Golang URL Encode

go-urlencode

func urlencode(s string) (result string){
  for _, c := range(s) {
    if c <= 0x7f { // single byte 
      result += fmt.Sprintf("%%%X", c)
    } else if c > 0x1fffff {// quaternary byte
      result += fmt.Sprintf("%%%X%%%X%%%X%%%X",
        0xf0 + ((c & 0x1c0000) >> 18),
        0x80 + ((c & 0x3f000) >> 12),
        0x80 + ((c & 0xfc0) >> 6),
        0x80 + (c & 0x3f),
      )
    } else if c > 0x7ff { // triple byte
      result += fmt.Sprintf("%%%X%%%X%%%X",
        0xe0 + ((c & 0xf000) >> 12),
        0x80 + ((c & 0xfc0) >> 6),
        0x80 + (c & 0x3f),
      )
    } else { // double byte
      result += fmt.Sprintf("%%%X%%%X",
        0xc0 + ((c & 0x7c0) >> 6),
        0x80 + (c & 0x3f),
      )
    }
  }

  return result
}

What if you need a path component with a slash? In Perl:

> perl -MURI -E 'use utf8; my $uri = URI->new("http://example.com"); $uri->path_segments(qw(tart user foo/bâr profile)); say $uri->canonical;'          
http://example.com/tart/user/foo%2Fb%C3%A2r/profile

This lets me specify each path part separately.

Why is this required? Golang already has QueryEscape() as well as PathEscape() functions. I've written about URL Encoding in Golang and URL Decoding in Golang. Also, there are tools to URL Encode online and URL Decode online.

@callicoder

It's needed because Golang does this:

1. use url.PathEscape and get space as %20, but leave : unencoded (should be %3A), or
2. use url.QueryEscape and get space as +, but encode : as %3A

So, if you integrate with something that goes by RFC in a certain manner and encodes %20 and %3A, you are out of luck.

P.S. Of course, : is semi-legal and perhaps shouldn't be encoded, but tell that to all kinds of Java/Python libs...

To add a cherry on top: QueryEscape incorrectly escapes already escaped strings over and over again, which is straight up ridiculous.

It's 2022 and it still does this.

https://go.dev/play/p/Lwu-ixIwMB3