From 6d551b47af4acae4f124ca6fad7ca4c67a8b3dd9 Mon Sep 17 00:00:00 2001
From: Kevin Veen-Birkenbach <kevin@veen.world>
Date: Fri, 12 Jan 2024 21:06:47 +0100
Subject: [PATCH] Implemented new deletion modell, which distributes the
 deletions over application. See also:
 https://chat.openai.com/share/c6c35642-276a-44db-abfd-31cb71c4cec9

---
 .../files/cleanup-backups.py                  | 81 ++++++++++++++-----
 1 file changed, 61 insertions(+), 20 deletions(-)

diff --git a/roles/cleanup-backups-service/files/cleanup-backups.py b/roles/cleanup-backups-service/files/cleanup-backups.py
index 87d0d702..9848aef6 100644
--- a/roles/cleanup-backups-service/files/cleanup-backups.py
+++ b/roles/cleanup-backups-service/files/cleanup-backups.py
@@ -42,27 +42,68 @@ def deleteVersion(version_path, backups_folder_path):
     new_disc_usage_percent=psutil.disk_usage(backups_folder_path).percent
     difference_percent=old_disc_usage_percent-new_disc_usage_percent
     print("{:6.2f} %% of drive freed".format(difference_percent))
-     
-backups_folder_path=args.backups_folder_path
-for host_backup_directory_name in os.listdir(backups_folder_path):
-    host_backup_directory_path = os.path.join(backups_folder_path, host_backup_directory_name)
-    for application_directory in os.listdir(host_backup_directory_path):
-        
-        # The directory which contains all backup versions of the application
-        versions_directory = os.path.join(host_backup_directory_path, application_directory) + "/"
-                    
-        versions = os.listdir(versions_directory)
-        versions.sort(reverse=False)
-        
-        print_used_disc_space(backups_folder_path)  
-        for version in versions:
-            version_path=os.path.join(versions_directory, version)
-            if not isDirectoryDeletable(version, versions_directory, version_path):
-                continue
 
-            if isSmallerThenMaximumBackupSize(maximum_backup_size_percent,backups_folder_path):
-                deleteVersion(version_path, backups_folder_path)
-                continue
+def count_total_application_directories(backups_folder_path):
+    total_app_directories = 0
+    for host_backup_directory_name in os.listdir(backups_folder_path):
+        host_backup_directory_path = os.path.join(backups_folder_path, host_backup_directory_name)
+        total_app_directories += sum(os.path.isdir(os.path.join(host_backup_directory_path, d)) for d in os.listdir(host_backup_directory_path))
+    return total_app_directories
+
+def count_total_version_folders(backups_folder_path):
+    total_version_folders = 0
+    for host_backup_directory_name in os.listdir(backups_folder_path):
+        host_backup_directory_path = os.path.join(backups_folder_path, host_backup_directory_name)
+        for application_directory in os.listdir(host_backup_directory_path):
+            versions_directory = os.path.join(host_backup_directory_path, application_directory)
+            total_version_folders += sum(os.path.isdir(os.path.join(versions_directory, d)) for d in os.listdir(versions_directory))
+    return total_version_folders
+
+def average_version_directories_per_application(backups_folder_path,blur=-1):
+    total_app_directories = count_total_application_directories(backups_folder_path)
+    total_version_folders = count_total_version_folders(backups_folder_path)
+    
+    if total_app_directories == 0:
+        return 0
+
+    average = total_version_folders / total_app_directories
+    return int(average) - blur
+
+def getAmountOfIterration(versions,average_version_directories_per_application):
+    return len(versions) - average_version_directories_per_application
+
+def deletInterration(backups_folder_path,average_version_directories_per_application):
+    for host_backup_directory_name in os.listdir(backups_folder_path):
+        host_backup_directory_path = os.path.join(backups_folder_path, host_backup_directory_name)
+        for application_directory in os.listdir(host_backup_directory_path):
             
+            # The directory which contains all backup versions of the application
+            versions_directory = os.path.join(host_backup_directory_path, application_directory) + "/"
+                        
+            versions = os.listdir(versions_directory)
+            versions.sort(reverse=False)
+            version_iterration=0 
+            while version_iterration < getAmountOfIterration(versions,average_version_directories_per_application):
+                print_used_disc_space(backups_folder_path)
+                version = versions[version_iterration]
+                version_path=os.path.join(versions_directory, version)
+                if isDirectoryDeletable(version, versions_directory, version_path):
+                    deleteVersion(version_path, backups_folder_path)
+                version_iterration += 1 
+                     
+backups_folder_path=args.backups_folder_path
+maximum_backup_size_percent=args.maximum_backup_size_percent
+
+itteration_counter = 1
+while isSmallerThenMaximumBackupSize(maximum_backup_size_percent, backups_folder_path):
+    if itteration_counter > 200:
+        raise Exception("Iteration limit exceeded")
+
+    print(f"Delete Iteration: {itteration_counter}")
+    average_version_directories = average_version_directories_per_application(backups_folder_path)
+    print(f"Average version directories per application directory: {average_version_directories}")
+    deleteIteration(backups_folder_path, average_version_directories)    
+    itteration_counter += 1
+
 print_used_disc_space(backups_folder_path)            
 print("Cleaning up finished.")
\ No newline at end of file