NAME
Catalyst::TraitFor::Request::StructuredParameters - Enforce structural rules on your body and data parameters
SYNOPSIS
For Catalyst v5.90090+
package MyApp;
use Catalyst;
MyApp->request_class_traits(['Catalyst::TraitFor::Request::StructuredParameters']);
MyApp->setup;
For Catalyst older than v5.90090
package MyApp;
use Catalyst;
use CatalystX::RoleApplicator;
MyApp->apply_request_class_roles('Catalyst::TraitFor::Request::StructuredParameters');
MyApp->setup;
In a controller:
package MyApp::Controller::User;
use Moose;
use MooseX::MethodAttributes;
extends 'Catalyst::Controller';
sub user :Local {
my ($self, $c) = @_;
# Basically this is like a whitelist for the allowed parameters. This is not a replacement
# for form validation but rather prior layer to make sure the incoming is semantically
# acceptable. It also does some sanity cleanup like flatten unexpected arrays. The following
# would accept body parameters like the following:
#
# $c->req->body_parameters == +{
# username => 'jnap',
# password => 'youllneverguess',
# password_confirmation => 'youllneverguess'
# 'name.first' => 'John',
# 'name.last' => 'Napiorkowski',
# 'email[0]' => 'jjn1056@example1.com',
# 'email[1]' => 'jjn1056@example2.com',
# }
my %body_parameters = $c->req->structured_body
->permitted('username', 'password', 'password_confirmation', name => ['first', 'last'], +{email=>[]} )
->to_hash;
# %body_parameters then looks like this, which is a form suitable for validating and creating
# or updating a database.
#
# %body_parameters == (
# username => 'jnap',
# password => 'youllneverguess',
# password_confirmation => 'youllneverguess'
# name => +{
# first => 'John',
# last => 'Napiorkowski',
# },
# email => ['jjn1056@example1.com', 'jjn1056@example2.com'],
# );
# Ok so now you know %body_parameters are 'well-formed', you can use them to do stuff like
# value validation and updating a databases, etc.
my $new_user = $c->model('Schema::User')->validate_and_create(\%body_parameters);
}
DESCRIPTION
This replaces Catalyst::TraitFor::Request::StrongParameters. If you were using that you should switch to this. This is right now just a name change but any bug fixes will happen here. Sooner or later I'll remove Catalyst::TraitFor::Request::StrongParameters from the indexes.
WARNING: This is a quick midnight hack and the code could have sharp edges. Happy to take broken test cases.
When your web application receives incoming POST body or data you should treat that data with suspicion. Even prior to validation you need to make sure the incoming structure is well formed (this is most important when you have deeply nested structures, which can be a major security risk both in parsing and in using that data to do things like update a database). Catalyst::TraitFor::Request::StructuredParameters offers a structured approach to whitelisting your incoming POSTed data, as well as a safe way to introduce nested data structures into your classic HTML Form posted data. It is also compatible for POSTed data (such as JSON POSTed data) although in the case of body data such as JSON we merely whitelist the fields and structure since JSON can already support nested data structures.
This is similar to a concept called 'strong parameters' in Rails although my implementation is somewhat different based on the varying needs of the Catalyst framework. However I consider this beta code and subject to change should real life use cases arise that indicate a different approach is warranted.
METHODS
This role defines the following methods:
structured_body
Returns an instance of Catalyst::Utils::StructuredParameters preconfigured with the current contents of ->body_parameters. Any arguments are passed to that instances "permitted" methods before return.
structured_query
Parses the URI query string; otherwise same as "structured_body".
structured_data
The same as "structured_body" except aimed at body data such as JSON post. Basically works the same except the default for handling array values is to leave them alone rather than to flatten.
PARAMETER OBJECT METHODS
The instance of Catalyst::Utils::StructuredParameters which is returned by any of the three builder methods above ("structured_body", "structured_query and ) supports the following methods."
namespace (\@fields)
Sets the current 'namespace' to start looking for fields and values. Useful when all the fields are under a key. For example if the value of ->body_parameters is:
+{
'person.name' => 'John',
'person.age' => 52,
}
If you set the namespace to ['person']
then you can create rule specifications that assume to be 'under' that key. See the "SYNOPSIS" for an example.
permitted (?\@namespace, @rules)
An array of rule specifications that are used to filter the current parameters as passed by the user and present a sanitized version that can safely be used in your code.
If the first argument is an arrayref, that value is used to set the starting "namespace".
required (?\@namespace, @rules)
An array of rule specifications that are used to filter the current parameters as passed by the user and present a sanitized version that can safely be used in your code.
If user submitted parameters do not match the spec an exception is throw (Catalyst::Exception::MissingParameter If you want to use required parameters then you should add code to catch this error and handle it (see below for more)
If the first argument is an arrayref, that value is used to set the starting "namespace".
flatten_array_value ($bool)
Switch to indicated if you want to flatten any arrayref values to 'pick last'. This is true by default for body and query parameters since its a common hack around quirks with certain types of HTML form controls (like checkboxes) which don't return a value when not selected or checked.
max_array_depth (number)
Prevent incoming parameters from having too many items in an array value. Default is 1,000. You may wish to set a different value based on your requirements. Throws Catalyst::Exception::InvalidArrayLength if violated.
to_hash
Returns the currently filtered parameters based on the current permitted and/or required specifications.
keys
Returns an unorderd list of all the top level keys
get (@key_names)
Given a list of key names, return values. Doesn't currently do anything if you use a key name that doesn't exist (you just get 'undef').
CHAINING
All the public methods for Catalyst::Utils::StructuredParameters return the current instance so that you can chain methods easily (except for "to_hash"). If you chain "permitted" and "required" the accepted hashrefs are merged.
RULE SPECIFICATIONS
Catalyst::TraitFor::Request::StructuredParameters offers a concise DSL for describing permitted and required parameters, including flat parameters, hashes, arrays and any combination thereof.
Given body_parameters of the following:
+{
'person.name' => 'John',
'person.age' => '52',
'person.address.street' => '15604 Harry Lind Road',
'person.address.zip' => '78621',
'person.email[0]' => 'jjn1056@gmail.com',
'person.email[1]' => 'jjn1056@yahoo.com',
'person.credit_cards[0].number' => '245345345345345',
'person.credit_cards[0].exp' => '2024-01-01',
'person.credit_cards[1].number' => '666677777888878',
'person.credit_cards[1].exp' => '2024-01-01',
'person.credit_cards[].number' => '444444433333',
'person.credit_cards[].exp' => '4024-01-01',
}
my %data = $c->req->strong_body
->namespace(['person'])
->permitted('name','age');
# %data = ( name => 'John', age => 53 );
my %data = $c->req->structured_body
->namespace(['person'])
->permitted('name','age', address => ['street', 'zip']); # arrayref means the following are subkeys
# %data = (
# name => 'John',
# age => 53,
# address => +{
# street => '15604 Harry Lind Road',
# zip '78621',
# }
# );
my %data = $c->req->structured_body
->namespace(['person'])
->permitted('name','age', +{email => []} ); # wrapping in a hashref mean 'under this is an arrayref
# %data = (
# name => 'John',
# age => 53,
# email => ['jjn1056@gmail.com', 'jjn1056@yahoo.com']
# );
# Combine hashref and arrayref to indicate array of subkeyu
my %data = $c->req->structured_body
->namespace(['person'])
->permitted('name','age', +{credit_cards => [qw/number exp/]} );
# %data = (
# name => 'John',
# age => 53,
# credit_cards => [
# {
# number => "245345345345345",
# exp => "2024-01-01",
# },
# {
# number => "666677777888878",
# exp => "2024-01-01",
# },
# {
# number => "444444433333",
# exp => "4024-01-01",
# },
# ]
# );
You can specify more than one specification for the same key. For example if body parameters are:
+{
'person.credit_cards[0].number' => '245345345345345',
'person.credit_cards[0].exp' => '2024-01-01',
'person.credit_cards[1].number' => '666677777888878',
'person.credit_cards[1].exp.year' => '2024',
'person.credit_cards[1].exp.month' => '01',
}
my %data = $c->req->structured_body
->namespace(['person'])
->permitted(+{credit_cards => ['number', 'exp', exp=>[qw/year month/] ]} );
# %data = (
# credit_cards => [
# {
# number => "245345345345345",
# exp => "2024-01-01",
# },
# {
# number => "666677777888878",
# exp => +{
# year => '2024',
# month => '01'
# },
# },
# ]
# );
ARRAY DATA AND ARRAY VALUE FLATTENING
Please note this only applies to "structured_body" / "structured_query"
In the situation when you have a array value for a given namespace specification such as the following :
'person.name' => 2,
'person.name' => 'John', # flatten array should jsut pick the last one
We automatically pick the last POSTed value. This can be a useful hack around some HTML form elements that don't set an 'off' value (like checkboxes).
'EMPTY' FINAL INDEXES
Please note this only applies to "structured_body" / "structured_query"
Since the index values used to sort arrays are not preserved (they indicate order but are not used to set the index since that could open your code to potential hackers) we permit final 'empty' indexes:
'person.credit_cards[0].number' => '245345345345345',
'person.credit_cards[0].exp' => '2024-01-01',
'person.credit_cards[1].number' => '666677777888878',
'person.credit_cards[1].exp' => '2024-01-01',
'person.credit_cards[].number' => '444444433333',
'person.credit_cards[].exp' => '4024-01-01',
This 'empty' index will always be considered the final element when sorting. You may have more than one final empty index as well when its either the last rule or the rule only contains a single index
'person.notes[]' => 'This is a note',
'person.notes[]' => 'This is another note',
'person.person_roles[1].role_id' => '1',
'person.person_roles[2].role_id' => '2',
'person.person_roles[].role_id' => '3',
'person.person_roles[].role_id' => '4',
Would produce:
+{
person => {
notes => [
'This is a note',
'This is another note',
],
person_roles => [
{
role_id => 1,
},
{
role_id => 2,
},
{
role_id => 3,
},
{
role_id => 4,
},
],
},
};
EXCEPTIONS
The following exceptions can be raised by these methods and you should add code to recognize and handle them. For example you can add a global or controller scoped 'end' action:
sub end :Action {
my ($self, $c) = @_;
if(my $error = $c->last_error) {
$c->clear_errors;
if(blessed($error) && $error->isa('Catalyst::Exception::StructuredParameter')) {
# Handle the error perhaps by forwarding to a view and setting a 4xx
# bad request response code.
}
}
}
Alternatively (and probably neater) just use CatalystX::Error.
sub end :Action Does(RenderError) { }
You'll need to add the included Catalyst::Plugin::Errors plugin to your application class in order to use this ActionRole.
Exception: Base Class
Catalyst::Exception::StructuredParameter
There's a number of different exceptions that this trait can throw but they all inherit from Catalyst::Exception::StructuredParameter so you can just check for that since those are all going to be considered 'Bad Request' type issues. This also inherits from CatalystX::Utils::HttpException so you can use the CatalystX::Errors package to neaten up / regularize your error control.
EXCEPTION: MISSING PARAMETER
Catalyst::Exception::MissingParameter ISA Catalyst::Exception::StructuredParameter
If you use "required" and a parameter is not present you will raise this exception, which will contain a message indicating the first found missing parameter. For example:
"Required parameter 'username' is missing."
This will not be an exhaustive list of the missing parameters and this feature in not intended to be used as a sort of form validation system.
AUTHOR
John Napiorkowski email:jjnapiork@cpan.org
SEE ALSO
COPYRIGHT & LICENSE
Copyright 20121, John Napiorkowski email:jjnapiork@cpan.org
This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.
1 POD Error
The following errors were encountered while parsing the POD:
- Around line 164:
Nested L<> are illegal. Pretending inner one is X<...> so can continue looking for other errors.
Unterminated L<...> sequence