HTTP requests with Req's arguments, answered with a Req.Response.

Each function takes the same arguments as its namesake in Req: get/2 as Req.get/2, request/2 as Req.request/2, a URL or a keyword list first and options after. Req's documentation describes the options, and the response's helpers work on the result. A function returns {:ok, response} or {:error, exception}; its bang variant returns the response or raises the exception.

Host.HTTP.get!("https://api.github.com/repos/elixir-lang/elixir").body["stargazers_count"]

case Host.HTTP.post("https://example.com/hooks", json: %{event: "deployed"}) do
  {:ok, %Req.Response{status: 200}} -> :ok
  {:ok, response} -> {:error, response.status}
  {:error, exception} -> {:error, Exception.message(exception)}
end

To stream a large body, pass an into function; into: :self is refused. The function receives each chunk and returns {:cont, acc} to go on or {:halt, acc} to stop:

Host.HTTP.get!(url,
  into: fn {:data, chunk}, acc ->
    Host.File.write!("export.csv", chunk, [:append])
    {:cont, acc}
  end
)

Requests reach the public internet. One to a host that is loopback, on a private network, link-local or another reserved address, through a redirect included, comes back as {:error, %Host.HTTP.BlockedError{}}. To call a route your beamlet serves, use Host.Router.call/4, which runs it in your process.

Options that would send a request anywhere but the network, or keep it on disk, are refused with an ArgumentError naming the option: plug, adapter, unix_socket, connect_options, finch and finch_request, the disk cache, and .netrc credentials. Req's own functions are not available to your code; these are how requests leave your beamlet.