FIX: validate VariationSelectorSmuggler base character

pyrit/converter/token_smuggling/variation_selector_smuggler_converter.py

@@ -47,10 +47,17 @@ def __init__(
                                     Default is True.
 
         Raises:
-            ValueError: If an unsupported action or ``encoding_mode`` is provided.
+            ValueError: If an unsupported action is provided, or ``base_char_utf8`` is not exactly one character
+                or is a variation selector used to encode payload bytes.
         """
         super().__init__(action=action)
-        self.utf8_base_char = base_char_utf8 if base_char_utf8 is not None else "😊"
+        base_char = base_char_utf8 if base_char_utf8 is not None else "😊"
+        if len(base_char) != 1:
+            raise ValueError("base_char_utf8 must be exactly one character.")
+        code_point = ord(base_char)
+        if 0xFE00 <= code_point <= 0xFE0F or 0xE0100 <= code_point <= 0xE01EF:
+            raise ValueError("base_char_utf8 must not be a variation selector.")
+        self.utf8_base_char = base_char
         self.embed_in_base = embed_in_base
 
     def _build_identifier(self) -> ComponentIdentifier:

tests/unit/converter/test_variation_selector_smuggler_converter.py

@@ -43,6 +43,31 @@ def test_variation_selector_invalid_action():
         VariationSelectorSmugglerConverter(action="invalid")
 
 
+@pytest.mark.parametrize("base_char", ["", "ab", "😊x"])
+def test_variation_selector_invalid_base_char(base_char: str) -> None:
+    with pytest.raises(ValueError, match="base_char_utf8 must be exactly one character"):
+        VariationSelectorSmugglerConverter(base_char_utf8=base_char)
+
+
+@pytest.mark.parametrize("base_char", ["\ufe00", "\ufe0f", "\U000e0100", "\U000e01ef"])
+def test_variation_selector_rejects_selector_base_char(base_char: str) -> None:
+    with pytest.raises(ValueError, match="base_char_utf8 must not be a variation selector"):
+        VariationSelectorSmugglerConverter(base_char_utf8=base_char)
+
+
+@pytest.mark.parametrize("base_char", ["A", "\ufdff", "\ufe10", "\U000e00ff", "\U000e01f0"])
+@pytest.mark.parametrize("embed_in_base", [True, False])
+async def test_variation_selector_custom_base_char_roundtrip(base_char: str, embed_in_base: bool) -> None:
+    encoder = VariationSelectorSmugglerConverter(action="encode", base_char_utf8=base_char, embed_in_base=embed_in_base)
+    encoded = await encoder.convert_async(prompt="test", input_type="text")
+
+    decoder = VariationSelectorSmugglerConverter(action="decode", base_char_utf8=base_char, embed_in_base=embed_in_base)
+    decoded = await decoder.convert_async(prompt=encoded.output_text, input_type="text")
+
+    assert encoded.output_text.startswith(base_char)
+    assert decoded.output_text == "test"
+
+
 async def test_variation_selector_input_not_supported():
     converter = VariationSelectorSmugglerConverter(action="encode")
     with pytest.raises(ValueError, match="Input type not supported"):