How to get the position of a pdf page shown on screen?

Question

I want to display some information on screen, based on the position of some pdf page being shown. For that, I need to get the position of the page's border, but right now, I only have a very naive and inefficient solution, which consists in taking vertical slices of the window until I detect a 'mostly white' slice, which I take as the searched border:

import win32gui
from win32api import GetSystemMetrics

dc = win32gui.GetDC(0)
width,height=GetSystemMetrics(0),GetSystemMetrics(1)

fraction=200 # This is to reduce the amount of info to be processed
p_height=height/fraction    
suma=0 # This variable holds the number of withe pixels of each slice

for i in range(width): # while in screen w
    for j in range(int(p_height)): # hile in screen h
        if win32gui.GetPixel(dc, i, j*fraction)!=16777215: 
            pass # pass if pixel is not withe
        else:
            suma+=1
    if suma/p_height>0.4: # if more than 40% of the pixels are white,
        print(i)
        break
    else:
        suma=0

This is both slow and not smart. I guess there's a way to get the information I'm looking for without having to literally 'look' at the screen. Any suggestions?

(I use google chrome to open the pdf.)

Unless you can interact directly with the application displaying the PDF directly through some API to get the information, my gess is there probably isn't any better way. — martineau, Aug 19 '20 at 01:55
You may need to think more. If the chrome is not full screen, perhaps it only occupies a quarter of the screen, then you must detect less than 40% of the white area (PDF file) of the screen. — Strive Sun, Aug 20 '20 at 09:16
If you are sure that chrome is full screen, then I can suggest you to use `CreateDIBSection` to get the pixel value, which is faster than `GetPixel`. Refer: [Get Pixel color fastest way?](https://stackoverflow.com/a/10516660/11128312) — Strive Sun, Aug 20 '20 at 09:18
Thanks for that, I'll take a look on CreateDIBSection, sorry I didn't get the notification of your comment before. About the full screen thing, I guess it won't be difficult to detect the size and position of the chrome screen to adapt my code, but I have to assume the possibility of the user reading on full screen, so that's why I started there. I did improve my code a little, I would appreciate any comments on that improvement. I'll upload it later today — José Chamorro, Aug 25 '20 at 21:24

score 0 · Answer 1 · answered Sep 23 '20 at 15:29

Just for the sake of keeping this question alive, here's a little improvement to my original code meant to save some time. In practice, it takes the code more than a second to find the page, which is not acceptable.

I just introduced a 'step' to ignore some screen slices in order to find the mostly-white section of the screen faster, and then go back slowly to find the exact starting point of that section.

import win32gui
from win32api import GetSystemMetrics

dc = win32gui.GetDC(0)
width,height=GetSystemMetrics(0),GetSystemMetrics(1)

fraction=200 # This is to reduce the amount of info to be processed
step=10 # Here, the code will look at one of every ten vertical slices

p_height=height/fraction    
suma=0 # This variable holds the number of withe pixels of each slice
for i in range(0,width,step): # while in screen w
    for j in range(int(p_height)): # while in screen h
        if win32gui.GetPixel(dc, i, j*fraction)!=16777215: 
            pass # pass if pixel is not withe
        else:
            suma+=1
    if suma/p_height>0.4: # if more than 40% of the pixels are white
        suma=0
        for k in range(step): # to cover the ignored slices
            for j in range(int(p_height)): # while in screen h
                if win32gui.GetPixel(dc, (i-k), j*fraction)!=16777215: 
                    pass # pass if pixel is not withe
                else:
                    suma+=1
            if suma/p_height<0.4: # If less than 40% of pixels are white, we found the edge
                print(i-k)
                break
            else:
                suma=0
         break
    else:
        suma=0

How to get the position of a pdf page shown on screen?

1 Answers1