applitools / eyes.selenium.python

Applitools Eyes SDK for Selenium Python WebDriver

DEPRECATED

This repository is deprecated. Check out our updated repo: https://github.com/applitools/eyes.sdk.python .

eyes.selenium

Applitools Eyes SDK For Selenium Python WebDriver

Installation

Install the eyes-selenium SDK.

pip install eyes-selenium

Usage

Please check the applitools website for usage instructions:

• Selenium Python example: https://applitools.com/resources/tutorial/selenium/python#step-2

• Python Appium native example: https://applitools.com/resources/tutorial/appium/native_python#step-2

• Python Appium web example: https://applitools.com/resources/tutorial/appium/python#step-2

eyes.selenium.python's Issues

Test is rerun multiple times when stitch mode is CSS

When I run below script, a lot of threads are created continuously and rerun the script.

from applitools.selenium import Eyes, StitchMode
from selenium.webdriver import Chrome
from selenium.webdriver.chrome.options import Options

eyes = Eyes()
eyes.api_key = '<my_api_key_here>'
eyes.force_full_page_screenshot = True
eyes.stitch_mode = StitchMode.CSS

try:
    driver = Chrome()
    eyes.open(driver, "My Project", "Test")
    driver.get('http://www.google.com')
    eyes.check_window("Home page")
finally:
    driver.close()
    eyes.abort_if_not_closed()

Currently I am using Python 3.7.3, eyes-selenium==3.16.2.

check_region of entire element fails in versions apper 3.12.0

Version 3.12.0 is able to take screenshot of the entire element while higher versions can't.

https://trello.com/c/zBqG4fkj/330-python-check-region

User can't use ActionChains with EyesWebDriver

Module versions:
selenium 3.3.3
eyes-selenium 3.7.1

When I initialize selenium ActionChains object with EyesWebDriver instance it fails with following error:

Eyes driver instance check
Traceback (most recent call last):
  File "w3c_attribute_missing.py", line 15, in <module>
    eyes_action = ActionChains(eyes_driver)
  File "C:\Python34\lib\site-packages\selenium\webdriver\common\action_chains.py", line 69, in __init__
    if self._driver.w3c:
AttributeError: 'EyesWebDriver' object has no attribute 'w3c'

Seems like new version of selenium is checking whether web driver is w3c compliant or not.
Please fix it.

Full example to reproduce the issue:

from applitools.eyes import EyesWebDriver, Eyes
from selenium import webdriver
from selenium.webdriver.common.action_chains import ActionChains

selenium_driver = webdriver.Chrome()
eyes_driver = EyesWebDriver(selenium_driver, Eyes())

#This passes
selenium_action = ActionChains(selenium_driver)

#This fails due to missing w3c attribute on driver instance
eyes_action = ActionChains(eyes_driver)

`StaleElementReferenceExceptions` occasionally due to some minor modification of the DOM

Originally posted by @Dude-x in #29 (comment)

Scaling malfunction

There is a problem with the scaling of the captured screenshots.
It seems that it takes the top left quarter of the actual image instead of scaling down the entire image.

https://trello.com/c/ao6gG6rq/332-scaling-malfunction-eyes-python

Bounds is defined for Eyes Web Element but is never used.

The bounds method is defined:
https://github.com/applitools/eyes.selenium.python/blob/master/applitools/selenium/webelement.py#L108

However it is not used anywhere in the latest codebase.

Compare against mockup image

Currently, there's not a direct way to upload a mockup image (from your local files) as a baseline image, which I find it would be a useful feature. Could it be a feature to take into consideration?

Add 'anonymous_children' to _READONLY_PROPERTIES

While attempting to finally get Firefox working with geckodriver, I ran into the issue where I got the message 'Strategy not supported: anon'.

I dived deeper into the source code and apparently the WebElement passed to EyesWebElement has the attribute 'anonymous_children', now I don't know if this is perhaps version incompatibility between my browser (50.0.2), selenium (3.0.2) and applitools (3.5), but if I add it to the read only properties list then this resolves my issue entirely.

I may be definitely not aware of the fact that perhaps this attribute is needed somehow, or shouldn't be there in the first place at all, but it seems to have resolved the issue. Would you know more about this, or if adding this doesn't have any negative consequences?

Stitch Mode - Only before Open Session

Currently, we can set force_full_page_screenshot whenever we want, either before or after opening the session. However, we can only set stitch_mode before opening session, otherwise it won't be taken into consideration when taking the screenshot.
Given that we can set force_full_page_screenshot freely, wouldn't it make sense to also allow to set stitch_mode after opening the session?