Re: Boost.URL: Something Wicked This Way Comes! (Was: Needs Review)

From: Alex Christensen (achristensen_at_[hidden])
Date: 2021-10-12 21:52:06

> On Oct 12, 2021, at 3:15 PM, Vinnie Falco <vinnie.falco_at_[hidden]> wrote:
> 
> On Tue, Oct 12, 2021 at 1:02 PM Alex Christensen <achristensen_at_[hidden]> wrote:
>> at some point you may run into issues with people trying to give
>> your library input like http://example.com/%f0%9f%92© and expecting the
>> URL parser to normalize it to http://example.com/%F0%9F%92%A9
> 
> Okay, I think what you're saying is that you will have this string literal:
> 
>    string_view s = "http://example.com/\xf0\x9f\x92\xa9";
> 
> Unfortunately, this is not a valid URL and I don't think that the
> library should accept this input.
It is perfectly valid input that some URL libraries I work with accept and percent encode, and some URL libraries I work with reject it as an invalid URL.  I think it’s a valid URL parser input that ought to produce a valid URL, but not everyone agrees on this yet.

> However, you could write this:
> 
>    url u = parse_uri( "http://example.com" ).value();
> 
>    u.set_path( "/\xf0\x9f\x92\xa9" );
> 
> This will produce:
> 
>    assert( u.encoded_url() == "http://example.com/%f0%9f%92%a9" );
> 
> Is this what you meant?
Welcome to the crazy world of URL parsing!

> I'm guessing that the turd emoji is inserted
> into the C++ source file as a utf-8 encoded code point, so that's what
> you get in the string literal.
> 
> Thanks