path: root/lib/Dancer2/Manual.pod

# ABSTRACT: A gentle introduction to Dancer2
package Dancer2::Manual;

__END__

=pod

=encoding UTF-8

=head1 NAME

Dancer2::Manual - A gentle introduction to Dancer2

=head1 VERSION

version 1.1.0

=head1 DESCRIPTION

Dancer2 is a free and open source web application framework written in Perl.

It's a complete rewrite of L<Dancer>, based on L<Moo> and using a more
robust and extensible fully-OO design.

It's designed to be powerful and flexible, but also easy to use - getting up
and running with your web app is trivial, and an ecosystem of adaptors for
common template engines, session storage, logging methods, serializers, and
plugins to make common tasks easy means you can do what you want to do, your
way, easily.

=head1 INSTALL

Installation of Dancer2 is simple, using your favourite method to install from
CPAN, e.g.:

    perl -MCPAN -e 'install Dancer2'

Thanks to the magic of cpanminus, if you do not have CPAN.pm configured, or
just want a quickfire way to get running, the following should work, at
least on Unix-like systems:

    wget -O - http://cpanmin.us | sudo perl - Dancer2

(If you don't have root access, omit the 'sudo', and cpanminus will install
Dancer2 and prereqs into C<~/perl5>.)

Dancer2 is also available as a package from the package repository of several
distributions, for example on Debian/Ubuntu you should be able to just:

    apt-get install libdancer2-perl

Do be aware, though, that distribution-packaged versions sometimes lag behind
the most recent version on CPAN.

=head1 BOOTSTRAPPING A NEW APP

Create a web application using the dancer script:

    $ dancer2 gen -a MyApp && cd MyApp
    + MyApp
    + MyApp/config.yml
    + MyApp/Makefile.PL
    + MyApp/MANIFEST.SKIP
    + MyApp/.dancer
    + MyApp/cpanfile
    + MyApp/bin
    + MyApp/bin/app.psgi
    + MyApp/environments
    + MyApp/environments/development.yml
    + MyApp/environments/production.yml
    + MyApp/lib
    + MyApp/lib/MyApp.pm
    + MyApp/public
    + MyApp/public/favicon.ico
    + MyApp/public/500.html
    + MyApp/public/dispatch.cgi
    + MyApp/public/404.html
    + MyApp/public/dispatch.fcgi
    + MyApp/public/css
    + MyApp/public/css/error.css
    + MyApp/public/css/style.css
    + MyApp/public/images
    + MyApp/public/images/perldancer.jpg
    + MyApp/public/images/perldancer-bg.jpg
    + MyApp/public/javascripts
    + MyApp/public/javascripts/jquery.js
    + MyApp/t
    + MyApp/t/001_base.t
    + MyApp/t/002_index_route.t
    + MyApp/views
    + MyApp/views/index.tt
    + MyApp/views/layouts
    + MyApp/views/layouts/main.tt

It creates a directory named after the name of the app, along with a
configuration file, a views directory (where your templates and layouts
will live), an environments directory (where environment-specific
settings live), a module containing the actual guts of your application, and
a script to start it. Finally, F<.dancer> indicates the root directory of
your app, making it easier for Dancer2 to determine the various paths it
needs for finding resources and code within your application.

A default skeleton is used to bootstrap the new
application, but you can use the C<-s> option to provide another skeleton.
For example:

    $ dancer2 gen -a MyApp -s ~/mydancerskel

For an example of a skeleton directory check the default one available in
the C<share/> directory of your Dancer2 distribution.

(In what follows we will refer to the directory in which you have created your
Dancer2 application -- I<e.g.,> what C<MyApp> was above -- as the
C<appdir>.)

Because Dancer2 is a L<PSGI> web application framework, you can use the
C<plackup> tool (provided by L<Plack>) for launching the application:

    plackup -p 5000 bin/app.psgi

View the web application at:

    http://localhost:5000

=head1 USAGE

When Dancer2 is imported to a script, that script becomes a webapp, and at
this point, all the script has to do is declare a list of B<routes>. A
route handler is composed by an HTTP method, a path pattern and a code
block. C<strict>, C<warnings> and C<utf8> pragmas are also imported with
Dancer2.

The code block given to the route handler has to return a string which will
be used as the content to render to the client.

Routes are defined for a given HTTP method. For each method supported, a
keyword is exported by the module.

=head2 HTTP Methods

Here are some of the standard HTTP methods which you can use to define your
route handlers.

=over 4

=item * B<GET> The GET method retrieves information, and is the most common

GET requests should be used for typical "fetch" requests - retrieving
information. They should not be used for requests which change data on the
server or have other effects.

When defining a route handler for the GET method, Dancer2 automatically
defines a route handler for the HEAD method (in order to honour HEAD
requests for each of your GET route handlers).

To define a GET action, use the L<get|Dancer2::Manual/get> keyword.

=item * B<POST> The POST method is used to create a resource on the server.

To define a POST action, use the L<post|Dancer2::Manual/post> keyword.

=item * B<PUT> The PUT method is used to replace an existing resource.

To define a PUT action, use the L<put|Dancer2::Manual/put> keyword.

a PUT request should replace the existing resource with that specified - for
instance - if you wanted to just update an email address for a user, you'd
have to specify all attributes of the user again; to make a partial update,
a PATCH request is used.

=item * B<PATCH> The PATCH method updates some attributes of an existing resource.

To define a PATCH action, use the L<patch|Dancer2::Manual/patch> keyword.

=item * B<DELETE> The DELETE method requests that the origin server delete the
resource identified by the Request-URI.

To define a DELETE action, use the L<del|Dancer2::Manual/del> keyword.

=back

=head3 Handling multiple HTTP request methods

Routes can use C<any> to match all, or a specified list of HTTP methods.

The following will match any HTTP request to the path C</myaction>:

    any '/myaction' => sub {
        # code
    }

The following will match GET or POST requests to C</myaction>:

    any ['get', 'post'] => '/myaction' => sub {
        # code
    };

For convenience, any route which matches GET requests will also match HEAD
requests.

=head2 Route Handlers

The route action is the code reference declared. It can access parameters
through the specific L<route_parameters|Dancer2::Manual::Keywords/route_parameters>,
L<query_parameters|Dancer2::Manual::Keywords/query_parameters>, and L<body_parameters|Dancer2::Manual::Keywords/body_parameters>
keywords, which return a L<Hash::MultiValue> object. 
This hashref is a merge of the route pattern matches and the request params.

You can find more details about how params are built and how to access them
in the L<Dancer2::Core::Request> documentation.

=head3 Declaring Routes

To control what happens when a web request is received by your webapp,
you'll need to declare C<routes>. A route declaration indicates which HTTP
method(s) it is valid for, the path it matches (e.g. C</foo/bar>), and a
coderef to execute, which returns the response.

    get '/hello/:name' => sub {
        return "Hi there " . route_parameters->get('name');
    };

The above route specifies that, for GET requests to C</hello/...>, the code
block provided should be executed.

You can also provide routes with a name:

    get 'hi_to' => '/hello/:name' => sub {...};

See L<Dancer2::Manual::Keywords/uri_for_route> on how this can be used.

=head3 Retrieving request parameters

The L<query_parameters|Dancer2::Manual/query_parameters>,
L<route_parameters|Dancer2::Manual/route_parameters>, and
L<body_parameters|Dancer2::Manual/body_parameters> keywords provide
a L<Hash::MultiValue> result from the three different parameters.

=head3 Named matching

A route pattern can contain one or more tokens (a word prefixed with ':').
Each token found in a route pattern is used as a named-pattern match. Any
match will be set in the route parameters.

    get '/hello/:name' => sub {
        return "Hey " . route_parameters->get('name') . ", welcome here!";
    };

Tokens can be optional, for example:

    get '/hello/:name?' => sub {
        my $name = route_parameters->get('name') // 'Whoever you are';
        return "Hello there, $name";
    };

=head3 Named matching with type constraints

Type constraints can be added to tokens.

    get '/user/:id[Int]' => sub {
        # matches /user/34 but not /user/jamesdean
        my $user_id = route_parameters->get('id');
    };

    get '/user/:username[Str]' => sub {
        # matches /user/jamesdean but not /user/34 since that is caught
        # by previous route
        my $username = route_parameters->get('username');
    };

You can even use type constraints to add a regexp check:

    get '/book/:date[StrMatch[qr{\d\d\d\d-\d\d-\d\d}]]' => sub {
        # matches /book/2014-02-04
        my $date = route_parameters->get('date');
    };

The default type library is L<Dancer2::Core::Types> but any type library
built using L<Type::Tiny>'s L<Type::Library> can be used instead.
If you'd like to use a different default type library you must declare it
in the configuration file, for example:

    type_library: My::Type::Library

Alternatively you can specify the type library in which the type is defined
as part of the route definition:

    get '/user/:username[My::Type::Library::Username]' => sub {
        my $username = route_parameters->get('username');
    };

This will load C<My::Type::Library> and from it use the type C<Username>. This
allows types to be used that are not part of the type library defined by config's
C<type_library>.

More complex constructs are allowed such as:

    get '/some/:thing[Int|MyDate]' => sub {
        ...;
    };

See L<Type::Registry/lookup($name)> for more details.

=head3 Wildcard Matching

A route can contain a wildcard (represented by a C<*>). Each wildcard match
will be placed in a list, which the C<splat> keyword returns.

    get '/download/*.*' => sub {
        my ($file, $ext) = splat;
        # do something with $file.$ext here
    };

An extensive, greedier wildcard represented by C<**> (A.K.A. "megasplat") can be
used to define a route. The additional path is broken down and returned as an
arrayref:

    get '/entry/*/tags/**' => sub {
        my ( $entry_id, $tags ) = splat;
        my @tags = @{$tags};
    };

The C<splat> keyword in the above example for the route F</entry/1/tags/one/two>
would set C<$entry_id> to C<1> and C<$tags> to C<['one', 'two']>.

=head3 Mixed named and wildcard matching

A route can combine named (token) matching and wildcard matching.
This is useful when chaining actions:

    get '/team/:team/**' => sub {
        var team => route_parameters->get('team');
        pass;
    };

    prefix '/team/:team';

    get '/player/*' => sub {
        my ($player) = splat;

        # etc...
    };

    get '/score' => sub {
        return score_for( vars->{'team'} );
    };

=head3 Regular Expression Matching

A route can be defined with a Perl regular expression.

In order to tell Dancer2 to consider the route as a real regexp, the route
must be defined explicitly with C<qr{}>, like the following:

    get qr{/hello/([\w]+)} => sub {
        my ($name) = splat;
        return "Hello $name";
    };

A route regex may use named capture groups. The C<captures>
keyword will return a reference to a copy of C<%+>.

=head3 Conditional Matching

Routes may include some matching conditions (on content_type, agent,
user_agent, content_length and path_info):

    get '/foo', {agent => 'Songbird (\d\.\d)[\d\/]*?'} => sub {
      'foo method for songbird'
    }

    get '/foo' => sub {
      'all browsers except songbird'
    }

=head2 Prefix

A prefix can be defined for each route handler, like this:

    prefix '/home';

From here, any route handler is defined to /home/*

    get '/page1' => sub {}; # will match '/home/page1'

You can unset the prefix value

    prefix '/'; # or: prefix undef;
    get '/page1' => sub {}; # will match /page1

Alternatively, to prevent you from ever forgetting to undef the prefix, you
can use lexical prefix like this:

    prefix '/home' => sub {
      get '/page1' => sub {}; # will match '/home/page1'
    }; ## prefix reset to previous value on exit

    get '/page1' => sub {}; # will match /page1

=head2 Delayed responses (Async/Streaming)

L<Dancer2> can provide delayed (otherwise known as I<asynchronous>) responses
using the C<delayed> keyword. These responses are streamed, although you can
set the content all at once, if you prefer.

    get '/status' => sub {
        delayed {
            response_header 'X-Foo' => 'Bar';

            # flush headers (in case of streaming)
            flush;

            # send content to the user
            content 'Hello, world!';

            # you can write more content
            # all streaming
            content 'Hello, again!';

            # when done, close the connection
            done;

            # do whatever you want else, asynchronously
            # the user socket closed by now
            ...
        };
    };

If you are streaming (calling C<content> several times), you must call
C<flush> first. If you're sending only once, you don't need to call C<flush>.

Here is an example of using delayed responses with L<AnyEvent>:

    use Dancer2;
    use AnyEvent;

    my %timers;
    my $count = 5;
    get '/drums' => sub {
        delayed {
            print "Stretching...\n";
            flush; # necessary, since we're streaming

            $timers{'Snare'} = AE::timer 1, 1, delayed {
                $timers{'HiHat'} ||= AE::timer 0, 0.5, delayed {
                    content "Tss...\n";
                };

                content "Bap!\n";

                if ( $count-- == 0 ) {
                    %timers = ();
                    content "Tugu tugu tugu dum!\n";
                    done;

                    print "<enter sound of applause>\n\n";
                    $timers{'Applause'} = AE::timer 3, 0, sub {
                        # the DSL will not available here
                        # because we didn't call the "delayed" keyword
                        print "<applause dies out>\n";
                    };
                }
            };
        };
    };

If an error happens during a write operation, a warning will be issued
to the logger.

You can handle the error yourself by providing an C<on_error> handler:

    get '/' => sub {
        delayed {
            flush;
            content "works";

            # ... user disconnected here ...

            content "fails";

            # ... error triggered ...

            done; # doesn't even get run
        } on_error => sub {
            # delayed{} not needed, DSL already available
            my ($error) = @_;
            # do something with $error
        };
    };

Here is an example that asynchronously streams the contents of a CSV file:

    use Dancer2;
    use Text::CSV_XS  qw< csv >;
    use Path::Tiny    qw< path >;
    use JSON::MaybeXS qw< encode_json >;
    # Create CSV parser
    my $csv = Text::CSV_XS->new({
        binary    => 1,
        auto_diag => 1,
    });
    get '/' => sub {
        # delayed response:
        delayed {
            # streaming content
            flush;
            # Read each row and stream it in JSON
            my $fh = path('filename.csv')->openr_utf8;
            while ( my $row = $csv->getline($fh) ) {
                content encode_json $row;
            }
            # close user connection
            done;
        } on_error => sub {
            my ($error) = @_;
            warning 'Failed to stream to user: ' . request->remote_address;
        };
    };

B<NOTE:> If you just want to send a file's contents asynchronously,
use C<send_file($filename)> instead of C<delayed>, as it will
automatically take advantage of any asynchronous capability.

=head2 Action Skipping

An action can choose not to serve the current request and ask Dancer2 to
process the request with the next matching route.

This is done with the B<pass> keyword, like in the following example

    get '/say/:word' => sub {
        pass if route_parameters->get('word') =~ /^\d+$/;
        "I say a word: " . route_parameters->get('word');
    };

    get '/say/:number' => sub {
        "I say a number: " . route_parameters->get('number');
    };

=head1 HOOKS

Hooks are code references (or anonymous subroutines) that are triggered at
specific moments during the resolution of a request.  They are set up using the
L<hook|Dancer2::Manual::Keywords/hook> keyword.

Many of them are provided by Dancer2's core, but plugins and engines can also
define their own.

=over 4

=item * C<before> hooks

C<before> hooks are evaluated before each request within the context of the
request and receives as argument the app (a L<Dancer2::Core::App> object).

It's possible to define variables which will be accessible in the action
blocks with the L<var keyword|Dancer2::Manual::Keywords/var>.

    hook before => sub {
        var note => 'Hi there';
    };

    get '/foo/*' => sub {
        my ($match) = splat; # 'oversee';
        vars->{note};        # 'Hi there'
    };

For another example, this can be used along with session support to easily
give non-logged-in users a login page:

    hook before => sub {
        if (!session('user') && request->path !~ m{^/login}) {
            # Pass the original path requested along to the handler:
            forward '/login', { requested_path => request->path };
        }
    };

The request keyword returns the current L<Dancer2::Core::Request> object
representing the incoming request.

=item * C<after> hooks

C<after> hooks are evaluated after the response has been built by a route
handler, and can alter the response itself, just before it's sent to the
client.

This hook runs after a request has been processed, but before the response
is sent.

It receives a L<Dancer2::Core::Response> object, which it can modify if it
needs to make changes to the response which is about to be sent.

The hook can use other keywords in order to do whatever it wants.

    hook after => sub {
        response->content(
            q{The "after" hook can alter the response's content here!}
        );
    };

=back

=head2 Templates

=over 4

=item * C<before_template_render>

C<before_template_render> hooks are called whenever a template is going to
be processed, they are passed the tokens hash which they can alter.

    hook before_template_render => sub {
        my $tokens = shift;
        $tokens->{foo} = 'bar';
    };

The tokens hash will then be passed to the template with all the
modifications performed by the hook. This is a good way to setup some
global vars you like to have in all your templates, like the name of the
user logged in or a section name.

=item * C<after_template_render>

C<after_template_render> hooks are called after the view has been rendered.
They receive as their first argument the reference to the content that has
been produced. This can be used to post-process the content rendered by the
template engine.

    hook after_template_render => sub {
        my $ref_content = shift;
        my $content     = ${$ref_content};

        # do something with $content
        ${$ref_content} = $content;
    };

=item * C<before_layout_render>

C<before_layout_render> hooks are called whenever the layout is going to be
applied to the current content. The arguments received by the hook are the
current tokens hashref and a reference to the current content.

    hook before_layout_render => sub {
        my ($tokens, $ref_content) = @_;
        $tokens->{new_stuff} = 42;
        $ref_content = \"new content";
    };

=item * C<after_layout_render>

C<after_layout_render> hooks are called once the complete content of the
view has been produced, after the layout has been applied to the content.
The argument received by the hook is a reference to the complete content
string.

    hook after_layout_render => sub {
        my $ref_content = shift;
        # do something with ${ $ref_content }, which reflects directly
        #   in the caller
    };

=back

=head2 Error Handling

Refer to L<Error Hooks|Dancer2::Manual/Error-Hooks>
for details about the following hooks:

=over 4

=item * C<init_error>

=item * C<before_error>

=item * C<after_error>

=item * C<on_route_exception>

=back

=head2 File Rendering

Refer to L<File Handler|Dancer2::Manual/File-Handler>
for details on the following hooks:

=over 4

=item * C<before_file_render>

=item * C<after_file_render>

=back

=head2 Serializers

=over 4

=item * C<before_serializer> is called before serializing the content, and receives
the content to serialize as an argument.

  hook before_serializer => sub {
    my $content = shift;
    ...
  };

=item * C<after_serializer> is called after the payload has been serialized, and
receives the serialized content as an argument.

  hook after_serializer => sub {
    my $serialized_content = shift;
    ...
  };

=back

=head1 HANDLERS

=head2 File Handler

Whenever a content is produced out of the parsing of a static file, the
L<Dancer2::Handler::File> component is used. This component provides two
hooks, C<before_file_render> and C<after_file_render>.

C<before_file_render> hooks are called just before starting to parse the
file, the hook receives as its first argument the file path that is going to
be processed.

    hook before_file_render => sub {
        my $path = shift;
    };

C<after_file_render> hooks are called after the file has been parsed and the
response content produced. It receives the response object
(L<Dancer2::Core::Response>) produced.

    hook after_file_render => sub {
       my $response = shift;
    };

=head2 Auto page

Whenever a page that matches an existing template needs to be served, the
L<Dancer2::Handler::AutoPage> component is used.

=head2 Writing your own

A route handler is a class that consumes the L<Dancer2::Core::Role::Handler>
role. The class must implement a set of methods: C<methods>, C<regexp> and
C<code> which will be used to declare the route.

Let's look at L<Dancer2::Handler::AutoPage> for example.

First, the matching methods are C<get> and C<head>:

    sub methods { qw(head get) }

Then, the C<regexp> or the I<path> we want to match:

    sub regexp { '/:page' }

Anything will be matched by this route, since we want to check if there's
a view named with the value of the C<page> token. If not, the route needs
to C<pass>, letting the dispatching flow to proceed further.

    sub code {
        sub {
            my $app = shift;
            my $prefix = shift;

            my $template = $app->template_engine;
            if ( !defined $template ) {
                $app->response->has_passed(1);
                return;
            }

            my $page       = $app->request->path;
            my $layout_dir = $template->layout_dir;
            if ( $page =~ m{^/\Q$layout_dir\E/} ) {
                $app->response->has_passed(1);
                return;
            }

            # remove leading '/', ensuring paths relative to the view
            $page =~ s{^/}{};
            my $view_path = $template->view_pathname($page);

            if ( ! $template->pathname_exists( $view_path ) ) {
                $app->response->has_passed(1);
                return;
            }

            my $ct = $template->process( $page );
            return ( $app->request->method eq 'GET' ) ? $ct : '';
        };
    }

The C<code> method passed the L<Dancer2::Core::App> object which provides
access to anything needed to process the request.

A C<register> is then implemented to add the route to the registry and if
the C<auto_page setting> is off, it does nothing.

    sub register {
        my ($self, $app) = @_;

        return unless $app->config->{auto_page};

        $app->add_route(
            method => $_,
            regexp => $self->regexp,
            code   => $self->code,
        ) for $self->methods;
    }

The config parser looks for a C<route_handlers> section and any handler defined
there is loaded. Thus, any random handler can be added to your app.
For example, the default config file for any Dancer2 application is as follows:

    route_handlers:
      File:
        public_dir: /path/to/public
      AutoPage: 1

=head1 ERRORS

=head2 Error Pages

When an HTTP error occurs (i.e. the action responds with a status code other
than 200), this is how Dancer2 determines what page to display.

=over

=item * Looks in the C<views/> directory for a corresponding template file
matching the error code (e.g. C<500.tt> or C<404.tt>). If such a file exists,
it's used to report the error.

=item * Next, looks in the C<public/> directory for a corresponding HTML file
matching the error code (e.g. C<500.html> or C<404.html>). If such a file
exists, it's used to report the error. (Note, however, that if B<show_stacktrace>
is set to true, in the case of a 500 error the static HTML page will not be
shown, but will be replaced with a default error page containing more
informative diagnostics. For more information see L<Dancer2::Config>.)

(In older versions, B<show_errors> was used instead of B<show_stacktrace>.
Both are supported, but B<show_errors> is deprecated.)

=item * As default, render a generic error page on the fly.

=back

=head2 Execution Errors

When an error occurs during the route execution, Dancer2 will render an
error page with the HTTP status code 500.

It's possible either to display the content of the error message or to hide
it with a generic error page. This is a choice left to the end-user and can
be controlled with the B<show_stacktrace> setting (see above).

=head2 Error Hooks

When an error is caught by Dancer2's core, an exception object is built (of
the class L<Dancer2::Core::Error>). This class provides a hook to let the
user alter the error workflow if needed.

C<init_error> hooks are called whenever an error object is built, the object
is passed to the hook.

    hook init_error => sub {
        my $error = shift;
        # do something with $error
    };

I<This hook was named B<before_error_init> in Dancer, both names currently
are synonyms for backward-compatibility.>

C<before_error> hooks are called whenever an error is going to be thrown, it
receives the error object as its sole argument.

    hook before_error => sub {
        my $error = shift;
        # do something with $error
    };

I<This hook was named B<before_error_render> in Dancer, both names currently
are synonyms for backward-compatibility.>

C<after_error> hooks are called whenever an error object has been thrown, it
receives a L<Dancer2::Core::Response> object as its sole argument.

    hook after_error => sub {
        my $response = shift;
    };

I<This hook was named B<after_error_render> in Dancer, both names currently
are synonyms for backward-compatibility.>

C<on_route_exception> is called when an exception has been caught, at the
route level, just before rethrowing it higher. This hook receives a
L<Dancer2::Core::App> and the error as arguments.

  hook on_route_exception => sub {
    my ($app, $error) = @_;
  };

=head1 SESSIONS

=head2 Handling sessions

It's common to want to use sessions to give your web applications state; for
instance, allowing a user to log in, creating a session, and checking that
session on subsequent requests.

By default Dancer 2 has L<Simple|Dancer2::Session::Simple> sessions enabled.
It implements a very simple in-memory session storage. This will be fast and
useful for testing, but such sessions will not persist between restarts of
your app.

If you'd like to use a different session engine you must declare it in the
configuration file.

For example to use YAML file base sessions you need to add the following
to your F<config.yml>:

    session: YAML

Or, to enable session support from within your code,

    set session => 'YAML';

(However, controlling settings is best done from your config file.)

The L<Dancer2::Session::YAML> backend implements a file-based YAML session
storage to help with debugging, but shouldn't be used on production systems.

There are other session backends, such as L<Dancer2::Session::Memcached>,
which are recommended for production use.

You can then use the L<session|Dancer2::Manual/session> keyword to manipulate the
session:

=head3 Storing data in the session

Storing data in the session is as easy as:

    session varname => 'value';

=head3 Retrieving data from the session

Retrieving data from the session is as easy as:

    session('varname')

Or, alternatively,

    session->read("varname")

=head3 Controlling where sessions are stored

For disc-based session backends like L<Dancer2::Session::YAML>,
session files are written to the session dir specified by
the C<session_dir> setting, which defaults to C<./sessions>
if not specifically set.

If you need to control where session files are created, you can do so
quickly and easily within your config file, for example:

    session: YAML
    engines:
      session:
        YAML:
          session_dir: /tmp/dancer-sessions

If the directory you specify does not exist, Dancer2 will attempt to create
it for you.

=head3 Changing session ID

If you wish to change the session ID (for example on privilege level change):

    my $new_session_id = app->change_session_id

=head3 Destroying a session

When you're done with your session, you can destroy it:

    app->destroy_session

=head2 Sessions and logging in

A common requirement is to check the user is logged in, and, if not, require
them to log in before continuing.

This can easily be handled using a before hook to check their session:

    use Dancer2;
    set session => "Simple";

    hook before => sub {
        if (!session('user') && request->path !~ m{^/login}) {
            forward '/login', { requested_path => request->path };
        }
    };

    get '/' => sub { return "Home Page"; };

    get '/secret' => sub { return "Top Secret Stuff here"; };

    get '/login' => sub {
        # Display a login page; the original URL they requested is available as
        # query_parameters->get('requested_path'), so could be put in a hidden field in the form
        template 'login', { path => query_parameters->get('requested_path') };
    };

    post '/login' => sub {
        # Validate the username and password they supplied
        if (body_parameters->get('user') eq 'bob' && body_parameters->get('pass') eq 'letmein') {
            session user => body_parameters->get('user');
            redirect body_parameters->get('path') || '/';
        } else {
            redirect '/login?failed=1';
        }
    };

    dance();

Here is what the corresponding C<login.tt> file should look like. You should
place it in a directory called C<views/>:

    <html>
      <head>
        <title>Session and logging in</title>
      </head>
      <body>
        <form action='/login' method='POST'>
            User Name : <input type='text' name='user'/>
            Password: <input type='password' name='pass' />

            <!-- Put the original path requested into a hidden
                       field so it's sent back in the POST and can be
                       used to redirect to the right page after login -->
            <input type='hidden' name='path' value='<% path %>'/>

            <input type='submit' value='Login' />
        </form>
      </body>
    </html>

Of course, you'll probably want to validate your users against a database
table, or maybe via IMAP/LDAP/SSH/POP3/local system accounts via PAM etc.
L<Authen::Simple> is probably a good starting point here!

A simple working example of handling authentication against a database table
yourself (using L<Dancer2::Plugin::Database> which provides the C<database>
keyword, and L<Crypt::SaltedHash> to handle salted hashed passwords (well,
you wouldn't store your users passwords in the clear, would you?)) follows:

    post '/login' => sub {
        my $user_value = body_parameters->get('user');
        my $pass_value = body_parameters->get('pass');

        my $user = database->quick_select('users',
            { username => $user_value }
        );
        if (!$user) {
            warning "Failed login for unrecognised user $user_value";
            redirect '/login?failed=1';
        } else {
            if (Crypt::SaltedHash->validate($user->{password}, $pass_value))
            {
                debug "Password correct";
                # Logged in successfully
                session user => $user;
                redirect body_parameters->get('path') || '/';
            } else {
                debug("Login failed - password incorrect for " . $user_value);
                redirect '/login?failed=1';
            }
        }
    };

=head3 Retrieve complete hash stored in session

Get complete hash stored in session:

    my $hash = session;

=head2 Writing a session engine

In Dancer 2, a session backend consumes the role
L<Dancer2::Core::Role::SessionFactory>.

The following example using the Redis session demonstrates how session
engines are written in Dancer 2.

First thing to do is to create the class for the session engine,
we'll name it C<Dancer2::Session::Redis>:

     package Dancer2::Session::Redis;
     use Moo;
     with 'Dancer2::Core::Role::SessionFactory';

we want our backend to have a handle over a Redis connection.
To do that, we'll create an attribute C<redis>

     use JSON;
     use Redis;
     use Dancer2::Core::Types; # brings helper for types

     has redis => (
         is => 'rw',
         isa => InstanceOf['Redis'],
         lazy => 1,
         builder => '_build_redis',
     );

The lazy attribute says to Moo that this attribute will be
built (initialized) only when called the first time. It means that
the connection to Redis won't be opened until necessary.

     sub _build_redis {
         my ($self) = @_;
         Redis->new(
             server => $self->server,
             password => $self->password,
             encoding => undef,
         );
     }

Two more attributes, C<server> and C<password> need to be created.
We do this by defining them in the config file. Dancer2 passes anything
defined in the config to the engine creation.

     # config.yml
     ...
     engines:
       session:
         Redis:
           server: foo.mydomain.com
           password: S3Cr3t

The server and password entries are now passed to the constructor
of the Redis session engine and can be accessed from there.

     has server => (is => 'ro', required => 1);
     has password => (is => 'ro');

Next, we define the subroutine C<_retrieve> which will return a session
object for a session ID it has passed. Since in this case, sessions are
going to be stored in Redis, the session ID will be the key, the session the value.
So retrieving is as easy as doing a get and decoding the JSON string returned:

     sub _retrieve {
         my ($self, $session_id) = @_;
         my $json = $self->redis->get($session_id);
         my $hash = from_json( $json );
         return bless $hash, 'Dancer2::Core::Session';
     }

The C<_flush> method is called by Dancer when the session needs to be stored in
the backend. That is actually a write to Redis. The method receives a C<Dancer2::Core::Session>
object and is supposed to store it.

     sub _flush {
         my ($self, $session) = @_;
         my $json = encode_json( { %{ $session } } );
         $self->redis->set($session->id, $json);
     }

For the C<_destroy> method which is supposed to remove a session from the backend,
deleting the key from Redis is enough.

     sub _destroy {
         my ($self, $session_id) = @_;
         $self->redis->del($session_id);
     }

The C<_sessions> method which is supposed to list all the session IDs currently
stored in the backend is done by listing all the keys that Redis has.

     sub _sessions {
         my ($self) = @_;
         my @keys = $self->redis->keys('*');
         return \@keys;
     }

The session engine is now ready.

=head3 The Session keyword

Dancer2 maintains two session layers.

The first layer, L<Dancer2::Core::Session> provides a session object
which represents the current session. You can read from it as many
times as you want, and write to it as many times as you want.

The second layer is the session engine (L<Dancer2::Session::Simple>
is one example), which is used in order to implement the reading and
writing from the actual storage. This is read only once, when a request
comes in (using a cookie whose value is C<dancer.session> by default).
At the end of a request, all the data you've written will be flushed
to the engine itself, which will do the actual write to the storage
(whether it's in a hash in memory, in Memcache, or in a database).

=head1 TEMPLATES

Returning plain content is all well and good for examples or trivial apps,
but soon you'll want to use templates to maintain separation between your
code and your content. Dancer2 makes this easy.

Your route handlers can use the L<template|Dancer2::Manual/template> keyword
to render templates.

=head2 Views

In Dancer2, a file which holds a template is called a I<view>.  Views are
located in the C<appdir/views> directory.

You can change this location by changing the setting 'views'. For instance
if your templates are located in the 'templates' directory, do the
following:

    set views => path( app->location , 'templates' );

By default, the internal template engine L<Dancer2::Template::Simple> is
used, but you may want to upgrade to L<Template
Toolkit|http://www.template-toolkit.org/>. If you do so, you have to enable
this engine in your settings as explained in
L<Dancer2::Template::TemplateToolkit> and you'll also have to install the
L<Template> module.

In order to render a view, just call the
L<template|Dancer2::Manual/template> keyword at the end of the action by
giving the view name and the HASHREF of tokens to interpolate in the view
(note that for convenience, the request, session, params and vars are
automatically accessible in the view, named C<request>, C<session>, C<params>,
and C<vars>) - for example:

    hook before => sub { var time => scalar(localtime) };

    get '/hello/:name' => sub {
        my $name = route_parameters->get('name');
        template 'hello.tt', { name => $name };
    };

The template C<hello.tt> could contain, for example:

    <p>Hi there, [% name %]!</p>
    <p>You're using [% request.user_agent %]</p>
    [% IF session.username %]
        <p>You're logged in as [% session.username %]</p>
    [% END %]
    It's currently [% vars.time %]

For a full list of the tokens automatically added to your template (like
C<session>, C<request>, and C<vars>, refer to
L<Dancer2::Core::Role::Template>).

By default, views use a F<.tt> extension. This can be overridden by setting
the C<extension> attribute in the template engine configuration:

    set engines => {
        template => {
            template_toolkit => {
                extension => 'foo',
            },
        },
    };

=head2 Layouts

A layout is a special view, located in the F<layouts> directory (inside the
views directory) which must have a token named C<content>. That token marks
the place where to render the action view. This lets you define a global
layout for your actions, and have each individual view contain only
specific content. This is a good thing and helps avoid lots of needless
duplication of HTML. :)

For example, the layout F<views/layouts/main.tt>:

    <html>
        <head>...</head>
        <body>
        <div id="header">
        ...
        </div>

        <div id="content">
        [% content %]
        </div>

        </body>
    </html>

You can tell your app which layout to use with C<layout: name> in the config
file, or within your code:

    set layout => 'main';

You can control which layout to use (or whether to use a layout at all) for
a specific request without altering the layout setting by passing an options
hashref as the third param to the template keyword:

    template 'index.tt', {}, { layout => undef };

If your application is not mounted under root (C</>), you can use a
C<before_template_render> hook instead of hardcoding the path into your
application for your CSS, images and JavaScript:

    hook before_template_render => sub {
        my $tokens = shift;
        $tokens->{uri_base} = request->base->path;
    };

Then in your layout, modify your CSS inclusion as follows:

    <link rel="stylesheet" href="[% uri_base %]/css/style.css" />

From now on you can mount your application wherever you want, without any
further modification of the CSS inclusion.

=head2 Encoding

If you use L<Plack> and have a Unicode problem with your Dancer2
application, don't forget to check if you have set your template engine to
use Unicode, and set the default charset to UTF-8. So, if you are using
template toolkit, your config file will look like this:

    charset: UTF-8
    engines:
      template:
        template_toolkit:
          ENCODING: utf8

=head2 Default Template Variables

Every template knows about the following variables, which are provided by
L<Dancer2::Core::Role::Template>. Some are similar to the keywords you can
use in the Perl part of your Dancer2 application.

=over 4

=item * B<perl_version>

Current version of perl, effectively
L<C<$^V>|http://perldoc.perl.org/perlvar.html#%24%5eV>.

=item * B<dancer_version>

Current version of Dancer2, effectively C<< Dancer2->VERSION >>.

=item * B<settings>

A hash of the application configuration. This is like
the L<config|Dancer2::Manual/config> keyword.

=item * B<request>

The current request object. This is like the L<request|Dancer2::Manual/request> keyword.

=item * B<params>

A hash reference of all the parameters.

Currently the equivalent of C<< $request->params >>, and like the
L<params|Dancer2::Manual/params> keyword.

=item * B<vars>

The list of request variables, which is what you would get if you
called the L<vars|Dancer2::Manual/vars> keyword.

=item * B<session>

The current session data, if a session exists. This is like
the L<session|Dancer2::Manual/session> keyword.

=back

=head1 STATIC FILES

=head2 Static Directory

Static files are served from the F<./public> directory. You can specify a
different location by setting the C<public_dir> option:

    set public_dir => path( app->location , 'static' );

When you modify default public_dir you have to set C<static_handler> option.

    set static_handler => true;

Note that the public directory name is not included in the URL. A file
F<./public/css/style.css> is made available as
L<http://example.com/css/style.css>.

=head2 Static File from a Route Handler

It's possible for a route handler to send a static file, as follows:

    get '/download/*' => sub {
        my ($file) = splat;

        send_file $file;
    };

Or even if you want your index page to be a plain old F<index.html> file,
just do:

    get '/' => sub {
        send_file '/index.html'
    };

=head1 FILE UPLOADS

Files are uploaded in Dancer2 using the class L<Dancer2::Core::Request::Upload>.
The objects are accessible within the route handlers using the C<upload>
keyword:

    post '/upload' => sub {
        my $upload     = upload('file_input_name');    # upload object
        $upload->copy_to('Uploads/');
    };

=head1 CONFIGURATION

=head2 Configuration and environments

Configuring a Dancer2 application can be done in many ways. The easiest one
(and maybe the dirtiest) is to put all your settings statements at the top
of your script, before calling the C<dance()> method.

Other ways are possible: for example, you can define all your settings in the file
C<appdir/config.yml>. For this, you must have installed the L<YAML> module, and
of course, write the config file in YAML.

That's better than the first option, but it's still not perfect as you can't
switch easily from an environment to another without rewriting the config
file.

A better solution is to have one F<config.yml> file with default global
settings, like the following:

    # appdir/config.yml
    logger: 'file'
    layout: 'main'

And then write as many environment files as you like in
C<appdir/environments>. That way, the appropriate environment config file
will be loaded according to the running environment (if none is specified,
it will be 'development').

You can change the running environment when starting your app using the
C<plackup> command's C<--env> or C<--E> switch:

    plackup -E production bin/app.psgi

Altenatively, you can set the
L<C<DANCER_ENVIRONMENT>|https://metacpan.org/pod/Dancer2::Config#DANCER_ENVIRONMENT>
environment variable in the shell or in your web server's configuration file.

Typically, you'll want to set the following values in a development config
file:

    # appdir/environments/development.yml
    log: 'debug'
    startup_info: 1
    show_stacktrace:  1

And in a production one:

    # appdir/environments/production.yml
    log: 'warning'
    startup_info: 0
    show_stacktrace:  0

Please note that you are not limited to writing configuration files in YAML.
Dancer2 supports any file format that is supported by L<Config::Any>, such
as JSON, XML, INI files, and Apache-style config files. See the Dancer2
L<configuration guide|https://metacpan.org/pod/Dancer2::Config#MANIPULATING-SETTINGS-VIA-CONFIGURATION-FILES>
for more information.

=head2 Accessing configuration information

A Dancer2 application can use the C<config> keyword to easily access the
settings within its config file, for instance:

    get '/appname' => sub {
        return "This is " . config->{appname};
    };

This makes keeping your application's settings all in one place simple and
easy - you shouldn't need to worry about implementing all that yourself. :)

=head2 Settings

It's possible to change almost every parameter of the application via the
settings mechanism.

A setting is a key/value pair assigned by the keyword B<set>:

    set setting_name => 'setting_value';

More usefully, settings can be defined in a configuration file.
Environment-specific settings can also be defined in environment-specific
files (for instance, you do not want to show error stacktraces in
production, and might want extra logging in development).

=head2 Serializers

When writing a webservice, data serialization/deserialization is a common
issue to deal with. Dancer2 can automatically handle that for you, via a
serializer.

When setting up a serializer, a new behaviour is authorized for any route
handler you define: any non-scalar response will be rendered as a serialized
string, via the current serializer.

Here is an example of a route handler that will return a hashref:

    use Dancer2;
    set serializer => 'JSON';

    get '/user/:id/' => sub {
        { foo => 42,
          number => 100234,
          list => [qw(one two three)],
        }
    };

Dancer2 will render the response via the current serializer.

Hence, with the JSON serializer set, the route handler above would result in
a content like the following:

    {"number":100234,"foo":42,"list":["one","two","three"]}

If you send a value which is validated serialized data, but is not in the
form a key and value pair (such as a serialized string or a JSON array), the
data will not be available in C<params> but will be available in
C<< request->data >>.

The following serializers are available, be aware they dynamically depend on
Perl modules you may not have on your system.

=over 4

=item * B<JSON>

Requires L<JSON>.

=item * B<YAML>

Requires L<YAML>,

=item * B<XML>

Requires L<XML::Simple>.

=item * B<Mutable>

Will try to find the appropriate serializer using the B<Content-Type> and
B<Accept-type> header of the request.

=back

=head2 Importing using Appname

An app in Dancer2 uses the class name (defined by the C<package> function) to
define the App name. Thus separating the App to multiple files, actually means
creating multiple applications. This means that any engine defined in an application,
because the application is a complete separate scope, will not be available to a
different application:

     package MyApp::User {
         use Dancer2;
         set serializer => 'JSON';
         get '/view' => sub {...};
     }

     package MyApp::User::Edit {
         use Dancer2;
         get '/edit' => sub {...};
     }

These are two different Dancer2 Apps. They have different scopes, contexts,
and thus different engines. While C<MyApp::User> has a serializer defined,
C<MyApp::User::Edit> will not have that configuration.

By using the import option C<appname>, we can ask Dancer2 to extend an
App without creating a new one:

     package MyApp::User {
         use Dancer2;
         set serializer => 'JSON';
         get '/view' => sub {...};
     }

     package MyApp::User::Edit {
         use Dancer2 appname => 'MyApp::User'; # extending MyApp::User
         get '/edit' => sub {...};
     }

The import option C<appname> allows you to seamlessly extend Dancer2 Apps
without creating unnecessary additional applications or repeat any definitions.
This allows you to spread your application routes across multiple files and allow
ease of mind when developing it, and accommodate multiple developers working
on the same codebase.

     # app.pl
     use MyApp::User;
     use MyApp::User::Edit;

     # single application composed of routes provided in multiple files
     MyApp::User->to_app;

This way only one class needs to be loaded while creating an app:

     # app.pl:
     use MyApp::User;
     MyApp::User->to_app;

=head1 LOGGING

=head2 Configuring logging

It's possible to log messages generated by the application and by Dancer2
itself.

To start logging, select the logging engine you wish to use with the
C<logger> setting; Dancer2 includes built-in log engines named C<file> and
C<console>, which log to a logfile and to the console respectively.

To enable logging to a file, add the following to your config file:

    logger: 'file'

Then you can choose which kind of messages you want to actually log:

    log: 'core'      # will log debug, info, warnings, errors,
                     #   and messages from Dancer2 itself
    log: 'debug'     # will log debug, info, warning and errors
    log: 'info'      # will log info, warning and errors
    log: 'warning'   # will log warning and errors
    log: 'error'     # will log only errors

If you're using the C<file> logging engine, a directory C<appdir/logs> will
be created and will host one logfile per environment. The log message
contains the time it was written, the PID of the current process, the
message and the caller information (file and line).

=head2 Logging your own messages

Just call L<debug|https://metacpan.org/pod/Dancer2::Manual#debug>,
L<info|https://metacpan.org/pod/Dancer2::Manual#info>,
L<warning|https://metacpan.org/pod/Dancer2::Manual#warning> or
L<error|https://metacpan.org/pod/Dancer2::Manual#error> with your message:

    debug "This is a debug message from my app.";

=head1 TESTING

=head2 Using Plack::Test

L<Plack::Test> receives a common web request (using standard L<HTTP::Request>
objects), fakes a web server in order to create a proper PSGI request, and sends it
to the web application. When the web application returns a PSGI response
(which Dancer applications do), it will then convert it to a common web response
(as a standard L<HTTP::Response> object).

This allows you to then create requests in your test, create the code reference
for your web application, call them, and receive a response object, which can
then be tested.

=head3 Basic Example

Assuming there is a web application:

     # MyApp.pm
     package MyApp;
     use Dancer2;
     get '/' => sub {'OK'};
     1;

The following test I<base.t> is created:

     # base.t
     use strict;
     use warnings;
     use Test::More tests => 2;
     use Plack::Test;
     use HTTP::Request;
     use MyApp;

Creating a coderef for the application using the C<to_app> keyword:

     my $app = MyApp->to_app;

Creating a test object from L<Plack::Test> for the application:

     my $test = Plack::Test->create($app);

Creating the first request object and sending it to the test object
to receive a response:

     my $request  = HTTP::Request->new( GET => '/' );
     my $response = $test->request($request);

It can now be tested:

     ok( $response->is_success, '[GET /] Successful request' );
     is( $response->content, 'OK', '[GET /] Correct content' );

=head3 Putting it together

     # base.t
     use strict;
     use warnings;
     use Test::More;
     use Plack::Test;
     use HTTP::Request::Common;
     use MyApp;

     my $test     = Plack::Test->create( MyApp->to_app );
     my $response = $test->request( GET '/' );

     ok( $response->is_success, '[GET /] Successful request' );
     is( $response->content, 'OK', '[GET /] Correct content' );

     done_testing();

=head3 Subtests

Tests can be separated using L<Test::More>'s C<subtest> functionality,
thus creating multiple self-contained tests that don't overwrite each other.

Assuming we have a different app that has two states we want to test:

     # MyApp.pm
     package MyApp;
     use Dancer2;
     set serializer => 'JSON';

     get '/:user' => sub {
         my $user = route_parameters->get('user');

         $user and return { user => $user };

         return {};
     };

     1;

This is a contrived example of a route that checks for a user
parameter. If it exists, it returns it in a hash with the key
'user'. If not, it returns an empty hash

     # param.t
     use strict;
     use warnings;
     use Test::More;
     use Plack::Test;
     use HTTP::Request::Common;
     use MyApp;

     my $test = Plack::Test->create( MyApp->to_app );

     subtest 'A empty request' => sub {
         my $res = $test->request( GET '/' );
         ok( $res->is_success, 'Successful request' );
         is( $res->content '{}', 'Empty response back' );
     };

     subtest 'Request with user' => sub {
         my $res = $test->request( GET '/?user=sawyer_x' );
         ok( $res->is_success, 'Successful request' );
         is( $res->content '{"user":"sawyer_x"}', 'Empty response back' );
     };

     done_testing();

=head3 Cookies

To handle cookies, which are mostly used for maintaining sessions,
the following modules can be used:

=over 4

=item * L<Test::WWW::Mechanize::PSGI>

=item * L<LWP::Protocol::PSGI>

=item * L<HTTP::Cookies>

=back

Taking the previous test, assuming it actually creates and uses
cookies for sessions:

     # ... all the use statements
     use HTTP::Cookies;

     my $jar  = HTTP::Cookies->new;
     my $test = Plack::Test->create( MyApp->to_app );

     subtest 'A empty request' => sub {
         my $res = $test->request( GET '/' );
         ok( $res->is_success, 'Successful request' );
         is( $res->content '{}', 'Empty response back' );
         $jar->extract_cookies($res);
         ok( $jar->as_string, 'We have cookies!' );
     };

     subtest 'Request with user' => sub {
         my $req = GET '/?user=sawyer_x';
         $jar->add_cookie_header($req);
         my $res = $test->request($req);
         ok( $res->is_success, 'Successful request' );
         is( $res->content '{"user":"sawyer_x"}', 'Empty response back' );
         $jar->extract_cookies($res);

         ok( ! $jar->as_string, 'All cookies deleted' );
     };

     done_testing();

Here a cookie jar is created, all requests and responses, existing
cookies, as well as cookies that were deleted by the response, are checked.

=head3 Accessing the configuration file

By importing Dancer2 in the command line scripts, there is full
access to the configuration using the imported keywords:

     use strict;
     use warnings;
     use Test::More;
     use Plack::Test;
     use HTTP::Request::Common;
     use MyApp;
     use Dancer2;

     my $appname = config->{'appname'};
     diag "Testing $appname";

     # ...

=head1 PACKAGING

=head2 Carton

=head3 What it does

L<Carton> sets up a local copy of your project prerequisites. You only
need to define them in a file and ask Carton to download all of them
and set them up.
When you want to deploy your app, you just carry the git clone and ask
Carton to set up the environment again and you will then be able to run it.

The benefits are multifold:

=over 4

=item * Local Directory copy

By putting all the dependencies in a local directory, you can make
sure they aren't updated by someone else by accident and their versions
locked to the version you picked.

=item * Sync versions

Deciding which versions of the dependent modules your project needs
allows you to sync this with other developers as well. Now you're all
using the same version and they don't change unless you want update the
versions you want. When updated everyone again uses the same new version
of everything.

=item * Carry only the requirement, not bundled modules

Instead of bundling the modules, you only actually bundle the requirements.
Carton builds them for you when you need it.

=back

=head3 Setting it up

First set up a new app:

     $ dancer2 gen -a MyApp
     ...

Delete the files that are not needed:

     $ rm -f Makefile.PL MANIFEST MANIFEST.SKIP

Create a git repo:

     $ git init && git add . && git commit -m "initial commit"

Add a requirement using the L<cpanfile> format:

     $ cat > cpanfile
     requires 'Dancer2' => 0.155000;
     requires 'Template' => 0;
     recommends 'URL::Encode::XS' => 0;
     recommends 'CGI::Deurl::XS' => 0;
     recommends 'HTTP::Parser::XS' => 0;

Ask carton to set it up:

     $ carton install
     Installing modules using [...]
     Successfully installed [...]
     ...
     Complete! Modules were install into [...]/local

Now we have two files: I<cpanfile> and I<cpanfile.snapshot>. We
add both of them to our Git repository and we make sure we don't
accidentally add the I<local/> directory Carton created which holds
the modules it installed:

     $ echo local/ >> .gitignore
     $ git add .gitignore cpanfile cpanfile.snapshot
     $ git commit -m "Start using carton"

When we want to update the versions on the production machine,
we simply call:

     $ carton install --deployment

By using --deployment we make sure we only install the modules
we have in our cpanfile.snapshot file and do not fallback to querying
the CPAN.

=head2 FatPacker

L<App::FatPacker> (using its command line interface, L<fatpack>) packs
dependencies into a single file, allowing you to carry a single file
instead of a directory tree.

As long as your application is pure-Perl, you could create a single
file with your application and all of Dancer2 in it.

The following example will demonstrate how this can be done:

Assuming we have an application in I<lib/MyApp.pm>:

     package MyApp;
     use Dancer2;
     get '/' => sub {'OK'};
     1;

And we have a handler in I<bin/app.pl>:

     use strict;
     use warnings;
     use FindBin;
     use lib "$FindBin::Bin/../lib";
     use MyApp;

     MyApp->to_app;

To fatpack it, we begin by tracing the script:

     $ fatpack trace bin/app.pl

This creates a I<fatpacker.trace> file. From this we create the packlists:

     $ fatpack packlists-for `cat fatpacker.trace` > packlists

The packlists are stored in a file called I<packlists>.

Now we create the tree using the following command:

     $ fatpack tree `cat packlists`

The tree is created under the directory I<fatlib>.

Now we create a file containing the dependency tree, and add our script to it,
using the following command:

     $ (fatpack file; cat bin/app.pl) > myapp.pl

This creates a file called I<myapp.pl> with everything in it. Dancer2 uses
L<MIME::Types> which has a database of all MIME types and helps translate those.
The small database file containing all of these types is a binary and therefore
cannot be fatpacked. Hence, it needs to be copied to the current directory so our
script can find it:

     $ cp fatlib/MIME/types.db .

=head1 MIDDLEWARES

=head2 Plack middlewares

If you want to use Plack middlewares, you need to enable them using
L<Plack::Builder> as such:

    # in app.psgi or any other handler
    use MyApp;
    use Plack::Builder;

    builder {
        enable 'Deflater';
        enable 'Session', store => 'File';
        enable 'Debug', panels => [ qw<DBITrace Memory Timer> ];
        MyApp->to_app;
    };

The nice thing about this setup is that it will work seamlessly through
Plack or through the internal web server.

    # load dev web server (without middlewares)
    perl -Ilib app.psgi

    # load plack web server (with middlewares)
    plackup -I lib app.psgi

You do not need to provide different files for either server.

=head3 Path-based middlewares

If you want to set up a middleware for a specific path, you can do that using
L<Plack::Builder> which uses L<Plack::App::URLMap>:

    # in your app.psgi or any other handler
    use MyApp;
    use Plack::Builder;

    my $special_handler = sub { ... };

    builder {
        mount '/special' => $special_handler;
        mount '/'        => MyApp->to_app;
    };

=head3 Removing default middlewares

By default, a Dancer2 app is automatically wrapped with the following middleware

=over 4

=item * L<Plack::Middleware::FixMissingBodyInRedirect>

=item * L<Plack::Middleware::Head>

=back

You can configure the setting C<no_default_middleware> to a true value to stop your
Dancer2 app being wrapped with these default middleware layers.

    # in you Dancer2 app or config.yml
    package MyApp;
    use Dancer2

    set no_default_middleware => true;

This is necessary if you need to add eTag or ContentMD5 headers to
C<HEAD> requests, and you are encouraged to manually add those default
middleware back into your PSGI stack.

=head3 Running on Perl web servers with plackup

A number of Perl web servers supporting PSGI are available on CPAN:

=over 4

=item * L<Starman>

C<Starman> is a high performance web server, with support for preforking,
signals, multiple interfaces, graceful restarts and dynamic worker pool
configuration.

=item * L<Twiggy>

C<Twiggy> is an C<AnyEvent> web server, it's light and fast.

=item * L<Corona>

C<Corona> is a C<Coro> based web server.

=back

To start your application, just run plackup (see L<Plack> and specific
servers above for all available options):

   $ plackup bin/app.psgi
   $ plackup -E deployment -s Starman --workers=10 -p 5001 -a bin/app.psgi

As you can see, the scaffolded Perl script for your app can be used as a
PSGI startup file.

=head4 Enabling content compression

Content compression (gzip, deflate) can be easily enabled via a Plack
middleware (see L<Plack/Plack::Middleware>): L<Plack::Middleware::Deflater>.
It's a middleware to encode the response body in gzip or deflate, based on the
C<Accept-Encoding> HTTP request header.

Enable it as you would enable any Plack middleware. First you need to
install L<Plack::Middleware::Deflater>, then in the handler (usually
F<app.psgi>) edit it to use L<Plack::Builder>, as described above:

    use Dancer2;
    use MyApp;
    use Plack::Builder;

    builder {
        enable 'Deflater';
        MyApp->to_app;
    };

To test if content compression works, trace the HTTP request and response
before and after enabling this middleware. Among other things, you should
notice that the response is gzip or deflate encoded, and contains a header
C<Content-Encoding> set to C<gzip> or C<deflate>.

=head3 Running multiple apps with Plack::Builder

You can use L<Plack::Builder> to mount multiple Dancer2 applications on a
L<PSGI> webserver like L<Starman>.

Start by creating a simple app.psgi file:

    use OurWiki;  # first app
    use OurForum; # second app
    use Plack::Builder;

    builder {
        mount '/wiki'  => OurWiki->to_app;
        mount '/forum' => OurForum->to_app;
    };

and now use L<Starman>

    plackup -a app.psgi -s Starman

Currently this still demands the same appdir for both (default circumstance)
but in a future version this will be easier to change while staying very
simple to mount.

=head3 Running from Apache with Plack

You can run your app from Apache using PSGI (Plack), with a config like the
following:

    <VirtualHost myapp.example.com>
        ServerName www.myapp.example.com
        ServerAlias myapp.example.com
        DocumentRoot /websites/myapp.example.com

        <Directory /home/myapp/myapp>
            AllowOverride None
            Order allow,deny
            Allow from all
        </Directory>

        <Location />
            SetHandler perl-script
            PerlResponseHandler Plack::Handler::Apache2
            PerlSetVar psgi_app /websites/myapp.example.com/app.psgi
        </Location>

        ErrorLog  /websites/myapp.example.com/logs/error_log
        CustomLog /websites/myapp.example.com/logs/access_log common
    </VirtualHost>

To set the environment you want to use for your application (production or
development), you can set it this way:

    <VirtualHost>
        ...
        SetEnv DANCER_ENVIRONMENT "production"
        ...
    </VirtualHost>

=head1 PLUGINS

=head2 Writing a plugin

See L<Dancer2::Plugin/Writing the plugin> for information on how to author
a new plugin for Dancer2.

=head1 EXPORTS

By default, C<use Dancer2> exports all the DSL keywords and sets up the
webapp under the name of the current package. The following tags control
exports and webapp namespace.

=over 4

=item * B<!keyword>

If you want to prevent Dancer2 from exporting specific keywords (perhaps you
plan to implement them yourself in a different way, or they clash with
another module you're loading), you can simply exclude them:

    use Test::More;
    use Dancer2 qw(!pass);

The above would import all keywords as usual, with the exception of C<pass>.

=item * B<appname>

A larger application may split its source between several packages to aid
maintainability. Dancer2 will create a separate application for each
package, each having separate hooks, config and/or engines. You can force
Dancer2 to collect the route and hooks into a single application with the
C<appname> tag; e.g.

    package MyApp;
    use Dancer2;
    get '/foo' => sub {...};

    package MyApp::Private;
    use Dancer2 appname => MyApp;
    get '/bar' => sub {...};

The above would add the C<bar> route to the MyApp application. Dancer2 will
I<not> create an application with the name C<MyApp::Private>.

=item * B<:nopragmas>

By default L<Dancer2> will import three pragmas: L<strict>, L<warnings>,
and L<utf8>. If you require control over the imported pragmas, you can add
B<:nopragmas> to the importing flags, in which case Dancer2 will not import
any pragmas:

    use strict;
    use warnings;
    no warnings 'experimental::smartmatch'; # for example...
    use Dancer2 ':nopragmas'; # do not touch the existing pragmas

This way importing C<Dancer2> does not change the existing pragmas setup
you have.

=back

When you C<use Dancer2>, you get an C<import> method added into the current
package. This B<will> override previously declared import methods from other
sources, such as L<Exporter>. Dancer2 applications support the following
tags on import:

=over 4

=item * B<with>

The C<with> tag allows an app to pass one or more config entries to another
app, when it C<use>s it.

    package MyApp;
    use Dancer2;

    BEGIN { set session => 'YAML' };
    use Blog with => { session => engine('session') };

In this example, the session engine is passed to the C<Blog> app. That way,
anything done in the session will be shared between both apps.

Anything that is defined in the config entry can be passed that way. If we
want to pass the whole config object, it can be done like so:

    use SomeApp with => { %{config()} };

=back

=head1 DSL KEYWORDS

Dancer2 provides you with a DSL (Domain-Specific Language) which makes
implementing your web application trivial.

For example, take the following example:

    use Dancer2;

    get '/hello/:name' => sub {
        my $name = route_parameters->get('name');
    };
    dance;

C<get> and C<route_parameters> are keywords provided by Dancer2.

See L<Dancer2::Manual::Keywords/DSL Keywords manual> for a complete list of 
keywords provided by Dancer2.

=head1 AUTHOR

Dancer Core Developers

=head1 COPYRIGHT AND LICENSE

This software is copyright (c) 2023 by Alexis Sukrieh.

This is free software; you can redistribute it and/or modify it under
the same terms as the Perl 5 programming language system itself.

=cut