Lab 2: Drive the desktop

Keys go to whatever has focus. If the terminal hadn't opened yet (or opened on another workspace), the text and Return would go into whatever window was focused, maybe someone's browser. An early version of the other book's Wayland lab did exactly that.

Text goes through wtype, whose own keymap can hold any character. Key combinations go through the uinput keyboard as physical key codes, which is what Omarchy's code: shortcuts and apps' shortcuts expect.

The region is in screenshot pixels (1280x720 space), but zoom captures at the monitor's full resolution (1920x1080 space): 1260 × 1.5 = 1890.

== 1. Switch to workspace 5 with Omarchy's shortcut: a key combination through the virtual keyboard
$ act '{"action": "key", "text": "super+5"}'
{}
$ vm_ssh_desktop 'hyprctl -j activeworkspace' | jq -c '{workspace: .id, windows}'
{"workspace":5,"windows":0}

== 2. Open a terminal (Super+Return) and find it in the window list
$ act '{"action": "key", "text": "super+Return"}'
{}
$ act '{"action": "wait", "duration": 2}'
{}
$ cu GET /windows
{"windows":[{"title":"omarchy@omarchy-auto:~","app":"foot","pid":20109,"workspace":5,"focused":true,"bounds":{"x":10,"y":32,"width":1260,"height":678}}]}
   Focused: foot. Only now is it safe to type: keys go to whatever has focus.

== 3. Type a command. Text goes through wtype, so any Unicode works, whatever the keyboard layout
$ act '{"action": "type", "text": "echo \"typed by an agent: café ✓ → 日本\""}'
{}
$ act '{"action": "key", "text": "Return"}'
{}
$ act '{"action": "wait", "duration": 1}'
{}

== 4. Keys by name (xdotool's names): type, jump to the start of the line with ctrl+a, comment it out
$ act '{"action": "type", "text": "hello"}'
{}
$ act '{"action": "key", "text": "ctrl+a"}'
{}
$ act '{"action": "type", "text": "# "}'
{}
$ act '{"action": "key", "text": "Return"}'
{}
$ act '{"action": "key", "text": "ctrl+nope"}'
{"error":"Unknown key 'nope'"}

== 5. Look closely: zoom into the top of the terminal (a region of the screen, in pixels)
$ shot $LABS/out/lab2-terminal.png 10 32 1270 202
saved $LABS/out/lab2-terminal.png (60517 bytes)
   zoom returns just that region, at full resolution: cheaper than a whole screenshot, and easier to read.

== 6. A click: right-click in the middle of the terminal (its context menu, if it has one), then Escape
$ act '{"action": "right_click", "coordinate": [640,371]}'
{}
$ act '{"action": "key", "text": "Escape"}'
{}

== 7. Close the terminal (Super+W) and go back to workspace 1
$ act '{"action": "key", "text": "super+w"}'
{}
$ act '{"action": "key", "text": "super+1"}'
{}
$ cu GET /windows
{"windows":[]}
$ lab_stop_forwards
port-forwards stopped