The HTTP client for Elixir
Latest commit dbddfc4 Oct 15, 2016 @ctreatma ctreatma committed with Fix #100: Follow redirects when extended (#101)
API clients that are built by extending `HTTPotion.Base`
can use the `process_options` hook to ensure that all
redirects will be followed automatically. A typo in that
module caused `follow_redirects` to be ignored.

This adds a test to specifically cover the behavior of
an API client using `process_options` to inject a `true`
value for `follow_redirects`, and fixes the typo.

HTTPotion version downloads API Docs Build Status unlicense

HTTP client for Elixir, based on ibrowse. Continues the HTTPun tradition of HTTParty, HTTPretty, HTTParrot and HTTPie.


Add HTTPotion to your project's dependencies in mix.exs:

  defp deps do
      {:httpotion, "~> 3.0.2"}

  def application do
    [ applications: [:httpotion] ]
    # Application dependency auto-starts it, otherwise: HTTPotion.start

And fetch your project's dependencies:

$ mix deps.get


Note: You can load HTTPotion into the Elixir REPL by executing this command from the root of your project:

$ iex -S mix

Some basic examples:

iex> response = HTTPotion.get ""
%HTTPotion.Response{body: "...", headers: [Connection: "keep-alive", ...], status_code: 200}

iex> HTTPotion.Response.success?(response)

# HTTPotion also supports querystrings like
iex> HTTPotion.get("", query: %{page: 2})
%HTTPotion.Response{body: "...", headers: [Connection: "keep-alive", ...], status_code: 200}

# Form data
iex> "", [body: "hello=" <> URI.encode_www_form("w o r l d !!"),
  headers: ["User-Agent": "My App", "Content-Type": "application/x-www-form-urlencoded"]]
%HTTPotion.Response{body: "...", headers: [Connection: "keep-alive", ...], status_code: 200}

iex> HTTPotion.request :propfind, "", [body: "I have no idea what I'm doing"]
%HTTPotion.Response{body: "...", headers: [Connection: "keep-alive", ...], status_code: 405}

iex> HTTPotion.get "", [basic_auth: {"foo", "bar"}]
%HTTPotion.Response{body: "...", headers: ["Access-Control-Allow-Credentials": "true", ...], status_code: 200}

# Passing options to ibrowse (note that it usually takes char_lists, not elixir strings)
iex> HTTPotion.get "", [ ibrowse: [ proxy_host: 'fc81:6134:ba6c:8458:c99f:6c01:6472:8f1e', proxy_port: 8118 ] ]
%HTTPotion.Response{body: "...", headers: [Connection: "keep-alive", ...], status_code: 200}

# The default timeout is 5000 ms, but can be changed
iex> HTTPotion.get "", [timeout: 10_000]

# If there is an error a `HTTPotion.ErrorResponse` is returned
iex> HTTPotion.get "http://localhost:1"
%HTTPotion.ErrorResponse{message: "econnrefused"}

# You can also raise `HTTPError` with the `bang` version of request
iex> HTTPotion.get! "http://localhost:1"
** (HTTPotion.HTTPError) econnrefused

The Response is a struct – you access its fields like this: response.body.

response.headers is a HTTPotion.Headers struct that provides case-insensitive access (so you can use response.headers[:authorization] and it doesn't matter if the server returned AuThOrIZatIOn or something).

HTTPError is an exception that happens when the request fails.

Note: the API changed in 2.0.0, body and headers are options now!

Available options and their default value:

  body: "",                # Request's body contents Ex.: "{json: \"string\"}"
  headers: [],             # Request's headers. Ex.: ["Accepts" => "application/json"]
  timeout: 5000,           # Timeout in milliseconds Ex: 5000
  ibrowse: [],             # ibrowse options
  follow_redirects: false, # Specify whether redirects should be followed
  stream_to: nil,          # Specify a process to stream the response to when performing async requests
  basic_auth: nil,         # Basic auth credentials. Ex.: {"username", "password"}

Metaprogramming magic

You can extend HTTPotion.Base to make cool API clients or something (this example uses jsx for JSON):

defmodule GitHub do
  use HTTPotion.Base

  def process_url(url) do
    "" <> url

  def process_request_headers(headers) do
    Dict.put headers, :"User-Agent", "github-potion"

  def process_response_body(body) do
    body |> IO.iodata_to_binary |> :jsx.decode
    |> fn ({k, v}) -> { String.to_atom(k), v } end
    |> :orddict.from_list
iex> GitHub.get("users/myfreeweb").body[:public_repos]

Read the source to see all the hooks. It's not intimidating at all, pretty easy to read actually :-)

Don't forget that IO.iodata_to_binary is called by default in process_response_body and process_response_chunk, you'll probably need to call it too.

Asynchronous requests

Hey, we're on the Erlang VM, right? Every serious OTP app probably makes a lot of these. It's easy to do in HTTPotion.

iex> HTTPotion.get "", [stream_to: self]
%HTTPotion.AsyncResponse{id: {1372,8757,656584}}

iex> flush
%HTTPotion.AsyncHeaders{id: {1372,8757,656584}, status_code: 200, headers: ["Transfer-Encoding": "chunked", ...]}
%HTTPotion.AsyncChunk{id: {1372,8757,656584}, chunk: "<!DOCTYPE html>\n..."}
%HTTPotion.AsyncEnd{id: {1372,8757,656584}}

Note that instead of process_response_body, process_response_chunk is called on the chunks before sending them out to the receiver (the stream_to process).

Direct access to ibrowse workers

ibrowse allows you to use its separate worker processes directly. We expose this functionality through the direct option.

Don't forget that you have to pass the URL to the worker process, which means the worker only communicates with one server (domain!)

iex> {:ok, worker_pid} = HTTPotion.spawn_worker_process("")

iex> HTTPotion.get "", [direct: worker_pid]
%HTTPotion.Response{body: "...", headers: ["Connection": "close", ...], status_code: 200}

You can even combine it with async!

iex> {:ok, worker_pid} = HTTPotion.spawn_worker_process("")

iex> "", [direct: worker_pid, stream_to: self, headers: ["User-Agent": "hello it's me"]]
%HTTPotion.AsyncResponse{id: {1372,8757,656584}}


Please feel free to submit pull requests!

By participating in this project you agree to follow the Contributor Code of Conduct.

The list of contributors is available on GitHub.


This is free and unencumbered software released into the public domain.
For more information, please refer to the UNLICENSE file or