How to get Get-ChildItem to handle path with non-breaking space

Question

I have the following code that works for most files. The input file (FoundLinks.csv) is a UTF-8 file with one file path per line. It is full paths of files on a particular drive that I need to process.

$inFiles = @()
$inFiles += @(Get-Content -Path "C:\Users\sw_admin\FoundLinks.csv")

foreach ($inFile in $inFiles) {
    Write-Host("Processing: " + $inFile)
    $objFile = Get-ChildItem -LiteralPath $inFile
    New-Object PSObject -Prop @{ 
        FullName = $objFile.FullName
        ModifyTime = $objFile.LastWriteTime
    }
} 

But even though I've used -LiteralPath, it continues to not be able to process files that have a non-breaking space in the file name.

Processing: q:\Executive\CLC\Budget\Co  2018 Budget - TO Bob (GA Prophix).xlsx
Get-ChildItem : Cannot find path 'Q:\Executive\CLC\Budget\Co  2018 Budget - TO Bob (GA Prophix).xlsx'
because it does not exist.
At ListFilesWithModifyTime.ps1:6 char:29
+     $objFile = Get-ChildItem <<<<  -LiteralPath $inFile
    + CategoryInfo          : ObjectNotFound: (Q:\Executive\CL...A Prophix).xlsx:String) [Get-ChildItem], ItemNotFound
   Exception
    + FullyQualifiedErrorId : PathNotFound,Microsoft.PowerShell.Commands.GetChildItemCommand

I know my input file has the non-breaking space in the path because I'm able to open it in Notepad, copy the offending path, paste into Word, and turn on paragraph marks. It shows a normal space followed by a NBSP just before 2018.

Is PowerShell not reading in the NBSP? Am I passing it wrong to -LiteralPath? I'm at my wit's end. I saw this solution, but in that case they are supplying the path as a literal in the script, so I can't see how I could use that approach.

I've also tried: -Encoding UTF8 parameter on Get-Content, but no difference.

I'm not even sure how I can check $inFile in the code just to confirm if it still contains the NBSP.

Grateful for any help to get unstuck!

Confirmed that $inFile has NBSP

Thank you all! As per @TheMadTechnician, I have updated the code like this, and also reduced my input file to only the one file having a problem.

$inFiles = @()
$inFiles += @(Get-Content -Path "C:\Users\sw_admin\FoundLinks.csv" -Encoding UTF8)

foreach ($inFile in $inFiles) {
    Write-Host("Processing: " + $inFile)

    # list out all chars to confirm it has an NBSP
    $inFile.ToCharArray()|%{"{0} -> {1}" -f $_,[int]$_}

    $objFile = Get-ChildItem -LiteralPath $inFile
    New-Object PSObject -Prop @{ 
        FullName = $objFile.FullName
        ModifyTime = $objFile.LastWriteTime
    }
} 

And so now I can confirm that $inFile in fact still contains the NBSP just as it gets passed to Get-ChildItem. Yet Get-ChildItem says the file does not exist.

More I've tried:

• Same if I use Get-Item instead of Get-ChildItem
• Same if I use -Path instead of -LiteralPath
• Windows explorer and Excel can deal with the file successfully.

I'm on a Windows 7 machine, Powershell 2.

Thanks again for all the responses!

Answer 1

It's still unclear why Sandra's code didn't work: PowerShell v2+ is capable of retrieving files with paths containing non-ASCII characters; perhaps a non-NTFS filesystem with different character encoding was involved?

However, the following workaround turned out to be effective:

$objFile = Get-ChildItem -Path ($inFile -replace ([char] 0xa0), '?')

• The idea is to replace the non-breaking space char. (Unicode U+00A0; hex. 0xa) in the input file path with wildcard character ?, which represents any single char.

• For Get-ChildItem to perform wildcard matching, -Path rather than -LiteralPath must be used (note that -Path is actually the default if you pass a path argument positionally, as the first argument).

• Hypothetically, the wildcard-based paths could match multiple files; if that were the case, the individual matches would have to be examined to identify the specific match that has a non-breaking space in the position of the ?.

Answer 2

Get-ChildItem is for listing children so you would be giving it a directory, but it seems you are giving it a file, so when it says it cannot find the path, it's because it can't find a directory with that name.

Instead, you would want to use Get-Item -LiteralPath to get each individual item (this would be the same items you would get if you ran Get-ChildItem on its parent.

I think swapping in Get-Item would make your code work as is.

After testing, I think the above is in fact false, so sorry for that, but I will leave the below in case it's helpful, even though it may not solve your immediate problem.

---

But let's take a look at how it can be simplified with the pipeline.

First, you're starting with an empty array, then calling a command (Get-Content) which likely already returns an array, wrapping that in an array, then concatenating it to the empty one.

You could just do:

$inFiles = Get-Content -Path "C:\Users\sw_admin\FoundLinks.csv"

Yes, there is a chance that $inFiles will contain only a single item and not an array at all.

But the nice thing is that foreach won't mind one bit!

You can do something like this and it just works:

foreach ($string in "a literal single string") {
    Write-Host $string
}

But Get-Item (and Get-ChildItem for that matter) accept pipeline input, so they accept multiple items.

That means you could do this:

$inFiles = Get-Content -Path "C:\Users\sw_admin\FoundLinks.csv" | Get-Item

foreach ($inFile in $inFiles) {
    Write-Host("Processing: " + $inFile)
    New-Object PSObject -Prop @{ 
        FullName = $inFile.FullName
        ModifyTime = $inFile.LastWriteTime
    }
} 

But even more than that, there is a pipeline-aware cmdlet for processing items, called ForEach-Object, to which you pass a [ScriptBlock], in which $_ represents the current item, so we could do it like this:

Get-Content -Path "C:\Users\sw_admin\FoundLinks.csv" | 
    Get-Item |
    ForEach-Object -Process {
        Write-Host("Processing: " + $_)
        New-Object PSObject -Prop @{ 
            FullName = $_.FullName
            ModifyTime = $_.LastWriteTime
        }
    }

All in one pipeline!

But further, you're creating a new object with the 2 properties you want.

PowerShell has a nifty cmdlet called Select-Object which takes an input object and returns a new object containing only the properties you want; this would make for a cleaner syntax:

Get-Content -Path "C:\Users\sw_admin\FoundLinks.csv" | 
    Get-Item |
    Select-Object -Property FullName,LastWriteTime

This is the power of the the pipeline passing real objects from one command to another.

I realize this last example does not write the processing message to the screen, however you could re-add that in if you wanted:

Get-Content -Path "C:\Users\sw_admin\FoundLinks.csv" | 
    Get-Item |
    ForEach-Object -Process {
        Write-Host("Processing: " + $_)
        $_ | Select-Object -Property FullName,LastWriteTime
    }

But you might also consider that many cmdlets support verbose output and try to just add -Verbose to some of your existing cmdlets. Sadly, it won't really help in this case.

One final note, when you pass items to the filesystem cmdlets via pipeline, the parameter they bind to is in fact -LiteralPath, not -Path, so your special characters are still safe.

Answer 3

I just run into the same issue. Looks like get-childitem ak gci expects the path in unicode (UTF-16). So either convert the csv file into unicode or convert the lines that include the path as unicode within your script. Testet on PS 5.1.22621.608