- Skeleton
- How to Start Apps
- How to Write Web Apps
- Subroutines for your Application
- Subroutines that should only be used inside the Code block of a Route
content_type(Str $type)
request()
uri-for(Str $path)
header(Str $name, Cool $value)
cookie(Str $name, Str $value, Str :$domain, Str :$path, DateTime :$expires, Bool :$http-only; Bool :$secure)
status(Int $code)
template(Str $template-name, *@params)
session()
to-json()
from-json()
render($content)
render(Int :$status, Str :$type is copy, :$content is copy)
render-file(Str $filename, Str :$mime-type)
- Advanced Concepts
- Templates
- Sessions
- Configuration
Run
$ bailador --name App-Name new
to create a skeleton project.
When you have installed Crust from the Perl 6 ecosystem there is a command called crustup
or crustup.bat
which can be used to launch your Bailador App. Bailador was developed and runs best on top of HTTP::Easy::PSGI. Always use baile()
at the end of your app, because in the default configuration it guesses whether your app is called via crustup
or directly with perl6
. Depending on that baile()
chooses the right Bailador::Command
to invoke your application with.
Store this Example in a file called example.p6w
use v6;
use Bailador;
# simple cases
get '/' => sub {
"hello world"
}
baile();
and then type this in your shell:
$ crustup --server HTTP:::Easy example.p6w
bailador
can be used to start your bailador web application. The command line tool converts the command line switches (e.g. --config=port:8080
) to environment variables (%*ENV
). There are different commands for bailador.
Takes a comma-separated list of parameters that configure various aspects of how Bailador will run. --config
overrides the BAILADOR environment variable.
For details of the available configuration parameters check the Configuration section of the documentation.
Currently available parameters:
$ bailador --config=host:0.0.0.0,port:3001 watch bin/your-bailador-app.p6
The command easy starts the the web application with the HTTP::Easy::PSGI backend.
The command watch watches the source code of your Bailador app for changes and automatically restart the app.
Takes comma-separated list of directories to watch. By default,
will watch the lib
, bin
, and views
directories.
If you have to watch a directory with a comma in its name, prefix it with a backslash:
$ bailador --w=x\\,y bin/app.p6 # watches directory "x,y"
$ bailador --w=lib,bin,views,public watch bin/your-bailador-app.p6
In order to invoke the Bailador App directly, you could simply call baile() in your script.
use v6;
use Bailador;
# simple cases
get '/' => sub {
"hello world"
}
baile;
This will launch the Bailador server on default port 3000.
Bailador offers two different approaches to write web applications. The first and classical approach is using the subs that are exported that you get when you use Bailador
. This API is meant to be stable and should not change much.
New features like nested routes and whatever is yet to come are implemented in Bailador::App
and can be used through the object oriented interface. Your own web application just inherits from Bailador::App
.
Sets a Bailador::App to be the default app for all the other exported subs described in Subroutines that should only be used inside the Code block of a Route.
Adds a route for get, post, put or delete requests. The key of the
Pair
is either a Str
or a Regex
. If a string is passed it is
automatically converted into a regex. The value of the pair must be a
Callable
. Whenever the route matches the requested URL the
callable is invoked with the list of the Match
as its
parameters. The return value of the callable will be auto rendered. So
it is the content of your response. The request is available via the
appropriately named variable request
; request.params
will contain
the route parameters, for instance; request.params<q>
will yield the
value for param q
.
The strings captured by the regular expression are available as subroutine parameters.
get "/foo/(.+)" => sub ( $route ) {
return "What a $route";
}
The prefix sets up a Nested Route. All other routes that will be added within the $code will be added to this nested route. With prefix-enter you can define code that will be called whenever the prefix matches your HTTP request. Only if this code returns True the routes within the prefix can be reached during request dispatching. Without using prefix-enter the routes in the prefix are reachable - this means the default code for a prefix route is sub { True }
.
prefix "/foo" => sub {
prefix-enter sub {
... something that returns True or False ...
}
get "/bar" => sub { ... }
get "/baz" => sub { ... }
}
Check out the prefix example and the corresponding tests.
A static file route can be used to serve files in a directory. This sets up a get and head route which checks for an existing file in the given directory. The path
( Regex
or Str
) must return a match with a single capture grupe, which will be turned into a .Str
. If there is a file in the directory with that name it will be rendered otherwise the route returns a False
, so in the end the route is left and maybe other routes can handle your request.
static-dir / (.+) / => 'data/';
Redirect to the specified location, can be relative or absolute URL. Adds Location-header to response and sets status code to 302.
Sets the Renderer that's being used to render your templates. See the Templates section for more details.
Returns the configuration. You can influence how sessions work, the mode, port and host of your Bailador app. See the Sessions and Configuration sections for details.
Returns the object from the collection class Bailador::Plugins it contains all configured and loaded plugins. See Configuration sections for how to configure a plugin. Following methods are available:
# public method to add a plugin on runtime.
app.plugins.add('Example', Bailador::Plugin::Example.new(config => {param => 'Test'}) );
# public method to get a plugin by name
app.plugins.get('Example');
# public method to detect configured plugins and add it to collection via the containing public method add
app.plugins.detect(app.config);
This is a simple way to set values in the config.
# this:
set("foo", True);
# is doing exactly the same as this:
config.foo = True;
or baile($command)
Let's enter the dance floor. ¡Olé!
Sets the Content Type for the response to $type.
Gets the object representing the current Request. For a list of methods check out the request example.
Constructs a URI String from the base and the passed $path.
Adds a Header to the Response.
cookie(Str $name, Str $value, Str :$domain, Str :$path, DateTime :$expires, Bool :$http-only; Bool :$secure)
Adds a Cookie to the response.
Sets the status code of a response.
Calls the template which is a file in the views folder. You can specify a $:layout if you want to override the settings in Bailador::Configuration. For more details see the Template section. Should only be used within the code block of a route.
Returns the Session Hash. Session Hashes are empty if you start a new session. For details see the Sessions section.
Check out the session example.
Converts your data into JSON format using JSON::Fast.
Check out the api example and the corresponding test case.
Converts your data from JSON format using JSON::Fast.
Renderes a result the http status code given in $status
, the header Content-Type: $type
and the body given in $content
. render($content)
is the same as render(content => $content)
.
Serve a file as-is to the client.
Routes can be nested and structured in a tree. If you just use the methods get, post, etc from the Bailador::App all the routes that you add are placed on the first level of the tree, so nothing is nested so far.
Nesting routes make sense if you want to enter routes just under conditions. The most perfect example when nested routes become handy is when you want to serve content just when someone is logged in. Instead of having the same check spread over and over in each and every sub you just create a Bailador::Route
and add it with self.add_route
. So the return value of the route now determines what to do.
Auto rendering means that whatever (except True
and False
) the return value of the sub is, it will be rendered. Using render($content)
or render(Int :$status, Str :$type is copy, :$content is copy)
will turn of auto rendering, because you obviously have rendered something manually.
-
False
The callable of a route works as a conditional check and the nested routes will not be checked nor invoked. It behaves as if the route would not have matched at all. So it will continue to look for a route that matches your request.
-
True
The callable of a route works as a conditional check and allows to go deeper into the tree. In case nothing in the tree matches the request an exception is thrown. That causes to leave the nested routes and continue checking for other routes. Of course, if this happens in the first level of the tree a
404
is created. -
Failure
s andException
sThis will cause a HTTP
500
-
anything that is not
defined
It is fine if a route returns nothing (e.g.
Nil
orAny
) or anything that is not defined as long as you have rendered something within the callable of the route. -
anything that is
defined
If anything defined is returned this will be the content of the response as long as you don't have rendered something in the callable of the route. If an
IO::Path
gets returned it automatically guesses its content type based on the filename extensions and renders the content of the file.
Instead of using simple subs for routes (e.g in combination with get
, post
, delete
etc) you could also use Controller Classes. Controller Classes are most likely the way to go for applications that grow bigger. You need to specify a class and a method name. The class gets instanciated for each HTTP request that matches the route definition. It is also possible to use an object instead of a class name, so you avoid the generation of new instances. In oder words your Controller is no longer stateless. You should still considder the controllers as glue code which glues your bailador-agnostic model, a model that is not even aware of bailador and does not use Bailadors DSL) to the HTTP requests. This allows to your model classes to be easily tested in tests. A small example can be found here.
get '/v1/data/:id' => { class => 'My::Controller::Class', method => 'get-data' };
Currently there are 2 different engines supported out of the box: Template::Mojo and Template::Mustache. Where Template::Mojo is the default engine but if you want to switch to Template::Mustache you just call
renderer(Bailador::Template::Mustache.new);
It is possible to user other template engines as well. Therefore you need to create a class that implements the role Bailador::Template. Its basically just required to implement the render method.
The template files should be placed by default in a folder named "views" which is located in the same directory as your application.pl file. If you want to override this, you just have to change the views
settings, and choose you own directory :
views: "templates"
When you call the subroutine
template 'template.tt', $name, $something;
the template (or in other words the file views/template.tt) gets invoked "as a subroutine" and the |@params get passed. This is a example of a template file with Template::Mojo:
% my ($name, $something) = @_;
<html ... codes goes here ...>
<h1><%= $name %></h1>
<html>
In order to use layouts you can pass a layout option to the template()
call.
template 'template.tt', layout => 'main', $name, $something;
First Bailador renders your template with its parameters, and then scans the 'layout' sub directory for another layout template. The same rendering engine will be used for the layout template. The result of your first template rendering will be passed as only option to layout template. If the specified layout was not found the result of the first template rendering will be returned.
In order to customize the error pages drop a template file with the filename of the HTTP status code and the suffix .xx
in your views directory. Currently there only two different error codes: 404
and 500
.
Sessions are implemented using a Session Cookie. If the browser rejects cookies, Bailador is not able to handle sessions.
In order to create a session just call the subroutine
session()
inside the code block of a route. This subroutine returns a Hash in which you can just toss in all data or objects that should be be in the session context. After your route code is finished the session will be stored automatically. How this should be done can be configured. The handling of sessions can be influenced with settings of Bailador::Configuration.
config()
inside the bailador script before you call baile
. As soon as you have requested the first session it is of no use to change the configure any further.
Following config options are available. Most of them should be self explanatory.
- cookie-name = 'bailador';
- cookie-path = '/';
- cookie-expiration = 3600;
- hmac-key = 'changeme';
- backend = "Bailador::Sessions::Store::Memory";
The Session-ID contains a HMAC to check if someone's trying to guess a Session-ID in order to hijack a session. This case it will create a warning which is printed to standard error.
The Session Data itself is stored by default in the memory, if you want to store the data on the disk or database of wherever, just implement a class which does the role Bailador::Sessions::Store and set backend to this class name.
Bailador uses a default configuration, but you can customize it, using the Bailador environment variable, or using configuration files. You can also change the configuration within your app. The Bailador::Configuration can also store custom-specific information, therefore please use the set
/ get
method. The order of adjusting the settings is: first the code in your app, second from the configuration files, third from the environment variables. So the settings from the environment variable BAILADOR
will finally override what you might have stated in the config files.
# directly
config.port = 8080;
config.hmack-key = 'xxxxxx';
# or via set()
config.set('port', 8080);
# set works for custom values as well
config.set('custom-key', 'value')
config.set('database-username', 'ufobat')
config.set('database-password', 'xxxxxx')
For now, Bailador only allows you to use YAML formatted configuration files. The default configuration file name is settings.yaml
, but you could change this if you like. Just write config.config-file = 'myconfig.yaml'
in the beginning of your app. Create at the root of your project directory a settings.yaml
file :
# settings.yaml
mode: "development"
port: 8080
plugins:
Example:
foo: bar
Bailador will now generate 2 more config file variants and process the settings from there. In our example settings-local.yaml
and, depending on our config.mode
which is development, a file named settings-development.yaml
. If our mode was production Bailador would have used settings-production.yaml
.
This allows you to have a general settings.yaml that you which to use on everywhere. Adaptions that only apply to a certain server could be placed into the -local
configuration file. And settings that only apply during development mode can be stored in the -development
file. As soon as you switch to production mode those settings will no longer be used.
Using the BAILADOR
environment variable is a comma separated list of key-value pairs.
$ BAILADOR=mode:development,host:0.0.0.0,port:5000 perl6 examples/app.pl6
Currently available parameters:
- config-file (defaults to 'settings.yaml')
- mode (defaults to 'production')
- port (defaults to 3000)
- host (defaults to 127.0.0.1)
- views
- layout
- command-detection (defaults to True)
- default-command
- watch-command (defaults to 'easy')
- watch-list
- cookie-name (defaults to 'bailador')
- cookie-path (defaults to = '/)
- cookie-expiration (defaults to 3600)
- hmac-key (defaults to 'changeme')
- backend (defaults to "Bailador::Sessions::Store::Memory")
- log-format (defaults to '\d (\s) \m'
- log-filter (defaults to
( 'severity' => '>=warning')
)
config.logs = [
# Prints HTTP requests on the standard terminal output using the default format.
'terminal:stdout' => { template-filter => 'http-requests' },
# Also prints debug logs to standard error terminal output
'terminal:stderr' => { severity => 'debug' },
# And insert logs into "access.log" file
'file:access.log' => { template-filter => 'http-requests' },
];
Logging can be configured by setting the logs
config key. The value is a Pair where the key represents an output (a file, the terminal, ...) and the value contains the specification about how to format the log and to filter them.
If multiples Pairs
are specified, they are all used.
The output is represented as an URI. Available values are:
- file:path/to/relative.log
- file:///path/to/absolute.log
- terminal:stdout
- terminal:stderr
- p6w:errors
Following parameters can be used to filter which log will be handled, and in which format.
Key | Desc | Examples |
---|---|---|
format | specify the format to use for the log | '\d \c \m' (date, category and message) |
template-format | predefined format strings | 'combined' (Apache combined format) |
template-filter | predefined filters | 'http-requests' HTTP request access |
category | filter on the log category | 'templates' logs related to templating |
severity | filter on the log severity | '>=debug' Debug logs |
The format string follows the definition of Log::Any (https://github.com/jsimonet/log-any#formatters). A work is in progress to describe all extra-fields that can be used, and in which contexts.
This parameter can be used to specify predefined formats like Apache "combined", or Apache "simple" error log.
Possible values :
- common
- combined
- simple
This parameter can be used to specify predefined filters.
Possible values:
- request
- templates
- request-error
The category to filter (based on Log::Any).
The severity of the log (based on Log::Any).
The default configuration prints all logs where the severity equals or is above "warning".
config.logs = [
'p6w:errors' => { 'severity' => '>=warning' },
]