Used a brutal hack that just inserts the current focused window id instead of tracing the window id from the PID.
Gnome terminal uses a single server instance for multiple windows and the algorithm fails to find the correct window. It often ends up drawing on some other terminal window that is open.
I install by cloning this repo
python3 -m pip install --user -e .
Überzug is a command line util which allows to draw images on terminals by using child windows.
Advantages to w3mimgdisplay:
- no race conditions as a new window is created to display images
- expose events will be processed,
so images will be redrawn on switch workspaces - tmux support (excluding multi pane windows)
- terminals without the WINDOWID environment variable are supported
- chars are used as position - and size unit
- no memory leak (/ unlimited cache)
Libraries used in the c extension:
- python
- X11
- Xext
There are also other direct dependencies,
but they will be installed by pip.
- Install by using pip:
The package is namedueberzug
- Packages of linux distributions:
At least one packager applies patches to my code.
So if there are issues uninstall it and install it via pip.
Actually I think it's not fine that they call it ueberzug after changing the code.
As bugs introduced by them look like they are part of my work.
Note: You can improve the performance of image manipulation functions by using pillow-simd instead of pillow.
The communication is realised via stdin.
A command is a request to execute a specific action with the passed arguments.
(Therefore a command has to contain a key value pair "action": action_name)
Commands are separated with a line break.
- json: Command as json object
- simple:
Key-value pairs seperated by a tab,
pairs are also seperated by a tab
⚠️ ONLY FOR TESTING!
Simple was never intended for the usage in production!
It doesn't support paths containing tabs or line breaks
which makes it error prone. - bash: dump of an associative array (
declare -p variable_name
)
Name: add
Description:
Adds an image to the screen.
If there's already an image with the same identifier
it will be replaced.
Key | Type | Description | Optional |
---|---|---|---|
identifier | String | a freely choosen identifier of the image | No |
x | Integer | x position | No |
y | Integer | y position | No |
path | String | path to the image | No |
width | Integer | desired width; original width will be used if not set | Yes |
height | Integer | desired height; original width will be used if not set | Yes |
Integer | Deprecated: replaced by scalers (this behavior is implemented by the default scaler contain) image will be resized (while keeping it's aspect ratio) if it's width is bigger than max width |
Yes | |
Integer | Deprecated: replaced by scalers (this behavior is implemented by the default scaler contain) image will be resized (while keeping it's aspect ratio) if it's height is bigger than max height |
Yes | |
draw | Boolean | redraw window after adding the image, default True | Yes |
synchronously_draw | Boolean | redraw window immediately | Yes |
scaler | String | name of the image scaler (algorithm which resizes the image to fit into the placement) |
Yes |
scaling_position_x | Float | the centered position, if possible Specified as factor of the image size, so it should be an element of [0, 1]. |
Yes |
scaling_position_y | Float | analogous to scaling_position_x | Yes |
ImageScalers:
Name | Description |
---|---|
crop | Crops out an area of the size of the placement size. |
distort | Distorts the image to the placement size. |
fit_contain | Resizes the image that either the width matches the maximum width or the height matches the maximum height while keeping the image ratio. |
contain | Resizes the image to a size <= the placement size while keeping the image ratio. |
forced_cover | Resizes the image to cover the entire area which should be filled while keeping the image ratio. If the image is smaller than the desired size it will be stretched to reach the desired size. If the ratio of the area differs from the image ratio the edges will be cut off. |
cover | The same as forced_cover but images won't be stretched if they are smaller than the area which should be filled. |
Name: remove
Description:
Removes an image from the screen.
Key | Type | Description | Optional |
---|---|---|---|
identifier | String | a previously used identifier | No |
draw | Boolean | redraw window after removing the image, default True | Yes |
Just a reminder: This is a GPLv3 licensed project, so if you use any of these libraries you also need to license it with a GPLv3 compatible license.
The library is deprecated.
Dump associative arrays if you want to use ueberzug with bash.
First of all the library doesn't follow the posix standard,
so you can't use it in any other shell than bash.
Executing ueberzug library
will print the path to the library to stdout.
Functions:
ImageLayer
starts the ueberzug process and uses bashs associative arrays to transfer commands.Also there will be a function namedImageLayer::{action_name}
declared for each action.
Each of this function takes the key values pairs of the respective action as arguments.
Every argument of these functions has to be an associative key value pair.
ImageLayer::{action_name} [{key0}]="{value0}" [{key1}]="{value1}" ...
Executing such a function builds the desired command string according to the passed arguments and prints it to stdout.
First of all everything which isn't mentioned here isn't safe to use and
won't necessarily shipped with new coming versions.
The library is included in ueberzug's package.
import ueberzug.lib.v0 as ueberzug
Classes:
-
Visibility:
An enum which contains the visibility states of a placement.- VISIBLE
- INVISIBLE
-
Placement:
A placement to put images on.Every key value pair of the add action is an attribute (except identifier).
Changing one of it will lead to building and transmitting an add command if the placement is visible.
The identifier key value pair is implemented as a property and not changeable.Constructor:
Name Type Optional Description canvas Canvas No the canvas where images should be drawn on identifier String No a unique string used to address this placement visibility Visibility Yes the initial visibility state
(if it's VISIBLE every attribute without a default value needs to be set)**kwargs dict Yes key value pairs of the add action Properties:
Name Type Setter Description identifier String No the identifier of this placement canvas Canvas No the canvas this placement belongs to visibility Visibility Yes the visibility state of this placement
- setting it to VISIBLE leads to the transmission of an add command
- setting it to INVISIBLE leads to the transmission of a remove commandWarning:
The transmission of a command can lead to an IOError.
(A transmission happens on assign a new value to an attribute of a visible Placement.
The transmission is delayed till leaving a with-statement if lazy_drawing is used.) -
ScalerOption:
Enum which contains the useable scaler names. -
Canvas:
Should either be used with a with-statement or with a decorated function.
(Starts and stops the ueberzug process)Constructor:
Name Type default Description debug bool False suppresses printing stderr if it's false Methods:
Name Returns Description create_placement Placement prevents the use of the same identifier multiple times,
takes the same arguments as the Placement constructor (excluding canvas parameter)__call__ Function Decorator which returns a function which calls the decorated function with the keyword parameter canvas=this_canvas_object.
Of course other arguments are also passed through.request_transmission - Transmits queued commands if automatic_transmission is enabled or force=True is passed as keyword argument. Properties / Attributes:
Name Type Setter Description lazy_drawing context manager factory No prevents the transmission of commands till the with-statement was left with canvas.lazy_drawing: pass
synchronous_lazy_drawing context manager factory No Does the same as lazy_drawing. Additionally forces the redrawing of the windows to happen immediately. automatic_transmission bool Yes Transmit commands instantly on changing a placement. If it's disabled commands won't be transmitted till a lazy_drawing or synchronous_lazy_drawing with-statement was left or request_transmission(force=True) was called. Default: True
Command formats:
- Json command format:
{"action": "add", "x": 0, "y": 0, "path": "/some/path/some_image.jpg"}
- Simple command format:
action add x 0 y 0 path /some/path/some_image.jpg
- Bash command format:
declare -A command=([path]="/some/path/some_image.jpg" [action]="add" [x]="0" [y]="0" )
Bash:
# process substitution example:
ueberzug layer --parser bash 0< <(
declare -Ap add_command=([action]="add" [identifier]="example0" [x]="0" [y]="0" [path]="/some/path/some_image0.jpg")
declare -Ap add_command=([action]="add" [identifier]="example1" [x]="10" [y]="0" [path]="/some/path/some_image1.jpg")
sleep 5
declare -Ap remove_command=([action]="remove" [identifier]="example0")
sleep 5
)
# group commands example:
{
declare -Ap add_command=([action]="add" [identifier]="example0" [x]="0" [y]="0" [path]="/some/path/some_image0.jpg")
declare -Ap add_command=([action]="add" [identifier]="example1" [x]="10" [y]="0" [path]="/some/path/some_image1.jpg")
read
declare -Ap remove_command=([action]="remove" [identifier]="example0")
read
} | ueberzug layer --parser bash
Python library:
- curses (based on https://docs.python.org/3/howto/curses.html#user-input):
import curses
import time
from curses.textpad import Textbox, rectangle
import ueberzug.lib.v0 as ueberzug
@ueberzug.Canvas()
def main(stdscr, canvas):
demo = canvas.create_placement('demo', x=10, y=0)
stdscr.addstr(0, 0, "Enter IM message: (hit Ctrl-G to send)")
editwin = curses.newwin(5, 30, 3, 1)
rectangle(stdscr, 2, 0, 2+5+1, 2+30+1)
stdscr.refresh()
box = Textbox(editwin)
# Let the user edit until Ctrl-G is struck.
box.edit()
# Get resulting contents
message = box.gather()
demo.path = ''.join(message.split())
demo.visibility = ueberzug.Visibility.VISIBLE
time.sleep(2)
if __name__ == '__main__':
curses.wrapper(main)
- general example:
import ueberzug.lib.v0 as ueberzug import time if __name__ == '__main__': with ueberzug.Canvas() as c: paths = ['/some/path/some_image.png', '/some/path/another_image.png'] demo = c.create_placement('demo', x=0, y=0, scaler=ueberzug.ScalerOption.COVER.value) demo.path = paths[0] demo.visibility = ueberzug.Visibility.VISIBLE for i in range(30): with c.lazy_drawing: demo.x = i * 3 demo.y = i * 3 demo.path = paths[i % 2] time.sleep(1/30) time.sleep(2)
Scripts:
- fzf with image preview: https://github.com/seebye/ueberzug/blob/master/examples/fzfimg.sh
- Mastodon viewer: https://github.com/seebye/ueberzug/blob/master/examples/mastodon.sh
- Fzf Mpd User Interface: https://github.com/seebye/fmui