Tkinter and multi-threading - python

I was using the following code to examine if Tkinter works along with multithreading. But the following code doesn't work (the Gui becomes unresponsive as soon as it runs). Can anyone please explain why it doesn't work?
from threading import Thread
import tkinter as tk
window = tk.Tk()
label = tk.Label(window, text='Hello')
label.pack()
def func():
i = 1
while True:
label['text'] = str(i)
i += 1
Thread(target=func).start()
Thread(target=window.mainloop).start()

It doesn't work because Tkinter doesn't support multithreading. Everything that interacts with a Tkinter widget needs to run in the main thread. If you want to use multithreading, put the GUI in the main thread and your other code in a worker thread, and communicate between them with a thread safe queue.

Related

Pygame mixer causes tkinter to freeze

to-speech module to say a list and the audio module I am using is mixer form pygame. When i start playing the audio the tkinter program can not be interacted with and if I move the window the audio stops. I think some sort of threading may fix this but I am not sure how that works.
Problem
Tkinter window freezes.
Goal
To be able to interact with the program when audio is playing.
Code
import tkinter as tk
from gtts import gTTS
from io import BytesIO
import pygame
window = tk.Tk()
window.geometry("800x600")
window.resizable(0, 0)
def speak(text,language="en",accent="com"):
mp3_fp = BytesIO()
phrase = gTTS(text=text,lang=language,tld=accent)
phrase.write_to_fp(mp3_fp)
pygame.init()
pygame.mixer.init()
pygame.mixer.music.load(mp3_fp,"mp3")
pygame.mixer.music.play()
while pygame.mixer.music.get_busy():
pygame.time.delay(10)
pygame.event.poll()
def play():
data = list([boy,girl,pink,blue])
for i in data:
speak(i)
window.mainloop()
Code Explanation
The play()passes each value in the dictionary E.g. Boy, Girl to speak() separately where they are played one by one consecutively
In the speak() the audio from the text-to-speech module (gTTS) gets passed to pygame.mixer where it is played when the last word has been said.
First of all you need to post the minimum code necessary to demonstrate your problem. You haven't done that. Fortunately, this is a well-known "problem" and easy to answer without importing four libraries and building an application around what you've provided in order to answer it.
Secondly -- in the code that you have provided, data = list[boy,girl,pink,blue] isn't even proper syntax. It should be data = list(["boy", "girl", "pink", "blue"]). You have to post running code to get the best answers.
Lecture over.
The issue is that conventional unmodified Python runs in a single thread. If you want to know why that is then I invite you to research the GIL (Global Interpreter Lock) for more background.
There's only one thread, and when PyGame is doing something then the thread is busy and TkInter stops responding to input, and vice versa -- when TkInter is in the middle of something you'll find that PyGame stops responding.
You can demonstrate this phenomenon with this:
import tkinter as tk
import time
def delay():
time.sleep(10)
def main():
root = tk.Tk()
tk.Button(root, text="Test Me", command=delay).pack(expand=True, fill=tk.BOTH)
root.mainloop()
if __name__ == "__main__":
main()
When you run this you'll see that the button is depressed, the button stays depressed while the application goes to sleep, and the button doesn't go back to its unclicked status until after it wakes up.
The only way I know of to get around your particular problem is by running TkInter and/or PyGame on separate threads.
You are going to have to read up on Python's Threading() module. You might start here. I've browsed it and it seems to be pretty complete.
Just to demonstrate the difference:
import tkinter as tk
import time
import threading
def delay():
print("Delay started...")
time.sleep(10)
print("... and finished.")
def dispatchDelayToThread():
t = threading.Thread(target=delay)
t.start()
def main():
root = tk.Tk()
tk.Button(root, text="Test Me", command=dispatchDelayToThread).pack(expand=True, fill=tk.BOTH)
root.mainloop()
if __name__ == "__main__":
main()
I didn't even really change the code any! I added a function to dispatch the code I'd already written then changed the button to call the dispatcher instead of the code. Very easy to implement.
Run this and you'll see that the button returns to ready right away. If you run this from the command line you'll see that it prints a line when you enter the thread, and another line when the thread completes. And an even cooler thing is that if you click the button three times in a row you'll get three "starting" message, followed by three "finished" messages shortly afterwards.
To demonstrate this threading using your own code:
import pygame
import tkinter as tk
import io
import gtts
import threading
def speak(text,language="en",accent="com"):
mp3_fp = io.BytesIO()
phrase = gtts.gTTS(text=text,lang=language,tld=accent)
phrase.write_to_fp(mp3_fp)
pygame.init()
pygame.mixer.init()
pygame.mixer.music.load(mp3_fp,"mp3")
pygame.mixer.music.play()
while pygame.mixer.music.get_busy():
pygame.time.delay(10)
pygame.event.poll()
def dispatchPlay():
t = threading.Thread(target=play)
t.start()
def play():
data = list(["boy", "girl", "pink", "blue"])
for i in data:
speak(i)
def main():
root = tk.Tk()
root.geometry('300x200')
tk.Button(root, text="This is a clicky button", command=dispatchPlay).pack(expand=True, fill=tk.BOTH)
root.mainloop()
if __name__ == "__main__":
main()
Generally you would have your user interface on one thread, frame updates on another thread if it's a game, sound stuff on yet another thread, networking connection on a thread, etc., with all of them tied together through some kind of messaging system.
Note well, however, that in conventional unmodified Python there is only ever one thread running at any one time! You can have a hundred threads spawned, but only one of them will run at a time. Python sucks at multithreading computationally-intensive tasks, but shines in threading out I/O stuff that you usually just spend your time waiting on.

How to properly close a tkinter window with multiple threads?

I'm writing a script that takes some time to complete, but want to give some feedback to the user showing that the program is actually running. I'm trying to use the progress bar from tkinter because it's simple, but when the window closes, the program continues running in the background.
import sys
from time import sleep
from threading import Thread
import tkinter as tk
import tkinter.ttk as ttk
def work(win):
sleep(10)
win.destroy()
root = tk.Tk()
p = ttk.Progressbar(root, mode="indeterminate")
p.pack()
t = Thread(target=lambda: work(root))
t.start() # start thread
p.start() # start progressbar
root.mainloop()
sys.exit()
Here's essentially what the code does. I've tried putting sys.exit() after the thread closes the window, but that yields the same result.
After some debugging it seems like the program keeps running the main loop of root even though I destroyed it, so how can I make sure the program terminates the script properly?

label.configure works sometimes why?

Part of my code is as follows:
def get_songs():
label6.configure(text='Wait')
os.system('/home/norman/my-startups/grabsongs')
label6.configure(text='Done')
The label is not updated at the first .configure() but is at the second one.
Except if I cause a deliberate error immediately after the first one at which point it is updated and then the program terminates.
The system call takes about 2 minutes to complete so it isn't as if there isn't time to display the first one.
I am using Python 2.7.6
Does anyone know why please?
I'm going to guess you're using Tkinter. If so, as #albert just suggested, you'll want to call label.update_idletasks() or label.update() to tell Tkinter to refresh the display.
As a very crude example to reproduce your problem, let's make a program that will:
Wait 1 second
Do something (sleep for 2 seconds) and update the text to "wait"
Display "done" afterwards
For example:
import Tkinter as tk
import time
root = tk.Tk()
label = tk.Label(root, text='Not waiting yet')
label.pack()
def do_stuff():
label.configure(text='Wait')
time.sleep(2)
label.configure(text='Done')
label.after(1000, do_stuff)
tk.mainloop()
Notice that "Wait" will never be displayed.
To fix that, let's call update_idletasks() after initially setting the text:
import Tkinter as tk
import time
root = tk.Tk()
label = tk.Label(root, text='Not waiting yet')
label.pack()
def do_stuff():
label.configure(text='Wait')
label.update_idletasks()
time.sleep(2)
label.configure(text='Done')
label.after(1000, do_stuff)
tk.mainloop()
As far as why this happens, it actually is because Tkinter doesn't have time to update the label.
Calling configure doesn't automatically force a refresh of the display, it just queues one the next time things are idle. Because you immediately call something that will halt execution of the mainloop (calling an executable and forcing python to halt until it finishes), Tkinter never gets a chance to process the changes to the label.
Notice that while the gui displays "Wait" (while your process/sleep is running) it won't respond to resizing, etc. Python has halted execution until the other process finishes running.
To get around this, consider using subprocess.Popen (or something similar) instead of os.system. You'll then need to perodically poll the returned pipe to see if the subprocess has finished.
As an example (I'm also moving this into a class to keep the scoping from getting excessively confusing):
import Tkinter as tk
import subprocess
class Application(object):
def __init__(self, parent):
self.parent = parent
self.label = tk.Label(parent, text='Not waiting yet')
self.label.pack()
self.parent.after(1000, self.do_stuff)
def do_stuff(self):
self.label.configure(text='Wait')
self._pipe = subprocess.Popen(['/bin/sleep', '2'])
self.poll()
def poll(self):
if self._pipe.poll() is None:
self.label.after(100, self.poll)
else:
self.label.configure(text='Done')
root = tk.Tk()
app = Application(root)
tk.mainloop()
The key difference here is that we can resize/move/interact with the window while we're waiting for the external process to finish. Also note that we never needed to call update_idletasks/update, as Tkinter now does have idle time to update the display.

Python Tkinter Program is crashing with forget()

In my Program I want to use forget() on a button. Now if I try that, the program crashes. I know that it has something to do with threading but I couldn't find a soultion yet. Thanks in advance. Here is my examplecode:
import Tkinter as tk
import thread
window = tk.Tk()
def ok():
pass
def voice():
button1.forget()
print("If you see this, it works!")
thread.start_new_thread(voice,())
button1=tk.Button(command=ok, text="PRESS")
button1.pack()
window.mainloop()
You can't access tkinter objects from any thread but the thread that creates the object. In other words, you can't call button1.forget() from a thread and expect it to work reliably.
The generally accepted solution is to have your thread(s) write information to a thread-safe queue, and have your GUI thread poll that queue perioducally, pull an item off, and do whatever that item is requesting.
So I solved this problem simply by using the module mtTkinter , which you can find here :
http://tkinter.unpythonic.net/wiki/mtTkinter To use it you only have to write import mtTkinter as Tkinter at the beginning. After that you can use your Tkinter normally. This module changes nothing in Tkinter, it only makes it thread-friendly.
Tkinter is notorious for the fact that its lack of thread safety means that code you have written can sometimes work, and sometimes cause the entire program to hang with no errors produced, which is a pain.
Luckily, Tkinter does have its own measure for dealing with the problem, so to start a a thread with voice in it, just call voice. However, at theend of voice make sure you have made use of the window.after method to call it again later down the line. For example:
import Tkinter as tk
import thread
window = tk.Tk()
def ok():
pass
def voice():
button1.forget()
print("If you see this, it works!")
window.after(10, voice())
voice()
button1=tk.Button(command=ok, text="PRESS")
button1.pack()
window.mainloop(

Scrolling Progress Bar in Tkinter

I have been trying to set up a progress bar in a python tkinter gui that shows that a process is running. The process is long and I have no way to really measure the progress, so I need to use an indeterminate progress bar. However, I really dislike the style of the ttk indeterminate progress bar that bounces back and forth. I want one that scrolls across the bar over and over again, kind of like this image
Is this possible with tkinter?
have you tried ttk's determinate Progressbar? You can make the progress just continuously scroll across the bar.
for example:
#!/usr/bin/env python3
import tkinter
import tkinter.ttk as ttk
root = tkinter.Tk()
frame = ttk.Frame()
pb = ttk.Progressbar(frame, length=300, mode='determinate')
frame.pack()
pb.pack()
pb.start(25)
root.mainloop()
I know its an old question, but I have found a way to do this for anyone else writing tkinter.
I've been working on a tkinter app for a bit now and have determined that to handle tkinter objects, you absolutely need a separate thread. Although it is apparently frowned upon to handle tkinter objects via something else than the mainloop() method, it has been working well for me. I've never had a main thread is not in main loop error and never experienced objects that didn't update correctly.
I edited Corey Goldberg's code a bit and got it working. Here's what I got (some explanations in the comments).
import tkinter
import tkinter.ttk as ttk
import threading
def mainProgram(): # secure the main program initialization in its own def
root = tkinter.Tk()
frame = ttk.Frame()
# You need to use indeterminate mode to achieve this
pb = ttk.Progressbar(frame, length=300, mode='indeterminate')
frame.pack()
pb.pack()
# Create a thread for monitoring loading bar
# Note the passing of the loading bar as an argument
barThread = threading.Thread(target=keepLooping, args=(pb,))
# set thread as daemon (thread will die if parent is killed)
barThread.daemon=True
# Start thread, could also use root.after(50, barThread.start()) if desired
barThread.start()
pb.start(25)
root.mainloop()
def keepLooping(bar):
# Runs thread continuously (till parent dies due to daemon or is killed manually)
while 1:
"""
Here's the tricky part.
The loading bar's position (for any length) is between 0 and 100.
Its position is calculated as position = value % 100.
Resetting bar['value'] to 0 causes it to return to position 0,
but naturally the bar would keep incrementing forever till it dies.
It works, but is a bit unnatural.
"""
if bar['value']==100:
bar.config(value=0) # could also set it as bar['value']=0
if __name__=='__main__':
mainProgram()
I've added if __name__=='__main__': because I feel it defines the scope a bit better.
As a side note I've found that running threads with while 1: will crank my CPU at about 20-30% usage for that one thread in particular. It's easily solvable by importing time and using time.sleep(0.05) thereafter significantly lowering the CPU usage.
Tested on Win8.1, Python 3.5.0.

Categories

Resources