Headers with byte values gets sent as strings which causes double encoding errors when other clients tries to decode it.

Question

Headers with byte values gets sent as strings which causes double encoding errors when other clients tries to decode it.

Closed this issue 10 years ago · 3 comments

For example if you read a pdf into bytes, send it as a header value, and then reads the data from the java RabbitMQ client then all non-ASCII characters are gone.

I worked around it by wrapping the bytes in a bytearray and applying the fix in #2

Answer 1 · 2014-11-12T14:54:11.000Z

Thanks for your rigorous testing of this stuff and sorry you're having so many issues. I'll be looking into these today/tomorrow.

Answer 2 · 2014-11-12T15:13:16.000Z

Don't worry :)

I have a few things I always test when I start using a new library. Dealing with strings and raw bytes is something most libraries have trouble with.

Heck, I just noticed PyCharm debugging on windows crashes if my python script have a variable with a non-ascii character in its name. print('ö'.encode().decode()) results in Ã¶ if I am debugging and ö if I am not.

Answer 3 · 2014-12-12T22:29:32.000Z

Ok, I believe I've addressed this with the last changeset I pushed. Turns out that in the AMQP spec short strings are supposed to be UTF-8 encoded, but long strings are not. I've removed the forcing of UTF-8 on long-strings in the latest change set which should address your issue.