Reputation: 1009
I'm writing a simple sample binary PowerShell module:
using System;
using System.Management.Automation;
using System.Management.Automation.Runspaces;
namespace MyModule
{
[Cmdlet(VerbsDiagnostic.Test,"SampleCmdlet")]
[OutputType(typeof(System.String))]
public class TestSampleCmdletCommand : PSCmdlet
{
[Parameter(
Mandatory = true,
Position = 0,
ValueFromPipeline = true,
ValueFromPipelineByPropertyName = true)]
public string Path { get; set; }
// This method gets called once for each cmdlet in the pipeline when the pipeline starts executing
protected override void BeginProcessing()
{
WriteVerbose("Begin!");
}
// This method will be called for each input received from the pipeline to this cmdlet; if no input is received, this method is not called
protected override void ProcessRecord()
{
WriteObject( Path );
WriteObject( System.IO.Path.GetFullPath(Path) );
}
// This method will be called once at the end of pipeline execution; if no input is received, this method is not called
protected override void EndProcessing()
{
WriteVerbose("End!");
}
}
}
When I import the module and run the code, GetFullPath() always returns a path relative to the user profile and not the real full path:
PS C:\Users\User\GitHub\MyModule\source\MyModule> Import-Module -Name .\bin\Debug\netstandard2.0\MyModule.dll
PS C:\Users\User\GitHub\MyModule\source\MyModule> cd \
PS C:\> Test-SampleCmdlet -Path .\Temp\
.\Temp\
C:\Users\User\Temp\
PS C:\>
How can I expand the path parameter properly?
Upvotes: 2
Views: 331
Reputation: 437248
.NET methods use the process-wide current directory as reflected in Environment.CurrentDirectory
, which is not the same as PowerShell's runspace-specific (thread-specific) current location.[1]
You cannot directly rely on System.IO.Path.GetFullPath()
, at least not without also specifying the runspace's current file-system location as the optional 2nd basePath
argument - which is only available in .NET Core / PowerShell [Core] 6+.
To robustly refer to the calling runspace's current file-system location, even when a different provider's location happens to be the current one[2], you can use .CurrentProviderLocation("FileSystem").ProviderPath
as follows in PowerShell [Core] 6+ / .NET Core (Path
is the property representing the cmdlet's -Path
parameter, i.e. the input path):
// Get the runspace's file-system location.
string currentDir = CurrentProviderLocation("FileSystem").ProviderPath;
// .NET Core ONLY: Resolve the given path to a full path as a native path.
// relative to the given base path.
string fullPath = System.IO.Path.GetFullPath(Path, currentDir);
When hosting Windows PowerShell more work is needed:
// Get the runspace's file-system location as a native path.
string currentDir = CurrentProviderLocation("FileSystem").ProviderPath;
// Requires `using System.Text.RegularExpressions;`
// Strips a ".\" or "./" prefix from a relative path.
string relativePathWithoutPrefix = Regex.Replace(Path, @"^.[\\/]", "")
// Caveat: Combine() will not resolve any additional, *interior* .
// or .. components, should they be present in relativePathWithoutPrefix.
string fullPath = System.IO.Path.Combine(currentDir, relativePathWithoutPrefix);
Both methods will pass a Path
value that already is a full path through.
Note:
The resulting path is by design a native file-system path, even if the runspace's current file-system location is based on a PowerShell-only drive (created with New-PSDrive
).
The approach implies that your cmdlet supports file-system paths only.
For information about how to work with PS provider locations more generally, including wildcard resolution, see this answer.
[1] This discrepancy is an unfortunate side effect of PowerShell supporting multiple runspaces (threads) inside a single process, all of which need to have their own current location - see this GitHub issue for background information.
[2] For instance, the runspace's current location may be a registry location such as HKCU:\Console
.
Upvotes: 4