NAME

Jifty::Dispatcher - The Jifty Dispatcher

SYNOPSIS

In MyApp::Dispatcher:

    package MyApp::Dispatcher;
    use Jifty::Dispatcher -base;

    under ['blog', 'wiki'] => [
        run {
            default model => "MyApp::Model::\u$1"
        },
        on PUT 'entries/*' => run {
            set entry_id => $1;
            show '/display/entry';
        },
        on '*/*' => run {
            my ($page, $op) = ($1, $2);
            my $item = get('model')->load($page) or next_rule;

            set item => $item;
            set page => $page;
            set op   => $op;

            show "/display/$op";
        },
        on '*' => run { dispatch "$1/view" },
        on ''  => show '/display/list',
    ];
    under qr{logs/(\d+)} => [
        when { $1 > 100 } => show '/error',
        set model => 'MyApp::Model::Log',
        run { dispatch "/wiki/LogPage-$1" },
    ];
    # ... more rules ...

DESCRIPTION

Jifty::Dispatcher takes requests for pages, walks through a dispatch table, possibly running code or transforming the request before finally handing off control to the templating system to display the page the user requested or whatever else the system has decided to display instead.

Generally, this is not the place to be performing model and user specific access control checks or updating your database based on what the user has sent in. You want to do that in your model classes. (Well, we want you to do that, but you're free to ignore our advice).

The Dispatcher runs rules in several stages:

before

before rules are run before Jifty evaluates actions. They're the perfect place to enable or disable Jifty::Actions using "allow" in Jifty::API and "deny" in Jifty::API or to completely disallow user access to private component templates such as the _elements directory in a default Jifty application. They're also the right way to enable Jifty::LetMe actions.

You can entirely stop processing with the redirect, tangent and abort directives, though "after" rules will still run.

on

on rules are run after Jifty evaluates actions, so they have full access to the results actions users have performed. They're the right place to set up view-specific objects or load up values for your templates.

Dispatcher directives are evaluated in order until we get to either a show, redirect, tangent or abort.

after

after rules let you clean up after rendering your page. Delete your cache files, write your transaction logs, whatever.

At this point, it's too late to show, redirect, tangent or abort page display.

Jifty::Dispatcher is intended to replace all the autohandler, dhandler and index.html boilerplate code commonly found in Mason applications, but there's nothing stopping you from using those features in your application when they're more convenient.

Each directive's code block runs in its own scope, but all share a common $Dispatcher object.

Plugins and rule ordering

By default, Jifty::Plugin dispatcher rules are added in the order they are specified in the application's configuration file; that is, after all the plugin dispatchers have run in order, then the application's dispatcher runs. It is possible to specify rules which should be reordered with respect to this rule, however. This is done by using a variant on the before and after syntax:

    before plugin NAME =>
        RULE(S);
    
    after plugin NAME =>
        RULE(S);

    after app,
        RULE(S)

NAME may either be a string, which must match the plugin name exactly, or a regular expression, which is matched against the plugin name. The rule will be placed at the first boundary that it matches -- that is, given a before plugin qr/^Jifty::Plugin::Auth::/ and both a Jifty::Plugin::Auth::Basic and a Jifty::Plugin::Auth::Complex, the rules will be placed before the first.

after app inserts the following RULES after the application's dispatcher rules, and is identical to, but hopefully clearer than, after plugin Jifty => RULES.

RULES may either be a single before, on, under, or after rule to change the ordering of, or an array reference of rules to reorder.

Data your dispatch routines has access to

request

The current Jifty::Request object.

$Dispatcher

The current dispatcher object.

get $arg

Return the argument value.

Things your dispatch routine might do

under $match => $rule

Match against the current requested path. If matched, set the current context to the directory and process the rule.

The $rule may be an array reference of more rules, a code reference, a method name of your dispatcher class, or a fully qualified subroutine name.

All wildcards in the $match string becomes capturing regex patterns. You can also pass in an array reference of matches, or a regex pattern.

The $match string may be qualified with a HTTP method name or protocol, such as

GET
POST
PUT
OPTIONS
DELETE
HTTPS
HTTP

on $match => $rule

Like under, except it has to match the whole path instead of just the prefix. Does not set current directory context for its rules.

before $match => $rule

Just like on, except it runs before actions are evaluated.

after $match => $rule

Just like on, except it runs after the page is rendered.

when {...} => $rule

Like under, except using an user-supplied test condition. You can stick any Perl you want inside the {...}; it's just an anonymous subroutine.

run {...}

Run a block of code unconditionally; all rules are allowed inside a run block, as well as user code. You can think of the {...} as an anonymous subroutine.

stream {...}

Run a block of code unconditionally, which should return a coderef that is a PSGI streamy response.

set $arg => $val

Adds an argument to what we're passing to our template, overriding any value the user sent or we've already set.

default $arg => $val

Adds an argument to what we're passing to our template, but only if it is not defined currently.

del $arg

Deletes an argument we were passing to our template.

show $component

Display the presentation component. If not specified, use the request path as the default page.

dispatch $path

Dispatch again using $path as the request path, preserving args.

next_rule

Break out from the current run block and go on the next rule.

last_rule

Break out from the current run block and stop running rules in this stage.

abort $code

Abort the request; this skips straight to the cleanup stage.

If $code is specified, it's used as the HTTP status code.

redirect $uri

Redirect to another URI.

tangent $uri

Take a continuation here, and tangent to another URI.

plugin

app

See "Plugins and rule ordering", above.

import

Jifty::Dispatcher is an Exporter, that is, part of its role is to blast a bunch of symbols into another package. In this case, that other package is the dispatcher for your application.

You never call import directly. Just:

    use Jifty::Dispatcher -base;

in MyApp::Dispatcher

rules STAGE

Returns an array of all the rules for the stage STAGE.

Valid values for STAGE are

SETUP
RUN
CLEANUP

new

Creates a new Jifty::Dispatcher object. You probably don't ever want to do this. (Jifty.pm does it for you)

handle_request

Actually do what your dispatcher does. For now, the right thing to do is to put the following two lines first:

    require MyApp::Dispatcher;
    MyApp::Dispatcher->handle_request;

_handle_stage NAME, EXTRA_RULES

Handles the all rules in the stage named NAME. Additionally, any other arguments passed after the stage NAME are added to the end of the rules for that stage.

This is the unit which calling "last_rule" skips to the end of.

_handle_rules RULESET

When handed an arrayref or array of rules (RULESET), walks through the rules in order, executing as it goes.

_handle_rule RULE

When handed a single rule in the form of a coderef, _handle_rule, calls _do_run on that rule and returns the result. When handed a rule that turns out to be an array of subrules, recursively calls itself and evaluates the subrules in order.

_do_under

This method is called by the dispatcher internally. You shouldn't need to.

_do_when

This method is called by the dispatcher internally. You shouldn't need to.

_do_before

This method is called by the dispatcher internally. You shouldn't need to.

_do_on

This method is called by the dispatcher internally. You shouldn't need to.

_do_after

This method is called by the dispatcher internally. You shouldn't need to.

already_run

Returns true if the code block has run once already in this request. This can be useful for 'after' rules to ensure that they only run once, even if there is a sub-dispatch which would cause it to run more than once. The idiom is:

    after '/some/path/*' => run {
        return if already_run;
        # ...
    };

_do_redirect PATH

This method is called by the dispatcher internally. You shouldn't need to.

Redirect the user to the URL provided in the mandatory PATH argument.

_do_tangent PATH

This method is called by the dispatcher internally. You shouldn't need to.

Take a tangent to the URL provided in the mandatory PATH argument. (See Jifty::Manual::Continuation for more about tangents.)

_do_stream CODE

The method is called by the dispatcher internally. You shouldn't need to.

Take a coderef that returns a PSGI streamy response code.

_do_abort

This method is called by the dispatcher internally. You shouldn't need to.

Don't display any page. just stop.

_do_show [PATH]

This method is called by the dispatcher internally. You shouldn't need to.

Render a template. If the scalar argument "PATH" is given, render that component. Otherwise, just render whatever we were going to anyway.

_do_dispatch [PATH]

First, this routine runs all the before dispatcher rules, then it runs Jifty->web->handle_request(), then it runs all the main on rules, evaluating each one in turn. If it gets through all the rules without running an abort, redirect or show directive, it shows the template originally requested.

Once it's done with that, it runs all the cleanup rules defined with after.

_match CONDITION

Returns the regular expression matched if the current request fits the condition defined by CONDITION.

CONDITION can be a regular expression, a "simple string" with shell wildcard characters (*, ?, #, [], {}) to match against, or an arrayref or hashref of those. It should even be nestable.

Arrayref conditions represents alternatives: the match succeeds as soon as the first match is found.

Hashref conditions are conjunctions: each non-empty hash key triggers a separate _match_$keyname call on the dispatcher object. For example, a method key would call _match_method with its value to be matched against. After each subcondition is tried (in lexicographical order) and succeeded, the value associated with the '' key is matched again as the condition.

_match_method METHOD

Takes an HTTP method. Returns true if the current request came in with that method.

_match_https

Returns true if the current request is under SSL.

_match_http

Returns true if the current request is not under SSL.

_compile_condition CONDITION

Takes a condition defined as a simple string and return it as a regex condition.

_compile_glob METAEXPRESSION

Private function.

Turns a metaexpression containing *, ? and # into a capturing regex pattern.

Also supports the non-capturing [] and {} notations.

The rules are:

  • A * between two / characters, or between a / and end of string, should match one or more non-slash characters:

        /foo/*/bar
        /foo/*/
        /foo/*
        /*
  • All other * can match zero or more non-slash characters:

        /*bar
        /foo*bar
        *
  • Two stars (**) can match zero or more characters, including slash:

        /**/bar
        /foo/**
        **
  • Consecutive ? marks are captured together:

        /foo???bar      # One capture for ???
        /foo??*         # Two captures, one for ?? and one for *
  • The # character captures one or more digit characters.

  • Brackets such as [a-z] denote character classes; they are not captured.

  • Braces such as {xxx,yyy}] denote alternations; they are not captured.

import_plugins

Imports rules from "plugins" in Jifty into the main dispatcher's space.