Validate data length in decode_value for NULL type

Every other fixed-size decode path (INTEGER, FLOAT, BOOLEAN, etc.)
validates that at least 8 bytes are available before returning a
consumed count of 8. The NULL path was missing this check, which
could cause misleading errors on truncated data — the error would
appear on the next column decode instead of the NULL column.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: antoineleclair

src/dqlitewire/types.py

@@ -275,6 +275,8 @@ def decode_value(data: bytes, value_type: ValueType) -> tuple[Any, int]:
     elif value_type == ValueType.BLOB:
         return decode_blob(data)
     elif value_type == ValueType.NULL:
+        if len(data) < 8:
+            raise DecodeError(f"Need 8 bytes for NULL value, got {len(data)}")
         return None, 8
     else:
         raise DecodeError(f"Unknown value type: {value_type}")

tests/test_types.py

@@ -636,3 +636,11 @@ def test_decode_null(self) -> None:
         value, consumed = decode_value(b"\x00" * 8, ValueType.NULL)
         assert value is None
         assert consumed == 8
+
+    def test_decode_null_truncated(self) -> None:
+        with pytest.raises(DecodeError, match="8 bytes"):
+            decode_value(b"\x00" * 4, ValueType.NULL)
+
+    def test_decode_null_empty(self) -> None:
+        with pytest.raises(DecodeError, match="8 bytes"):
+            decode_value(b"", ValueType.NULL)