Array not creating correctly in PowerShell - arrays

Question

I have a script where I get all of the folders in d:\folder\*\*\ where the name is -like "*\Log". I then split the folder paths apart to run through wmi to get the corresponding services. After that I'm wanting to split apart the PathName property from $Services so I get everything before the \xxxxx.exe and add \log to the end of the result. Eventually I'll then use those paths to do some compression and archiving of files via a gci.
For whatever reason when I run the script below I the previous loops $LocalLogVar without "log" appended and the current loops LocalLogVar with log appended. I'm sure I'm doing something wrong that's blatantly obvious to somebody out there. If somebody could point me in the right direction on this it'd be much appreciated! I also apologize for the word vomit here, I've been looking at this script all day and my brain's pretty much used up.
A couple of notes:
The number of words in the paths vary which is why I can't manually do $LocalLogVar = "$Var1\$Var2\$Var3\Log"
If I don't have the [array] in front of $LogFolders the object type becomes a string and I get the previous loop's $LocalLogVar without "log" appended combined with the current loop's $LocalLogVar
I tried doing [collections.arraylist]$LogFolders=#() with no success
c:\folder is a shortcut to d:\folder, which is why there's c:\folder\xxx and d:\folder\xxx in the list below
SplitCount is -1 because I don't want the .exe from the path, I just want the folder structure
The naming convention for the string before .exe varies so I can't use an enumerated counter.
Example of first bullet:
word7-word8 #This is the previous loop's $LocalLogVar w/o "log" appended
C:\folder\word5\word6\word9-word8\log #This is the current loop's $LocalLogVar w/ "log" appended.
Example of the second bullet:
word7-word8C:\folder\word5\word6\word9-word8\log
What I should be getting:
D:\folder\word-anotherword\word7-word8\log
D:\folder\word-anotherword\word9-word8\log
C:\folder\word1\word7-word8\log
C:\folder\word1\word9-word8\log
C:\folder\word2\word7-word8\log
C:\folder\word2\word9-word8\log
D:\folder\word2\word10-word11\log
D:\folder\word2\word12-word8\log
C:\folder\word3\word7-word8\log
C:\folder\word3\word9-word8\log
D:\folder\word4\word7-word8\log
C:\folder\word4\word9-word8\log
C:\folder\word5\word6\word7-word8\log
C:\folder\word5\word6\word9-word8\log
C:\folder\word5\word6\word7-word8\log
C:\folder\word5\word6\word9-word8\log
$Folders = Get-ChildItem D:\folder\*\*\ -Directory -Recurse -Verbose `
| Where-Object { $_.fullname -like "*\Log" }
$2 = #()
$LogFolders = #()
foreach ($folder in $folders) {
$ServName = $folder.fullname.split('\')[2]
$ServType = $folder.fullname.split('\')[3]
$ServNameCheck = "*$($ServName.replace('-',' '))*"
$ServTypeCheck = "*$($ServType.replace('-',' '))*"
$PathName = Get-WmiObject -ClassName Win32_Service `
| Where-Object { $_.caption -like "$ServNameCheck" -and $_.caption -like "$ServTypeCheck" } `
| Select-Object Name, Caption, #{n = 'PathName'; e = { ($_.PathName).trim('"') } }
$2 += $PathName
}
$Services = $2 | Sort-Object pathname | Get-Unique -AsString
foreach ($ServPath in $services.pathname) {
$LocalLogVar = #()
if (Get-Variable `
| Where-Object { $_.name -match "^Split([0-9]|10)$" }) {
Get-Variable `
| Where-Object { $_.name -match "^Split([0-9]|10)$" } | Remove-Variable -Force
}
[int]$SplitCount = $ServPath.split('\').count
[int]$SplitCountCheck = $SplitCount - 1
$x = 0
do {
New-Variable -Name "Split$x" -Value "$($ServPath.split('\')[$x])"
$RegEx = "Split$x"
$LogFolderName = Get-Variable | Where-Object { $_.name -match $RegEx } | Select-Object value
[string]$LogFolders = $LogFolderName.value.ToString()
$LocalLogVar += $LogFolders + '\'
$x++
} until ($x -eq $SplitCountCheck)
$LocalLogVar = $LocalLogVar
$LocalLogVar = $LocalLogVar + "log"
[array]$LogFolders += $LocalLogVar
}

Answer 1

Wow, so that's a script. Kind of hard to follow, since some of it seems needlessly complex. I'm not sure if it will accomplish what you're looking for, but that's because you were super vague with your folder descriptions. Do the folders always start like this:
D:\folder<Service Short Name><Service Long Name>...\logs
If not you could be in trouble. The last four items on your example list of what you expect to see don't look like they are like that. I think the way your folders are laid out are like this:
D:\folder...<Service Short Name><Service Long Name>\logs
The difference being where the extra folders are located. If they're before the service like I think they are your script will miss things.
Now, on to getting your list that you want. What I see from looking at your script is that you get a folder list for all folders under D:\folder\*\*\ named 'log'. Then you split out the 3rd and 4th folders to get a service's short name, and long name respectively. Then one by one you pull a list of all services from WMI, and filter for just the service that matches the name and caption (short name, and long name) referred to by the folders. After that you make sure you only have one listing of any given service.
Regarding this first part of the script, you can make it faster by letting the file system provider filter things for you. Instead of pulling a folder list of everything and then filtering for paths that end in '\log', you should use the -filter parameter of the Get-ChildItem cmdlet like this:
$Folders = Get-ChildItem C:\temp\*\*\ -Directory -Recurse -Verbose -Filter 'log'
Then you should query WMI one time, save the result, then pick and choose from there based on your folders. Something like:
[array]$2 = foreach ($folder in $folders) {
$ServName,$ServType = $folder.fullname.split('\')[2,3] -replace '-',' '
$PathName = $AllServices |
Where-Object { $_.caption -like "*$ServName*" -and $_.caption -like "*$ServType*" } |
Select-Object Name, Caption, #{n = 'PathName'; e = { $_.PathName -replace '^(\w\S+) .*','$1' -replace '^([''"])([^\1]+)\1.*','$2' } }
}
$Services = $2 | Sort-Object pathname | Get-Unique -AsString
I did a little regex magic to clean up the pathname instead of just .trim('"') since this gets rid of parameters in the service execution, and cleans paths that are enclosed in single quotes not just double quotes. If what you have works for you feel free to keep it, but this is a little more capable. It may be worth noting that Get-Unique is case sensitive, so 'C:\folder\word3\word9-word8' and 'C:\folder\word3\word9-Word8' are different. You might want to do a .ToUpper() on your paths before you look for unique ones.
Once you have your array of services you loop through them, splitting the file path, reassembling it, and finally adding 'log' to the end of it. That was your way to remove the executable from the path. There's a cmdlet that was designed to do just that: split-path. Use that with Join-Path and that whole last loop gets much simpler:
[array]$LogFolders = foreach ($ServPath in $services.pathname) {
Join-Path (Split-Path $ServPath) 'log'
}
Lastly, try not to use +=, since PowerShell has to rebuild the whole array each time you do that. You'll notice I moved the $Variable = bit outside the loop in places that you do that.