As per #1105, certifi is being end-of-lifed. Requests will use either
its own vendored bundle, or possibly (when packaged with OS distributions)
an externally packaged bundle, which can be enabled by patching
requests.certs.where().
This will permit the deletion of just that one file in order
to fall back to the [probably more accurate but less consistent]
Distro provided CA certs.
JSON *must* be encoded using UTF-8, UTF-16 or UTF-32 (see the [RFC][1]; detect the encoding based on the fact that JSON always starts with 2 ASCII characters.
[1]: http://tools.ietf.org/html/rfc4627#section-3
compat.py: relevant imports added
utils.py: then those imports are imported here.
Nothing else was changed, but I saw mentions of quoting later in
utils.py but I wasnt sure what to do about that.
This allows iter_content and iter_lines to succeed without
crashing even after the response content has been fetched
(iter_content gives you an iterator over the prefetched
content)
It uses the same logic whenever we're trying to use a list of key values and
avoid what I had started doing -- using the same pattern everywhere.
proxies, and headers should both be covered by this commit.
This adds support for lowercase environment proxy variables (which are
quite popular too). It returns proxies in a format compatible with
request's proxy parameter.
Moreover, it can be used in the request models for proxy defaults.
Signed-off-by: Rohan Jain <crodjer@gmail.com>
This is almost entirely taken from the unquote implementation in urllib,
slightly modified for the case in hand. It now deals with invalid %encodings
rather than exploding, and the code is somewhat simpler.
It's possible in some shared hosting environments that the .netrc file isn't readable by the user for the calling process. In that circumstance, just forgo the netrc authentication.
oczkers