Historically, I have had a hard time making my mind up between maintaining the Selenium extension to CSF.Screenplay in the main Screenplay repository and this separate one. For the sake of simplicity and discoverability, I have decided to move all of this logic into the main Screenplay repo for version 2.0 and above. This repository is now archived and will represent v1 of CSF.Screenplay.Selenium only. It will not receive any further updates.
This is an add-in to CSF.Screenplay for controlling a Selenium WebDriver from within Screenplay-based BDD tests.
The Screenplay pattern, previously known as Journey, originally came about to address some of the limitations of the well-known Page Object pattern. In many ways, it is the result of aggressively refactoring Page Objects following the SOLID principles.
Screenplay, applied to a Selenium WebDriver, describes individual user interactions as Action classes, but through the Screenplay architecture, makes it simple to compose those actions into reusable Task classes, representing higher-level interactions. In turn, there are no limits on the composition of task classes.
In order to perform test assertions, create Question types, which represent reading information from the browser. Like tasks, questions may be composed of other tasks and questions, making both high and low-level questions available with code reuse between them.
The CSF.Screenplay.Selenium library comes with a wide number of action classes built-in, which permit a wide range of low-level interactions between a 'testing user' and a web application.
Browser support for Selenium WebDriver isn't perfect. A number of web browsers have less-than-perfect support; for example, Apple Safari version 11.0 cannot make selections from HTML <select> elements, due a bug in their WebDriver implementation. In order to work around known issues, this library ships with JavaScript workarounds built-in.
Using the action classes included, if it is detected that you are requesting an action which is unsupported in the current browser, a JavaScript-based workaround is automatically used instead. This will be visible in the Screenplay Report generated at the end of the test run (assuming that you are using reporting).
CI builds are configured via both Travis (for build & test on Linux/Mono) and AppVeyor (Windows/.NET). Below are links to the most recent build statuses for these two CI platforms.
The Travis build also includes browser-based testing on the "big 5" browsers; AppVeyor tests only on the built-in installation of Firefox.
- Edge
- Internet Explorer
- Chrome
- Firefox
- Safari
| Platform | Status |
|---|---|
| Linux/Mono (Travis) |  |
| Windows/.NET (AppVeyor) |