use Asterisk::AMI;
my $astman = Asterisk::AMI->new(PeerAddr => '127.0.0.1',
PeerPort => '5038',
Username => 'admin',
Secret => 'supersecret'
);
die "Unable to connect to asterisk" unless ($astman);
my $action = $astman->({ Action => 'Command',
Command => 'sip show peers'
});
DESCRIPTION
This module provides an interface to the Asterisk Manager Interface. It's goal is to provide a flexible, powerful, and reliable way to interact with Asterisk upon which other applications may be built. It utilizes AnyEvent and therefore can integrate very easily into event-based applications, but it still provides blocking functions for us with standard scripting.
SSL SUPPORT INFORMAION
For SSL support you will also need the module that AnyEvent::Handle uses for SSL support, which is not a required dependency. Currently that module is 'Net::SSLeay' (AnyEvent:Handle version 5.251) but it may change in the future.
CentOS/Redhat
If the version of Net:SSLeay included in CentOS/Redhat does not work try installing an updated version from CPAN.
Constructor
new([ARGS])
Creates a new AMI object which takes the arguments as key-value pairs.
Key-Value Pairs accepted:
PeerAddr Remote host address <hostname>
PeerPort Remote host port <service>
Events Enable/Disable Events 'on'|'off'
Username Username to access the AMI
Secret Secret used to connect to AMI
AuthType Authentication type to use for login 'plaintext'|'MD5'
UseSSL Enables/Disables SSL for the connection 0|1
BufferSize Maximum size of buffer, in number of actions
Timeout Default timeout of all actions in seconds
Handlers Hash reference of Handlers for events { 'EVENT' => /&somesub };
Keepalive Interval (in seconds) to periodically send 'Ping' actions to asterisk
TCP_Keepalive Enables/Disables SO_KEEPALIVE option on the socket 0|1
Blocking Enable/Disable blocking connects 0|1
on_connect A subroutine to run after we connect
on_connect_err A subroutine to call if we have an error while connecting
on_error A subroutine to call when an error occurs on the socket
on_disconnect A subroutine to call when the remote end disconnects
on_timeout A subroutine to call if our Keepalive times out
OriginateHack Changes settings to allow Async Originates to work 0|1
'PeerAddr' defaults to 127.0.0.1.
'PeerPort' defaults to 5038.
'Events' default is 'off'. May be anything that the AMI will accept as a part of the 'Events' parameter for the
login action.
'Username' has no default and must be supplied.
'Secret' has no default and must be supplied.
'AuthType' sets the authentication type to use for login. Default is 'plaintext'. Use 'MD5' for MD5 challenge
authentication.
'UseSSL' defaults to 0 (no ssl). When SSL is enabled the default PeerPort changes to 5039.
'BufferSize' has a default of 30000. It also acts as our max actionid before we reset the counter.
'Timeout' has a default of 0, which means no timeout on blocking.
'Handlers' accepts a hash reference setting a callback handler for the specified event. EVENT should match
the contents of the {'Event'} key of the event object will be. The handler should be a subroutine reference that
will be passed the a copy of the AMI object and the event object. The 'default' keyword can be used to set
a default event handler. If handlers are installed we do not buffer events and instead immediately dispatch them.
If no handler is specified for an event type and a 'default' was not set the event is discarded.
'Keepalive' only works when running with an event loop. Used with on_timeout, this can be used to detect if
asterisk has become un-responsive.
'TCP_Keepalive' default is disabled. Activates the tcp keep-alive at the socket layer. This does not require
an event-loop and is lightweight. Useful for applications that use long-lived connections to Asterisk but
do not run an event loop.
'Blocking' has a default of 1 (block on connecting). A value of 0 will cause us to queue our connection
and login for when an event loop is started. If set to non blocking we will always return a valid object.
'on_connect' is a subroutine to call when we have successfully connected and logged into the asterisk manager.
it will be passed our AMI object.
'on_connect_err', 'on_error', 'on_disconnect'
These three specify subroutines to call when errors occur. 'on_connect_err' is specifically for errors that
occur while connecting, as well as failed logins. If 'on_connect_err' or 'on_disconnect' it is not set,
but 'on_error' is, 'on_error' will be called. 'on_disconnect' is not reliable, as disconnects seem to get lumped
under 'on_error' instead. When the subroutine specified for any of theses is called the first argument is a copy
of our AMI object, and the second is a string containing a message/reason. All three of these are 'fatal', when
they occur we destroy our buffers and our socket connections.
'on_timeout' is called when a keep-alive has timed out, not when a normal action has. It is non-'fatal'.
The subroutine will be called with a copy of our AMI object and a message.
'OriginateHack' defaults to 0 (off). This essentially enables 'call' events and says 'discard all events
unless the user has explicitly enabled events' (prevents a memory leak). It does its best not to mess up
anything you have already set. Without this, if you use 'Async' with an 'Originate' the action will timeout
or never callback. You don't need this if you are already doing work with events, simply add 'call' events
to your eventmask.
If you are running an event loop and use blocking methods (e.g. get_response, check_response, action,
simple_action, connected) the outcome is unspecified. It may work, it may lock everything up, the action may
work but break something else. I have tested it and behavior seems unpredictable at best and is very
circumstantial.
If you are running an event-loop use non-blocking callbacks! It is why they are there!
However if you do play with blocking methods inside of your loops let me know how it goes.
Actions
Construction
No matter which method you use to send an action (send_action(), simple_action(), or action()), they all accept actions in the same format, which is a hash reference. The only exceptions to this rules are when specifying a callback and a callback timeout, which only work with send_action.
To build and send an action you can do the following:
%action = ( Action => 'Command',
Command => 'sip show peers'
);
$astman->send_action(/%action);
Alternatively you can also do the following to the same effect:
$astman->send_action({ Action => 'Command',
Command => 'sip show peers'
});
Additionally the value of the hash may be an array reference. When an array reference is used, every value in the array is append as a different line to the action. For example:
{ Variable => [ 'var1=1', 'var2=2' ] }
Will become:
Variable: var1=1
Variable: var2=2
When the action is sent.
Sending and Retrieving
More detailed information on these individual methods is available below
The send_action() method can be used to send an action to the AMI. It will return a positive integer, which is the ActionID of the action, on success and will return undef in the event it is unable to send the action.
After sending an action you can then get its response in one of two methods.
The method check_response() accepts an actionid and will return 1 if the action was considered successful, 0 if it failed and undef if an error occurred or on timeout.
The method get_response() accepts an actionid and will return a Response object (really just a fancy hash) with the contents of the Action Response as well as any associated Events it generated. It will return undef if an error occurred or on timeout.
All responses and events are buffered, therefor you can issue several send_action()s and then retrieve/check their responses out of order without losing any information. In-fact, if you are issuing many actions in series you can get much better performance sending them all first and then retrieving them later, rather than waiting for responses immediately after issuing an action.
Alternatively you can also use simple_action() and action(). simple_action() combines send_action() and check_response(), and therefore returns 1 on success and 0 on failure, and undef on error or timeout. action() combines send_action() and get_response(), and therefore returns a Response object or undef.
Examples
Send and retrieve and action:
my $actionid = $astman->send_action({ Action => 'Command',
Command => 'sip show peers'
});
my $response = $astman->get_response($actionid)
This is equivalent to the above:
my $response = $astman->action({ Action => 'Command',
Command => 'sip show peers'
});
The following:
my $actionid1 = $astman->send_action({ Action => 'Command',
Command => 'sip show peers'
});
my $actionid2 = $astman->send_action({ Action => 'Command',
Command => 'sip show peers'
});
my $actionid3 = $astman->send_action({ Action => 'Command',
Command => 'sip show peers'
});
my $response3 = $astman->get_response($actionid3);
my $response1 = $astman->get_response($actionid1);
my $response2 = $astman->get_response($actionid2);
Can be much faster than:
my $response1 = $astman->action({ Action => 'Command',
Command => 'sip show peers'
});
my $response2 = $astman->action({ Action => 'Command',
Command => 'sip show peers'
});
my $response3 = $astman->action({ Action => 'Command',
Command => 'sip show peers'
});
Originate Examples
I see enough searches hit my site for this that I figure it should be included in the documentation.
These don't include non-blocking examples, please read the section on 'Callbacks' below for information
on using non-blocking callbacks and events.
NOTE: Please read about the 'OriginateHack' option for the constructor above if you plan on using the 'Async'
option in your Originate command, as it may be required to properly retrieve the response.
In these examples we are dialing extension '12345' at a sip peer named 'peer' and when the call connects
we drop the channel into 'some_context' at priority 1 for extension 100.
Example 1 - A simple non-ASYNC Originate
my $response = $astman->action({Action => 'Originate',
Channel => 'SIP/peer/12345',
Context => 'some_context',
Exten => 100,
Priority => 1});
And the contents of respone will look similiar to the following:
{
'Message' => 'Originate successfully queued',
'ActionID' => '3',
'GOOD' => 1,
'COMPLETED' => 1,
'Response' => 'Success'
};
Example 2 - Originate with multiple variables
This will set the channel variables 'var1' and 'var2' to 1 and 2, respectfully.
The value for the 'Variable' key should be an array reference or an anonymous array in order
to set multiple variables.
my $response = $astman->action({Action => 'Originate',
Channel => 'SIP/peer/12345',
Context => 'some_context',
Exten => 100,
Priority => 1,
Variable = [ 'var1=1', 'var2=2' ]});
Example 3 - An Async Originate
If youre Async Originate never returns please read about the 'OriginateHack' option for the constructor.
my $response = $astman->action({Action => 'Originate',
Channel => 'SIP/peer/12345',
Context => 'some_context',
Exten => 100,
Priority => 1,
Async => 1});
And the contents of response will look similiar to the following:
{
'Message' => 'Originate successfully queued',
'EVENTS' => [
{
'Exten' => '100',
'CallerID' => '<unknown>',
'Event' => 'OriginateResponse',
'Privilege' => 'call,all',
'Channel' => 'SIP/peer-009c5510',
'Context' => 'some_context',
'Response' => 'Success',
'Reason' => '4',
'CallerIDName' => '<unknown>',
'Uniqueid' => '1276543236.82',
'ActionID' => '3',
'CallerIDNum' => '<unknown>'
}
],
'ActionID' => '3',
'GOOD' => 1,
'COMPLETED' => 1,
'Response' => 'Success'
};
More Info:
Check out the voip-info.org page for more information on the Originate action.
http://www.voip-info.org/wiki/view/Asterisk+Manager+API+Action+Originate
Callbacks
You may also specify a method to callback when using send_action as well as a timeout.
An example of this would be:
$astman->send_action({ Action => 'Ping',
CALLBACK => /&somemethod,
TIMEOUT => 7 });
Equivalent in the new alternative sytanx:
$astman->send_action({ Action => 'Ping' }, /&somemethod, 7);
In this example once the action 'Ping' finishes we will call somemethod() and pass it the a copy of our AMI object and the Response Object for the action. If TIMEOUT is not specified it will use the default set. A value of 0 means no timeout. When the timeout is reached somemethod() will be called and passed a reference to the our $astman and the uncompleted Response Object, therefore somemethod() should check the state of the object. Checking the key {'GOOD'} is usually a good indication if the response is useable.
Callback Caveats
Callbacks only work if we are processing packets, therefore you must be running an event loop. Alternatively, we run mini-event loops for our blocking calls (e.g. action(), get_action()), so in theory if you set callbacks and then issue a blocking call those callbacks should also get triggered. However this is an unsupported scenario.
Timeouts are done using timers and they are set as soon as you send the object. Therefore if you send an action with a timeout and then monkey around for a long time before getting back to your event loop (to process input) you can time out before ever even attempting to receive the response.
A very contrived example:
$astman->send_action({ Action => 'Ping',
CALLBACK => /&somemethod,
TIMEOUT => 3 });
sleep(4);
#Start loop
$astman->loop;
#Oh no we never even tried to get the response yet it will still time out
ActionIDs
This module handles ActionIDs internally and if you supply one in an action it will simply be ignored and overwritten.
Responses and Events
NOTE: Empty fields sent by Asterisk (e.g. 'Account: ' with no value in an event) are represented by the hash
value of null string, not undef. This means you need to test for ''
(e.g. if ($response->{'Account'} ne '')) ) for any values that might be possibly be empty.
Responses
Responses are returned as response objects, which are hash references, structured as follows:
$response->{'Response'} Response to our packet (Success, Failed, Error, Pong, etc).
{'ActionID'} ActionID of this Response.
{'Message'} Message line of the response.
{'EVENTS'} Array reference containing Event Objects associated with this actionid.
{'PARSED'} Hash reference of lines we could parse into key->value pairs.
{'CMD'} Contains command output from 'Action: Command's. It is an array reference.
{'COMPLETED'} 1 if completed, 0 if not (timeout)
{'GOOD'} 1 if good, 0 if bad. Good means no errors and COMPLETED.
Events
Events are turned into event objects, these are similar to response objects, but their keys vary much more
depending on the specific event.
Some common contents are:
$event->{'Event'} The type of Event
{'ActionID'} Only available if this event was caused by an action
Event Handlers
Here is a very simple example of how to use event handlers. Please note that the key for the event handler
is matched against the event type that asterisk sends. For example if asterisk sends 'Event: Hangup' you use a
key of 'Hangup' to match it. This works for any event type that asterisk sends.
my $astman = Asterisk::AMI->new(PeerAddr => '127.0.0.1',
PeerPort => '5038',
Username => 'admin',
Secret => 'supersecret',
Events => 'on',
Handlers => { default => /&do_event,
Hangup => /&do_hangup };
);
die "Unable to connect to asterisk" unless ($astman);
sub do_event {
my ($asterisk, $event) = @_;
print 'Yeah! Event Type: ' . $event->{'Event'} . "/r/n";
}
sub do_hangup {
my ($asterisk, $event) = @_;
print 'Channel ' . $event->{'Channel'} . ' Hungup because ' . $event->{'Cause-txt'} . "/r/n";
}
#Start some event loop
someloop;
How to use in an event-based application
Getting this module to work with your event based application is really easy so long as you are running an
event-loop that is supported by AnyEvent. Below is a simple example of how to use this module with your
preferred event loop. We will use EV as our event loop in this example. I use subroutine references in this
example, but you could use anonymous subroutines if you want to.
#Use your preferred loop before our module so that AnyEvent will auto-detect it
use EV;
use Asterisk::AMI:
#Create your connection
my $astman = Asterisk::AMI->new(PeerAddr => '127.0.0.1',
PeerPort => '5038',
Username => 'admin',
Secret => 'supersecret',
Events => 'on',
Handlers => { default => /&eventhandler }
);
#Alternatively you can set Blocking => 0, and set an on_error sub to catch connection errors
die "Unable to connect to asterisk" unless ($astman);
#Define the subroutines for events
sub eventhandler { my ($ami, $event) = @_; print 'Got Event: ',$event->{'Event'},"/r/n"; }
#Define a subroutine for your action callback
sub actioncb { my ($ami, $response) = @_; print 'Got Action Reponse: ',$response->{'Response'},"/r/n"; }
#Send an action
my $action = $astman->({ Action => 'Ping',
CALLBACK => /&actioncb });
#Do all of you other eventy stuff here, or before all this stuff, whichever
#..............
#Start our loop
EV::loop
That's it, the EV loop will allow us to process input from asterisk. Once the action completes it will
call the callback, and any events will be dispatched to eventhandler(). As you can see it is fairly
straight-forward. Most of the work will be in creating subroutines to be called for various events and
actions that you plan to use.
Methods
send_action ( ACTION, [ [ CALLBACK ], [ TIMEOUT ] ] )
Sends the action to asterisk, where ACTION is a hash reference. If no errors occurred while sending it returns
the ActionID for the action, which is a positive integer above 0. If it encounters an error it will return undef.
You may specify a callback function and timeout either in the ACTION hash or in the method call. CALLBACK is
optional and should be a subroutine reference or any anonymous subroutine. TIMEOUT is optional and only has an
affect if a CALLBACK is specified. CALLBACKs and TIMEOUTs specified during a method call override any found in
the ACTION hash.
check_response( [ ACTIONID ], [ TIMEOUT ] )
Returns 1 if the action was considered successful, 0 if it failed, or undef on timeout or error. If no ACTIONID
is specified the ACTIONID of the last action sent will be used. If no TIMEOUT is given it blocks, reading in
packets until the action completes. This will remove a response from the buffer.
get_response ( [ ACTIONID ], [ TIMEOUT ] )
Returns the response object for the action. Returns undef on error or timeout.
If no ACTIONID is specified the ACTIONID of the last action sent will be used. If no TIMEOUT is given it
blocks, reading in packets until the action completes. This will remove the response from the buffer.
action ( ACTION [, TIMEOUT ] )
Sends the action and returns the response object for the action. Returns undef on error or timeout.
If no ACTIONID is specified the ACTIONID of the last action sent will be used.
If no TIMEOUT is given it blocks, reading in packets until the action completes. This will remove the
response from the buffer.
simple_action ( ACTION [, TIMEOUT ] )
Sends the action and returns 1 if the action was considered successful, 0 if it failed, or undef on error
and timeout. If no ACTIONID is specified the ACTIONID of the last action sent will be used. If no TIMEOUT is
given it blocks, reading in packets until the action completes. This will remove the response from the buffer.
disconnect ()
Logoff and disconnects from the AMI. Returns 1 on success and 0 if any errors were encountered.
get_event ( [ TIMEOUT ] )
This returns the first event object in the buffer, or if no events are in the buffer it reads in packets
waiting for an event. It will return undef if an error occurs.
If no TIMEOUT is given it blocks, reading in packets until an event arrives.
amiver ()
Returns the version of the Asterisk Manager Interface we are connected to. Undef until the connection is made
(important if you have Blocking => 0).
connected ( [ TIMEOUT ] )
This checks the connection to the AMI to ensure it is still functional. It checks at the socket layer and
also sends a 'PING' to the AMI to ensure it is still responding. If no TIMEOUT is given this will block
waiting for a response.
Returns 1 if the connection is good, 0 if it is not.
error ()
Returns 1 if there are currently errors on the socket, 0 if everything is ok.
destroy ( [ FATAL ] )
Destroys the contents of all buffers and removes any current callbacks that are set. If FATAL is true
it will also destroy our IO handle and its associated watcher. Mostly used internally. Useful if you want to
ensure that our IO handle watcher gets removed.
loop ()
Starts an eventloop via AnyEvent.
See Also
