MessagePack strips newlines in binary attachment content

[lazka]

Using zeep 4.3.2:

I'm not that deep into soap etc so I hope I'm not missing something, or my SOAP server is not doing something stupid. I get a binary response from a SOAP service which with zeep is missing a trailing newline (I know this because I also get a signature of the content in the response, and adding a newline makes the signature verification pass).

Here is an example based on the zeep test suite:

from requests_toolbelt.multipart.decoder import MultipartDecoder
from pretend import stub

from zeep.wsdl.attachments import MessagePack

data = "\r\n".join(
    line.strip()
    for line in """
    --MIME_boundary
    Content-Type: application/soap+xml; charset=UTF-8
    Content-Transfer-Encoding: 8bit
    Content-ID: <root@test>

    <soap:Envelope xmlns:soap="http://www.w3.org/2003/05/soap-envelope"/>

    --MIME_boundary
    Content-Type: application/octet-stream
    Content-Transfer-Encoding: binary
    Content-ID: <attachment@test>

    hello world

    --MIME_boundary--
""".splitlines()
).encode("utf-8")

response = stub(
    status_code=200,
    content=data,
    encoding=None,
    headers={
        "Content-Type": 'multipart/related; boundary=MIME_boundary; type="application/soap+xml"; start="<root@test>"'
    },
)

decoder = MultipartDecoder(response.content, response.headers["Content-Type"], "utf-8")
part = decoder.parts[1]
actual = MessagePack(decoder.parts[1:]).attachments[0].content

expected = b"hello world\r\n"
assert actual == expected, f"Expected {expected!r}, got {actual!r}"

I see that these are removed in

python-zeep/src/zeep/wsdl/attachments.py

Line 78 in 368861c

return content.strip(b"\r\n")

but maybe that is not considering that the binary content itself could contain those characters at the end?

thanks for zeep!