Constructing an array from a text file in Powershell - powershell

Question

I have an annoying report output (curse be to HP) that is not something I can shape by a query - it's all or nothing more or less. I would like to take two lines from each "chunk" of output and construct an array from this. I figured it would be a simple split() operation but no such luck. A sample of the output is as so:
Medium identifier : 1800010a:54bceddd:1d8c:0007
Medium label : [ARJ170L6] ARJ170L6
Location : [TapeLibrary: 24]
Medium Owner : wfukut01
Status : Poor
Blocks used [KB] : 2827596544
Blocks total [KB] : 2827596544
Usable space [KB] : 1024
Number of writes : 16
Number of overwrites : 4
Number of errors : 0
Medium initialized : 19 January 2015, 11:43:32
Last write : 26 April 2016, 21:02:12
Last access : 26 April 2016, 21:02:12
Last overwrite : 24 April 2016, 04:48:55
Protected : Permanent
Write-protected : No
Medium identifier : 1800010a:550aa81e:3a0c:0006
Medium label : [ARJ214L6] ARJ214L6
Location : External
Medium Owner : wfukut01
Status : Poor
Blocks used [KB] : 2904963584
Blocks total [KB] : 2904963584
Usable space [KB] : 0
Number of writes : 9
Number of overwrites : 7
Number of errors : 0
Medium initialized : 19 March 2015, 10:42:45
Last write : 30 April 2016, 22:14:19
Last access : 30 April 2016, 22:14:19
Last overwrite : 29 April 2016, 13:41:35
Protected : Permanent
Write-protected : No
What would be ideal is if the final output of this work would create an array somewhat similar to this:
Location UsableSpace
--------- -------------
External 0
TapeLibrary 1024
So I can (for example) query the output so that I can do operations on the data within the array:
$myvar | where-object { $_.Location -eq "TapeLibrary" }
Perhaps there are better approaches? I would be more than happy to hear them!

Answer 1

If the command is not a Powershell cmdlet like Kolob Canyon's answer then you would need to parse the text. Here's an inelegant example using -match and regex to find the lines with Location and Usable space [KB] and find the word characters after the colon.
((Get-Content C:\Example.txt -Raw) -split 'Medium identifier') | ForEach-Object {
[void]($_ -match 'Location\s+:\s(.*?(\w+).*)\r')
$Location = #($Matches.values | Where-Object {$_ -notmatch '\W'})[0]
[void]($_ -match 'Usable\sspace\s\[KB\]\s+:\s(.*?(\w+).*)\r')
$UsableSpace = #($Matches.values | Where-Object {$_ -notmatch '\W'})[0]
if ($Location -or $UsableSpace){
[PSCustomObject]#{
Location = $Location
UsableSpace = $UsableSpace
}
}
}
As this is extremely fragile and inelegant, it's much better to interact with an object where ever possible.

Answer 2

Assuming the data is as regular as it looks, you could use multiple assignment to extract the data from the array as in:
$data = 1, 2, "ignore me", 3, 10, 22, "ignore", 30
$first, $second, $null, $third, $data = $data
where the first, second and fourth array elements go into the variables, "ignore me" gets discarded in $null and the remaining data goes back into data. In your case, this would look like:
# Read the file into an array
$data = Get-Content data.txt
# Utility to fix up the data row
function FixUp ($s)
{
($s -split ' : ')[1].Trim()
}
# Loop until all of the data is processed
while ($data)
{
# Extract the current record using multiple assignment
# $null is used to eat the blank lines
$identifier,$null,$label,$location,$owner,$status,
$used,$total,$space,$writes,$overwrites,
$errors, $initialized, $lastwrite, $lastaccess,
$lastOverwrite, $protected, $writeprotected,
$null, $null, $data = $data
# Convert it into a custom object
[PSCustomObject] [ordered] #{
Identifier = fixup $identifier
Label = fixup $label
location = fixup $location
Owner = fixup $owner
Status = fixup $status
Used = fixup $used
Total = fixup $total
Space = fixup $space
Write = fixup $writes
OverWrites = fixup $overwrites
Errors = fixup $errors
Initialized = fixup $initialized
LastWrite = [datetime] (fixup $lastwrite)
LastAccess = [datetime] (fixup $lastaccess)
LastOverWrite = [datetime] (fixup $lastOverwrite)
Protected = fixup $protected
WriteProtected = fixup $writeprotected
}
}
Once you have the data extracted, you can format it any way you want

Answer 3

That looks a very regular pattern, so I'd say there are three typical approaches to this.
First: your own bucket-fill-trigger-empty parser, load lines in until you reach the next trigger ("Medium identifier"), then empty out the bucket to the pipeline and start a new one.
Something like:
$bucket = #{}
foreach ($line in Get-Content -LiteralPath C:\path\data.txt)
{
# if full, empty bucket to pipeline
if ($line -match '^Medium identifier')
{
[PSCustomObject]$bucket
$bucket = #{}
}
# add line to bucket (unless it's blank)
if (-not [string]::IsNullOrWhiteSpace($line))
{
$left, $right = $line.Split(':', 2)
$bucket[$left.Trim()] = $right.Trim()
}
}
# empty last item to pipeline
[PSCustomObject]$bucket
Adjust to taste for identifying numbers, dates, etc.
Second: a multiline regex: I tried, but can't. It would look something like:
# Not working, but for example:
$r = #'
Medium identifier : (?<MediumIdentifier>.*)
\s*
Write-protected : (?<WriteProtected>.*)
Blocks used [KB] : (?<BlockesUsed>.*)
Medium label : (?<MediumLabel>.*)
Last write : (?<LastWrite>.*)
Medium Owner : (?<MediumOwner>.*)
Usable space [KB] : (?<UsableSpaceKB>.*)
Number of overwrites : (?<NumberOfOverwrites>.*)
Last overwrite : (?<LastOverwrite>.*)
Medium identifier : (?<MediumIdentifier>.*)
Blocks total [KB] : (?<BlocksTotalKB>.*)
Number of errors : (?<NumberOfErrors>.*)
Medium initialized : (?<MediumInitialized>.*)
Status : (?<Status>.*)
Location : (?<Location>.*)
Protected : (?<Protected>.*)
Number of writes : (?<NumberOfWrites>.*)
Last access : (?<LastAccess>.*)
\s*
'#
[regex]::Matches((get-content C:\work\a.txt -Raw), $r,
[System.Text.RegularExpressions.RegexOptions]::IgnoreCase +
[System.Text.RegularExpressions.RegexOptions]::Singleline
)
Third: ConvertFrom-String - http://www.lazywinadmin.com/2014/09/powershell-convertfrom-string-and.html or https://blogs.technet.microsoft.com/ashleymcglone/2016/09/14/use-the-new-powershell-cmdlet-convertfrom-string-to-parse-klist-kerberos-ticket-output/ then after you made the template
Get-Content data.txt | ConvertFrom-String -TemplateFile .\template.txt

Answer 4

The easiest way is to use the command itself and select certain properties. If the command is a powershell cmdlet, it should return an Object.
$output = Some-HPCommand | select 'Medium label', 'Location'
Then you can access specific properties:
$output.'Medium label'
$output.Location
If you can provide the exact command, I can write this more accurately.
The biggest issue when people are learning powershell is they treat output like a String... Everything in PowerShell is object-oriented, and once you begin to think in terms of Objects, it becomes much easier to process data; in other words, always try to handle output as objects or arrays of objects. It will make your life a hell of a lot easier.

Answer 5

If each section is in the same format, i.e. the Usable space section is always 5 lines down from the location then you can use the Select-String in combination with the context parameter. Something like this:
Select-String .\your_file.txt -Pattern '(?<=Location\s*:\s).*' -Context 0, 5 | % {
New-Object psobject -Property #{
Location = (($_.Matches[0] -replace '\[|\]', '') -split ':')[0]
UsableSpace = ($_.Context.PostContext[4] -replace '^\D+(\d+)$', '$1' )
}
}