feat: support list attribute

[kkb912002]

No description provided.

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: kkb912002 (3b9648d, 2f08b5a, cbb8244)

[KazuCocoa]

Could you add test code like

python-client/test/unit/webdriver/webelement_test.py

Line 66 in 876233e

def test_get_attribute_with_dict(self):

?
Please sign the CLA as well.

[mykola-mokhnach]

I assume the return type must be updated as well and then (probably) mypy output fixed

[mykola-mokhnach]

also isinstance may accept a tuple as the second argument

[mykola-mokhnach]

Could you please provide some particular examples of where list is returned as attribute value? In java client all attributes are converted to strings: https://github.com/SeleniumHQ/selenium/blob/36585d189b2e9f2ced136a7e6c456ffe53604141/java/src/org/openqa/selenium/remote/RemoteWebElement.java#L149

[kkb912002]

@mykola-mokhnach
If the 'value' field is parseable as JSON, it might be automatically parsed by the HTTP library used by the client in each language.
If there is a standard, we should follow it. This change follows the context of returning the dictionary as is. If it needs to be returned as a string, using json.dumps would be necessary for dict or list in the current situation.

ex) original value(string): '["abc", "def"]'
Parsed as List => ['abc', 'def']

• str() => "['abc', 'def']" (value changed can't be parsed as json)
• json.dumps() => '["abc", "def"]'

What should we do?

[mykola-mokhnach]

If there is a standard, we should follow it. This change follows the context of returning the dictionary as is. If it needs to be returned as a string, using json.dumps would be necessary for dict or list in the current situation.

W3C Spec needs the result to be a string. I don't know why somebody added dict there as a special return case. I assume it would be smart to perform json.dumps on list values for this particular case without changing the return type. I would also do the same for dictionaries, but this will be a breaking change, so maybe in the next major client update

[KazuCocoa]

@kkb912002 Could you give us what driver and element caused the case? Perhaps we could check the behavior as well

[kkb912002]

@kkb912002 Could you give us what driver and element caused the case? Perhaps we could check the behavior as well

This is specific to occurrences within my custom UI library. I plan to make changes to comply with the W3C spec, returning a string. Feel free to close this request.