URL.path: don't unquote slashes as this corrupts the path

[vdwees]

This PR addresses issue #1405

Using unquote in a blanket way on the path can result in corrupting the path if the path contains encoded slashes. This change has some precedent in https://docs.python.org/3/library/urllib.parse.html#urllib.parse.quote, where the default behaviour is to treat the slash as a special case.

Personally, I am not sure URL.path should be doing unquote at all. My preference would be to drop the unquote behaviour entirely. Nevertheless, with this PR at least URL.path does not corrupt paths.

I'd be happy to write some tests if this approach is approved.

[vdwees]

I think this change is ugly, and #1407 solves #1405 much better, so I am closing it. However, I still am concerned that if URL.path was accidentally used for constructing paths within httpx itself, wouldn't users be likely to make the mistake also?

[tomchristie]

We're very careful and deliberate around our URL API.

.path returns a string, without the query, and with any percent-encoded characters being URL escaped. That'll give you the same behaviour you'd typically see within a web application framework.

.raw_path returns full raw bytes, of the entire path (including any query string component.) No url escaping. No nonsense. This is what gets used for the TARGET part when constructing an HTTP request.