fix: raise EncodeError for non-str input to encode_text

encode_text relied on `"\x00" in value` to detect embedded nulls, which
leaks raw TypeError for non-iterable inputs (None, int, float) and
leaks misleading TypeErrors for bytes/bytearray/memoryview. Lone
surrogates additionally leak UnicodeEncodeError from .encode("utf-8").

Add an isinstance(value, str) guard at the top of encode_text and wrap
the UTF-8 encode step so both failure paths surface as EncodeError,
matching the library's documented exception contract.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: antoineleclair

src/dqlitewire/types.py

@@ -88,12 +88,17 @@ def pad_to_word(size: int) -> int:
 
 def encode_text(value: str) -> bytes:
     """Encode text as null-terminated UTF-8, padded to 8-byte boundary."""
+    if not isinstance(value, str):
+        raise EncodeError(f"encode_text expected str, got {type(value).__name__}")
     if "\x00" in value:
         raise EncodeError(
             f"Text value contains embedded null byte at position {value.index(chr(0))}; "
             "null-terminated encoding would lose data"
         )
-    encoded = value.encode("utf-8") + b"\x00"
+    try:
+        encoded = value.encode("utf-8") + b"\x00"
+    except UnicodeEncodeError as e:
+        raise EncodeError(f"Text contains invalid UTF-8: {e}") from e
     padding = pad_to_word(len(encoded))
     return encoded + (b"\x00" * padding)
 

tests/test_types.py

@@ -234,6 +234,27 @@ def test_embedded_null_raises_encode_error(self) -> None:
         with pytest.raises(EncodeError, match="embedded null byte"):
             encode_text("\x00")
 
+    @pytest.mark.parametrize(
+        "bad",
+        [None, 42, b"x", 3.14, ["x"], bytearray(b"x"), memoryview(b"x")],
+    )
+    def test_non_str_raises_encode_error(self, bad: object) -> None:
+        """234: Non-string inputs must raise EncodeError, not raw TypeError.
+
+        Callers following the library's documented exception contract
+        (``except EncodeError``) would otherwise miss these failures.
+        """
+        with pytest.raises(EncodeError, match="expected str"):
+            encode_text(bad)  # type: ignore[arg-type]
+
+    def test_lone_surrogate_raises_encode_error(self) -> None:
+        """234: Lone UTF-16 surrogates are legal Python str but not UTF-8-encodable.
+
+        Must raise EncodeError (not the stdlib's UnicodeEncodeError).
+        """
+        with pytest.raises(EncodeError, match="invalid UTF-8"):
+            encode_text("\ud800")
+
     def test_decode_not_terminated_fails(self) -> None:
         with pytest.raises(DecodeError):
             decode_text(b"hello")