Recover backup capacity with verified adaptive retention

Resume pending publications under local pressure, allow verified offsite-only history as a final local recovery stage, and prune old complete remote sets against actual upload quota while preserving the newest verified copies. Document defaults, recovery limits and validation; add pressure and quota regression coverage.

lib/space.py

@@ -5,7 +5,9 @@ Only private, recognized backup scratch is disposable. Recovery journals and
 manual restore extractions always win over attempts to reclaim disk space.
 """
 import argparse
+import json
 import os
+import sys
 from pathlib import Path
 import re
 import shutil
@@ -105,6 +107,48 @@ def eligible(archives, prefix, keep, protected=False):
         print(name)
 
 
+def pending(archives, prefix):
+    if not private(archives):
+        raise ValueError('Pending uploads require private canonical archives')
+    for path in sorted(archives.iterdir()):
+        if not private(path) or not path.name.startswith(prefix+'_') or (path/'UPLOADED').exists() or (path/'UPLOADED').is_symlink():
+            continue
+        try:
+            data = backup_set.load_set(path, metadata_only=True)
+            if data['run_id'] == path.name:
+                print(path.name)
+        except (ValueError, OSError, KeyError, TypeError):
+            continue
+
+
+def remote_valid(folder, name):
+    data = backup_set.load_set(folder, metadata_only=True)
+    if data['run_id'] != name:
+        raise ValueError('Remote directory/manifest identity mismatch')
+    entries = json.loads((folder/'listing.json').read_text())
+    expected = {a['file']: a['size'] for a in data['archives']}
+    expected.update({p: (folder/p).stat().st_size for p in ('manifest.json', 'COMPLETE')})
+    actual = {}
+    for entry in entries:
+        path = entry['Path']
+        if path in actual or entry['IsDir'] or type(entry['Size']) is not int:
+            raise ValueError('Invalid remote file inventory')
+        actual[path] = entry['Size']
+    if actual != expected:
+        raise ValueError('Remote files missing, unexpected or wrong sizes')
+
+
+def quota():
+    try:
+        data = json.load(sys.stdin)
+    except (ValueError, TypeError):
+        sys.exit(1)
+    free = data.get('free') if isinstance(data, dict) else None
+    if type(free) is not int or not 0 <= free <= 10**18:
+        sys.exit(1)
+    print(free)
+
+
 def inventory(archives, prefix, keep):
     valid = uploaded_sets(archives, prefix)
     protected = set(valid[:keep])
@@ -127,10 +171,16 @@ def inventory(archives, prefix, keep):
 
 if __name__ == '__main__':
     parser = argparse.ArgumentParser()
-    parser.add_argument('mode', choices=('clean', 'eligible', 'protected', 'inventory', 'remove'))
-    parser.add_argument('paths', nargs='+')
+    parser.add_argument('mode', choices=('clean', 'eligible', 'protected', 'inventory', 'remove', 'pending', 'remote-valid', 'quota'))
+    parser.add_argument('paths', nargs='*')
     args = parser.parse_args()
-    if args.mode == 'clean':
+    if args.mode == 'quota':
+        quota()
+    elif args.mode == 'remote-valid':
+        remote_valid(Path(args.paths[0]), args.paths[1])
+    elif args.mode == 'pending':
+        pending(Path(args.paths[0]), args.paths[1])
+    elif args.mode == 'clean':
         clean(*(Path(p) for p in args.paths[:3]), args.paths[3], Path(args.paths[4]) if len(args.paths) > 4 else None)
     elif args.mode in ('eligible', 'protected'):
         eligible(Path(args.paths[0]), args.paths[1], int(args.paths[2]), args.mode == 'protected')