This is similar to Quickly place a window to another screen using only the keyboard , but I want to be able to use the command line (so that all I need to do is to recall the command line from the bash history).
E.g., send
eDP1, VGA1, and HDMI1 (and maximize them after moving - but not the crazy F11 way, the normal window-manager-style maximization).
I would like to specify windows by the executable name.
The script below will send windows, belonging to a specific WM_CLASS (application), to a specific screen, by the screen's name. How that is done is explained in the script and also further below.
The script assumes the screens are arranged horizontally, and more or less top- aligned (with a difference < 100 PX).
#!/usr/bin/env python3
import subprocess
import sys
# just a helper function, to reduce the amount of code
get = lambda cmd: subprocess.check_output(cmd).decode("utf-8")
# get the data on all currently connected screens, their x-resolution
screendata = [l.split() for l in get(["xrandr"]).splitlines() if " connected" in l]
screendata = sum([[(w[0], s.split("+")[-2]) for s in w if s.count("+") == 2] for w in screendata], [])
def get_class(classname):
# function to get all windows that belong to a specific window class (application)
w_list = [l.split()[0] for l in get(["wmctrl", "-l"]).splitlines()]
return [w for w in w_list if classname in get(["xprop", "-id", w])]
scr = sys.argv[2]
try:
# determine the left position of the targeted screen (x)
pos = [sc for sc in screendata if sc[0] == scr][0]
except IndexError:
# warning if the screen's name is incorrect (does not exist)
print(scr, "does not exist. Check the screen name")
else:
for w in get_class(sys.argv[1]):
# first move and resize the window, to make sure it fits completely inside the targeted screen
# else the next command will fail...
subprocess.Popen(["wmctrl", "-ir", w, "-e", "0,"+str(int(pos[1])+100)+",100,300,300"])
# maximize the window on its new screen
subprocess.Popen(["xdotool", "windowsize", "-sync", w, "100%", "100%"])
The script needs both wmctrl and xdotool:
sudo apt-get install xdotool wmctrl
Copy the script below into an empty file, save it as move_wclass.py
Run it by the command:
python3 /path/to/move_wclass.py <WM_CLASS> <targeted_screen>
for example:
python3 /path/to/move_wclass.py gnome-terminal VGA-1
For the WM_CLASS, you may use part of the WM_CLASS, like in the example. The screen's name needs to be the exact and complete name.
The explanation is mostly on the concept, not so much on the coding.
In the output of xrandr, for every connected screen, there is a string/line, looking like:
VGA-1 connected 1280x1024+1680+0
This line gives us information on the screen's position and its name, as explained here.
The script lists the information for all screens. When the script is run with the screen and the window class as arguments, it looks up the (x-) position of the screen, looks up all windows (-id's) of a certain class (with the help of wmctrl -l and the output of xprop -id <window_id>.
Subsequently, the script moves all windows, one by one, to a position on the targeted screen (using wmctrl -ir <window_id> -e 0,<x>,<y>,<width>,<height>) and maximizes it (with xdotool windowsize 100% 100%).
The script worked fine on the tests I ran it with. Using wmctrl, and even xdotool, on Unity can have some stubborn peculiarities however that sometimes need to be solved by experiment rather than reasoning. If you might run into exceptions, please mention.
I've rewrited @jacobs python code to simple bash and make it works (I tested this on ubuntu 16 cinnamon).
I had to add remove,maximized_vert, remove,maximized_horz without that windows didn't move.
#!/bin/bash
if [ ! -z "$1" ] || [ -z "$2" ]; then
command=$(wmctrl -l | grep $1 | cut -d" " -f1)
if [ ! -z "$command" ]; then
position=$(xrandr | grep "^$2" | cut -d"+" -f2)
if [ ! -z "$position" ]; then
for window in $command; do
wmctrl -ir $window -b remove,maximized_vert
wmctrl -ir $window -b remove,maximized_horz
wmctrl -ir $window -e 0,$position,0,1920,1080
wmctrl -ir $window -b add,maximized_vert
wmctrl -ir $window -b add,maximized_horz
done
else
echo -e "not found monitor with given name"
fi
else
echo -e "not found windows with given name"
fi
else
echo -e "specify window and monitor name;\nmove.sh window-name monitor-name"
fi
sudo apt-get install xdotool wmctrl/path/to/script.sh "window-name" "monitor-name"
Based on @AndrzejPiszczek's answer, here's a way to move all windows to a specific screen:
function move_win {
if [ -z "$1" ]; then
echo -e "Specify a screen, possible options: "
echo -e $(xrandr | grep " connected " | cut -d'-' -f1)
return
fi
MONITOR=$1
# get all relevant windows on all screens
windows=$(wmctrl -l | egrep -v " -1 " | cut -d" " -f1)
if [ ! -z "$windows" ]; then
# get the necessary metrics from the screen the windows should be moved to
# will contain: width, height, offsetX, offsetY
screen_values=($(xrandr | grep "^$MONITOR-.* connected" | grep -Eo '[0-9]+x[0-9]+\+[0-9]+\+[0-9]+' | sed 's/x/ /g; s/+/ /g'))
if (( ${#screen_values[@]} )); then
# get the start/end position of the screen so we can later determine
# if the window is already on the screen or not
screen_start_pos=$(( ${screen_values[2]} ))
screen_end_pos=$(( ${screen_values[2]} + ${screen_values[0]} ))
for window in $windows; do
# get the window name
window_name=$(wmctrl -lG | grep "$window" | awk -F "$HOSTNAME " '{print $2}')
# extract relevant window geometry values such as x, y, width, height
window_values=($(wmctrl -lG | grep "$window" | awk -F " " '{print $3, $5, $6}'))
# if the window's X origin position is already inside the screen's
# total width then don't move it (this won't work exactly for windows only partially on the screen)
if (( ${window_values[0]} >= $screen_end_pos || ${window_values[0]} < $screen_start_pos )); then
echo -e "Moving to screen $MONITOR: $window_name"
wmctrl -ir $window -b remove,maximized_vert
wmctrl -ir $window -b remove,maximized_horz
# the -e parameters are gradient,x,y,width,height
# move window to (X,Y) -> (0,0) of new screen and the same window dimensions
wmctrl -ir $window -e 0,$screen_start_pos,0,${window_values[1]},${window_values[2]}
else
echo -e "Already on screen $MONITOR: $window_name"
fi
done
else
echo -e "No screen found"
fi
else
echo -e "No windows found"
fi
}
For the record, here is what I use for the combination of this question and Restore multiple monitor settings:
# configure multiple displays and
# move the windows to their appropriate displays
import subprocess
import os
import wmctrl
import re
mydisplays = [("VGA1",0,"left"),
("eDP1",1080,"normal"),
("HDMI1",3000,"left")]
https://askubuntu.com/questions/702002/restore-multiple-monitor-settings
def set_displays ():
subprocess.check_call(" && ".join([
"xrandr --output %s --pos %dx0 --rotate %s" % d for d in mydisplays]),
shell=True)
https://askubuntu.com/questions/702071/move-windows-to-specific-screens-using-the-command-line
mywindows = [("/emacs$","VGA1"),
("/chrome$","HDMI1"),
("gnome-terminal","eDP1")]
def max_windows ():
didi = dict([(d,x) for d,x,_ in mydisplays])
for w in wmctrl.Window.list():
try:
exe = os.readlink("/proc/%d/exe" % (w.pid))
for (r,d) in mywindows:
if re.search(r,exe):
x = didi[d]
print "%s(%s) --> %s (%d)" % (r,exe,d,x)
w.set_properties(("remove","maximized_vert","maximized_horz"))
w.resize_and_move(x,0,w.w,w.h)
w.set_properties(("add","maximized_vert","maximized_horz"))
break
except OSError:
continue
def cmdlines (cmd):
return subprocess.check_output(cmd).splitlines()
def show_displays ():
for l in cmdlines(["xrandr"]):
if " connected " in l:
print l
if name == 'main':
show_displays()
set_displays()
show_displays()
max_windows()
you would need to use wmctrl version 0.3 or later (because of my pull request).
