Added cleaning fuction

config.yml.sample

@@ -4,6 +4,8 @@ input:
     - "~/tilde/is/supported"
 output:
   output_root: "/path/should/be/on/the/same/filesystem"
-
+cleaning:
+  min_days_to_keep: 6
+  target_size: 40G
 customization:
   time_format: "%Y-%m-%d_%H-%M-%S"

requirements.txt

@@ -1,5 +1,6 @@
-pytz
-python-dateutil
 PyYAML
 appdirs
+humanfriendly
+python-dateutil
+pytz
 reflink

tibi_hardlinks/__init__.py

@@ -1,6 +1,7 @@
 from argparse import ArgumentParser
 from pathlib import Path
 
+import humanfriendly
 import yaml
 
 from tibi_hardlinks.config import CONFIG_DATA, CONFIG_FILE_PATH
@@ -8,12 +9,17 @@ from tibi_hardlinks.backups import Backup, create_backup_directory_name
 from tibi_hardlinks.cache import cache_manager
 
 
-def find_all_property_files():
+def find_all_property_files_in_backup_path():
     for path in map(Path, CONFIG_DATA["input"]["backup_paths"]):
         if path.exists():
             yield from path.absolute().glob("*.properties")
 
 
+def find_all_property_files_in_backup_root():
+    root = Path(CONFIG_DATA["output"]["output_root"]).absolute()
+    yield from root.glob("**/*.properties")
+
+
 def hardlink_all_files(files, dry=False):
     for prop_file in files:
         try:
@@ -36,13 +42,14 @@ def hardlink_all_files(files, dry=False):
             backup.hardlink_to(dest_dir, dry)
 
 
-def clean_out(backups: list[Backup], before_date, target_size):
+def get_backups_to_remove(backups: list[Backup], before_date, target_size):
+    before_date = before_date.timestamp()
     total_size = sum(backup.size for backup in backups)
-    target_removal_size = target_size - total_size
+    target_removal_size = total_size - target_size
     if target_removal_size <= 0:
         return ()
     backups_eligable = (
-        backup for backup in backups if backup.properties["backup_time"] <= before_date
+        backup for backup in backups if backup.backup_info["backup_time"] <= before_date
     )
     sorted_backups_eligable = sorted(
         backups_eligable, key=lambda backup: backup.size, reverse=True
@@ -52,7 +59,22 @@ def clean_out(backups: list[Backup], before_date, target_size):
         to_remove.append(backup)
         target_removal_size -= backup.size
         if target_removal_size <= 0:
-            return to_remove
+            break
+
+    return to_remove
+
+
+def clean_out(property_files, days_before, target_size, dry=True):
+    before_date = datetime.datetime.utcnow() - datetime.timedelta(days=days_before)
+    to_remove = get_backups_to_remove(
+        list(map(Backup, property_files)), before_date, target_size
+    )
+    for backup in to_remove:
+        path_to_remove = backup.property_file.parent
+        if dry:
+            print(f"remove '{path_to_remove}'")
+        else:
+            shutil.rmtree(path_to_remove)
 
 
 def main():
@@ -70,6 +92,13 @@ def main():
         action="store_true",
         help="Clear the cache entirely",
     )
+    subparsers = parser.add_subparsers(dest="action")
+    subparsers.add_parser(
+        "link",
+    )
+    subparsers.add_parser(
+        "clean",
+    )
     args = parser.parse_args()
     if args.dump_config:
         print(CONFIG_FILE_PATH)
@@ -79,4 +108,13 @@ def main():
         cache_manager.invalidate_all()
     elif args.clear_already_processed:
         cache_manager.invalidate_already_processed()
-    hardlink_all_files(find_all_property_files(), args.dry)
+
+    if args.action == "link":
+        hardlink_all_files(find_all_property_files_in_backup_path(), args.dry)
+    elif args.action == "clean":
+        clean_out(
+            find_all_property_files_in_backup_root(),
+            CONFIG_DATA["cleaning"]["min_days_to_keep"],
+            humanfriendly.parse_size(CONFIG_DATA["cleaning"]["target_size"], True),
+            args.dry,
+        )

tibi_hardlinks/backups.py

@@ -97,13 +97,17 @@ class Backup:
         self.property_file = property_file
         self.backup_info = read_backup_properties(property_file)
         self.related_files = find_backup_data(self.property_file, self.backup_info)
+        self._size = None
 
     @property
     def size(self):
-        total = 0
-        for f in self.related_files.values():
-            total += f.stat().st_size
-        return total
+        if self._size is None:
+            total = 0
+            for f in self.related_files.values():
+                if f is not None:
+                    total += f.stat().st_size
+            self._size = total
+        return self._size
 
     def hardlink_to(self, dest_dir, dry):
         for filepath in self.related_files.values():