How to use test_sync_download method in Playwright Python

afaik, playwright uses a virtual keyboard and does not exactly send the same signal as a real keyboard to the browser application.

consider going native and using something closer to kernel like pyautogui, after you've focused you window:

import pyautogui
pyautogui.press('f5')

Just use some condition instead of print method, for example you could check if response contains some key in its json:

def run(playwright):
    chromium = playwright.chromium
    browser = chromium.launch()
    page = browser.new_page()
    # Subscribe to "request" and "response" events.
    page.on("request", lambda request: print(">>", request.method, request.url))
    page.on("response", lambda response: print("<<", response.status, response.url))
    page.goto("https://example.com")
    browser.close()

For Example:

page.on("response", lambda response: response if key in response.body())

There should be waitForResponse for python too, and you could use that.

You can define your own function like this one:

def my_own_wait_for_selector(page, selector, time_out):
    try:
        page.wait_for_selector(selector, timeout=time_out)
        return True
    except:
        return False

That function will wait for a selector and if given some time (miliseconds) the element is not there it will return False instead of trowing an error.

Then in your code you could try something like:

while not my_own_wait_for_selector(page, "my_button_selector", 5000):
    page.reload()
page.locator("my_button_selector").click()

As you can see we are checking if our button is present, and we will wait only 5 seconds, if after 5 seconds the button is not there, we will reload, and again and again until our button is there, once our button is there code will continue and then we will click the button.

Be careful, because if your button is never there, your script will reload and your script will never finish, so you should add some other condition to your while

It looks like there is some kind of lazy load, at first, first 8 elements are loaded, if you scroll down, more content will load

For the 1st form:

def get_playwright():
    with sync_playwright() as playwright:
        yield playwright

print(get_playwright())  # <generator object get_playwright at 0x108aac580>

That is expected because get_playwright is a generator, which returns a generator iterator, which you have to call next(...) on to get each yielded value from the iterator.

Consider a simpler, non-playwright example:

In [14]: def generate_nums():
    ...:     for num in range(10):
    ...:         yield num
    ...: 

In [15]: nums = generate_nums()

In [16]: nums
Out[16]: <generator object generate_nums at 0x11115e6d0>

In [17]: next(nums)
Out[17]: 0

In [18]: next(nums)
Out[18]: 1

In [19]: next(nums)
Out[19]: 2

For more examples, see Understanding generators in Python.

Since your get_playwright returns an iterator, you need to call next() once to get the actual object:

from playwright.sync_api import sync_playwright


def get_playwright():
    with sync_playwright() as playwright:
        yield playwright

playwright_generator = get_playwright()
print(playwright_generator)  # <generator object get_playwright at 0x104031580>

playwright = next(playwright_generator)
print(playwright)  # <playwright._impl._playwright.Playwright object at 0x1041aabb0>

For the 2nd form:

@pytest.fixture()
def get_playwright():
    with sync_playwright() as playwright:
        yield playwright

def test(get_playwright):
    print(get_playwright)

It should be the same case, but it's just that pytest automatically calls next() on the fixture value if it's a generator. I could not find documentation for this behavior from the pytest docs, but it was mentioned by one of the pytest author's/maintainer's in a different answer:

Here's roughly the execution here

• pytest notices your fixture is used for the test function
• pytest calls the fixture function
  • since it is a generator, it returns immediately without executing code
• pytest notices it is a generator, calls next(...) on it
  • this causes the code to execute until the yield and then "pausing". you can think of it kind of as a co-routine ...
• pytest then executes your test function

...which is probably why the value passed to your test function is already the next-ed value, the playwright object.

I ran into this problem as well where my test would run in headed mode but not headless mode. I had a window of Chromium open when trying to run my test and it will fail. Once I closed all Chromium windows it worked in headless mode.

The issue is the docker image you have selected (alpine) architecture tags dont match any that playwright supports.

python:3.9:alpine

/code/src # pip debug --verbose | grep "py3-"
WARNING: This command is only meant for debugging. Do not use this with automation for parsing and getting these details, since the output and options of this command may change without notice.
  py3-none-musllinux_1_2_x86_64
  py3-none-musllinux_1_1_x86_64
  py3-none-musllinux_1_0_x86_64
  py3-none-linux_x86_64
  py3-none-any

playwright tags

/code/src # pip install playwright -vv | grep "Skipping link" | cut -f2 -d'(' | cut -f1 -d')' | grep "^py3" | sort -u
ERROR: Could not find a version that satisfies the requirement playwright (from versions: none)
ERROR: No matching distribution found for playwright
py3-none-macosx_10_13_x86_64
py3-none-macosx_11_0_arm64
py3-none-macosx_11_0_universal2
py3-none-manylinux1_x86_64
py3-none-manylinux2014_aarch64, py3-none-manylinux_2_17_aarch64
py3-none-manylinux_2_27_x86_64
py3-none-win32
py3-none-win_amd64

This is why pip is telling you there is no matching distribution because its checking all the builds for playwright against the builds alpine linux wants and there isnt a common one, so it cannot install. You will need to use a docker image with an architecture that playwright is built against and supports. You can see this in more detail if you run pip install playwright -vv in the alpine linux image

  Skipping link: none of the wheel's tags (py3-none-manylinux2014_aarch64, py3-none-manylinux_2_17_aarch64) are compatible (run pip debug --verbose to show compatible tags): https://files.pythonhosted.org/packages/a6/c5/142449e50253fef
ff667584793ef303c1bb1ecf1cb849cd8b83cd9a653c1/playwright-1.27.1-py3-none-manylinux_2_17_aarch64.manylinux2014_aarch64.whl (from https://pypi.org/simple/playwright/) (requires-python:>=3.7)
  Skipping link: none of the wheel's tags (py3-none-win32) are compatible (run pip debug --verbose to show compatible tags): https://files.pythonhosted.org/packages/79/54/cc0b824dd92507f4ed48e2a71424af2d0b4f91d72f3932ad4fefa7c68981/pla
ywright-1.27.1-py3-none-win32.whl (from https://pypi.org/simple/playwright/) (requires-python:>=3.7)
  Skipping link: none of the wheel's tags (py3-none-win_amd64) are compatible (run pip debug --verbose to show compatible tags): https://files.pythonhosted.org/packages/5d/a0/e7009f47b7fef143617a2907fad14170a5ae765fe7bad8050402addd3cab
/playwright-1.27.1-py3-none-win_amd64.whl (from https://pypi.org/simple/playwright/) (requires-python:>=3.7)

Skipping link: not a file: https://pypi.org/simple/playwright/
Given no hashes to check 0 links for project 'playwright': discarding no candidates
ERROR: Could not find a version that satisfies the requirement playwright (from versions: none)
ERROR: No matching distribution found for playwright

You can see the architectures it builds against on there github https://github.com/microsoft/playwright-python/blob/release-1.27/setup.py#L79

I had a similar problem: I had PM2 run by an Apache Airflow Celery worker running a playwright task, hourly. The job hung.

I suspect it is the environment variables littered by Byobu and PM2. What worked for me was wrapping the Python script in a bash script and sanitizing the environment variable like the following:

#!/bin/bash
[ "$HOME" != "" ] && exec -c $0
# Your python script should come below:
/home/ubuntu/usr/venv-3.10-airflow/bin/airflow celery worker

The first line clears the env vars (as suggested in Sanitize environment with command or bash script?). Now you can run this script with pm2 start command. My worker ran fine for about 24 hours as of now. I hope this works for you too.

I see what the problem here is. It does have to do with pyautogui and playwright not functioning together correctly. I tried searching around to find how to actually get the current mouse position with playwright, but I did not find anything.

So, here is my conclusion:

How did I figure out the problem, and what actually is the problem?

The problem here is that pyautogui actually controls the computer's mouse in this case. It seems like playwright only controls that of the browser. If you add a time.sleep(10) (In order to actually see if the cursor has changed position and to have time to move the cursor at a different position) at the end of the function like so:

def test_simple_move():
    mouse_start_position = pyautogui.position()
    print(mouse_start_position)
    
    with sync_playwright() as playwright:
        browser = playwright.chromium.launch(headless=False, slow_mo=10)
        page = browser.new_page()
        page.goto(r"http://www.uitestingplayground.com/")
        page.mouse.move(100,400)
        mouse_final_position = pyautogui.position()
        print(mouse_final_position)

        time.sleep(10)

You will then see that the actual cursor does not move. It seems like what playwright is doing here is just moving a "virtual" cursor in its browser. In fact, what I did to prove this was:

1. Launch the code
2. While the delay of 10 seconds is going, move the cursor to another point on the screen.
3. Wait for the code to finish.
4. You will now see how the points are different. Here are my points