_widgets:
|
||||
widget = overlay_widgets[node_id]
|
||||
x, y, w, h = widget.get_geometry()
|
||||
|
||||
print(f"[Capture] Screenshot task {node_id} at ({x},{y},{w},{h})")
|
||||
img = ImageGrab.grab(bbox=(x, y, x + w, y + h))
|
||||
buffer = BytesIO()
|
||||
img.save(buffer, format="PNG")
|
||||
encoded = base64.b64encode(buffer.getvalue()).decode("utf-8")
|
||||
|
||||
sio.emit("agent_screenshot_task", {
|
||||
"agent_id": AGENT_ID,
|
||||
"node_id": node_id,
|
||||
"image_base64": encoded
|
||||
while True:
|
||||
x, y, w, h = get_overlay_geometry(node_id)
|
||||
grab = partial(ImageGrab.grab, bbox=(x, y, x + w, y + h))
|
||||
img = await loop.run_in_executor(executor, grab)
|
||||
buf = BytesIO()
|
||||
img.save(buf, format='PNG')
|
||||
encoded = base64.b64encode(buf.getvalue()).decode('utf-8')
|
||||
await sio.emit('agent_screenshot_task', {
|
||||
'agent_id': AGENT_ID,
|
||||
'node_id': node_id,
|
||||
'image_base64': encoded
|
||||
})
|
||||
|
||||
# Schedule any visibility or geometry updates in GUI thread
|
||||
QtCore.QTimer.singleShot(
|
||||
0,
|
||||
lambda nid=node_id, xx=x, yy=y, ww=w, hh=h, vis=visible: gui_update_widget(nid, xx, yy, ww, hh, vis)
|
||||
)
|
||||
await asyncio.sleep(interval)
|
||||
except asyncio.CancelledError:
|
||||
return
|
||||
except Exception as e:
|
||||
print(f"[ERROR] Screenshot task {node_id} failed: {e}")
|
||||
|
||||
time.sleep(interval / 1000)
|
||||
# ---------------- Config Watcher ----------------
|
||||
async def config_watcher():
|
||||
while True:
|
||||
if CONFIG.watch():
|
||||
# settings updated, e.g., executor pool size will apply on next task run
|
||||
pass
|
||||
await asyncio.sleep(CONFIG.data.get('config_watch_interval', 2))
|
||||
|
||||
# ---------------- Main ----------------
|
||||
if __name__ == "__main__":
|
||||
app_instance = QtWidgets.QApplication(sys.argv)
|
||||
retry_interval = 5 # seconds between connection attempts
|
||||
|
||||
def connect_loop():
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
# CORE SECTION: MAIN & EVENT LOOP (do not modify unless you know what you’re doing)
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
async def connect_loop():
|
||||
retry = 5
|
||||
while True:
|
||||
try:
|
||||
print(f"[WebSocket] Connecting to {SERVER_URL}...")
|
||||
sio.connect(SERVER_URL, transports=["websocket"], wait=False)
|
||||
print(f"[WebSocket] Connecting to {CONFIG.data['SERVER_URL']}...")
|
||||
await sio.connect(CONFIG.data['SERVER_URL'], transports=['websocket'])
|
||||
break
|
||||
except Exception:
|
||||
print(f"[WebSocket] Borealis Server is Not Running - Retrying in {retry_interval} seconds...")
|
||||
time.sleep(retry_interval)
|
||||
print(f"[WebSocket] Server not available, retrying in {retry}s...")
|
||||
await asyncio.sleep(retry)
|
||||
|
||||
threading.Thread(target=connect_loop, daemon=True).start()
|
||||
sys.exit(app_instance.exec_())
|
||||
if __name__ == '__main__':
|
||||
app = QtWidgets.QApplication(sys.argv)
|
||||
loop = QEventLoop(app)
|
||||
asyncio.set_event_loop(loop)
|
||||
with loop:
|
||||
loop.create_task(config_watcher())
|
||||
loop.create_task(connect_loop())
|
||||
loop.run_forever()
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
# END CORE SECTION: MAIN & EVENT LOOP
|
||||
# //////////////////////////////////////////////////////////////////////////
|
||||
|
||||
Reference in New Issue
Block a user