From d863a0f0300f453cc446af33edc78466cd150291 Mon Sep 17 00:00:00 2001
From: rlogwood <richard.logwood@gmail.com>
Date: Fri, 7 Jun 2024 16:36:38 -0400
Subject: [PATCH] delete all duplicates for a hash, not just the first. Control
 execution with options. Add usage message when no options specified. Make
 deletion optional and backup mandetory when performing deletion

---
 windows/frd.ps1 | 95 ++++++++++++++++++++++++++++++++++++++++---------
 1 file changed, 78 insertions(+), 17 deletions(-)

diff --git a/windows/frd.ps1 b/windows/frd.ps1
index 02bbe24..f14fe83 100755
--- a/windows/frd.ps1
+++ b/windows/frd.ps1
@@ -2,15 +2,66 @@
 
 # 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"
 
+#$check_dir = "C:\Program Files\Microsoft Office\Updates\Download\PackageFiles"
+#$backup_dir = "C:\Users\richa\tmp\backups3"
 
+# = "C:\Program Files\Microsoft Office\Updates\Download\PackageFiles",
+ # = "C:\Users\richa\tmp\backups3"
+param (
+    [string]$check_dir,
 
-# Methodolgy
-# A hash value is computed for each file, files with the same hash are considered duplicate
-# Sort the duplicate files found for a hash by their creation date in ascending order
-# Deletion canidate will be the first file in the list
+    [string]$backup_dir,
+
+    [switch]$show_dups,
+
+    [switch]$delete
+)
+
+function ShowUsage {
+    $message = @"
+
+Find and Remove Duplicates usage:
+    frd.ps1 -check_dir [-delete -backup_dir (backup location)] [-show_dups]
+     -check_dir (where to look for duplicates)
+     -backup_dir (location to backup files before deletion)
+     -delete # will delete the duplicates
+     -show_dups # will show the duplicates
+"@      
+
+    Write-Output $message
+    exit 1
+}
+
+function CheckUsage {
+
+    if (-not $delete -and -not $show_dups) {
+        Write-Output "`n*** ERROR: No work requested, bye"
+        ShowUsage
+    }
+        
+    if (-not $check_dir) {
+        Write-Output "`n*** ERROR: -check_dir option not specified"
+        ShowUsage
+    }
+
+    if ($delete -and -not $backup_dir) {
+        Write-Output "`n*** ERROR: When -delete is specified, you need to also specify the -backup_dir option"
+        ShowUsage
+    }
+
+    Write-Output "> Checking for duplicates in: $check_dir"
+    if ($delete) {
+        Write-Output "> Looking for duplicates to delete..."
+    }
+    if ($show_dups) {
+        Write-Output "> Showing Duplicates..."
+    }
+    if ($backup_dir) {
+        Write-Output "> Backing up to: $backup_dir"
+    }
+
+}
 
 
 class FoundFile {
@@ -134,7 +185,7 @@ function Get-Duplicates {
         $dup_files = $entry.Value
     
         if ($dup_files.Count -gt 1) {
-            $dups_hash[$entry.key] = $dup_files | Sort-Object -Property Creation        
+            $dups_hash[$entry.key] = $dup_files | Sort-Object -Property Creation -Descending
             #$dup_files = $dups_hash[$entry.key]
         }
     }
@@ -229,7 +280,7 @@ function DeleteDuplicateFile {
     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 
+    Remove-Item -Path $File.FullName -Force 
     Write-Output "   - removed from: $($File.Directory)"
 }
 
@@ -252,12 +303,17 @@ function ProcessDuplicates {
         if ($dup_files.Count -lt 2) {
             throw "duplicates collection contains non duplicate for entry: $($entry.Key))"
         }
+        
+        # the first file is the newest, array is sorted in descending order by creation date
+        # delete all the duplicates older than the first file, element 0
+        for ($i = 1; $i -lt $dup_files.Count; $i++) {
+            $file_to_delete = $dup_files[$i]
+            DeleteDuplicateFile -File $file_to_delete
+            $total_deleted_size += $file_to_delete.Size
+            $num_deleted += 1
+
+        }        
 
-        # 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 "Deleted $num_deleted duplicate files"
@@ -265,12 +321,17 @@ function ProcessDuplicates {
 }
 
 
-
-
+CheckUsage
 
 $files = Get-AllFilesByHash($check_dir) | Select-Object -First 10
 $dups = Get-Duplicates $files
-Show-Duplicates $dups
-ProcessDuplicates $dups
+
+if ($show_dups) {
+    Show-Duplicates $dups
+}
+
+if ($delete) {
+    ProcessDuplicates $dups
+}