rfunix/httpoison

Yet Another HTTP client for Elixir powered by hackney

HTTPoison

HTTP client for Elixir, based on HTTPotion (documentation).

Note about broken ssl in Erlang 19

Until this issue is fixed ssl handshakes may fail. If you receive this error:

{:error, %HTTPoison.Error{id: nil, reason: :closed}}

Try the following fix:

HTTPoison.get("https://example.com/", [], [ ssl: [{:versions, [:'tlsv1.2']}] ])

But... why something so similar to HTTPotion?

HTTPoison uses hackney to execute HTTP requests instead of ibrowse. I like hackney 👍

Using hackney we work only with binaries instead of string lists.

Installation

First, add HTTPoison to your mix.exs dependencies:

def deps do
  [{:httpoison, "~> 1.0"}]
end

and run $ mix deps.get. Add :httpoison to your applications list if your Elixir version is 1.3 or lower:

def application do
  [applications: [:httpoison]]
end

Usage

iex> HTTPoison.start
iex> HTTPoison.get! "http://httparrot.herokuapp.com/get"
%HTTPoison.Response{
  body: "{\n  \"args\": {},\n  \"headers\": {} ...",
  headers: [{"Connection", "keep-alive"}, {"Server", "Cowboy"},
  {"Date", "Sat, 06 Jun 2015 03:52:13 GMT"}, {"Content-Length", "495"},
  {"Content-Type", "application/json"}, {"Via", "1.1 vegur"}],
  status_code: 200
}
iex> HTTPoison.get! "http://localhost:1"
** (HTTPoison.Error) :econnrefused
iex> HTTPoison.get "http://localhost:1"
{:error, %HTTPoison.Error{id: nil, reason: :econnrefused}}

iex> HTTPoison.post "http://httparrot.herokuapp.com/post", "{\"body\": \"test\"}", [{"Content-Type", "application/json"}]
{:ok, %HTTPoison.Response{body: "{\n  \"args\": {},\n  \"headers\": {\n    \"host\": \"httparrot.herokuapp.com\",\n    \"connection\": \"close\",\n    \"accept\": \"application/json\",\n    \"content-type\": \"application/json\",\n    \"user-agent\": \"hackney/1.6.1\",\n    \"x-request-id\": \"4b85de44-6227-4480-b506-e3b9b4f0318a\",\n    \"x-forwarded-for\": \"76.174.231.199\",\n    \"x-forwarded-proto\": \"http\",\n    \"x-forwarded-port\": \"80\",\n    \"via\": \"1.1 vegur\",\n    \"connect-time\": \"1\",\n    \"x-request-start\": \"1475945832992\",\n    \"total-route-time\": \"0\",\n    \"content-length\": \"16\"\n  },\n  \"url\": \"http://httparrot.herokuapp.com/post\",\n  \"origin\": \"10.180.37.142\",\n  \"form\": {},\n  \"data\": \"{\\\"body\\\": \\\"test\\\"}\",\n  \"json\": {\n    \"body\": \"test\"\n  }\n}",
    headers: [{"Connection", "keep-alive"}, {"Server", "Cowboy"},
    {"Date", "Sat, 08 Oct 2016 16:57:12 GMT"}, {"Content-Length", "681"},
    {"Content-Type", "application/json"}, {"Via", "1.1 vegur"}],
status_code: 200}}

You can also easily pattern match on the HTTPoison.Response struct:

case HTTPoison.get(url) do
  {:ok, %HTTPoison.Response{status_code: 200, body: body}} ->
    IO.puts body
  {:ok, %HTTPoison.Response{status_code: 404}} ->
    IO.puts "Not found :("
  {:error, %HTTPoison.Error{reason: reason}} ->
    IO.inspect reason
end

Options

There are a number of supported options(not to be confused with the HTTP options method), documented here, that can be added to your request. The example below shows the use of the :ssl and :recv_timeout options for a post request to an api that requires a bearer token. The :ssl option allows you to set options accepted by th Erlang SSL module, and :recv_timeout sets a timeout on receiving a response, the default is 5000ms.

token = "some_token_from_another_request"
url = "https://example.com/api/endpoint_that_needs_a_bearer_token"
headers = ["Authorization": "Bearer #{token}", "Accept": "Application/json; Charset=utf-8"]
options = [ssl: [{:versions, [:'tlsv1.2']}], recv_timeout: 500]
{:ok, response} = HTTPoison.get(url, headers, options)

And the example below shows the use of the :ssl options for a post request to an api that requires a client certification.

url = "https://example.org/api/endpoint_that_needs_client_cert"
options = [ssl: [certfile: "certs/client.crt"]]
{:ok, response} = HTTPoison.post(url, [], options)

Wrapping HTTPoison.Base

You can also use the HTTPoison.Base module in your modules in order to make cool API clients or something. The following example wraps HTTPoison.Base in order to build a client for the GitHub API (Poison is used for JSON decoding):

defmodule GitHub do
  use HTTPoison.Base

  @expected_fields ~w(
    login id avatar_url gravatar_id url html_url followers_url
    following_url gists_url starred_url subscriptions_url
    organizations_url repos_url events_url received_events_url type
    site_admin name company blog location email hireable bio
    public_repos public_gists followers following created_at updated_at
  )

  def process_url(url) do
    "https://api.github.com" <> url
  end

  def process_response_body(body) do
    body
    |> Poison.decode!
    |> Map.take(@expected_fields)
    |> Enum.map(fn({k, v}) -> {String.to_atom(k), v} end)
  end
end

iex> GitHub.start
iex> GitHub.get!("/users/myfreeweb").body[:public_repos]
37

It's possible to extend the functions listed below:

defp process_request_body(body), do: body

defp process_response_body(body), do: body

defp process_request_headers(headers) when is_map(headers) do
  Enum.into(headers, [])
end

defp process_request_headers(headers), do: headers

defp process_request_options(options), do: options

defp process_response_chunk(chunk), do: chunk

defp process_headers(headers), do: headers

defp process_status_code(status_code), do: status_code

defp process_url(url), do: url

Async requests

HTTPoison now comes with async requests!

iex> HTTPoison.get! "https://github.com/", %{}, stream_to: self
%HTTPoison.AsyncResponse{id: #Reference<0.0.0.1654>}
iex> flush
%HTTPoison.AsyncStatus{code: 200, id: #Reference<0.0.0.1654>}
%HTTPoison.AsyncHeaders{headers: %{"Connection" => "keep-alive", ...}, id: #Reference<0.0.0.1654>}
%HTTPoison.AsyncChunk{chunk: "<!DOCTYPE html>...", id: #Reference<0.0.0.1654>}
%HTTPoison.AsyncEnd{id: #Reference<0.0.0.1654>}
:ok

Warning: this option can flood a receiver in messages.

If a server may send very large messages the async: :once option should be used. This will send only a single chunk at a time the receiver can call HTTPoison.stream_next/1 to indicate ability to process more chunks.

Cookies

HTTPoison allows you to send cookies:

iex> HTTPoison.get!("http://httparrot.herokuapp.com/cookies", %{}, hackney: [cookie: ["session=a933ec1dd923b874e691; logged_in=true"]])
%HTTPoison.Response{body: "{\n  \"cookies\": {\n    \"session\": \"a933ec1dd923b874e691\",\n    \"logged_in\": \"true\"\n  }\n}",
 headers: [{"Connection", "keep-alive"}, ...],
 status_code: 200}

You can also receive cookies from the server by reading the "set-cookie" headers in the response:

iex(1)> response = HTTPoison.get!("http://httparrot.herokuapp.com/cookies/set?foo=1")
iex(2)> cookies = Enum.filter(response.headers, fn
...(2)> {"Set-Cookie", _} -> true
...(2)> _ -> false
...(2)> end)
[{"Set-Cookie", "foo=1; Version=1; Path=/"}]

You can see more usage examples in the test files (located in the test/) directory.

Connection Pools

Normally hackney opens and closes connections on demand, but it also creates a default pool of connections which are reused for requests to the same host. If the connection and host support keepalive, the connection is kept open until explicitly closed.

To use the default pool, you can just declare it as an option:

HTTPoison.get("httpbin.org/get", [], hackney: [pool: :default])

It is possible to use different pools for different purposes when a more fine grained allocation of resources is necessary.

Simple pool declaration

The easiest way is to just pass the name of the pool, and hackney will create it if it doesn't exist. Pools are independent from each other (they won't compete for connections) and are created with the default configuration.

HTTPoison.get("httpbin.org/get", [], hackney: [pool: :first_pool])
HTTPoison.get("httpbin.org/get", [], hackney: [pool: :second_pool])

Explicit pool creation

If you want to use different configuration options you can create a pool manually when your app starts with :hackney_pool.start_pool/2.

:ok = :hackney_pool.start_pool(:first_pool, [timeout: 15000, max_connections: 100])

From the already linked hackney's readme:

timeout is the time we keep the connection alive in the pool, max_connections is the number of connections maintained in the pool. Each connection in a pool is monitored and closed connections are removed automatically.

Pools as supervised processes

A third option is to add the pool as part of your supervision tree:

children = [
  :hackney_pool.child_spec(:first_pool, [timeout: 15000, max_connections: 100])
]

Add that to the application supervisor and first_pool will be available to be used by HTTPoison/hackney.

License

Copyright © 2013-2014 Eduardo Gurgel <eduardo@gurgel.me>

This work is free. You can redistribute it and/or modify it under the
terms of the MIT License. See the LICENSE file for more details.