Search code examples
powershellreplacepathenvironment-variablespath-manipulation

How can I find the potential source environment variable for a partial path in PowerShell?


I want to write a function that converts regular path to path that includes environment variables:

For example:

C:\Windows\SomePath

convert to:

%Windir%\SomePath

How would I do that and is this possible?

Here is what I'm trying to do, but problem is, I need to check the string for all possible variables, is there some more automatic way? such that -replace operator wont be needed

function Format-Path
{
    param (
        [parameter(Mandatory = $true)]
        [string] $FilePath
    )

    if (![System.String]::IsNullOrEmpty($FilePath))
    {
        # Strip away quotations and ending backslash
        $FilePath = $FilePath.Trim('"')
        $FilePath = $FilePath.TrimEnd('\\')
    }

$FilePath = $FilePath -replace "C:\\Windows", "%Windir%"
$FilePath = $FilePath -replace "C:\\ProgramFiles", "%ProgramFiles%"
$FilePath = $FilePath -replace "C:\\ProgramFiles (x86)", "%ProgramFiles (x86)%"
# ETC.. the list goes on..

return $FilePath
}

# test case
Format-Path '"C:\Windows\SomePath\"'

Output is:

%Windir%\SomePath

EDIT: Invalid input or bad code isn't really the problem because in the end the $Path can be easily checked with:

Test-Path -Path ([System.Environment]::ExpandEnvironmentVariables($FilePath))

Solution

  • The code below is my take on this. There's some peculiarities with the path and backslash manipulations, so I tried to explain everything in comments.

    There is one key take away and that is that unbounded string searches such as those performed by -replace, -like, .Contains(), etc. and can produce undesirable results when the value of one variable's path is a substring of either another variable's path or a directory's path. For example, given %ProgramFiles% (C:\Program Files) and %ProgramFiles(x86)% (C:\Program Files (x86)), the path C:\Program Files (x86)\Test could be transformed into %ProgramFiles% (x86)\Test instead of %ProgramFiles(x86)%\Test if %ProgramFiles% happens to be tested before %ProgramFiles(x86)%.

    The solution is to only compare a variable's path to a complete path segment. That is, in the case of the path C:\Program Files (x86)\Test, the comparisons would go like this...

    • Test for equality with the original path C:\Program Files (x86)\Test. No variables match.
    • Test for equality with the parent path C:\Program Files (x86). %ProgramFiles(x86)% matches. No further ancestor paths (i.e. C:) are tested.
    • %ProgramFiles% will never match because the partial path C:\Program Files is not tested.

    By only testing against complete path segments it does not matter in what order the variables are compared to the candidate path.

    New-Variable -Name 'VariablesToSubstitute' -Option Constant -Value @(
        # Hard-code system variables that contain machine-wide paths
        'CommonProgramFiles',
        'CommonProgramFiles(x86)',
        'ComSpec',
        'ProgramData',            # Alternatively: ALLUSERSPROFILE
        'ProgramFiles',
        'ProgramFiles(x86)',
        'SystemDrive'
        'SystemRoot'              # Alternatively: WinDir
    
        'MyDirectoryWithoutSlash' # Defined below
        'MyDirectoryWithSlash'    # Defined below
    );
    
    function Format-Path
    {
        param (
            [parameter(Mandatory = $true)]
            [string] $FilePath
        )
    
        if (![System.String]::IsNullOrEmpty($FilePath))
        {
            # Strip away quotations
            $FilePath = $FilePath.Trim('"')
            # Leave trailing slashes intact so variables with a trailing slash will match
            #$FilePath = $FilePath.TrimEnd('\')
        }
    
        # Initialize this once, but only after the test code has started
        if ($null -eq $script:pathVariables)
        {
            $script:pathVariables = $VariablesToSubstitute | ForEach-Object -Process {
                $path = [Environment]::GetEnvironmentVariable($_)
                if ($null -eq $path)
                {
                    Write-Warning -Message "The environment variable ""$_"" is not defined."
                }
                else
                {
                    return [PSCustomObject] @{
                        Name = $_
                        Path = $path
                    }
                }
            }
        }
    
        # Test against $FilePath and its ancestors until a match is found or the path is empty.
        # Only comparing with complete path segments prevents performing partial substitutions
        # (e.g. a path starting with %ProgramFiles(x86)% being substituted with %ProgramFiles%, 
        #       or "C:\Windows.old" being transformed to "%SystemRoot%.old")
        for ($filePathAncestorOrSelf = $FilePath;
            -not [String]::IsNullOrEmpty($filePathAncestorOrSelf);
            # Split-Path -Parent removes the trailing backslash on the result *unless* the result
            # is a drive root.  It'd be easier to normalize all paths without the backslash, but
            # Split-Path throws an error if the input path is a drive letter with no slash, so
            # normalize everything *with* the backslash and strip it off later.
            $filePathAncestorOrSelf = EnsureTrailingBackslash (
                # Protect against the case where $FilePath is a drive letter with no backslash
                # We have to do this here because we want our initial path above to be
                # exactly $FilePath, not (EnsureTrailingBackslash $FilePath).
                Split-Path -Path (EnsureTrailingBackslash $filePathAncestorOrSelf) -Parent
            )
        )
        {
            # Test against $filePathAncestorOrSelf with and without a trailing backslash
            foreach ($candidatePath in $filePathAncestorOrSelf, $filePathAncestorOrSelf.TrimEnd('\'))
            {
                foreach ($variable in $pathVariables)
                {
                    if ($candidatePath -ieq $variable.Path)
                    {
                        $variableBasePath = "%$($variable.Name)%"
                        # The rest of the path after the variable's path
                        $pathRelativeToVariable = $FilePath.Substring($variable.Path.Length)
    
                        # Join-Path appends a trailing backslash if the child path is empty - we don't want that
                        if ([String]::IsNullOrEmpty($pathRelativeToVariable))
                        {
                            return $variableBasePath
                        }
                        # Join-Path will join the base and relative path with a slash,
                        # which we don't want if the variable path already ends with a slash
                        elseif ($variable.Path -like '*\')
                        {
                            return $variableBasePath + $pathRelativeToVariable
                        }
                        else
                        {
                            return Join-Path -Path $variableBasePath -ChildPath $pathRelativeToVariable
                        }
                    }
                }
            }
        }
    
        return $FilePath
    }
    
    function EnsureTrailingBackslash([String] $path)
    {
        return $(
            # Keep an empty path unchanged so the for loop will terminate properly
            if ([String]::IsNullOrEmpty($path) -or $path.EndsWith('\')) {
                $path
            } else {
                "$path\"
            }
        )
    }
    

    Using this test code...

    $Env:MyDirectoryWithoutSlash = 'C:\My Directory'
    $Env:MyDirectoryWithSlash    = 'C:\My Directory\'
    
    @'
    X:
    X:\Windows
    X:\Windows\system32
    X:\Windows\system32\cmd.exe
    X:\Windows.old
    X:\Windows.old\system32
    X:\Windows.old\system32\cmd.exe
    X:\Program Files\Test
    X:\Program Files (x86)\Test
    X:\Program Files (it's a trap!)\Test
    X:\My Directory
    X:\My Directory\Test
    '@ -split "`r`n?" `
        | ForEach-Object -Process {
            # Test the path with the system drive letter
            $_ -replace 'X:', $Env:SystemDrive
    
            # Test the path with the non-system drive letter
            $_
        } | ForEach-Object -Process {
            $path = $_.TrimEnd('\')
    
            # Test the path without a trailing slash
            $path
    
            # If the path is a directory (determined by the
            # absence of an extension in the last segment)...
            if ([String]::IsNullOrEmpty([System.IO.Path]::GetExtension($path)))
            {
                # Test the path with a trailing slash
                "$path\"
            }
        } | ForEach-Object -Process {
            [PSCustomObject] @{
                InputPath  = $_
                OutputPath = Format-Path $_
            }
        }
    

    ...I get this result...

    InputPath                             OutputPath
    ---------                             ----------
    C:                                    %SystemDrive%
    C:\                                   %SystemDrive%\
    X:                                    X:
    X:\                                   X:\
    C:\Windows                            %SystemRoot%
    C:\Windows\                           %SystemRoot%\
    X:\Windows                            X:\Windows
    X:\Windows\                           X:\Windows\
    C:\Windows\system32                   %SystemRoot%\system32
    C:\Windows\system32\                  %SystemRoot%\system32\
    X:\Windows\system32                   X:\Windows\system32
    X:\Windows\system32\                  X:\Windows\system32\
    C:\Windows\system32\cmd.exe           %ComSpec%
    X:\Windows\system32\cmd.exe           X:\Windows\system32\cmd.exe
    C:\Windows.old                        %SystemDrive%\Windows.old
    X:\Windows.old                        X:\Windows.old
    C:\Windows.old\system32               %SystemDrive%\Windows.old\system32
    C:\Windows.old\system32\              %SystemDrive%\Windows.old\system32\
    X:\Windows.old\system32               X:\Windows.old\system32
    X:\Windows.old\system32\              X:\Windows.old\system32\
    C:\Windows.old\system32\cmd.exe       %SystemDrive%\Windows.old\system32\cmd.exe
    X:\Windows.old\system32\cmd.exe       X:\Windows.old\system32\cmd.exe
    C:\Program Files\Test                 %ProgramFiles%\Test
    C:\Program Files\Test\                %ProgramFiles%\Test\
    X:\Program Files\Test                 X:\Program Files\Test
    X:\Program Files\Test\                X:\Program Files\Test\
    C:\Program Files (x86)\Test           %ProgramFiles(x86)%\Test
    C:\Program Files (x86)\Test\          %ProgramFiles(x86)%\Test\
    X:\Program Files (x86)\Test           X:\Program Files (x86)\Test
    X:\Program Files (x86)\Test\          X:\Program Files (x86)\Test\
    C:\Program Files (it's a trap!)\Test  %SystemDrive%\Program Files (it's a trap!)\Test
    C:\Program Files (it's a trap!)\Test\ %SystemDrive%\Program Files (it's a trap!)\Test\
    X:\Program Files (it's a trap!)\Test  X:\Program Files (it's a trap!)\Test
    X:\Program Files (it's a trap!)\Test\ X:\Program Files (it's a trap!)\Test\
    C:\My Directory                       %MyDirectoryWithoutSlash%
    C:\My Directory\                      %MyDirectoryWithSlash%
    X:\My Directory                       X:\My Directory
    X:\My Directory\                      X:\My Directory\
    C:\My Directory\Test                  %MyDirectoryWithSlash%Test
    C:\My Directory\Test\                 %MyDirectoryWithSlash%Test\
    X:\My Directory\Test                  X:\My Directory\Test
    X:\My Directory\Test\                 X:\My Directory\Test\
    

    Note that candidate ancestor paths are always searched first with a trailing slash and then without. This means that in the unlikely event there are two variable paths that differ only by the presence or absence of a trailing slash, the variable with the trailing slash will be matched. Thus, as seen above, C:\My Directory\Test will become %MyDirectoryWithSlash%Test, which looks a little strange. By reversing the order of the first foreach loop in the function from...

    foreach ($candidatePath in $filePathAncestorOrSelf, $filePathAncestorOrSelf.TrimEnd('\'))
    

    ...to...

    foreach ($candidatePath in $filePathAncestorOrSelf.TrimEnd('\'), $filePathAncestorOrSelf)
    

    ...the relevant output changes to this...

    InputPath                             OutputPath
    ---------                             ----------
    ...                                   ...
    C:\My Directory\                      %MyDirectoryWithoutSlash%\
    ...                                   ...
    C:\My Directory\Test                  %MyDirectoryWithoutSlash%\Test
    C:\My Directory\Test\                 %MyDirectoryWithoutSlash%\Test\
    ...                                   ...