--- name: "Jifty::Action" section: "3pm" platform: "linux" package: "libjifty-perl" version: "1.10518+dfsg-2" release: "Debian 7 (Wheezy)" platforms: ["linux"] url: "https://new.carta.tech/Jifty::Action" --- # Jifty::Action(3pm) The ability to Do Things in the framework ## NAME Jifty::Action - The ability to Do Things in the framework ## SYNOPSIS ``` package MyApp::Action::Foo; use Jifty::Param::Schema; use Jifty::Action schema { param bar => type is 'checkbox', label is 'Want Bar?', hints is 'Bar is this cool thing that you really want.', default is 0; }; sub take_action { ... } 1; ``` ## DESCRIPTION `"Jifty::Action"` is the superclass for all actions in Jifty. Action classes form the meat of the Jifty framework; they control how form elements interact with the underlying model. See also Jifty::Action::Record for data-oriented actions, Jifty::Result for how to return values from actions. See Jifty::Param::Schema for more details on the declarative syntax. See Jifty::Manual::Actions for examples of using actions. ## COMMON METHODS These common methods provide the basic guts for the action. ### new **Do not call this directly**; always go through `"Jifty->web->new_action"`! This method constructs a new action. Subclasses who need do custom initialization should start with: ``` my $class = shift; my $self = $class->SUPER::new(@_) ``` The arguments that this will be called with include: *Arguments* - moniker The moniker of the action. Defaults to an autogenerated moniker. - order An integer that determines the ordering of the action's execution. Lower numbers occur before higher numbers. Defaults to 0. - arguments A hash reference of default values for the arguments of the action. Defaults to none. - sticky\_on\_failure A boolean value that determines if the form fields are sticky when the action fails. Defaults to true. - sticky\_on\_success A boolean value that determines if the form fields are sticky when the action succeeds. Defaults to false. ### \_generate\_moniker Construct a moniker for a new (or soon-to-be-constructed) action that did not have an explicit moniker specified. The algorithm is simple: We snapshot the call stack, prefix it with the action class, and then append it with an per-request autoincrement counter in case the same class/stack is encountered twice, which can happen if the programmer placed a `"new_action"` call inside a loop. Monikers generated this way are guaranteed to work across requests. ### arguments **Note**: this API is now deprecated in favour of the declarative syntax offered by Jifty::Param::Schema. This method, along with "take\_action", is the most commonly overridden method. It should return a hash which describes the arguments this action takes: ``` { argument_name => {label => "properties go in this hash"}, another_argument => {mandatory => 1} } ``` Each argument listed in the hash will be turned into a Jifty::Web::Form::Field object. For each argument, the hash that describes it is used to set up the Jifty::Web::Form::Field object by calling the keys as methods with the values as arguments. That is, in the above example, Jifty will run code similar to the following: ``` # For 'argument_name' $f = Jifty::Web::Form::Field->new; $f->name( "argument_name" ); $f->label( "Properties go in this hash" ); ``` If an action has parameters that **must** be passed to it to execute, these should have the constructor property set. This is separate from the mandatory property, which deal with requiring that the user enter a value for that field. ### run This routine, unsurprisingly, actually runs the action. If the result of the action is currently a success (validation did not fail), `"run"` calls "take\_action", and finally "cleanup". If you're writing your own actions, you probably want to override `"take_action"` instead. ### validate Checks authorization with "check\_authorization", calls `"/setup"`, canonicalizes and validates each argument that was submitted, but doesn't actually call "take\_action". The outcome of all of this is stored on the "result" of the action. ### check\_authorization Returns true if whoever invoked this action is authorized to perform this action. By default, always returns true. ### setup `"setup"` is expected to return a true value, or "run" will skip all other actions. By default, does nothing. ### take\_action Do whatever the action is supposed to do. This and "arguments" are the most commonly overridden methods. By default, does nothing. The return value from this method is NOT returned. (Instead, you should be using the "result" object to store a result). ### cleanup Perform any action-specific cleanup. By default, does nothing. Runs after "take\_action" -- whether or not "take\_action" returns success. ### moniker Returns the moniker for this action. ### argument\_value ARGUMENT \[VALUE\] Returns the value from the argument with the given name, for this action. If *VALUE* is provided, sets the value. ### has\_argument ARGUMENT Returns true if the action has been provided with an value for the given argument, including a default\_value, and false if none was ever passed in. ### form\_field ARGUMENT Returns a Jifty::Web::Form::Field object for this argument. If there is no entry in the "arguments" hash that matches the given `"ARGUMENT"`, returns `"undef"`. ### form\_value ARGUMENT Returns a Jifty::Web::Form::Field object that renders a display value instead of an editable widget for this argument. If there is no entry in the "arguments" hash that matches the given `"ARGUMENT"`, returns `"undef"`. ### hidden ARGUMENT VALUE A shortcut for specifying a form field `"ARGUMENT"` which should render as a hidden form field, with the default value `"VALUE"`. ### order \[INTEGER\] Gets or sets the order that the action will be run in. This should be an integer, with lower numbers being run first. Defaults to zero. ### result \[RESULT\] Returns the Jifty::Result method associated with this action. If an action with the same moniker existed in the **last** request, then this contains the results of that action. ### register Registers this action as being present, by outputting a snippet of HTML. This expects that an HTML form has already been opened. Note that this is not a guarantee that the action will be run, even if the form is submitted. See Jifty::Request for the definition of "active" actions. Normally, "new\_action" in Jifty::Web takes care of calling this when it is needed. ### render\_errors Render any the "error" in Jifty::Result of this action, if any, as HTML. Returns nothing. ### button arguments =\> { KEY =\> VALUE }, PARAMHASH Create and render a button. It functions nearly identically like "link" in Jifty::Web, except it takes `"arguments"` in addition to `"parameters"`, and defaults to submitting this Jifty::Action. Returns nothing. Recommended reading: Jifty::Web::Form::Element, where most of the cool options to button and other things of its ilk are documented. *return PARAMHASH* Creates and renders a button, like "button", which additionally defaults to calling the current continuation. Takes an additional argument, `"to"`, which can specify a default path to return to if there is no current continuation. ## NAMING METHODS These methods return the names of HTML form elements related to this action. ### register\_name Returns the name of the "registration" query argument for this action in a web form. ### form\_field\_name ARGUMENT Turn one of this action's arguments into a fully qualified name; takes the name of the field as an argument. ### fallback\_form\_field\_name ARGUMENT Turn one of this action's arguments into a fully qualified "fallback" name; takes the name of the field as an argument. This is specifically to support checkboxes, which only show up in the query string if they are checked. Jifty creates a checkbox with the value of form\_field\_name as its name and a value of 1, and a hidden input with the value of fallback\_form\_field\_name as its name and a value of 0; using this information, Jifty::Request can both determine if the checkbox was present at all in the form, as well as its true value. ### error\_div\_id ARGUMENT Turn one of this action's arguments into the id for the div in which its errors live; takes name of the field as an argument. ### warning\_div\_id ARGUMENT Turn one of this action's arguments into the id for the div in which its warnings live; takes name of the field as an argument. ### canonicalization\_note\_div\_id ARGUMENT Turn one of this action's arguments into the id for the div in which its canonicalization notes live; takes name of the field as an argument. ## VALIDATION METHODS ### argument\_names Returns the list of argument names. This information is extracted from "arguments". ### \_canonicalize\_arguments Canonicalizes each of the arguments that this action knows about. This is done by calling "\_canonicalize\_argument" for each field described by "arguments". ### \_canonicalize\_argument ARGUMENT Canonicalizes the value of an argument. If the argument has an attribute named **canonicalizer**, call the subroutine reference that attribute points points to. If it doesn't have a **canonicalizer** attribute, but the action has a `"canonicalize_`*ARGUMENT*`"` function, also invoke that function. If neither of those are true, by default canonicalize dates using \_canonicalize\_date Note that it is possible that a canonicalizer will be called multiple times on the same field -- canonicalizers should be careful to do nothing to already-canonicalized data. ### \_canonicalize\_date DATE Parses and returns the date using Jifty::DateTime::new\_from\_string. ### \_validate\_arguments Validates the form fields. This is done by calling "\_validate\_argument" for each field described by "arguments" ### \_validate\_argument ARGUMENT Validate your form fields. If the field `"ARGUMENT"` is mandatory, checks for a value. If the field has an attribute named **validator**, call the subroutine reference validator points to. If the action doesn't have an explicit **validator** attribute, but does have a `"validate_`*ARGUMENT*`"` function, invoke that function. ### \_extra\_validator\_args Returns a list of extra arguments to pass to validators. By default, an empty hash reference, but subclasses can override it to pass, say, a better `"for"`. ### \_extra\_canonicalizer\_args Returns a list of extra arguments to pass to canonicalizers. By default, an empty hash reference, but subclasses can override it to pass, say, a better `"for"`. ### \_extra\_autocompleter\_args Returns a list of extra arguments to pass to autocompleters. By default, an empty hash reference, but subclasses can override it to pass, say, a better `"for"`. ### \_autocomplete\_argument ARGUMENT Get back a list of possible completions for `"ARGUMENT"`. The list should either be a list of scalar values or a list of hash references. Each hash reference must have a key named `"value"`. There can also additionally be a key named `"label"` which, if present, will be used as the user visible label. If `"label"` is not present then the contents of `"value"` will be used for the label. If the field has an attribute named **autocompleter**, call the subroutine reference **autocompleter** points to. If the field doesn't have an explicit **autocompleter** attribute, but does have a `"autocomplete_`*ARGUMENT*`"` function, invoke that function. ### valid\_values ARGUMENT Given an parameter name, returns the list of valid values for it, based on its `"valid_values"` field. This method returns a hash reference with a `"display"` field for the string to display for the value, and a `"value"` field for the value to actually send to the server. (Avoid using this -- this is not the appropriate place for this logic to be!) ### available\_values ARGUMENT Just like valid\_values, but if our action has a set of available recommended values, returns that instead. (We use this to differentiate between a list of acceptable values and a list of suggested values) ### validation\_error ARGUMENT =\> ERROR TEXT, \[OPTIONS\] Used to report an error during validation. Inside a validator you should write: ``` return $self->validation_error( $field => "error"); ``` ..where `$field` is the name of the argument which is at fault. Any extra `"OPTIONS"` are passed through to "field\_error" in Jifty::Result. ### validation\_warning ARGUMENT =\> WARNING TEXT, \[OPTIONS\] Used to report a warning during validation. Inside a validator you should write: ``` return $self->validation_warning( $field => _("warning")); ``` ..where `$field` is the name of the argument which is at fault. Any extra `"OPTIONS"` are passed through to "field\_warning" in Jifty::Result. ### validation\_ok ARGUMENT, \[OPTIONS\] Used to report that a field **does** validate. Inside a validator you should write: ``` return $self->validation_ok($field); ``` Any extra `"OPTIONS"` are passed through to "field\_warning" in Jifty::Result and "field\_error" in Jifty::Result when unsetting them. ### canonicalization\_note ARGUMENT =\> NOTE Used to send an informational message to the user from the canonicalizer. Inside a canonicalizer you can write: ``` $self->canonicalization_note( $field => _("I changed $field for you")); ``` ..where `$field` is the name of the argument which the canonicalizer is processing ### deny REASON When access to an action is denied by Jifty::API::is\_allowed the request handler calls this with a message. This should mark the action as failed and store the message but may also want to do other things (such as providing a nicer message or logging somewhere other than the jifty logs) ## CUSTOMIZATION ### Canonicalization If you wish to have the data in a field normalized into a particular format (such as changing a date into `"YYYY-MM-DD"` format, adding commas to numbers, capitalizing words, or whatever you need) you can do so using a canonicalizer. This is just a method titled `"canonicalize_FIELD"` where `"FIELD"` is the name of the field be normalized. Here is an example: ``` sub canonicalize_foo { my ($self, $value, $other, $metadata) = @_; # do something to canonicalize the value my $normal_form = lc($value) . '-' . $other->{other_field}; $normal_form .= '-update' if $metadata->{for} eq 'update'; return $normal_form; } ``` The first parameter to your canonicalizer will be the value to be canonicalized. The next will be a hash reference of all the parameters submitted with this canonicalization, so you can be smarter. Finally, the third parameter is a hash reference of other metadata, such as `"for"`, whose value will be `"create"` or `"update"`. While doing this you might also want to call the "canonicalization\_note" to inform the client of the modification: ``` my $normal_form = lc($value); $self->canonicalization_note( foo => _('Foo values are always in lowercase.')); ``` If the "foo" field has "ajax canoncalizes" set in the action schema, then this process will be performed automatically as the form is being filled without reloading the page. ### Validation If a value must follow a certain format, you can provide a validation method for fields to make sure that no value enters the database until it is in a valid form. A validation method is one named `"validate_FIELD"` where `"FIELD"` is the name of the field being checked. Here is an example: ``` sub validate_foo { my ($self, $value, $other, $metadata) = @_; # Check for uppercase letters if ($value =~ /\p{Lu}/) { return $self->validation_warning( foo => _("Foo cannot contain uppercase letters.")); } # Check for -, *, +, and ? elsif ($value =~ /[\-\*\+\?]/) { return $self->validation_error( foo => _("Foo cannot contain -, *, +, or ?.")); } return 1; } ``` Here the "foo" field should not contain uppercase letters and must not contain the characters '-', '\*', '+', or '?'. You can use "validation\_error" and "validation\_warning" to return the results of your validation to the user or simply return 1 to indicate a valid value. Note that the parameters are the same as in "Canonicalization". If you just have a list of valid values, you may want to use the `"valid_values"` schema parameter to perform this task instead. ### Autocompletion Autocompletion provides a way of suggesting choices to the client based upon partial data entry. This doesn't necessarily force the client to use one of the choices given but gives hints in an application specific way. To create an autocompletion field, you implement a method named `"autocomplete_FIELD"` where `"FIELD"` is the field to autocomplete. This is generally done with fields rendered as 'Text'. Here is an example: ``` sub autocomplete_foo { my ($self, $value) = @_; # Be careful to validate your input! You don't want a malicious user # hacking your system. my ($match_value) = $value =~ /^(\w+)$/; my $foos = MyApp::Model::FooCollection->new; $foos->limit( column => 'name', operator => 'LIKE', value => '%$value%', ); return map { $_->name } @{ $foos->items_array_ref }; } ``` In this example, the "foo" field is autocompleted from names matched from the `"MyApp::Model::Foo"` table. The match, in this case, matches any substring found in the database. I could have matched any item that starts with the string, ends with the string, matches other fields than the one returned, etc. It's up to you to decide. Note also that I have untainted the value coming in to make sure a malicious user doesn't get anyway. You should always perform a check like this when data is coming in from an outside source. If you need a more complicated solution, you can return the autocompletion values as a list of hash references containing the keys `"value"` and (optionally) `"label"`: ``` return map { { value => $_->name, label => $_->label } } @{ $foos->items_array_ref }; ``` In this case, the labels will be shown to the client, but the selected value would be returned to your application. ## SEE ALSO Jifty, Jifty::API, Jifty::Action::Record, Jifty::Result, Jifty::Param::Schema, Jifty::Manual::Actions ## LICENSE Jifty is Copyright 2005-2010 Best Practical Solutions, LLC. Jifty is distributed under the same terms as Perl itself.