P6

computer-use/py/atspi_snapshot.py

@@ -0,0 +1,36 @@
+#!/usr/bin/env python3
+import json
+import sys
+from gi.repository import Atspi
+
+mode, limit = sys.argv[1], int(sys.argv[2])
+Atspi.init()
+desktop = Atspi.get_desktop(0)
+nodes = []
+def walk(node, depth=0):
+    if len(nodes) >= limit or node is None or depth > 30:
+        return
+    try:
+        role = node.get_role_name() or ''
+        name = node.get_name() or ''
+        component = node.get_component()
+        rect = component.get_extents(Atspi.CoordType.SCREEN) if component else None
+        if role and (name or rect):
+            nodes.append({'role': role, 'name': name, 'rect': [rect.x, rect.y, rect.width, rect.height] if rect else None, 'state': [str(s) for s in node.get_state_set().get_states()]})
+        for i in range(node.get_child_count()):
+            walk(node.get_child_at_index(i), depth + 1)
+    except Exception:
+        return
+if mode == 'focused':
+    for i in range(desktop.get_child_count()):
+        app = desktop.get_child_at_index(i)
+        try:
+            for j in range(app.get_child_count()):
+                window = app.get_child_at_index(j)
+                if window.get_state_set().contains(Atspi.StateType.ACTIVE):
+                    walk(window)
+        except Exception:
+            continue
+else:
+    walk(desktop)
+print(json.dumps(nodes, ensure_ascii=False))

computer-use/py/normalize_frame.py

@@ -0,0 +1,14 @@
+#!/usr/bin/env python3
+import sys
+from PIL import Image
+
+source, target, cap, quality = sys.argv[1], sys.argv[2], int(sys.argv[3]), int(sys.argv[4])
+with Image.open(source) as image:
+    image = image.convert('RGB')
+    if len(sys.argv) == 9:
+        x, y, width, height = [int(v) for v in sys.argv[5:9]]
+        image = image.crop((x, y, x + width, y + height))
+    scale = min(1.0, cap / max(image.width, image.height))
+    if scale < 1.0:
+        image = image.resize((round(image.width * scale), round(image.height * scale)), Image.Resampling.LANCZOS)
+    image.save(target, 'WEBP', quality=quality, method=4)

computer-use/py/portal_screenshot.py

@@ -0,0 +1,27 @@
+#!/usr/bin/env python3
+"""Capture one frame through the GNOME Screenshot portal and print its path."""
+import sys
+from gi.repository import Gio, GLib
+
+target = sys.argv[1]
+bus = Gio.bus_get_sync(Gio.BusType.SESSION, None)
+proxy = Gio.DBusProxy.new_sync(bus, Gio.DBusProxyFlags.NONE, None, 'org.freedesktop.portal.Desktop', '/org/freedesktop/portal/desktop', 'org.freedesktop.portal.Screenshot', None)
+request = proxy.call_sync('Screenshot', GLib.Variant('(a{sv})', ({'interactive': GLib.Variant('b', False)},)), Gio.DBusCallFlags.NONE, -1, None).unpack()[0]
+loop = GLib.MainLoop()
+result = {'uri': None, 'code': 1}
+def response(_conn, _sender, _path, _interface, _member, params):
+    code, values = params.unpack()
+    if code == 0:
+        result['uri'] = values.get('uri')
+        result['code'] = 0
+    loop.quit()
+subscription = bus.signal_subscribe(None, 'org.freedesktop.portal.Request', 'Response', request, None, Gio.DBusSignalFlags.NONE, response)
+loop.run()
+bus.signal_unsubscribe(subscription)
+if not result['uri']:
+    raise SystemExit('Screenshot portal returned no image URI')
+ok, contents, _etag = Gio.File.new_for_uri(result['uri']).load_contents(None)
+if not ok:
+    raise SystemExit('could not read screenshot portal URI')
+Gio.File.new_for_path(target).replace_contents(contents, None, False, Gio.FileCreateFlags.REPLACE_DESTINATION, None)
+print(target)