Screenpy

------------------

New Features

- (h/t langgaibo and sramdas-dod) Actors can now clean up after themselves! Give 'em a task with `.has_cleanup_task(CleanUpTask())` and they'll do the `CleanUpTask()` task as they exit. Or sooner, if you call `Actor.cleans_up()` yourself! (`CleanUpTask` not included.)
- (h/t langgaibo) Actors can now take notes and use them... anywhere. The new **MakeNote** Action allows you to save the answer to a Question under a handy key, which can then be recalled later using the `noted_under` direction! Which brings me to...
- The Director! The Director is omnipresent and singular. They are always watching, waiting to take or give a needed note. Ask for one by using a `the_noted(key)` or `noted_under(key)` direction, wherever you need the information they hold. 🔺👁🔺
- Added the **See**, **SeeAllOf**, and **SeeAnyOf** Actions, to relieve Actors of the burden of asserting. Take a look at the new [Deprecations](https://screenpy-docs.readthedocs.io/en/latest/deprecations.html#id3) docs page to learn more.
- Added the **Attribute** Question, to ask about an element's attributes. Like "value", which is the one i'll be asking about.

Improvements

- (h/t WandyYing and bandophahita) SOLIDified Actors a little bit more by deprecating their asserting responsibility. This was part of adding the new **See** etc. Actions mentioned above. Actors now have a `should` method, which is really just an alias for `was_able_to` and `attempts_to`, but is meant to be used with that **See** Action. Or not, it's really up to you.
- Targets can now be used as if they themselves were locator tuples. You can do `target[0]` and pass in `*target`s. It's wild! And cleans up a really gross list comprehension from the **Wait** source code.
- As ever, another huge documentation update. Things should be more consistent now. Maybe. (I'll probably find more things to smooth out as soon as i commit this.) (*Edit*: i did.)

Deprecations

- Deprecated `Actor.should_see_the` and `Actor.should_see_any_of` in favor of using `Actor.should` with the new **See**, **SeeAllOf**, and **SeeAnyOf** Actions. These methods will be removed in 4.0.0. Instructions for modifying your tests can be found in the [3.1.0 Deprecations documentation](https://screenpy-docs.readthedocs.io/en/latest/deprecations.html#id3)

------------------

Bug Fixes

- Missed a `from typing import Protocol`, that's fixed now. Woops!

------------------

Improvements

- (h/t bandophahita) CLI logging now has indentation levels to help show which statements are encapsulated by Questions, tasks, Actions, etc.!
- Reverted the `Protocol` import from `typing` back to `typing_extensions` for better compatibility. This will be revisited when Python 4 is released, probably.

New Features

- (h/t bandophahita) New **IsClickable** Resolution to check if an element is clickable!

------------------

Improvements

- **Wait** now only puts *one* "..." at the end of its messages. :eye_roll:
- `mypy` will now be looking at the `examples/` directory as well, so it can catch errors like...

Bugfixes

- (h/t bandophahita) Removed the **BaseAction**, **BaseAbility**, and **BaseQuestion** classes, for *sure* this time.
- `mypy` should no longer complain whenever you use any of the **SendMETHODRequest** API request Actions (e.g. **SendGETRequest**, **SendPOSTRequest**, etc.).

------------------

Improvements

- **Wait** can do custom log messages again, to make the reports look nicer.

Bug Fixes

- **Element** now catches the correct exception type, now that Target's the one doing the finding.
- **Wait** now formats its custom exception correctly.
- What's a release without some same-day bug fixes, right?

------------------

Breaking Changes

- **BrowseTheWeb** no longer has `.to_find()` or `.to_wait_for()`. I realized these methods actually break the S in SOLID, and these methods were limiting Actions from really doing what they're supposed to do. See the Improvements section below for the direct benefit of this change!

Improvements

- **Wait** can now use strategies that don't use a Target, or have multiple arguments. You can now use any of Selenium's URL-monitoring expected conditions or any other custom strategies through `Wait().using(strategy).with_("any", "number", "of", "args")`!
- **AddHeader** can now accept new forms of header-setting arguments, like a dict or just alternating pairs. Now `AddHeader(Cookie="yes=please")` == `AddHeader({"Cookie": "yes=please"})` == `AddHeader("Cookie", "yes=please")`!
- Added **SetHeaders** Action, if you want to set the headers *exactly* (and remove all unmentioned headers).
- **AddHeader**, **SetHeaders**, and **Send\[METHOD\]Request** Actions now all have a way to mark their contents as secret, so they won't be logged. Add a `.which_should_be_kept_secret()` or `.secretly()` to the end and the log won't mention the unmentionables.
- I did another docs overhaul, this time making them read in a way that might please [Hemingway](http://www.hemingwayapp.com/).

Bug Fixes

- **Click** is now able to be chained without a Target to click. This was preventing clicking at an offset, which annoyed *me* at least.