Always encode forward slashes as %2F in query parameters

[zanieb]

Summary

Closes #2721

Encodes / as %2F in query parameters for more robust behavior matching browser implementations. See issue for discussion.

Checklist

• I understand that this PR may be closed in case there was no previous discussion. (This doesn't apply to typos!)
• I've added a test for each change that was introduced, and I tried as much as possible to make a single atomic change.
• I've updated the documentation accordingly.

[tomchristie]

Okay, so we also need to change this...

def urlencode(items: typing.List[typing.Tuple[str, str]]) -> str:
    # We can use a much simpler version of the stdlib urlencode here because
    # we don't need to handle a bunch of different typing cases, such as bytes vs str.
    #
    # https://github.com/python/cpython/blob/b2f7b2ef0b5421e01efb8c7bee2ef95d3bab77eb/Lib/urllib/parse.py#L926
    #
    # Note that we use '%20' encoding for spaces, and treat '/' as a safe
    # character. This means our query params have the same escaping as other
    # characters in the URL path. This is slightly different to `requests`,
    # but is the behaviour that browsers use.
    #
    # See https://github.com/encode/httpx/issues/2536 and
    # https://docs.python.org/3/library/urllib.parse.html#urllib.parse.urlencode
    return "&".join([quote(k) + "=" + quote(v) for k, v in items])

To not use the default safe="/" value...

def urlencode(items: typing.List[typing.Tuple[str, str]]) -> str:
    # We can use a much simpler version of the stdlib urlencode here because
    # we don't need to handle a bunch of different typing cases, such as bytes vs str.
    #
    # Note that we use '%20' encoding for spaces, instead of "+".
    return "&".join([quote(k, safe="") + "=" + quote(v, safe="") for k, v in items])

[zanieb]

Thanks for the hint! I'm pretty confused how it manifested that way still :D

[ttys0dev]

This needs to be reverted, see this comment and #2883.

[zanieb]

Hi @ttys0dev — I appreciate the sentiment but that's not a great way to collaborate with us. We're considering reverting this and the existing discussion is the place to weigh in with details of why this matters to you.

[ttys0dev]

existing discussion

Oh, I had looked for existing open issues and didn't originally find one tracking this issue yet, I hadn't realized this project often puts issues under discussions(I've never really used the discussions feature before) which is a bit different from other projects.

[elupus]

This change was unnecessary. The fix was the changes below. We dont need to change an already encoded query to encode slashes. We do need to handle it when passed in as separate parameter.