Import-CSV in powershell fails to treat as csv array when importing one value

Question

Import-CSV in powershell fails to treat as csv array when importing one value

MrE-2124 26

I have tested this on
PSVersion 5.1.19041.1682
PSEdition Desktop

and

PSVersion 5.1.17763.2867
PSEdition Desktop

Can anyone mimic the following results? I think there is an issue with how powershell natively treats a one line csv file with the import-csv command.

$csv_location = "C:\Scripts\CSV"  
  
### Sort and remove duplicates and write to a temp.csv  
Import-Csv -Path $csv_location\new.csv | sort user -Unique | Export-Csv -Path $csv_location\temp.csv -NoTypeInformation  
  
### Retrieve temp file to make updates  
$temp = Import-CSV $csv_location\temp.csv  
  
Write-Host $temp.count

If the new.csv has two value lines, the returned value is 2. With one value line, it cannot evaluate $temp.count and returns nothing. I think the devs need to evaluate this.

Attached is a sample file I am using. Please rename to new.csv as Microsoft does not permit .csv uploads and renames files uploaded(?). 210973-new.txt

MrE-2124 26 Reputation points

2022-06-13T16:43:25.987+00:00

If the csv has 0 values, the write-host returns 0; so it is clearly an issue with a one value line file.

Accepted answer

3 additional answers

Your answer

MrE-2124 26 Reputation points

2022-06-13T16:43:25.987+00:00

If the csv has 0 values, the write-host returns 0; so it is clearly an issue with a one value line file.

Answer 1

Rich Matheisen 47,901

If the CSV file is completely empty, the Import-CSV returns a $null. The "count" property should return a zero.
If the CSV file contains only a header row, the Import-CSV returns a $null. The "count" property should return a zero
If the CSV file contains a header row and one data row, the Import-CSV returns an object of type PSCustomObject. The "count" property should return nothing because that object type contains no "count" property and neither does its parent which is a System.Object.
If the CSV file contains one header row and two data rows, the Import-CSV returns an object of type Object[] (i.e., an array of objects). The "count" property will have a value of two.

If your code depends on an object always having a "count" property then cast the variable to an appropriate type. E.g., [array]$temp

MrE-2124 26

I wish this and Andreas' answers were it. This is what will still happen when trying to treat the one line import as an array though:

$csv_location = "C:\Scripts\CSV"  
  
### Sort and remove duplicates and write to a temp.csv  
Import-Csv -Path $csv_location\new.csv | sort user -Unique | Export-Csv -Path $csv_location\temp.csv -NoTypeInformation  
  
### Retrieve temp file to make updates  
[arrray]$temp = Import-CSV $csv_location\temp.csv  
  
Write-Host $temp.count  
  
$temp.user[0]

1
j

MrE-2124 26 Reputation points

2022-06-13T19:29:21.197+00:00

RichMatheisen-8856 answered
If the CSV file contains a header row and one data row, the Import-CSV returns an object of type PSCustomObject. The "count" property should return nothing because that object type contains no "count" property and neither does its parent which is a System.Object
If the CSV file contains one header row and two data rows, the Import-CSV returns an object of type Object[] (i.e., an array of objects). The "count" property will have a value of two.

Do you know why someone would want to use two different object types for imported csvs? How does that help anyone? Do you have an example where that would be useful?
Rich Matheisen 47,901 Reputation points

2022-06-13T19:30:16.537+00:00
You realize you're indexing the wrong object, right?

Is this what you intended -- to take the 1st item in the array and get the property named "user" from it?

$temp[0].user
Rich Matheisen 47,901 Reputation points

2022-06-13T20:07:07.907+00:00

I suppose that would depend on how you intend to treat the returned object. If your code depends on an object having a "count" property, you'll have to ensure that the object always has that property. Scalar values don't have a "count'". Try 9.count -- does that work? How about $x = "123"; $x.count? The variable "$x" does have a count property, but it isn't "3"!
MrE-2124 26 Reputation points

2022-06-13T21:10:47.557+00:00

Yes to this - 100%.

I ran both examples, but copied my response to Andreas and pasted, then wrote [arrray] into the forum script for your response, hence the typographical error.

The incorrect positional array works for a multi-line csv so I didn't notice that I had it in the wrong spot when declaring the csv as an array with the one line value.

I might be concerned about writing back to a csv with export-csv, but it is not something I am doing with the data post the import on the $temp values, and do not need to investigate further, though it probably works fine.

I appreciate the information. Thank you!

Just to reiterate what Rich said, $temp[x].user works way better than $temp.user[x] after [array]$temp has been declared for all csv size lists.

Answer 2

Andreas Baumgarten 123.4K MVP Volunteer Moderator

Hi @MrE-2124 ,

instead of $temp.count you could use Import-Csv -Path .\Junk\new.csv | Measure-Object
The value count is 1 with your example csv file.

Or this is an option as well and shows 1:

$temp = Import-Csv -Path .\Junk\new.csv | Measure-Object  
$temp.Count

----------

(If the reply was helpful please don't forget to upvote and/or accept as answer, thank you)

Regards
Andreas Baumgarten

MrE-2124 26 Reputation points

2022-06-13T18:33:56.557+00:00
Thank you for replying.

While that is a workaround to calculate and determine the count of the object. I would still need to write in exception statements for evaluating and manipulating the csv post calculation.

The rest of the script manipulates the data based on the array of $temp. If I have to write an exception because $temp is not an array during a $temp.count value of 1, I would still have to modify my script block using your workaround.

The fact of the matter is: $temp should always be an array; not get treated as a weird string value when one csv value line is presented. Here is a rough example of what occurs after the value is read:

$a= 0 if( $temp.Count -gt 0 ){ while ($a -lt $temp.count) { $reverse_sn = Reverse-CertificateSerialNumber $temp.serialnumber[$a] $a+= 1 } }

The whole block is 42 lines long, so I would have to add 42 more lines of code for the one instance of a csv line value of 1 because I cannot treat it as an array. The $temp.count could be modified, but $temp.serialnumber[$a] would return one character of the string instead of the whole value that is in the array as it does if there are two or more values.

It is a major inconvenience for csv script manipulation to behave differently for one value.
Rich Matheisen 47,901 Reputation points

2022-06-13T21:19:47.713+00:00
You're still indexing the wrong variable!

$temp[$a].serialnumber

Is the number of elements in the array important to you? You can do this in a much simpler way by making use of PowerShell's streams:

$temp | ForEach-Object{ $reverse_sn = Reverse-CertificateSerialNumber $_.serialnumber }

PowerShell's smart enough to figure out that $temp is either a scalar or an array. No need to index anything.

Answer 3

Andreas Baumgarten 123.4K MVP Volunteer Moderator

Hi @MrE-2124 ,

if you use this the BaseType of $temp is a System.Object :

$temp = Import-Csv -Path .\Junk\new.csv  
$temp.GetType()  
$temp.Count

If you want an array (BaseType = System.Array) please use this and count will work correct:

$temp = @(Import-Csv -Path .\Junk\new.csv)  
$temp.GetType()  
$temp.Count

----------

(If the reply was helpful please don't forget to upvote and/or accept as answer, thank you)

Regards
Andreas Baumgarten

MrE-2124 26

I wish this and Rich's answers were it. This is what will still happen when trying to treat the one line import as an array though:

$csv_location = "C:\Scripts\CSV"  
  
### Sort and remove duplicates and write to a temp.csv  
Import-Csv -Path $csv_location\new.csv | sort user -Unique | Export-Csv -Path $csv_location\temp.csv -NoTypeInformation  
  
### Retrieve temp file to make updates  
$temp = @(Import-CSV $csv_location\temp.csv)  
  
Write-Host $temp.count  
  
$temp.user[0]

1
j

Answer 4

Rich Matheisen 47,901

For what it's worth . . .

Here's the input CSV:

user  
User1  
User2  
User1

Here's the code (after correcting the spelling of "arrray" on line 7:

$csv_location = "C:\Junk"  
      
 ### Sort and remove duplicates and write to a temp.csv  
 Import-Csv -Path $csv_location\new.csv | sort user -Unique | Export-Csv -Path $csv_location\temp.csv -NoTypeInformation  
      
 ### Retrieve temp file to make updates  
 [array]$temp = Import-CSV $csv_location\temp.csv  
      
 Write-Host $temp.count  
      
 $temp[0].user

And here's the output:

2  
User1

Andreas Baumgarten 123.4K Reputation points MVP Volunteer Moderator

2022-06-13T20:01:34.277+00:00
Hi @Rich Matheisen ,

just because I like to learn from you:
Is there a difference between this:

$temp = @(Import-Csv -Path .\Junk\new.csv)

and that:

[array]$temp = Import-Csv -Path .\Junk\new.csv

The result of $temp.GetType() looks the same -> BaseType = System.Array

Kind regards
Andreas
Rich Matheisen 47,901 Reputation points

2022-06-13T20:17:03.31+00:00

Yes, there is a difference. The first example doesn't strongly type the variable $temp. You can, in another part of the script do this without error: $temp = @{} , making $temp an empty hash.

The second example does strongly type the variable $temp and $temp = @{} results in an array that contains a hash.
Andreas Baumgarten 123.4K Reputation points MVP Volunteer Moderator

2022-06-13T20:28:58.76+00:00

Hi @Rich Matheisen ,

thank you!

Kind regards
Andreas

Share via

Import-CSV in powershell fails to treat as csv array when importing one value

3 additional answers

Your answer