Initial remote commit

restore_meta.py

@@ -0,0 +1,117 @@
+#!/usr/bin/env python3
+"""
+Stage 4 – Restore filesystem metadata (permissions, ownership, timestamps) to
+an already-extracted directory tree.
+
+Run this after extract_tree.py if you did not use --restore-meta there, or to
+re-apply metadata without re-extracting files.
+
+Usage:
+    python3 restore_meta.py [options] <inode> <dest_dir>
+
+Options:
+    --device DEV          Block device  [/dev/dm-0]
+    --backup-sb BLOCK     Backup superblock block number  [32768]
+    --db PATH             Optional: read metadata from SQLite DB instead of
+                          re-reading the device (faster)  [inodes.db if present]
+"""
+import argparse, ctypes, ctypes.util, os, stat, sys
+import ext4lib
+
+DEFAULT_DEV       = '/dev/dm-0'
+DEFAULT_BACKUP_SB = 32768
+
+# ── libc lutimes ──────────────────────────────────────────────────────────────
+
+_libc = ctypes.CDLL(ctypes.util.find_library('c'), use_errno=True)
+
+class _Timeval(ctypes.Structure):
+    _fields_ = [('tv_sec', ctypes.c_long), ('tv_usec', ctypes.c_long)]
+
+def _lutimes(path, atime, mtime):
+    times = (_Timeval * 2)((_Timeval(atime, 0)), (_Timeval(mtime, 0)))
+    _libc.lutimes(path.encode(), ctypes.byref(times))
+
+
+# ── per-path restore ──────────────────────────────────────────────────────────
+
+def _apply_meta(dest_path, perms, uid, gid, atime, mtime):
+    is_link = os.path.islink(dest_path)
+    try:
+        os.lchown(dest_path, uid, gid)
+    except OSError:
+        pass
+    if not is_link:
+        try:
+            os.chmod(dest_path, perms)
+        except OSError:
+            pass
+    try:
+        _lutimes(dest_path, atime, mtime)
+    except Exception:
+        pass
+
+
+def restore_one(f, sb, gdt_data, inum, dest_path):
+    try:
+        idata, slot = ext4lib.read_inode(f, sb, gdt_data, inum)
+        perms, uid, gid, atime, mtime = ext4lib.get_inode_meta(idata, slot, sb)
+        _apply_meta(dest_path, perms, uid, gid, atime, mtime)
+    except Exception as e:
+        print(f"  WARN {dest_path}: {e}", file=sys.stderr)
+
+
+# ── recursive walker ──────────────────────────────────────────────────────────
+
+def walk_and_restore(f, sb, gdt_data, inum, dest_dir, visited=None):
+    if visited is None:
+        visited = set()
+    if inum in visited:
+        return
+    visited.add(inum)
+
+    # Restore the directory itself first (to set ownership before descending)
+    restore_one(f, sb, gdt_data, inum, dest_dir)
+
+    try:
+        entries = ext4lib.read_dir_entries(f, sb, gdt_data, inum)
+    except Exception:
+        return
+
+    for name, (child_inum, ftype) in entries.items():
+        if name in ('.', '..'):
+            continue
+        safe_name = name.replace('/', '_').replace('\x00', '')
+        dest = os.path.join(dest_dir, safe_name)
+
+        if not os.path.lexists(dest):
+            continue
+
+        if os.path.isdir(dest) and not os.path.islink(dest):
+            walk_and_restore(f, sb, gdt_data, child_inum, dest, visited)
+        else:
+            restore_one(f, sb, gdt_data, child_inum, dest)
+
+    # Re-apply directory timestamps after children are written
+    restore_one(f, sb, gdt_data, inum, dest_dir)
+
+
+# ── main ──────────────────────────────────────────────────────────────────────
+
+def main():
+    parser = argparse.ArgumentParser(description='Restore ext4 metadata to extracted tree (Stage 4)')
+    parser.add_argument('inode',       type=int)
+    parser.add_argument('dest_dir')
+    parser.add_argument('--device',    default=DEFAULT_DEV)
+    parser.add_argument('--backup-sb', type=int, default=DEFAULT_BACKUP_SB)
+    args = parser.parse_args()
+
+    with open(args.device, 'rb') as f:
+        sb, gdt_data, _ = ext4lib.load_fs(f, args.backup_sb)
+        print(f"Restoring metadata: inode {args.inode} → {args.dest_dir}")
+        walk_and_restore(f, sb, gdt_data, args.inode, args.dest_dir)
+        print("Done")
+
+
+if __name__ == '__main__':
+    main()