FileServerDedupe/Get-ProjectFolders.ps1

<#
Script should:
1) get list of all Project Folders 
2) Determine if there's content in these folders (exlude shortcuts, lnks, single files)
3) Flag folders with content 
4) Output results to file for each server.

Each Output file should be read by the analysis script to determine locations with duplicate data. 

Output File should show:
X:\CG   |   ProjectNumber    |  FullPath | filecount

#>


$Refresh = $true #refresh all data - causes the DB to be wiped clean and all file folders to be re-analyzed. 


$PathToDB = "M:\IT\Egnyte\DuplicateFiles\WorkingRun\Dedupe.SQLite"
$OutputFolder = "M:\IT\Egnyte\DuplicateFiles\WorkingRun\Output"
$BaseDrive = 'X:\'
$ProjectRegex = '^X:\\[A-Z]{2}\W\d{2}\\\d{2}'
$OfficeList = Get-ChildItem -Path $BaseDrive -Directory -Depth 0

if ($Refresh) {
    # Reset the DB, start fresh
    $DBConnect = New-SqliteConnection -DataSource $PathToDB
    $ClearQuery = 'DELETE FROM FolderData;'
    Invoke-SqliteQuery -DataSource $PathToDB -Query $ClearQuery

    $DBConnect.close()
}




$OfficeList | ForEach-Object -parallel {
    $PathToDB_Copy_Main = $USING:PathToDB
    $ServerProgress = @{
        ID       = 1
        Activity = "Processing: " + $_.FullName
        # PercentComplete = 100
        # CurrentOperation = $_.Project
    }
    Write-Progress @ServerProgress
    #Action that will run in Parallel. Reference the current object via $PSItem and bring in outside variables with $USING:varname

    $XDrivePath = $_.FullName
    # Write-Host "Processing " $_.FullName -ForegroundColor Green
    $ProjectFolders = get-childitem $_.FullName -Directory -Depth 2 | Where-Object FullName -Match $USING:ProjectRegex

    # $ProjectFolders now contains all folders down to the 3rd level of project number - for Example, 1234-567-001. We can now generate a project number for the output file. 
    # We need to get a file count for each "last" folder - IE, 001 in the example 1234-567-001. 

    $ProjectFolders | ForEach-Object -ThrottleLimit 5 -Parallel {
        $PathToDB_Copy = $USING:PathToDB_Copy_Main

        $JobID = Get-Random
        $CurentLoopProgress = @{
            ID              = $JobID
            Activity        = "Analyzing " + $_.FullName
            PercentComplete = 10
        }

        # Write-Progress @CurentLoopProgress


        $projectNumber = ''
        # thank you ChatGPT
        $string = $_.FullName
        $regex = "\\(\d+(-\d+)*)"
        $matches = [regex]::Matches($string, $regex)
        $result = ""
        foreach ($match in $matches) {
            $result += $match.Groups[1].Value
            if ($match.Groups[2].Success) {
                $result += $match.Groups[2].Value
            }
            # if ($match.Groups[3].Success) {
            #     $result += $match.Groups[3].Value
            # }
        }
        # / Thank you Chat GPT
        # $result = $result -replace ('-','')


        if ($result.Length -lt 5 ) {
            # $CurentLoopProgress = @{
            #     ID       = $JobID
            #     Activity = $_.FullName + " Not a Project"
            # }

            # Write-Progress @CurentLoopProgress
        }
        elseif ($result.Length -ge 5) {

            $projectnumber = $result
            $DBConnect = New-SqliteConnection -DataSource $PathToDB_Copy 
            $Query = "Select Project FROM FolderData where Project = '$ProjectNumber'"
            $sql_result = Invoke-SqliteQuery -DataSource $PathToDB_Copy -Query $Query

            if (($sql_result.project -eq $null)) {
                $CurentLoopProgress = @{
                    ID              = $JobID
                    Activity        = "Loading Telemetry for " + $_.FullName
                    PercentComplete = 75
                }
        
                Write-Progress @CurentLoopProgress
    

                $result3 = [Regex]::Matches($_.FullName, "^(.*[\\\/])") # This gets everything up the last slash, thus the "parent"
                $FolderPath_Data = Get-ChildItem -Path $_.FullName -File -Depth 50 -Recurse
                $folder_Parent = $result3.value
                $folder_FileSize = ($FolderPath_Data | Measure -sum Length).sum / 1024 / 1024 / 1024
                $folder_FileCount = ($FolderPath_Data | measure).Count
                $folder_LastWrite = ($FolderPath_Data | measure LastWriteTime -Maximum).Maximum
                $folder_LastAccess = ($FolderPath_Data | measure LastAccessTime -Maximum).Maximum
                
                if ($folder_FileSize -le 0 -or $folder_FileCount -eq $null) {
                    
                    #Let's ignore Null Values
                }
                else {
                    
                    $out = [PSCustomObject]@{
                        Server         = $_.FullName.Substring(1, 4) -replace (':|\\', '')
                        Project        = $ProjectNumber.ToString()
                        Parent         = $folder_Parent
                        Path           = $_.FullName
                        FileCount      = $folder_FileCount
                        FileSize       = $folder_FileSize
                        FileLastWrite  = $folder_LastWrite
                        FileLastAccess = $folder_LastAccess 
                    } | Out-DataTable
                    
                    Invoke-SQLiteBulkCopy -DataSource $PathToDB_Copy -Table "FolderData" -DataTable $out -Force

                }
            }
            $DBConnect.Close()
        }
        Write-Progress @CurentLoopProgress -Completed
    }
    Write-Progress @ServerProgress -Completed
} -ThrottleLimit 5

$ServerProgress = @{
    ID              = 1
    Activity        = "Database Work" 
    PercentComplete = 80
    # CurrentOperation = $_.Project
}
Write-Progress @ServerProgress


## Kick off GenTables

$SQLFile_to_Run = $PSScriptRoot + '\GenTables.sql'
Invoke-SqliteQuery -DataSource $PathToDB -InputFile $SQLFile_to_Run
$SQLFile_to_Run = $PSScriptRoot + '\DuplicatePairing.sql'
Invoke-SqliteQuery -DataSource $PathToDB -InputFile $SQLFile_to_Run

$DBConnect = New-SqliteConnection -DataSource $PathToDB

foreach ($office in $OfficeList) {
    $OfficeServer = $office.Substring(1, 4) -replace (':|\\', '')

    $Query = "select * from OfficeDuplicates where Server = '$OfficeServer';"
    $sql_result = Invoke-SqliteQuery -DataSource $PathToDB -Query $Query

    if ($sql_result) {
        $OutFile = $OutputFolder + "\$OfficeServer Duplicates.xlsx"
        $sql_result | Export-Excel -Path $OutFile -AutoFilter

    }
}
# Generate MPE All Xlsx
$Query = "select * from OfficeDuplicates;"
$sql_result = Invoke-SqliteQuery -DataSource $PathToDB -Query $Query

if ($sql_result) {
    $OutFile = $OutputFolder + "\MPE All Duplicates.xlsx"
    $sql_result | Export-Excel -Path $OutFile -AutoFilter

}

# Generate Parent Non-Duplicates

$Query = "select * from NonDupParent;"
$sql_result = Invoke-SqliteQuery -DataSource $PathToDB -Query $Query

if ($sql_result) {
    $OutFile = $OutputFolder + "\MPE Non-Duplicates - Parent Folders.xlsx"
    $sql_result | Export-Excel -Path $OutFile -AutoFilter

}


# Generate Project Non-Duplicates

$Query = "select * from NonDupProject where Path NOT IN (select Path from NonDupParent);"
$sql_result = Invoke-SqliteQuery -DataSource $PathToDB -Query $Query

if ($sql_result) {
    $OutFile = $OutputFolder + "\MPE Non-Duplicates - Project Folders.xlsx"
    $sql_result | Export-Excel -Path $OutFile -AutoFilter

}


$DBConnect.Close()