Fix binary data corruption in vBinary for non-UTF-8 byte input

news/1356.bugfix

@@ -0,0 +1 @@
+Refactored :class:`~icalendar.prop.binary.vBinary` to store raw ``bytes`` internally, ensuring lossless round-tripping of non-UTF-8 binary data, and updated :class:`~icalendar.prop.image.Image` to handle this change correctly. @uwezkhan

src/icalendar/prop/binary.py

@@ -7,18 +7,19 @@
 from icalendar.compatibility import Self
 from icalendar.error import JCalParsingError
 from icalendar.parser import Parameters
-from icalendar.parser_tools import to_unicode
 
 
 class vBinary:
     """Binary property values are base 64 encoded."""
 
     default_value: ClassVar[str] = "BINARY"
-    params: Parameters
-    obj: str
+    obj: bytes
 
     def __init__(self, obj: str | bytes, params: dict[str, str] | None = None) -> None:
-        self.obj = to_unicode(obj)
+        if isinstance(obj, str):
+            self.obj = obj.encode("utf-8")
+        else:
+            self.obj = obj
         self.params = Parameters(encoding="BASE64", value="BINARY")
         if params:
             self.params.update(params)
@@ -27,7 +28,7 @@ def __repr__(self) -> str:
         return f"vBinary({self.to_ical()})"
 
     def to_ical(self) -> bytes:
-        return binascii.b2a_base64(self.obj.encode("utf-8"))[:-1]
+        return base64.b64encode(self.obj)
 
     @staticmethod
     def from_ical(ical: str | bytes) -> bytes:
@@ -57,12 +58,17 @@ def to_jcal(self, name: str) -> list:
         if params.get("encoding") == "BASE64":
             # BASE64 is the only allowed encoding
             del params["encoding"]
-        return [name, params, self.VALUE.lower(), self.obj]
+        return [
+            name,
+            params,
+            self.VALUE.lower(),
+            base64.b64encode(self.obj).decode("ascii"),
+        ]
 
     @property
     def ical_value(self) -> bytes:
         """The bytes value of the BINARY property."""
-        return self.from_ical(self.obj)
+        return self.obj
 
     @classmethod
     def from_jcal(cls, jcal_property: list) -> Self:
@@ -77,7 +83,7 @@ def from_jcal(cls, jcal_property: list) -> Self:
         JCalParsingError.validate_property(jcal_property, cls)
         JCalParsingError.validate_value_type(jcal_property[3], str, cls, 3)
         return cls(
-            jcal_property[3],
+            cls.from_ical(jcal_property[3]),
             params=Parameters.from_jcal_property(jcal_property),
         )
 

src/icalendar/prop/image.py

@@ -56,9 +56,13 @@ def from_property_value(cls, value: vUri | vBinary | vText) -> Image:
         if value_type == "URI" or isinstance(value, vUri):
             params["uri"] = str(value)
         elif isinstance(value, vBinary):
-            params["b64data"] = value.obj
+            params["b64data"] = base64.b64encode(value.obj).decode("ascii")
         elif value_type == "BINARY":
-            params["b64data"] = str(value)
+            params["b64data"] = (
+                value
+                if isinstance(value, str)
+                else base64.b64encode(value).decode("ascii")
+            )
         else:
             raise TypeError(
                 f"The VALUE parameter must be URI or BINARY, not {value_type!r}."

src/icalendar/tests/prop/test_vBinary.py

@@ -1,7 +1,5 @@
 """Test vBinary"""
 
-import base64
-
 import pytest
 
 from icalendar import vBinary
@@ -23,8 +21,8 @@ def test_binary():
 
 
 def test_param():
-    assert isinstance(vBinary("txt").params, Parameters)
-    assert vBinary("txt").params == {"VALUE": "BINARY", "ENCODING": "BASE64"}
+    assert isinstance(vBinary(b"txt").params, Parameters)
+    assert vBinary(b"txt").params == {"VALUE": "BINARY", "ENCODING": "BASE64"}
 
 
 def test_long_data():
@@ -61,14 +59,9 @@ def test_from_ical_rejects_non_base64_characters(value):
 
 
 def test_ical_value():
-    """ical_value property returns the string value."""
-    magic_string = base64.b64encode(b"magic string")
-    assert vBinary(magic_string).ical_value == base64.b64decode(magic_string)
-
-
-def test_ical_value_rejects_non_base64_characters():
-    with pytest.raises(ValueError, match=r"Not valid base 64 encoding\."):
-        vBinary("!!!!dGV4dA==@@@@").ical_value
+    """ical_value property returns the binary value."""
+    raw_data = b"magic string"
+    assert vBinary(raw_data).ical_value == raw_data
 
 
 def test_hash():

src/icalendar/tests/rfc_7265_jcal/test_section_3_6_values.py

@@ -30,7 +30,7 @@
 )
 
 JCAL_PAIRS = [
-    (["attach", {}, "binary", "SGVsbG8gV29ybGQh"], vBinary("SGVsbG8gV29ybGQh")),
+    (["attach", {}, "binary", "SGVsbG8gV29ybGQh"], vBinary(b"Hello World!")),
     (["x-non-smoking", {}, "boolean", True], vBoolean(True)),
     (["x-non-smoking", {}, "boolean", False], vBoolean(False)),
     (

src/icalendar/tests/test_image.py

@@ -124,8 +124,7 @@ class DummyValue:
 
 def test_create_with_vBinary():
     """Test creating an Image from a vBinary property."""
-    b64data = base64.b64encode(TRANSPARENT_PIXEL).decode("ascii")
-    vbin = vBinary(b64data, params={"FMTTYPE": "image/png"})
+    vbin = vBinary(TRANSPARENT_PIXEL, params={"FMTTYPE": "image/png"})
     image = Image.from_property_value(vbin)
     assert image.uri is None
     assert image.data == TRANSPARENT_PIXEL
@@ -147,7 +146,7 @@ def test_create_with_vUri():
 
 
 def test_create_image_with_vText_as_uri():
-    """Test that creating an image with vText but VALUE URI or BINARY raises TypeError."""
+    """Test that creating an image with vText and VALUE=URI works."""
     img = Image.from_property_value(
         vText("http://example.com/image.png", params={"VALUE": "URI"})
     )
@@ -159,7 +158,7 @@ def test_create_image_with_vText_as_uri():
 
 
 def test_create_image_with_vText_as_binary():
-    """Test that creating an image with vText but VALUE URI or BINARY raises TypeError."""
+    """Test that creating an image with vText and VALUE=BINARY works."""
     b64data = base64.b64encode(TRANSPARENT_PIXEL).decode("ascii")
     img = Image.from_property_value(vText(b64data, params={"VALUE": "BINARY"}))
     assert img.uri is None