通过 pscustomobject 导出 csv 显示不正确的 powershell

Question

I have the following code:

    function realtest
    {
         $files = Get-ChildItem -Path 'D:\data\' -Filter *.csv
         $tester = [PSCustomObject]@{

         foreach ($file in $files)
         {
             $tempName = $file.BaseName
             $temp = Import-Csv $file
             $tester | Add-Member -MemberType NoteProperty -Name $tempName -Value $temp.$tempName
         }
         $tester
         $tester | Export-Csv "D:\result.csv" -NoTypeInformation
    }

I am trying to export a bunch of data to CSV however when it is display the data on csv it is shown as below

"E0798T102","E0798T103"
"System.Object[]","System.Object[]"

but when i do it as a print on console it displays as the below

E0798T102       E0798T103
---------       ---------
{0, 0, 0, 0...} {0, 0, 0, 0...}

Ultimately, I want E0798T102 and E0798T103 as seperate columns in the result.csv

just to note, I will have 50 csv to loop through and each should display as its own column

Answer 1

Based on your snippet, this can be significantly simplified:

function Get-Csv {
    $col = foreach ($file in Get-ChildItem -Path D:\data -Filter *.csv) {
        $csv = Import-Csv -Path $file.FullName
        [pscustomobject]@{
            $csv.($file.BaseName) = $csv
        }
    }
    $col | Export-Csv D:\result.csv -NoTypeInformation

    return $col
}

However , a csv file seems like the wrong approach because you're trying to embed objects under a header. This doesn't really work in a tabular format as you only get one layer of depth. You should either expand all the properties on your objects, or use a different format that can represent depth, like json .

The reason for your formatting woes is due to how the serialization works. You're getting a string representation of your objects.

Converting to json isn't difficult, you just trade your Export-Csv call:

$col | ConvertTo-Json -Depth 100 | Out-File -FilePath D:\result.json

Note: I specify -Depth 100 because the cmdlet's default Depth is 2 .

Answer 2

Here is an incredibly inefficient answer to your question. If left as is, it assumes your CSV files already have a header with the CSV file basename:

$CSVs = Get-ChildItem -path 'D:\data\' -filter "*.csv" -file
$headers = $CSVs.basename
$table = [System.Data.DataTable]::new("Files")
foreach ($header in $headers) {
    $table.Columns.Add($header) | out-null
}
foreach ($CSV in $CSVs) {
    #$contents = Import-Csv $CSV -Header $CSV.basename # If CSV has no header
    $contents = Import-Csv $CSV # If CSV contains header
    $rowNumber = 0
    foreach ($line in $Contents) {
        $rowcount = $table.rows.count
        if ($rowNumber -ge $rowCount) {
            $row = $table.NewRow()
            $row[$CSV.basename] = $line.$($CSV.basename)
            $table.Rows.Add($row)

        }
        else {
            $row = $table.rows[$rowNumber]
            $row[$CSV.basename] = $line.$($CSV.basename)

        }

        $rowNumber++
    }
}
$table | Export-Csv output.csv -NoTypeInformation   

You can uncomment the commented $contents line if your CSV files do not have a header. You will just have to comment out the next $contents variable assignment if you uncomment the first.