Last week I was formally invited to become a member of the JetBrains Development Academy Board – to celebrate, I decided to give something back to the community that has a JetBrains flavour. As I mentioned in my last post – we've been doing a lot of ALM / DevOps work in the last year and some of those projects have involved implementing TeamCity and other have involved using a lot of PowerShell – so I thought it would be a good idea to combine to two.
Rather than implement a PowerShell API from scratch I decided to "work smarter" and stand on the shoulders of giants – in this case fellow JetBrains Academy Member Paul Stack, who created a very nice C# TeamCity Wrapper called TeamCitySharp.
PowerShell is hosted on the CLR – so calling .NET types is a breeze – it was very straightforward to wrap the C# API. Firstly I used dotPeek to list all the method names and I pasted these into a new PowerShell script and converted each of them into the vanilla Pester BDD format and outlined some behaviours:
The next step was to create the parameters to pass into the cmdlet. To start with I mimicked the C# API and had expressive function signatures – but after writing the first couple of cmdlets did a small refactoring and switched to using splatting (one of the most unknown / underused features of PowerShell), which allowed me to create a pseudo ConnectionDetails object that is in fact a hashtable – which is a much nicer data structure for describing all the different connection options. The final step is to test the results of your cmdlet.
This test will obviously fail as we haven't implemented the Get-AllAgents cmdlet – but that's simple enough to fix:
Note the use of @PSBoundParameters – this allows you to splat the parameters passed into the current cmdlet into a nested cmdlet. Very cool indeed and saves a lot of typing.
I repeated the process for the rest of the TeamCitySharp API and created the following cmdlets:
- Get-ArtifactsAsArchive *
- Get-AllRolesByUserName *
- Get-AllServerPlugins *
- Get-AllUsers *
- Get-BuildsByBuildLocator *
- Get-ErrorBuildsByBuildConfigId *
- Get-FailedBuildsByBuildConfigId *
- Get-LastErrorBuildByBuildConfigId *
* denotes a cmdlet that has been implemented but doesn't have a passing test (mainly due to a lack of rights on the http://teamcity.codebetter.com server).
Although the tests are integration tests and a little slow to run – there is nothing more reassuring than having a full suite of specifications:
A simple example of using the TeamCityPowerShell API is as follows:
Very straight forward.
If you don't want to store your TeamCity credentials in plain text you can either enter them interactively using the following code:
Alternatively you can retrieve them disk using this PowerShell Cookbook recipe: Importing and Exporting Credentials in PowerShell
One item to note - TeamCityPowerShell depends on TeamCitySharp which is a .NET 4.0 application. By default PowerShell only supports .NET 2.0 - to enable .NET 4.0 support copy TeamCityPowerShell\SetUp\PowerShell.exe.config to the PowerShell install directory - this allows PowerShell to host the .NET 4.0 runtime.
You can find TeamCityPowershell on GitHub. If you have any feedback – please get in contact.