encodeURIcomponent

encode_uri_component.jl

using Printf

UNESCAPED = Set(codeunits("0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ-_.!~*'()"))

# https://tc39.es/ecma262/#sec-encodeuricomponent-uricomponent
function encode_uri_component(uri)
    isvalid(uri) || throw(ArgumentError("`encode_uri_component` can only handle valid UTF8 strings."))

    io = IOBuffer()
    for cp in codeunits(uri)
        if cp in UNESCAPED
            print(io, Char(cp))
        else
            print(io, '%')
            @printf(io, "%2X", cp)
        end
    end
    return String(take!(io))
end

@test encode_uri_component(";,/?:@&=+\$") == "%3B%2C%2F%3F%3A%40%26%3D%2B%24"
@test encode_uri_component("-_.!~*'()") == "-_.!~*'()"
@test encode_uri_component("#") == "%23"
@test encode_uri_component("ABC abc 123") == "ABC%20abc%20123"
@test encode_uri_component("шеллы") == "%D1%88%D0%B5%D0%BB%D0%BB%D1%8B"