NAME

Reply::Plugin - base class for Reply plugins

VERSION

version 0.28

SYNOPSIS

package Reply::Plugin::Foo;
use strict;
use warnings;

use base 'Reply::Plugin';

# ...

DESCRIPTION

A Reply plugin is an object which adds some functionality to a Reply instance by implementing some specific methods which the Reply object will call at various points during execution. Plugins may implement as many callback methods as necessary to implement their functionality (although the more callbacks a given plugin implements, the more likely it is that the plugin may be more useful as multiple independent plugins).

Callback methods have three potential calling conventions:

wrapped

Wrapped callbacks receive a coderef as their first argument (before any arguments to the callback itself), and that coderef can be used to call the next callback in the list (if more than one plugin implements a given callback). In particular, this allows calling the next plugin multiple times, or not at all if necessary. Wrapped plugins should always call their coderef in list context. All plugins listed below are wrapped plugins unless indicated otherwise.

chained

Chained callbacks receive a list of arguments, and return a new list of arguments which will be passed to the next plugin in the chain. This allows each plugin a chance to modify a value before it's actually used by the repl.

concatenate

Concatenate callbacks receive a list of arguments, and return a list of response values. Each plugin that implements the given callback will be called with the same arguments, and the results will be concatenated together into a single list, which will be returned. Callbacks for published messages are of this type.

CALLBACKS

prompt

Called to determine the prompt to use when reading the next line. Takes no arguments, and returns a single string to use as the prompt. The default implementation returns ">"

read_line

Called to actually read a line from the user. Takes no arguments, and returns a single string. The default implementation uses the <> operator to read a single line from the user.

command_$name (chained)

If the line read from the user is of the form "#foo args...", then plugins will be searched for a callback method named command_foo. This callback takes a single string containing the provided arguments, and returns a new line to evaluate instead, if any.

mangle_line (chained)

Modifies the line read from the user before it's evaluated. Takes the line as a string and returns the modified line.

compile

Compiles the string of Perl code into a coderef. Takes the line of code as a string and a hash of extra parameters, and returns the coderef to be executed. The default implementation uses Eval::Closure to compile the given string.

The hash of extra parameters is passed directly to eval_closure.

execute

Executes the coderef which has just been compiled. Takes the coderef and a list of parameters to pass to it, and returns the list of results returned by calling the coderef. The default implementation just calls the coderef directly.

mangle_error (chained)

If the compile or execute callbacks throw an exception, this callback will be called to modify the exception before it is passed to print_error. It receives the exception and returns the modified exception.

print_error

If the compile or execute callbacks throw an exception, this callback will be called to display it to the user. It receives the exception and returns nothing. The default implementation just uses print to print it to the screen.

mangle_result (chained)

This callback is used to modify the result of evaluating the line of code before it is displayed. It receives the list of results and returns a modified list of results.

print_result

This callback displays to the user the results of evaluating the given line of code. It receives the list of results, and returns nothing. The default implementation just uses print to print them to the screen.

loop (chained)

This callback is called at the end of each evaluation. It receives whether the repl has been requested to terminate so far, and returns whether the repl should terminate.

Reply plugins can also communicate among each other via a pub/sub mechanism. By calling the publish method, all plugins which respond to the given message (implement a method of the given name) will have that method called with the given arguments, and all of the responses will be collected and returned. Some messages used by the default plugins are:

tab_handler ($line)

Plugins can publish this message when they want to attempt tab completion. Plugins that respond to this message should return a list of potential completions of the line which is passed in.

lexical_environment

Plugins which wish to modify the lexical environment should do so by implementing this message, which should return a hashref of variable names (including sigils) to value references. There can be more than one lexical environment (each maintained by a different plugin), so plugins that wish to inspect the lexical environment should do so by calling $self->publish('lexical_environment'), and then merging together all of the hashrefs which are returned.

package

Plugins which wish to modify the currently active package should do so by implementing this message, which should return the name of the current package. Then, to access the currently active package, a plugin can call ($self->publish('package'))[-1].

Your plugins, however, are not limited to these messages - you can use whatever messages you want to communicate.

METHODS

publish ($name, @args)

Publish a message to other plugins which respond to it. All loaded plugins which implement a method named $name will have it called with @args as the parameters. Returns a list of everything that each plugin responded with.

commands

Returns the names of the # commands that this plugin implements. This can be used in conjunction with publish - $plugin->publish('commands') will return a list of all commands which are available in the current Reply session.

AUTHOR

Jesse Luehrs <doy@tozt.net>

COPYRIGHT AND LICENSE

This software is Copyright (c) 2013 by Jesse Luehrs.

This is free software, licensed under:

The MIT (X11) License