Script for removing duplicate files from OneDrive Documents folder [For Abhishek]

removeoddups.ps1

class RemoveDuplicates
{
    [string]$backup_dir
    [string]$suffix
    
    
    RemoveDuplicates(
        [string]$backupFolder,
        [string]$suffixString
    ) {
        $this.backup_dir = $backupFolder
        $this.suffix = $suffixString

        if (-not (Test-Path -Path $this.backup_dir)) {
            Write-Host "❕ [INFO] Creating a backup folder in $($this.backup_dir)"
            New-Item $this.backup_dir -ItemType Directory -Verbose
        }
        else {
            Write-Host "⚠️ [INFO] Backup folder $($this.backup_dir) already exists" -ForegroundColor Yellow
        }
    }
    
    [void] StartProcess([string] $folderName)
    {
        $entities = Get-ChildItem -Path $folderName -ErrorAction SilentlyContinue -Force

        $fileset = New-Object 'System.Collections.Generic.HashSet[string]'

        $non_copy_files = $entities | Where-Object { ($_ -is [System.IO.FileInfo]) -and (-not $_.BaseName.EndsWith($this.suffix)) }

        foreach ($file in $non_copy_files) {
            $fileset.Add($file.FullName)
        }

        foreach ($entry in $entities) {
            if ($entry -is [System.IO.DirectoryInfo]) {
                $this.StartProcess($entry.FullName)
            }
            else { # [System.IO.FileInfo]
                
                if ($entry.BaseName.EndsWith($this.suffix)) {
                    Write-Host "❕ [INFO] Found file: $($entry.FullName)"
                    $non_copy_name = $entry.DirectoryName+"\"+$entry.BaseName.Replace($this.suffix, "")+$entry.Extension

                    if ($fileset.Contains($non_copy_name)) { # a corresponding non_copy exists in hashset
                        # Assert that the non_copy file exists
                        if (-not (Test-Path -Path $non_copy_name)) {
                            Write-Host "❌ [ERROR] File $non_copy_name not found. Aborting." -ForegroundColor Red # This should not trigger!
                            Exit
                        }

                        Write-Host "❕ [INFO] Found duplicate file: $non_copy_name"

                        # Backup the older non_copy file
                        $non_copy_file = Get-Item -Path $non_copy_name
                        $non_copy_name_with_ts = $non_copy_file.BaseName+"_"+$non_copy_file.GetHashCode()+$non_copy_file.Extension
                        Move-Item -Path $non_copy_name -Destination "$($this.backup_dir)\$non_copy_name_with_ts" -Verbose

                        # Rename the newer _copy file
                        Rename-Item -Path $entry.FullName -NewName $non_copy_name -Verbose
                    }
                    else {
                        Write-Host "⚠️ [INFO] No corresponding duplicate file." -ForegroundColor Yellow
                    }
                }
            }
        }
    }
}

Write-Host "==== Triaging OneDrive Documents folder ====" -ForegroundColor DarkMagenta

$desktop = [System.Environment]::GetFolderPath("Desktop")
$documents = [System.Environment]::GetFolderPath("MyDocuments")

Write-Host "Found OneDrive Documents folder in $documents"
$in = Read-host "Is this correct❔ (y/n)"
if ($in.ToLower().StartsWith('n'))
{
    Write-Host "❌ [ERROR] Oops, unable to proceed. Contact Asmit." -ForegroundColor Red
    Exit
}

$backup_folder_name = "duplicate_files"
$backup_path = Join-Path -Path $desktop -ChildPath $backup_folder_name

[RemoveDuplicates]$rmdup = [RemoveDuplicates]::new($backup_path, "_copy")

$rmdup.StartProcess($documents)

Write-Host "✔️ [DONE] Triaging files completed. Check OneDrive Documents folder. Older files are available in $backup_path" -ForegroundColor Green
Write-Host "👉 [TODO] Give Asmit a treat 😎" -ForegroundColor DarkYellow