Comment 7 for bug 1641239

Reviewed: https://review.openstack.org/396816
Committed: https://git.openstack.org/cgit/openstack/python-glanceclient/commit/?id=03900522d4816fe5dc2958fa1eb30ab447cc8ee5
Submitter: Jenkins
Branch: master

commit 03900522d4816fe5dc2958fa1eb30ab447cc8ee5
Author: ckonstanski <email address hidden>
Date: Fri Nov 11 18:39:34 2016 -0700

    v2: Content-Type: application/octet-stream header always added

    The bug: any existing Content-Type header cannot be found because the
    call to headers.get() fails. Therefore we end up with two Content-Type
    headers because a new one (applicaion/octet-stream) gets added
    unconditionally. The cause: the strings (keys and values) in the headers
    dict are converted from unicode sequences of type <str> to utf-8
    sequences of type <bytes>. This happens in safe_encode()
    (oslo_utils/encodeutils.py:66). <str> != <bytes> even if they appear to
    have the same characters.

    Hence, for python 3.x, _set_common_request_kwargs() adds content-type
    to header even if custom content-type is set in the request.

    This results in unsupported media type exception when glance client
    is used with keystoneauth and python 3.x

    The fix: follow the directions in encode_headers().
    It says to do this just before sending the request. Honor this principle;
    do not encode headers and then perform more business logic on them.

    Change-Id: Idf6079b32f70bc171f5016467048e917d42f296d
    Closes-bug: #1641239
    Co-Authored-By: Pushkar Umaranikar <email address hidden>