Saving screen-shot of tkinter window as image

[Eclair_de_XII]

TL;DR Summary

Using PIL and win32gui in order to capture tkinter window as image, then save to file. It's always off-center, and I haven't a clue why.

from PIL import ImageGrab
from win32gui import GetWindowRect

def find_img_loc():
    k=1
    folder=sep.join((getcwd(),'screenshots'))
    initial_file='screenshot'+str(k)
    filename=lambda x: sep.join((folder,x+'.png'))
    while exists(filename(initial_file)):
        k+=1
        n=len(str(k-1))
        initial_file=initial_file[:-n]+str(k)
    kw={
        'initialfile':initial_file,\
        'defaultextension':'png',\
        'initialdir':folder,\
        'filetypes':[('.png files only','*.png')]
        }
    filename=filedialog.asksaveasfilename(**kw)
    return filename

def save_image(root):
    if not exists('screenshots'):
        mkdir('screenshots')
    root.update()
    HWND = root.winfo_id()  # get the handle of the canvas
    rect = GetWindowRect(HWND)  # get the coordinate of the canvas
    I am = ImageGrab.grab(rect)  # get image of the current location
    filename=find_img_loc()
    if not filename:
        return
    else:
        im.save(filename,format='png')
  
if __name__ == '__main__':
    root=Tk()
    lines='Sample text here','More sample text...','One more for the road.'
    for n, text in enumerate(lines):
        Label(root,text=text).grid(row=n)
    save_image(root)
    root.destroy()

This code was actually from:

https://stackoverflow.com/questions/9886274/how-can-i-convert-canvas-content-to-an-image

 

[jedishrfu]

how is it off center? horizontally or vertically or both?

 

[Eclair_de_XII]

It was both.

The actual window looks like this:


but the saved image looks like this:

In any case, I resolved the issue (partially) by using pyscreenshot.grab and manually tinkering with the capture dimensions. It was not efficient, I confess, but it did the trick alright

from os import mkdir, getcwd
from os.path import exists, sep

from pyscreenshot import grab

def find_img_loc():
    k=1
    folder=sep.join((getcwd(),'screenshots'))
    initial_file='screenshot'+str(k)
    filename=lambda x: sep.join((folder,x+'.png'))
    while exists(filename(initial_file)):
        k+=1
        n=len(str(k-1))
        initial_file=initial_file[:-n]+str(k)
    kw={
        'initialfile':initial_file,\
        'defaultextension':'png',\
        'initialdir':folder,\
        'filetypes':[('.png files only','*.png')]
        }
    filename=filedialog.asksaveasfilename(**kw)
    return filename

def save_image(im):
    if not exists('screenshots'):
        mkdir('screenshots')
    filename=find_img_loc()
    if not filename:
        return
    else:
        im.save(filename,format='png')

def save_image2(root):
    root.update()
    geometry=root.winfo_geometry()
    position=geometry.split('+')
    dimension=position[0].split('x')
    xlen=dimension[0]
    ylen=dimension[1]
    xpos=position[1]
    ypos=position[2]
    geo_list=(xpos,ypos,xlen,ylen)
    geometry=()
    k=1.25
    for n,g in enumerate(geo_list):
        g=int(g)
        g=k*g
        if n == 0:
            g+=10
        g=int(g)
        if n>1:
            h=geo_list[n-2]
            h=int(h)
            h=k*h
            k*=1.075
            h=int(h)
            geometry+=(g+h,)
        else:
            geometry+=(g,)
    im=grab(bbox=geometry)
    save_image(im)

if __name__ == '__main__':
    root=Tk()
    root.geometry('400x400+400+100')
    Label(root,text='Yabba dabba doo').grid()
    save_image1(root)

 

[Mark44]

Could it be that the screenshot you captured included more than just that one text window? If you manually capture a screenshot of that window, and then paste it into, say, Paint or other graphics app, do you get just that single window or do you get more?

Some documentation I found here says this. Note the last sentence.

PIL.ImageGrab.grab(bbox=None, include_layered_windows=False, all_screens=False, xdisplay=None)
Take a snapshot of the screen. The pixels inside the bounding box are returned as an “RGBA” on macOS, or an “RGB” image otherwise. If the bounding box is omitted, the entire screen is copied.