NAME

Bot::ChatBots::Minion - Minion-based pipeline breaker for Bot::ChatBots

VERSION

This document describes Bot::ChatBots::Minion version 0.004.

Build Status Perl Version Current CPAN version Kwalitee CPAN Testers CPAN Testers Matrix

SYNOPSIS

# Just send received records straight to a Minion worker
use Mojolicious::Lite;
plugin 'Bot::ChatBots::Minion' => Minion => [...];
plugin 'Bot::ChatBots::Telegram' => sources => [
   'WebHook',
   processor => app->chatbots->minion->wrapper($processor_in_worker),
   ...
];
...
app->start;
# now you will have to both start the Mojolicious::Lite app and the
# Minion worker!

DESCRIPTION

(Note: you are supposed to be familiar with Data::Tubes terminology).

This module allows you to break a potentially blocking long pipeline of operations into two parts, shifting the second part for execution in a Mojolicious Minion.

For example, suppose you are using both Data::Tubes and Bot::ChatBots::Telegram::WebHook in a Mojolicious::Lite app like this:

use Mojolicious::Lite;
use Data::Tubes qw< pipeline >;

my $pipeline = pipeline(
   \&simple_operation_1,
   \&simple_operation_2,
   \&long_running_operation,
   \&simple_operation_3,
   {tap => sink},
);

plugin Bot::ChatBots::Telegram => sources => [
   'Bot::ChatBots::Telegram::WebHook',
   processor => $pipeline,
   ...
];

app->start;

When a new update comes, it will eventually hit long_running_operation and block your frontend process. Ouch! This is what you can do instead:

use Mojolicious::Lite;
use Data::Tubes qw< pipeline >;

# configure Minion before calling Bot::ChatBots::Minion
plugin Minion => ...;
plugin 'Bot::ChatBots::Minion';

my $pipeline = pipeline(
   \&simple_operation_1,
   \&simple_operation_2,
   app->chatbots->minion->wrapper(
      downstream => pipeline(
         \&long_running_operation,
         \&simple_operation_3,
         {tap => sink},
      )
   ),
   {tap => sink},
);

# the rest is as before
plugin Bot::ChatBots::Telegram => sources => [
   'Bot::ChatBots::Telegram::WebHook',
   processor => $pipeline,
   ...
];

app->start;

So the trick is to divide the long-running pipeline into two separate pipelines, one to be executed in the main process with the first two simple operations and ending with a wrapper for the second pipeline, which includes the long operation. The wrapping mechanism takes care to send the received record along to the Minion worker, where the second pipeline will be executed.

You don't actually have to call pipeline inside the wrapper invocation, because it will be called for you if the parameter is an array reference:

use Mojolicious::Lite;
use Data::Tubes qw< pipeline >;

# configure Minion before calling Bot::ChatBots::Minion
plugin Minion => ...;
plugin 'Bot::ChatBots::Minion';

my $pipeline = pipeline(
   \&simple_operation_1,
   \&simple_operation_2,
   app->chatbots->minion->wrapper(
      downstream => [
         \&long_running_operation,
         \&simple_operation_3,
         {tap => sink},
      ],
   ),
   {tap => sink},
);

# the rest is as before
plugin Bot::ChatBots::Telegram => sources => [
   'Bot::ChatBots::Telegram::WebHook',
   processor => $pipeline,
   ...
];

app->start;

The only difference in this case is that if you do not pass ready-made tubes (i.e. sub references) but expressions that can be turned into tubes, they will be transformed using prefix Bot::ChatBots instead of the default Data::Tubes (and the transformation will be subject to the rules set for "resolve_module" in Bot::ChatBots::Utils.

METHODS

dequeuer

my $sub_reference = $obj->dequeuer(%args); # OR
   $sub_reference = $obj->dequeuer(\%args);

Generate a dequeuer sub reference, which is suitable to be used in a Minion worker process for receiving records and sending them to a downstream pipeline.

The %args MUST contain a downstream parameter (or its lower-precedence alias processor) with a tube-compliant sub reference or anything that can be transformed into one via "pipeline" in Bot::ChatBots::Utils, which will be invoked inside the Minion worker.

The following keys are recognised in %args:

downstream

mandatory parameter (unless processor is provided) carrying a tube or tubifiable definition via Bot::ChatBots::Utils/pipeline (the latter case assumes that you also have Data::Tubes installed);

name

set an alternative name for enqueuing/dequeuing stuff via Minion, defaults to "name";

prefix

set a prefix for automatic transformation of module names via Bot::ChatBots::Utils/pipeline, defaults to what set for "prefix" (i.e. Bot::ChatBots);

processor

low-priority alias for downstream, see above.

This method just returns the sub reference, so you will want it if you want to manage the Minion worker by yourself (e.g. explicitly calling "process_commands" in Minion::Worker). If you just want to install a task, you might want to consider the convenience function "install_dequeuer" described below.

enqueue

$obj->enqueue($record); # OR
$obj->enqueue($record, $name);

Send a $record to the Minion, optionally specifing the $name of the task (which defaults to "name"). It's actually just a convenience wrapper around "enqueuer".

enqueuer

my $sub_reference = $obj->enqueuer();         # OR
   $sub_reference = $obj->enqueuer($name);

Generate an enqueuer tube, most probably to be used as the last tube in a pipeline.

You can optionally pass a $name that will be used for enqueuing tasks towards the Minion; by default it takes the value from "name".

It is not mandatory that "minion" is defined at the time this method is called, but it MUST be defined when the $sub_reference is called.

helper_name

my $helper_name = $self->helper_name;

Default helper name. This is used by "register" in case $conf does not contain a helper_name itself.

install_dequeuer

$obj->install_dequeuer(%args); # OR
$obj->install_dequeuer(\%args);

Generate a dequeuer subroutine via "dequeuer" and set it as a task in Minion (via "add_task" in Minion). If you are defining a custom Minion worker process that relies on Mojolicious::Lite, you can e.g. do this:

use Mojolicious::Lite;
plugin 'Bot::ChatBots::Minion' ...;
app->chatbots->minion->install_dequeuer(downstream => sub { ... });
app->start;

minion

my $minion = $obj->minion;
$obj->minion($new_minion_ref);

Accessor for the minion object. If you pass an array reference, it will be used for instantiating a new Minion object passing its contents as the list provided to the Minion's new method.

name

my $name = $obj->name;
$obj->name('new name');

Accessor for the name of the object, also used as topic for queuing tasks. Defaults to the value of "typename".

register

$obj->register($app, $conf);

# implicitly called when you load the class as a Mojolicious::Plugin
plugin 'Bot::ChatBots::Minion' => %conf;

Plugin registration method, consumed by Mojolicious when loading this class as a plugin. The Mojolicious composite helper chatbots.minion is set to a subroutine reference that returns an instance of the Bot::ChatBots::Minion object, for possible further manipulation.

The argument $conf is a hash reference, the following keys are supported:

minion

set "minion". Do not confuse this with Minion below!

Minion

load the Minion plugin in the $app and use it for setting "minion". This option points to an array reference that contains the configuration for instantiating the Minion plugin

name

set "name"

prefix

set "prefix"

typename

set "typename"

wrapper

my $sub_reference = $obj->wrapper(%args); # OR
   $sub_reference = $obj->wrapper(\%args);

Wrap a tube (or a sequence that can be transformed into a tube) in a delayed execution via Minion. It returns a sub reference that is a valid tube where records can be sent to a Minion worker.

The %args MUST contain a downstream parameter (or its lower-precedence alias processor) with a tube-compliant sub reference or anything that can be transformed into one via "pipeline" in Bot::ChatBots::Utils, which will be invoked inside the Minion worker.

The following keys are recognised in %args:

downstream

mandatory parameter (unless processor is provided) carrying a tube or tubifiable definition via Bot::ChatBots::Utils/pipeline (the latter case assumes that you also have Data::Tubes installed);

name

set an alternative name for enqueuing/dequeuing stuff via Minion, defaults to "name"

prefix

set a prefix for automatic transformation of module names via Bot::ChatBots::Utils/pipeline, defaults to what set for "prefix" (i.e. Bot::ChatBots);

processor

low-priority alias for downstream, see above.

BUGS AND LIMITATIONS

Report bugs through GitHub (patches welcome).

SEE ALSO

Bot::ChatBots, Bot::ChatBots::Telegram.

AUTHOR

Flavio Poletti <polettix@cpan.org>

COPYRIGHT AND LICENSE

Copyright (C) 2016 by Flavio Poletti <polettix@cpan.org>

This module is free software. You can redistribute it and/or modify it under the terms of the Artistic License 2.0.

This program is distributed in the hope that it will be useful, but without any warranty; without even the implied warranty of merchantability or fitness for a particular purpose.