This is a working version without significant leak.
import photos import time import concurrent.futures import objc_util import ui @objc_util.on_main_thread def update(image_view, ui_image): image_view.image = ui_image def loop_photos(): all_photos = photos.get_assets() while True: for p in all_photos: with objc_util.autoreleasepool(): ui_image = p.get_ui_image(size=(300,300)) #image_data = p.get_image_data() update(v['img'], ui_image) time.sleep(.5) time.sleep(5) v = ui.View() v.add_subview(ui.ImageView(name='img')) v.present() with concurrent.futures.ThreadPoolExecutor(max_workers=1) as executor: executor.submit(loop_photos)
Thanks very much for all the suggestions. I'm grateful for the tip about threading and will try to incorporate it into the rest of the project but the thing that does the lifting here is the autoreleasepool context manager; with just this addition and removing the threading it works just fine.
@mrjk have you tried the same code, but without the imageview? I.e just creating the ui.Image's?
Thanks. I hadn't but it does suffer from the same issue.
import photos import time all_photos = photos.get_assets() while True: for p in all_photos: image = p.get_ui_image(size=(300,300)) time.sleep(.5) time.sleep(5)
Yes, the use of
allin my example was very sloppy. I don’t know what came over me.
Thanks for all your suggestions.
I’ve tried setting image to None and deleting the looping variable p but the increasing memory consumption continues.
I was trying to avoid using the objc because I am much more comfortable in Python but I’ll look into it this evening.
I've written a script which loops through a list of photos, displaying the current one while some processing happens behind the scenes. I've wrapped this up into an app in Xcode using the Pythonista3AppTemplate and have noticed that the memory consumption grows every time the photo changes. Minimum code to reproduce:
# Code 1 import photos import ui import time v = ui.View() v.add_subview(ui.ImageView(name = 'img')) v.present() all = photos.get_assets() while True: for p in all: v['img'].image = p.get_ui_image(size=(300,300)) time.sleep(.5) time.sleep(5)
Xcode shows memory usage steps up every time get_ui_image() is called. The rate of increase is linked to the size argument.
I tried a few things to narrow down where the issue is. This one still loops and changes the image forever but it gets all the ui_images once at the start.
# Code 2 import photos import ui import time v = ui.View() img = ui.ImageView(name='img') v.add_subview(img) v.present() all = photos.get_assets()[0:5] images = [p.get_ui_image(size=(300,300)) for p in all] while True: for i in images: v['img'].image = i time.sleep(.5) time.sleep(2)
The memory usage jumps up at the beginning, as you'd expect, but then it stays constant while it loops. This is ok for a small number of photos but impractical if you want to look at the whole photo roll.
I changed the original code so it removes the ImageView entirely to try to get some garbage collection.
import photos import ui import time v = ui.View() v.present() all = photos.get_assets() while True: for p in all: for sv in v.subviews: v.remove_subview(sv) img = ui.ImageView(name = 'img') img.image = p.get_ui_image(size=(300,300)) v.add_subview(img) time.sleep(.5) time.sleep(5)
The memory usage still grows with every change.
I think this shows it's caused by calling get_ui_image(). But is there anything I should (or could) be doing to kick start any garbage collection here? Is this a memory leak in the Pythonista photos module? And can I mitigate it?
Grateful for any advice. J