testing deletion and backup process

windows/frd.ps1

@@ -3,6 +3,9 @@
 # This is the directory we'll check for duplicates
 # TODO: make this a parameter
 $check_dir = "C:\Program Files\Microsoft Office\Updates\Download\PackageFiles"
+$backup_dir = "C:\Users\richa\tmp\backups"
+
+
 
 # Methodolgy
 # A hash value is computed for each file, files with the same hash are considered duplicate
@@ -12,6 +15,7 @@ $check_dir = "C:\Program Files\Microsoft Office\Updates\Download\PackageFiles"
 
 class FoundFile {
     [string]$Name
+    [string]$FullName
     [string]$Directory
     [int]$Size
     [double]$fileSizeInMB
@@ -21,6 +25,7 @@ class FoundFile {
 
     FoundFile([System.IO.FileInfo] $file, [string]$hash) {
         $this.Name = $file.Name
+        $this.FullName = $file.FullName
         $this.Directory = $file.DirectoryName
         $this.Size = $file.Length
         $this.fileSizeInMB = [math]::Round($this.Size / 1MB, 4)
@@ -143,7 +148,6 @@ function Show-Duplicates {
         [hashtable] $files
     )
 
-    #$hash_num = 0
     $num_dups = 0
     $total_dup_size = 0
     
@@ -161,71 +165,103 @@ function Show-Duplicates {
             $total_dup_size += $dup_files[$i].Size
             Write-Output "$($hash_num): $($dup_files[$i].ToString())"
         }        
-    
-        #$hash_num += 1
     }
     Write-Output ""
     Write-Output "Found $num_dups duplicate file hashes."
     ShowSizes "Duplicate Files Size" $total_dup_size
-    #Write-Output "total dup size = $(Format-NumberWithCommas $total_dup_size) Bytes,  $(Format-NumberWithCommas $total_dup_mb) MB"
 }
 
-function Show-Duplicates-v1 {    
+
+function BackupDuplicateFile {
+    param (
+        [Parameter(Mandatory = $true)]
+        [FoundFile] $file,
+
+        [Parameter(Mandatory = $true)]
+        [string] $backupDirectory
+    )
+
+    # Define the source file path
+    $sourceFilePath = Join-Path -Path $file.Directory -ChildPath $file.Name
+
+    # Get the FileInfo object for the source file
+    $fileInfo = Get-Item -Path $sourceFilePath
+
+    # Define the backup directory
+    Write-Output "backup directory is: $backupDirectory"
+
+    # Ensure the backup directory exists
+    if (-not (Test-Path -Path $backupDirectory)) {
+        New-Item -Path $backupDirectory -ItemType Directory
+    }
+
+    # Remove the drive letter using Split-Path
+    $pathWithoutDrive = Split-Path -Path $File.Directory -NoQualifier
+
+    # Construct the full backup file path
+    #$fullBackupDirectory = Join-Path -Path $backupDirectory -ChildPath $File.Directory
+    $fullBackupDirectory = Join-Path -Path $backupDirectory -ChildPath $pathWithoutDrive
+    
+
+
+    if (-not (Test-Path -Path $fullBackupDirectory)) {
+        New-Item -Path $fullBackupDirectory -ItemType Directory
+    }
+
+    # Copy the file to the backup directory
+
+    $fullDestinationPath = Join-Path -Path $fullBackupDirectory -ChildPath $file.Name
+    $fileInfo.CopyTo($fullDestinationPath, $true)
+
+    # Output the result
+    # Write-Output "File '$($fileInfo.FullName)'"
+    Write-Output "   + backed up to: $fullBackupDirectory"
+}
+
+function DeleteDuplicateFile {
+    param (
+        [Parameter(Mandatory = $true)]
+        [FoundFile] $File
+    )
+    
+    Write-Output ""
+    Write-Output "Deleting File: $($File.Name)  $(Format-NumberWithCommas $File.fileSizeInKB) KB"
+    BackupDuplicateFile -file $File -backupDirectory $backup_dir    
+    # Force delete a read-only or protected file
+    Write-Output "  ... would deleting: $($File.FullName)"
+    #Remove-Item -Path $File.FullName -Force 
+    Write-Output "   - removed from: $($File.Directory)"
+}
+
+
+
+
+# NOTE: Duplicates are pre-sorted by creation date
+function ProcessDuplicates {
     param (
         [Parameter(Mandatory = $true)]
         [hashtable] $files
     )
 
-    $hash_num = 0
-    $total_dup_size = 0
-    $num_dups = 0
+    $num_deleted = 0
+    $total_deleted_size = 0
+    
+    Write-Output ""
     foreach ($entry in $files.GetEnumerator()) {
         $dup_files = $entry.Value
+        if ($dup_files.Count -lt 2) {
+            throw "duplicates collection contains non duplicate for entry: $($entry.Key))"
+        }
 
-        if ($dup_files.Count -gt 1) {
-            $num_dups++
+        # delete the first duplicate in the array of dups, they are sorted by creation date ascending
+        $oldest_dup = $dup_files[0]
+        DeleteDuplicateFile -File $oldest_dup
+        $total_deleted_size += $oldest_dup.Size
+        $num_deleted += 1
+    }
     Write-Output ""
-            Write-Output "Found duplicate $($num_dups): $($entry.Key)"
-            for ($i = 0; $i -lt $dup_files.Count; $i++) {
-                $total_dup_size += $dup_files[$i].Size
-                Write-Output "$($hash_num): $($dup_files[$i].ToString())"
-            }    
-        }
-        $hash_num += 1
-    }
-    #$total_dup_mb = [math]::Round($total_dup_size / 1MB, 4)
-
-    Write-Output ""
-    Write-Output "Found $num_dups duplicate file hashes."
-    ShowSizes "Duplicate Files Size" $total_dup_size
-    #Write-Output "total dup size = $(Format-NumberWithCommas $total_dup_size) Bytes,  $(Format-NumberWithCommas $total_dup_mb) MB"
-    
-}
-
-function TBD_Get-Duplicates {
-    throw "not implemented"
-    # need to sort the dups by date before deleteing 
-
-    $dupsTable = @{}
-
-    $total_dup_size = 0
-
-    # Find and remove duplicate files
-    foreach ($hash in $hashTable.Keys) {
-        $fileGroup = $hashTable[$hash]
-        if ($fileGroup.Count -gt 1) {
-            # Keep the first file and delete the rest
-            $fileGroup[1..($fileGroup.Count - 1)] | ForEach-Object {
-                Write-Output "Would Delete duplicate file: $_"
-                $file = Get-Item -Path $_
-                $fileSize = $file.Length
-                $fileName = $file.Name
-                $total_dup_size += $fileSize
-                $dupsTable[$fileName] = $fileSize
-                # Remove-Item -Path $_ -Force
-            }
-        }
-    }
+    Write-Output "Deleted $num_deleted duplicate files"
+    ShowSizes "Recovered Size" $total_deleted_size
 }
 
 
@@ -235,4 +271,6 @@ function TBD_Get-Duplicates {
 $files = Get-AllFilesByHash($check_dir) | Select-Object -First 10
 $dups = Get-Duplicates $files
 Show-Duplicates $dups
+ProcessDuplicates $dups
+