1

I'm experiencing a weird issue where I need to remove "blank characters" which are probably not spaces or tabs.

It's information from WMI. The "SerialNumberID" property is stored as a UInt16[] Array.

I gather it like this :

$SERIAL = (Get-WmiObject -ComputerName XXXX WmiMonitorID -Namespace root\wmi ).SerialNumberID

This now contains the following array:

72
86
76
81
66
48
50
53
52
50
0
0
0
0
0
0

To convert that to a readable format, I'm doing :

$SERIAL_AS_STRING = [System.Text.Encoding]::Default.GetString($SERIAL)

"+$SERIAL_AS_STRING+"

(I'm using the + character to easily identify if there are characters after the data)

This returns :

+HVLQB02542      +

I want to have just the serial number, so i'm trying to get rid of the spaces like this :

$SERIAL_AS_STRING = $SERIAL_AS_STRING.Trim()

"+$SERIAL_AS_STRING+"

and it STILL returns with the blank spaces! :

+HVLQB02542      +

Even if I try something like :

$SERIAL_AS_STRING = $SERIAL_AS_STRING -replace '(^\s+|\s+$)','' -replace '\s+',' '

the spaces are always there.

Now I see that in the System.UInt16 array I have SIX trailing zeros, which precisely fit with the SIX trailing spaces in my result string....

So I tried a different method to convert the array which is :

$SERIAL2 = ($SERIAL -notmatch 0 | ForEach{[char]$_}) -join ""

"+$SERIAL2+"

This works to remove the trailing spaces however it's unnaceptable because it messes up the serial like this :

+HVLQB054+

Which should be : +HVLQB02542+

What am I doing wrong here? It's strange because if i specially ask to replace SIX spaces, it doesn't work either. This is why i'm suspecting these "spaces" are not really space characters!

For now I'm making it work by using RegEx, which removes anything that is not letters or numbers :

$FINAL = $SERIAL_AS_STRING -replace "[^A-Za-z0-9]"

But how can it be done properly, as this feels like a workaround solution?

Dharman
  • 30,962
  • 25
  • 85
  • 135
Rakha
  • 1,874
  • 3
  • 26
  • 60

3 Answers3

2

What .SerialNumberID outputs is an int [uint16[]] array (see the docs).

If you want to convert these uint16 instances to a string, you can simply interpret them directly as Unicode UTF-16 code points (the Unicode equivalent of an ASCII code, of which it is a superset), which is how characters ([char]) instances are represented in memory in .NET.

Thus, you can simply cast to [char[]] and join -join the resulting character array to form a [string].

(That is, there is no need for [System.Text.Encoding]::Default.GetString(), which, incidentally, would interpret the values as bytes.[1])

Your .SerialNumberID has trailing 0 instances, however, which both [char[]] and [System.Text.Encoding]::Default.GetString() retain in the conversion, as NUL characters.

You can remove the unwanted 0 / NUL instances by simply filtering them out from the input array with -ne 0 before the conversion.

To put it all together:

# Simulate the result of your (Get-WmiObject ...).SerialNumberID call
$SERIAL = [uint16[]] (72, 86, 76, 81, 66, 48, 50, 53, 52, 50, 0, 0, 0, 0, 0, 0)

$SERIAL_AS_STRING = -join [char[]] ($SERIAL -ne 0)

# Show the resulting string
"+$SERIAL_AS_STRING+"

The above yields +HVLQB02542+, as desired.

[1] It is only if the input array represents a specific character encoding other than UTF-16 that you need the .GetString() method of the appropriate [System.Text.Encoding] instance; note that that method requires a [byte[]] array, and while PowerShell will try to automatically coerce a [uint16[]] to that, doing so will fail if any of the array elements have a value greater than 255 (0xFF), i.e., are too large to fit into a [byte].

mklement0
  • 382,024
  • 64
  • 607
  • 775
  • @Rakha: My pleasure. It was I who flagged your comment as no longer needed, and a moderator removed it as a result. To be clear, I genuinely appreciated your nice feedback; since I perceived your comment as a personal thank-you, I thought it had served its purpose once I had read it. To future readers it is only comments with technical information that are of interest. Similarly, if I should thank _you_ in a comment in the future or again respond to a thank-you from you, I encourage you to flag that comment as no longer needed as well after reading it - including this one. – mklement0 Mar 06 '19 at 13:16
1

You can check whether or not those are actually spaces by converting the type of one of the space characters and checking the ASCII code:

$a = '+HVLQB02542      +'

Get the second-to-last character and check it's return value:

[byte][char]$a[16]

which returns 32, which is the ASCII code for a space

With the other method you had using the -replace operator, if you take out the space between the last two single quotes, you should end up with the desired result:

$a -replace '(^\s+|\s+$)','' -replace '\s+',''

returns: +HVLQB02542+

You could also do $a.replace(' ','') and should yield the same as above

I get errors when trying to do [System.Text.Encoding]::Default.GetString($a) though $a initially does return as a UInt16[]

trebleCode
  • 2,134
  • 19
  • 34
  • Hmm thanks for the help but it's not 32, it returns 0. If you copy paste from my post I bet it would return as a space (32) but when I actually run the commands, [byte][char]$SERIAL_AS_STRING[16] returns the ASCII code 0 so all the code you give as examples is not working. I guess my question is how can I use -replace and specify a ASCII code value to replace? – Rakha Mar 01 '19 at 18:35
  • This previous post suggests some options that might work: https://stackoverflow.com/questions/15343442/powershell-find-replace-pattern-of-ascii-control-characters – trebleCode Mar 01 '19 at 18:36
1

On a similar note... Convert System.UInt16[] array to integer example:
$rv=(Get-CIMInstance Win32_SystemEnclosure).ChassisTypes; [int]$as= -join [int[]] ($rv -ne 0);

Cyprian Wyatt
  • 11
  • 1