q2ebanking / boa-constrictor Goto Github PK

Use Moq to write unit tests for WebDriver Questions. These unit tests should not initialize a real WebDriver. Instead, they should mock WebDriver and make sure calls are successful.

There are multiple WebDriver Questions, so this issue might require multiple pull requests.

We should set the NuGet project logo to the new Boa Constrictor logo!

The repository has a security vulnerability:
https://github.com/q2ebanking/boa-constrictor/security/dependabot/docs/Gemfile.lock/nokogiri/open

Unfortunately, GitHub Pages current supports nokogiri = 1.11.7:
https://pages.github.com/versions/

I opened a GitHub Support ticket to get this updated:
https://support.github.com/ticket/personal/0/1332394

Once the support ticket is resolved, we should update Boa Constrictor.

Use Moq to write unit tests for WebDriver Tasks. These unit tests should not initialize a real WebDriver. Instead, they should mock WebDriver and make sure calls are successful.

There are multiple WebDriver Tasks, so this issue might require multiple pull requests.

We should create a GitHub Action to run the unit tests in Boa.Constrictor.UnitTests for every pull request.

Note: Do not include Boa.Constrictor.Example tests in this issue. (We could include them in a separate issue, but they require more dependencies because they need a WebDriver + a browser.)

We should do tagging for releases and stuff.

Do the pages need to be static classes?

Does it support parallel execution?

The current Nuget package (0.2.2) doesn't include the XML documentation:

Assuming you want the docs included, they can be enabled by adding <GenerateDocumentationFile>true</GenerateDocumentationFile> to the <PropertyGroup> in the csproj file.

However, after enabling the above, quite a few warnings of different types are raised such as codes 1591, 1572, 1573, ...
Specific codes can be ignored at a project level by adding another PropertyGroup item: <NoWarn>$(NoWarn);1591;1572;1573</NoWarn>

If you like I can submit a PR, but I'm not sure how you would like the warnings to be handled.

PR #23 added a GitHub Action to automatically run unit tests for pull requests. However, the unit test check is not "Required" like the other checks:

I'd like to make it required for main. I read this guide, but it looks like I don't have the ability to do so:
https://docs.github.com/en/free-pro-team@latest/github/administering-a-repository/enabling-required-status-checks

@dnunez32, are you able to help with this once #23 gets approved and merged? Is this just a feature that certain users don't have, or is it unavailable for the project?

Adding Equals and GetHashCode methods to RestSharp interactions will enable the answer cache to use them.

If the RestSharp project completes the following issue, then adding these methods to Boa Constrictor RestSharp interactions will be much easier: restsharp/RestSharp#1554.

Background

Boa Constrictor should maintain a change log file for every released version. This issue was created as the result of #17 for exploring release notes.

TODO

• Name it CHANGELOG.md
• Put it in the project root directory
• Follow the conventions from this web page: https://keepachangelog.com/en/1.0.0/
• Add a line in the file denoting that the file should follow that format
• Use an Unreleased section for upcoming changes
• Add instructions to the contribution guidelines to update the Unreleased section for any changes
• Update the change log file for every version released
• Make a one-time update to the .csproj file's PackageReleaseNotes setting to say "read the change log"
  • See https://docs.microsoft.com/en-us/dotnet/core/tools/csproj

Versioning

Furthermore, Boa Constrictor should use semantic versioning. Add documentation about this in the contribution guidelines.

Suggestions

It might make sense to complete #30 (making a CONTRIBUTING.md file) before making the change log file.

Add this line to the README. It's a soundbite for the project.

We should also consider using the related graphic from the Selenium Day talk, too.

We shouldn't hide the RestClient object from the caller. It should be able to be directly injected into the AbstractRestSharpAbility class.

This is a follow-up to #18. It would be nice to run code coverage and share the report.

Check out this article:
https://codeburst.io/code-coverage-in-net-core-projects-c3d6536fd7d7

Questions:

• What code coverage tool should we use?
• Where should the report be published? (in the log, uploaded elsewhere, etc.)
• Is code coverage worthwhile for this project?

ToString methods for Screenplay interactions vary in format. We should standardize them.

This ticket covers ToString method standardization for WebDriver Tasks.

• Each Question should explicitly define a ToString method instead of relying upon base classes.
• Questions should fit the phrasing "Actor attempts to ...". The first word should not be capitalized (unless it's a proper noun).

We should learn how to do release notes and add them to every NuGet package release.

Polly is a .NET resilience and transient-fault-handling library that allows developers to express policies such as Retry, Circuit Breaker, Timeout, Bulkhead Isolation, and Fallback in a fluent and thread-safe manner.

Should we use Polly for Boa Constrictor's waiting and retries? Maybe. Let's discuss that here in this issue.

Hey,

I'm using gecko driver for a number of tests and it doesn't seem to support MoveToElement (probably more).

A number of the tasks use this and I end up with error OpenQA.Selenium.MoveTargetOutOfBoundsException.

I can write my own tasks to run javascript to do the scroll but it's a bit annoying that I can't use Boa's inbuilt tasks.

Am I just missing something?

Hi there!

Would there be a recommendation for or against utilizing and creating Abilities/Interactions for the Actor to use RestSharp to set up the environment for a test? I would imagine the use case for the current implementation is for testing the Service/API layer, but it seems to me that it would be OK to use it for setup as well.

There are test automation approaches that encourage the use of code seams to write more efficient tests by setting up the test conditions thru the Service layer and then finish your test in the UI. For example... If I want to test removing a product from my Shopping Basket...

1. Navigate to web site.
2. Search for an item.
3. Add it to my cart.
4. Go to my cart.
5. Remove the item from my cart.

Steps 1-4 can all be bypassed (and tested in separate test cases as necessary) by using an API to directly add a product to the cart and then navigating directly to the cart to complete the UI automation of step 5.

Beyond that - are there any Examples kicking around of the RestSharp classes being used?

ToString methods for Screenplay interactions vary in format. We should standardize them.

This ticket covers ToString method standardization for WebDriver Questions.

• Each Question should explicitly define a ToString method instead of relying upon base classes.
• Questions should fit the phrasing "Actor asks for ...". The first word should not be capitalized (unless it's a proper noun).

With commit "17022664772fc071114eef81eb820f5c2535180c" two tests in IsDateAsStringTests are failing

I assume this is because I am running on german region settings.

Hello! and thank you for the repo! Really great to see Screenplay making it into the .NET space.

Coming from SerenityBDD, one feature that I'm missing the most in this repo is being able to pass in multiple tasks in a single AttemptsTo() call.

In other words, I want to be able to turn this:

actor.AttemptsTo(SendKeys.To(LoginPage.UserField, userName));
actor.AttemptsTo(SendKeys.To(LoginPage.PasswordField, password));
actor.AttemptsTo(Click.On(LoginPage.LogInButton));

into this:

actor.AttemptsTo(
    SendKeys.To(LoginPage.UserField, userName),
    SendKeys.To(LoginPage.PasswordField, password),
    Click.On(LoginPage.LogInButton)
);

Saves me a few keystrokes, typing out actor.AttemptsTo() over and over again and it's a little more readable imo.

The implementation would look something like this:

public class Actor {

...

public void AttemptsTo(params ITask[] tasks) {
    foreach(ITask task in tasks) {
        AttemptsTo(task);
    }
}

Hi there,

I was wondering what I would need to do to add additional properties to an Actor? I'd like to currently give them a username/password property, and may also want to describe them with other attributes down the road for other tests.

I can see that there are a Name and Logger property, but as a C# novice, I'm afraid I'm not sure how to build up from the class in the Boa Constrictor NuGet package.

Shadow DOM allows hidden DOM trees to be attached to elements in the regular DOM tree. It can be very tricky to handle with test automation. Info about Shadow DOM can be found here: https://developer.mozilla.org/en-US/docs/Web/Web_Components/Using_shadow_DOM.

We should answer these questions:

• Is Shadow DOM support worth pursuing?
• Can Boa Constrictor's current Web UI interactions handle Shadow DOM?
• If not, then can they be updated seamlessly to support Shadow DOM?
• If not, then how should we handle Shadow DOM?

Problem

Writing interactions for commands is difficult because the existing nomenclature feels clunky:

• Actor.AsksFor(JavaScriptCall.To(...))
• Actor.AsksFor(RestApiResponse.From(...))

Solution

It would be nice to have a method named Actor.Calls(...) for any interaction, Task or Question. "Calls" is generic enough to typically work in both cases. It would also be a good fit for command-like interactions.

• Actor.Calls(Rest.Request(...))
• Actor.Calls(Rest.Get(...))
• Actor.Calls(JavaScript.InBrowser(...))

We should think about this a bit more before proceeding.

A good logo would give the project a visual identity.

I delivered a talk to introduce Boa Constrictor at the Selenium Day - North America conference on October 16, 2020. However, that talk was not recorded and shared publicly. I'd like to record the talk, upload the video to YouTube, write a blog article about it, and link it from the project README. That way, people can get a higher-level intro to Boa Constrictor than the tutorial.

Under a doc section named "Usage", write a doc page to show how to use Boa Constrictor with SpecFlow.

The page should include:

1. Preparing Visual Studio for SpecFlow
2. Creating a SpecFlow project
3. Adding the Boa Constrictor NuGet package
4. Writing a basic test in Gherkin
5. Writing step definitions with Boa Constrictor

This page should be more of a quick reference than a lengthy tutorial. It should somewhat mirror the structure of the NUnit page for #89.

Playwright has a C# implementation. Perhaps we should add Screenplay interactions for it. I heard it's much faster than Selenium WebDriver.

https://playwright.dev/
https://github.com/microsoft/playwright-sharp

I just needed an IsNullOrWhitespace condition, because I had to wait until an HTML attribute was available.

Would you be interested in adding it to boa itself? I can submit a PR with it later.

Under a doc section named "Usage", write a doc page to show how to use Boa Constrictor with NUnit.

The page should include:

1. Creating an NUnit project
2. Adding the Boa Constrictor NuGet package
3. Writing a basic test with Boa Constrictor

Reference the tutorial, since it uses NUnit. The purpose of this page is to be a quick reference, not a lengthy tutorial itself. It should somewhat mirror the SpecFlow page for #88.

It would be nice to have a "file drop" task. I could really use this for my tests.

The open source code is great, but people need a NuGet package if they really want to use Boa Constrictor.

Questions:

• Name? (Boa.Constrictor seems good.)
• Versioning?
• How to build and release? (GitHub actions?)
• NuGet.org account?

Some messages are very, very long. Perhaps it would be nice to have logs automatically wrap long lines. This could be an optional feature that does not happen by default.

Sarah Watkins (https://github.com/Swatkins18) is a new member of PL's Testing & Automation Team. She should be added as a "Member" of the Boa Constrictor repository so that she can make and review code changes.

Hello,

Two things that I have noticed that I thought were worth bringing to your attention:

1. The first link in the Screenplay pattern section in the Readme redirects to a login page.
2. SwitchWindowToLatest.cs isn't sitting in the Boa.Constrictor namespace like every other type.

I have to run tests in a flakey environment. I know I shouldn't but it is what it is.

During login I occasionally get the "The HTTP request to the remote WebDriver server for URL timed out after 60 seconds" error and attempts to re-navigate trying to recover from this fail.

I'd like to be able to close the driver, start a new one and update it on the BrowseTheWeb ability or remove the BrowseTheWeb ability and readd with the new driver.

The Webdriver on BrowseTheWeb has no setter.

Am I missing an easier way of doing this?

Improvements:

1. Move request logging to the CallRestApi Ability.
2. Improve REST API interaction names using the Actor's new Calls methods.
3. Possibly add new REST API interactions.

Recent runs of the GitHub Action that publishes the Boa Constrictor NuGet package have reported errors. Thankfully, it appears that packages still get published successfully, but we should still resolve the error.

Log: https://github.com/q2ebanking/boa-constrictor/actions/runs/775107523

These metadata badges might be of interest to potential users and contributors if included at the top of the project README. Perhaps things like "does it build", downloads, contributions, issues and dependencies could be added for a glanceable impression of the project state. https://shields.io/

Currently, CodeQL analysis runs as a GitHub Action: .github/workflows/codeql-analysis.yml

It has the following warnings:

It's a common industry practice to put info about contributing to an open source project in a file named "CONTRIBUTING.md". See https://github.blog/2012-09-17-contributing-guidelines/.

We should move our info about contributing from the README (where it currently exists) to a CONTRIBUTING file. We should keep the original README section as per Q2 guidelines but simply put a link there to the CONTRIBUTING file.

Serilog is a popular .NET logging library.

Should we use Serilog as part of Boa Constrictor? Maybe. Let's discuss that possibility here in this issue.

Pros:

• Serilog is very popular and well supported.
• Boa Constrictor wouldn't need to maintain its own logging pattern.

Cons:

• What if projects using Boa Constrictor don't want to use Serilog or already use a different logger?

While updating the doc site on March 29, 2021, GitHub's Dependabot issued an alert for the kramdown package:
https://github.com/q2ebanking/boa-constrictor/security/dependabot/docs/Gemfile.lock/kramdown/open

The Boa Constrictor doc site has a dependency on kramdown 2.3.0. Unfortunately, that version has a security vulnerability. The fix is to update to 2.3.1.

Unfortunately, the github-pages gem, which is required for hosting the site on GitHub Pages, declares an explicit dependency on kramdown (= 2.3.0):

> bundle update kramdown
Fetching gem metadata from https://rubygems.org/...........
Resolving dependencies....
Bundler could not find compatible versions for gem "kramdown":
  In Gemfile:
    kramdown (>= 2.3.1)

    github-pages (~> 213) was resolved to 213, which depends on
      kramdown (= 2.3.0)

I verified that explicit dependency here:
https://rubygems.org/gems/github-pages

GitHub Pages requires the following package versions:
https://pages.github.com/versions/

At the time of opening this issue:

• github-pages --> 213
• kramdown --> 2.3.0

I do not believe that I can "hack" kramdown 2.3.1 into the doc site. I tried playing with it locally to see if I could make something work cleanly, but that did not work. I think we must wait for GitHub Pages to update the gem versions it supports and to release an update to the github-pages gem that supports the kramdown update.

It looks like kramdown 2.3.1 was released very recently:
https://kramdown.gettalong.org/news.html

I suspect that the Dependabot security alert is very new. Hopefully, if everyone in the world is receiving this alert, GitHub Pages will put an update in place very soon. Once they do, I'll update our doc site.

Currently, new NuGet packages are built and pushed to NuGet using a private pipeline within Q2. I want to replace that with a GitHub Action to build and push a new NuGet package every time the version is updated in the .csproj file. However, the GitHub Action will need a NuGet API key to authenticate as my NuGet.org user.

This issue covers the work to add that key as a secret to this repository. Instructions for adding secrets are here:
https://docs.github.com/en/free-pro-team@latest/actions/reference/encrypted-secrets

Please name the key NUGET_API_KEY. I'll send the value for the key via direct message.

The README and TUTORIAL are great, but we should invest in more comprehensive documentation for the project. GitHub wikis would be great.

Topics to cover:

• The Screenplay Pattern
• Using Boa Constrictor
• Available Questions and Tasks
• Project information

Update the .NET projects to target .NET 5. We might want to have multiple targets.

Dependabot discovered a security vulnerability in nokogiri:
https://github.com/q2ebanking/boa-constrictor/security/dependabot/docs/Gemfile.lock/nokogiri/open

Unfortunately, GitHub Pages support only up to 1.11.2 at the time of opening this issue:
https://pages.github.com/versions/

Once GitHub Pages supports an updated version of the package, we should update the Gemfile for the Boa Constrictor doc site.

The IDumper interface doesn't include a shared Dump method. This means that users cannot genuinely make a custom dumper. We should improve that. Users might want to dump in different ways. Update the docs, too.