NAME

SYNOPSIS

  package Foo;
  use Class::Trigger;

  sub foo {
      my $self = shift;
      $self->call_trigger('before_foo');
      # some code ...
      $self->call_trigger('middle_of_foo');
      # some code ...
      $self->call_trigger('after_foo');
  }

  package main;
  Foo->add_trigger(before_foo => \&sub1);
  Foo->add_trigger(after_foo => \&sub2);

  my $foo = Foo->new;
  $foo->foo;            # then sub1, sub2 called

  # triggers are inheritable
  package Bar;
  use base qw(Foo);

  Bar->add_trigger(before_foo => \&sub);

  # triggers can be object based
  $foo->add_trigger(after_foo => \&sub3);
  $foo->foo;            # sub3 would appply only to this object

DESCRIPTION

METHODS

add_trigger

  Foo->add_trigger($triggerpoint => $sub);
  $foo->add_trigger($triggerpoint => $sub);


  Foo->add_trigger( name => $triggerpoint,
                    callback => sub {return undef},
                    abortable => 1); 

  # no further triggers will be called. Undef will be returned.
    

Adds triggers for trigger point. You can have any number of triggers for each point. Each coderef will be passed a reference to the calling object, as well as arguments passed in via call_trigger. Return values will be captured in list context.

If add_trigger is called with named parameters and the "abortable" parameter is passed a true value, a false return value from trigger code will stop processing of this trigger point and return a "false" value to the calling code.

If "add_trigger" is called without the "abortable" flag, return values will be captured by call_trigger, but failures will be ignored.

If "add_trigger" is called as object method, whole current trigger table will be copied onto the object and the new trigger added to that. (The object must be implemented as hash.)

  my $foo = Foo->new;

  # this trigger ($sub_foo) would apply only to $foo object
  $foo->add_trigger($triggerpoint => $sub_foo);
  $foo->foo;

  # And not to another $bar object
  my $bar = Foo->new;
  $bar->foo;
    

call_trigger

  $foo->call_trigger($triggerpoint, @args);
    

Calls triggers for trigger point, which were added via "add_trigger" method. Each triggers will be passed a copy of the object as the first argument. Remaining arguments passed to "call_trigger" will be passed on to each trigger. Triggers are invoked in the same order they were defined.

If there are no "abortable" triggers or no "abortable" trigger point returns a false value, "call_trigger" will return the number of triggers processed.

If an "abortable" trigger returns a false value, call trigger will stop execution of the trigger point and return undef.

last_trigger_results

    my @results = @{ $foo->last_trigger_results };
    

Returns a reference to an array of the return values of all triggers called for the last trigger point. Results are ordered in the same order the triggers were run.

TRIGGER POINTS

  package Foo;
  use Class::Trigger qw(foo bar baz);

  package main;
  Foo->add_trigger(foo  => \&sub1); # okay
  Foo->add_trigger(hoge => \&sub2); # exception

FAQ

Q.

This module lets me add subs to be run before/after a specific subroutine is run. Yes?

A.

You put various call_trigger() method in your class. Then your class users can call add_trigger() method to add subs to be run in points just you specify (exactly where you put call_trigger()).

Q.

Are you aware of the perl-aspects project and the Aspect module? Very similar to Class::Trigger by the look of it, but its not nearly as explicit. Its not necessary for foo() to actually say "triggers go *here*", you just add them.

A.

Yep ;)

But the difference with Aspect would be that Class::Trigger is so simple that it's easy to learn, and doesn't require 5.6 or over.

Q.

How does this compare to Sub::Versive, or Hook::LexWrap?

A.

Very similar. But the difference with Class::Trigger would be the explicitness of trigger points.

In addition, you can put hooks in any point, rather than pre or post of a method.

Q.

It looks interesting, but I just can't think of a practical example of its use...

A.

(by Tony Bowden)

I originally added code like this to Class::DBI to cope with one particular case: auto-upkeep of full-text search indices.

So I added functionality in Class::DBI to be able to trigger an arbitary subroutine every time something happened - then it was a simple matter of setting up triggers on INSERT and UPDATE to reindex that row, and on DELETE to remove that index row.

See Class::DBI::mysql::FullTextSearch and its source code to see it in action.

AUTHORS

Code by Tatsuhiko Miyagawa <miyagawa@bulknews.net>.

Jesse Vincent added a code to get return values from triggers and abortable flag.

LICENSE

SEE ALSO