Apns4erl

This lib is intended to allow you to write an APNs provider for Apple Push Notificaion services (APNs) in Erlang.

Copyright (c) 2010 Inaka Labs SRL support@inaka.net, released under the MIT license

Example

Using apns4erl is quite simple. First, setup something similar to this in your sys.config:

    {apns, [
      {apple_host, "gateway.sandbox.push.apple.com"},
      {apple_port, 2195},
      {cert_file, "/etc/certs/mycert_dev.pem"},
      {key_file, undefined},
      {cert_password, undefined},
      {timeout, 30000},
      {feedback_port, 2196},
      {feedback_host, "feedback.sandbox.push.apple.com"},
      {feedback_timeout, 18000000}
    ]}

NOTE: The apple_host to use will depend on your environment (production or development). Remember to always use the correct certificate, device tokens, and apns hostname for production or development environments.

NOTE 2: To generate the .pem file, from the .cer an .p12 files provided by Apple, you can use this script

Then, once you've started the apns application, you can connect to the APNS network using:

      apns:connect(
        %% your connection identifier:
        my_connection_name,
        %% called in case of a "hard" error:
        fun ?MODULE:handle_apns_error/2,
        %% called if the device uninstalled the application:
        fun ?MODULE:handle_apns_delete_subscription/1
      ).

As a result, you will get a tuple:

• {ok, Pid}
• {error, {already_started, Pid}}
• {error, Reason}

Pid is the Pid of the apns_connection process spawned to handle the connection.

CAUTION: It is highly recommended to pass a fully qualified function for callbacks, using the fun M:F/A syntax. If you pass in a local fun using fun F/A, any subsequent code upgrades to the module where that local fun was defined, will cause the code server to kill processes that are holding on to references to the old code (eg. the apns connection processes, thus killing the apns4erl application).

To send a notification

apns:send_message(my_connection_name, "this_is_a_valid_device_token", "hello world").

That's it!

A little more about what's going on

Actually, send_message/3, send_message/4, send_message/5, send_message/6, send_message/7, and send_message/8 are calling send_message/2, which takes a #apns_msg record as its 2nd argument. Thus, you can also create the message customized with your own needs, by using a #apns_msg record:

    -include_lib("apns/include/apns.hrl").

    apns:send_message(my_connection_name, #apns_msg{
      alert  = "alert" ,
      badge  = 1,
      sound  = "sound" ,
      expiry = 1348000749,
      device_token = "this_is_a_valid_device_token"
    }).

Feedback Channel and Getting Errors

Notice how we are passing 2 funs to the connect function. These are used as callbacks:

If there was an error while sending a message, the first fun will be called.

If there were no errors, but Apple reported that the user removed the application from the device, the 2nd fun will be used (this is effectively the feedback channel).

    handle_apns_error(MsgId, Status) ->
      error_logger:error_msg("error: ~p - ~p~n", [MsgId, Status]).

    handle_apns_delete_subscription(Data) ->
      error_logger:info_msg("delete subscription: ~p~n", [Data]).