From 897cbdc08277ad966283fdb16a1021ef2d58673e Mon Sep 17 00:00:00 2001
From: Nick Sweeting <git@sweeting.me>
Date: Fri, 31 Jul 2026 22:36:40 -0700
Subject: [PATCH] Select archivebox account for root-owned collections

---
 archivebox/config/permissions.py     | 47 ++++++++++++++++++++++------
 archivebox/tests/test_permissions.py | 46 +++++++++++++++++++++++++++
 2 files changed, 84 insertions(+), 9 deletions(-)
 create mode 100644 archivebox/tests/test_permissions.py

diff --git a/archivebox/config/permissions.py b/archivebox/config/permissions.py
index 8667f1d0..b20fcd25 100644
--- a/archivebox/config/permissions.py
+++ b/archivebox/config/permissions.py
@@ -13,6 +13,32 @@ from rich import print
 
 #############################################################################################
 
+
+def select_archivebox_user(
+    *,
+    running_uid: int,
+    running_gid: int,
+    effective_uid: int,
+    effective_gid: int,
+    sudo_uid: int,
+    sudo_gid: int,
+    data_dir_uid: int,
+    data_dir_gid: int,
+    account_uid: int | None,
+    account_gid: int | None,
+) -> tuple[int, int]:
+    if running_uid == 0:
+        if data_dir_uid != 0:
+            return data_dir_uid, data_dir_gid
+        if account_uid is not None and account_gid is not None:
+            return account_uid, account_gid
+
+    return (
+        effective_uid or running_uid or sudo_uid,
+        effective_gid or running_gid or sudo_gid,
+    )
+
+
 DATA_DIR = Path(os.getcwd())
 
 try:
@@ -44,15 +70,18 @@ try:
 except KeyError:
     ARCHIVEBOX_ACCOUNT = None
 
-if RUNNING_AS_UID == 0 and DATA_DIR_UID == 0 or DATA_DIR_UID != 0:
-    ARCHIVEBOX_USER = DATA_DIR_UID
-    ARCHIVEBOX_GROUP = DATA_DIR_GID
-elif RUNNING_AS_UID == 0 and ARCHIVEBOX_ACCOUNT is not None:
-    ARCHIVEBOX_USER = ARCHIVEBOX_ACCOUNT.pw_uid
-    ARCHIVEBOX_GROUP = ARCHIVEBOX_ACCOUNT.pw_gid
-else:
-    ARCHIVEBOX_USER = EUID or RUNNING_AS_UID or FALLBACK_UID
-    ARCHIVEBOX_GROUP = EGID or RUNNING_AS_GID or FALLBACK_GID
+ARCHIVEBOX_USER, ARCHIVEBOX_GROUP = select_archivebox_user(
+    running_uid=RUNNING_AS_UID,
+    running_gid=RUNNING_AS_GID,
+    effective_uid=EUID,
+    effective_gid=EGID,
+    sudo_uid=SUDO_UID,
+    sudo_gid=SUDO_GID,
+    data_dir_uid=DATA_DIR_UID,
+    data_dir_gid=DATA_DIR_GID,
+    account_uid=ARCHIVEBOX_ACCOUNT.pw_uid if ARCHIVEBOX_ACCOUNT is not None else None,
+    account_gid=ARCHIVEBOX_ACCOUNT.pw_gid if ARCHIVEBOX_ACCOUNT is not None else None,
+)
 if not USER:
     try:
         # alternative method 1 to get username
diff --git a/archivebox/tests/test_permissions.py b/archivebox/tests/test_permissions.py
new file mode 100644
index 00000000..de0aecf2
--- /dev/null
+++ b/archivebox/tests/test_permissions.py
@@ -0,0 +1,46 @@
+from archivebox.config.permissions import select_archivebox_user
+
+
+def test_root_uses_archivebox_account_for_root_owned_data_dir():
+    assert select_archivebox_user(
+        running_uid=0,
+        running_gid=0,
+        effective_uid=0,
+        effective_gid=0,
+        sudo_uid=0,
+        sudo_gid=0,
+        data_dir_uid=0,
+        data_dir_gid=0,
+        account_uid=911,
+        account_gid=911,
+    ) == (911, 911)
+
+
+def test_root_preserves_existing_non_root_data_dir_owner():
+    assert select_archivebox_user(
+        running_uid=0,
+        running_gid=0,
+        effective_uid=0,
+        effective_gid=0,
+        sudo_uid=0,
+        sudo_gid=0,
+        data_dir_uid=1001,
+        data_dir_gid=1002,
+        account_uid=911,
+        account_gid=911,
+    ) == (1001, 1002)
+
+
+def test_non_root_uses_current_effective_identity():
+    assert select_archivebox_user(
+        running_uid=501,
+        running_gid=20,
+        effective_uid=501,
+        effective_gid=20,
+        sudo_uid=0,
+        sudo_gid=0,
+        data_dir_uid=0,
+        data_dir_gid=0,
+        account_uid=None,
+        account_gid=None,
+    ) == (501, 20)