Parsing HTTP Link header with NimbleParsec

Questions & Help Questions
1

Hi - I implemented HTTP Link header parser with nimble_parsec. I know there is ex_link_header but just took it as a chance to use the parser library :slight_smile:

Here is the gist link - any feedback would be welcome!

  • As , is not allowed, I may just split the whole string with , - but not sure I can leverage existing tool to handle optionally escaped string (e.g. Plug.Conn.Utils.params)
  • Probably I can switch all string with ascii_char since only ascii chars are allowed in the value
  • Not sure I’m using wrap/unwrap correctly :wink:
  • I do not want to create another “micro” library. What would be better place to put this? plug, phoenix, standalone package to cover HTTP headers
2

Okay, now I added LinkParser.parse_naive/1 to just use core modules (e.g. String, Enum, …).

It passes some tests, but it doesn’t implement something - e.g. it allows any chars in the token (compared to NimbleParsec version checks allowed chars)

3

NimbleParsec can output the compiled module, so users don’t need nimble_parsec as dependency, if that’s the reasoning here.

4

haha, I already did (initially published to hex @ March 30, 2019):

but it has a generator as well as a parser. The forum showed me this thread as related when I was creating that thread about it :slight_smile:

5

Another naive example at Tesla.Middleware.DecodeRels:"

src

  defp rels(link) do
    link
    |> String.split(",")
    |> Enum.map(&String.trim/1)
    |> Enum.map(&rel/1)
    |> Enum.into(%{})
  end

  defp rel(item) do
    Regex.run(~r/\A<(.+)>; rel="(.+)"\z/, item, capture: :all_but_first)
    |> Enum.reverse()
    |> List.to_tuple()
  end