I'm trying to encode a binary file with MIMEApplication in Python 3.3, as part of a multi-part MIME HTTP POST. I have a problem that character 0x0d gets reinterpreted as a newline 0xa, despite everything being set to binary bytes.
Here's a minimal test scenario, with a binary string with a 0x0d in it, getting misinterpreted:
from email.encoders import encode_noop
from email.generator import BytesGenerator
from email.mime.application import MIMEApplication
import io
app = MIMEApplication(b'Q\x0dQ', _encoder=encode_noop)
b = io.BytesIO()
g = BytesGenerator(b)
g.flatten(app)
for i in b.getvalue()[-3:]:
print("%02x " % i, end="")
print()
Output is: 51 0a 51 when it should be 51 0d 51
Note that this is to generate a binary part for a multipart http POST message.
I was able to solve my problem by putting a dummy 'marker' in as the MIMEApplication contents, then substituting in the real binary text after generating the MIME message:
from email.encoders import encode_noop
from email.generator import BytesGenerator
from email.mime.application import MIMEApplication
import io
# Actual binary "file" I want to encode (in real life, this is a file read from disk)
bytesToEncode = b'Q\x0dQ'
# unique marker that we can find and replace after message generation
binarymarker = b'GuadsfjfDaadtjhqadsfqerasdfiojBDSFGgg'
app = MIMEApplication(binarymarker, _encoder=encode_noop)
b = io.BytesIO()
g = BytesGenerator(b)
g.flatten(app, linesep='\r\n') # linesep for HTTP-compliant header line endings
# replace the marker with the actual binary data, then you have the output you want!
body = b.getvalue().replace(binarymarker, bytesToEncode)
After this, body has the value I want, without messing up the binary bytes:
b'Content-Type: application/octet-stream\r\nMIME-Version: 1.0\r\n\r\nQ\rQ'
For a multi-part message, you just assemble the multipart message first, then do the replace() at the end.