fix(netlink): guard string.decode() against AttributeError in NLA array mode

pyroute2/netlink/__init__.py

@@ -2355,10 +2355,11 @@ def encode(self):
         def decode(self):
             nla_base_string.decode(self)
             self.value = self['value']
-            try:
-                self.value = self.value.decode('utf-8')
-            except UnicodeDecodeError:
-                pass  # Failed to decode, keep undecoded value
+            if isinstance(self.value, bytes):
+                try:
+                    self.value = self.value.decode('utf-8')
+                except UnicodeDecodeError:
+                    pass  # Failed to decode, keep undecoded value
 
     class asciiz(string):
         '''

tests/test_unit/test_nlmsg/test_string_nla_array.py

@@ -0,0 +1,67 @@
+# SPDX-License-Identifier: Apache-2.0
+#
+# Regression test for https://github.com/svinota/pyroute2/issues/1330
+#
+# NL80211_ATTR_SCAN_SSIDS is declared as *string (an NLA array of strings).
+# When the outer NLA is decoded with _nla_array=True, self.value is set to a
+# list of child cells by nla_base.decode(). The string.decode() method then
+# tried to call self.value.decode('utf-8') on that list, raising:
+#
+#     AttributeError: 'list' object has no attribute 'decode'
+#
+# The fix: guard the utf-8 decode with isinstance(self.value, bytes).
+
+import struct
+
+from pyroute2.netlink import nlmsg_atoms
+
+
+def _build_string_nla_array(*strings):
+    """Build a raw NLA payload for an array of strings.
+
+    The outer NLA header is prepended so the result can be passed directly
+    to nlmsg_atoms.string(data=...) for decoding.
+    """
+    payload = bytearray()
+    for i, s in enumerate(strings):
+        encoded = s.encode('utf-8') if s else b''
+        inner_length = 4 + len(encoded)  # 4-byte NLA header
+        payload += struct.pack('HH', inner_length, i)
+        payload += encoded
+        # Pad to 4-byte boundary
+        pad = (4 - (inner_length % 4)) % 4
+        payload += b'\x00' * pad
+
+    total_length = 4 + len(payload)
+    return bytearray(struct.pack('HH', total_length, 0)) + payload
+
+
+def test_string_nla_array_decode_no_attributeerror():
+    """string.decode() with _nla_array=True must not raise AttributeError."""
+    buf = _build_string_nla_array('', 'ssid_a', 'ssid_b')
+    instance = nlmsg_atoms.string(data=buf, offset=0)
+    instance._nla_array = True
+    instance.decode()  # must not raise
+
+
+def test_string_nla_array_value_is_list():
+    """After decoding a *string NLA, value must be a list, not raw bytes."""
+    buf = _build_string_nla_array('', 'ssid_a')
+    instance = nlmsg_atoms.string(data=buf, offset=0)
+    instance._nla_array = True
+    instance.decode()
+    assert isinstance(
+        instance.value, list
+    ), f"Expected list, got {type(instance.value).__name__}"
+
+
+def test_string_nla_plain_decode_still_returns_str():
+    """Ensure the fix does not break plain (non-array) string decode."""
+    text = 'hello'
+    encoded = text.encode('utf-8')
+    length = 4 + len(encoded)
+    buf = bytearray(struct.pack('HH', length, 0)) + bytearray(encoded)
+    instance = nlmsg_atoms.string(data=buf, offset=0)
+    instance._nla_array = False
+    instance.decode()
+    assert instance.value == text, f"Expected {text!r}, got {instance.value!r}"