i have a complex form. This form is created in the controller A.
All actions, like: edit, add, load, index
are in the controller A defined and the method actionIndex is essential for all other requests.
Like this schema:
public class controllerA {
function actionIndex(Requerst r){
r.handleRequest();
if ($form->get('index')->isClicked()){
// Index Action
}
if ($form->get('add')->isClicked()){
$this->actionAdd();
}
// ...
}
function actionAdd(){}
}
}
How do i seperate the requests in different controllers?
Actually it doesnt feel good.
Here my actual code of the controller:
http://pastebin.com/HuXhV37q
I have skimmed through your code and I speak german so I understand it.
It would be too much to cover here so I just give you some "keywords".
Outsource business logic from controllers to symfony services to keep the controllers light. ($zusaetzeArray etc.)
Create Entities and FormTypes for the form data if it makes sense economically.
Create multiple controller actions with corresponding routes for the different form states.
Symfony2 Service: https://stackoverflow.com/a/13099900/982075
Related
Simple problem but can't find a solution: I have a Thymeleaf form used to add a new object, say of a Book class. It works perfectly well and I only need that particular form for adding new objects, not editing the existing ones. The question is: how can I put several objects of the Book class in the same single form? So, purely for convenience, instead of filling form for a single book and clicking Send you can fill form for several books at once and only then click Send, have them all inserted into the database (in whatever order) and also have the option to fill the form partially (e.g. the form has room for 5 books but it will also accept 1, 2, 3 or 4 and you can leave the rest blank).
Edit: I've tried passing a list of object to the Thymeleaf template with the form bound to the whole list and iteration inside, but Thymeleaf throws BingingResultError upon rendering it.
You need to use a wrapper object to realize what you want.
Something like:
public class BooksCreationDto {
private List<Book> books;
// default and parameterized constructor
public void addBook(Book book) {
this.books.add(book);
}
// getter and setter
}
Then you need to pass this object as a model attribute in your controller:
BooksCreationDto booksForm = new BooksCreationDto();
model.addAttribute("form", booksForm);
bind fields using index property
th:field="*{books[__${itemStat.index}__].title}"
and get back the result with
#ModelAttribute BooksCreationDto form
in your controller.
For a complete and detailled explaination visit: https://www.baeldung.com/thymeleaf-list
I am currently working on a project developed using Zend Framework, based on the structure of my web page design I have reached a point where I have to pass a small number of variables to my layout from each Controller/Action. These variables are:
<?php Zend_Layout::getMvcInstance()->assign('pageId', 'page1'); ?>
<?php Zend_Layout::getMvcInstance()->assign('headerType', '<header id="index">'); ?>
The reason for passing this information is firstly, I pass the page id as the multi column layout may change depending on the content being displayed, thus the page id within the body tag links the appropriate CSS to how the page should be displayed. Secondly I display a promotional jQuery slider only on the index page, but I need the flexibility to have it displayed on potentially multiple pages in case the wind changes and the client changes their mind.
My actual question: Is there a more appropriate method of passing this information to the Layout that I am overlooking?
I am not really questioning whether the information has to be sent, rather is there some Zend Framework feature that I have, in my haste, overlooked which would reduce the amount of repetitive redundant code which may very well be repeated in multiple Actions within the same controller?
You could turn that logic into an action helper than you can call from your controllers in a more direct way. You could also make a view helper to accomplish the same thing but view helpers usually generate data for the view rather than set properties.
// library/PageId.php
class Lib_PageId extends Zend_Controller_Action_Helper_Abstract
{
public function direct($title, $pageId, $headerType)
{
$view = $this->getActionController()->view;
$view->headTitle()->append($title);
$view->pageId = $pageId;
$view->headerType = $headerType;
}
}
In your controller actions you can now do this:
$this->_helper->PageId('Homepage', 'page1', 'index');
// now pageId and headerType are available in the view and
// Homepage has been appended to the title
You will also need to register the helper path in your Bootstrap like this:
protected function _initActionHelpers()
{
Zend_Controller_Action_HelperBroker::addPrefix('Lib');
}
Doing it like that can reduce the amount of repetitive code and remove needing to assign the values from the view. You can do it in the controller very quickly. You can also have default values in the case that the helper hasn't been called.
You shoudn't really be passing anything from the view to the layout, for a start the view should be included IN the layout, not the other way around.
So, setting your page title should be done using similar code to what you have, but inside the controller action being called:
$this->view->headTitle()->append('Homepage');
And the other two issues - you need to rethink as I stated to begin with. Maybe you're misunderstanding the layout/view principle? If you include the different views per action, then you simply change the div id when needed, and include the header for your banner only in the index.phtml file.
I have come from WPF (MVVM) background and trying to shift to MVC 2. Is there any pattern in MVC2 where you can use Commanding/Command buttons like <input> which you use to submit the form so that you can hide/disable when you try to Render the View.
In MVVM world, your commands could implement ICommand interface, and it had CanExecute method which was quite useful. I was wondering if there is anything similar in ASP MVC 2 ?
The only way I can think of, is to do it in the View, so that I can check the flag on ViewModel (CanSave) and depending on that show/hide the <input> tag.
Basically I want to have 2 version of the website running, one in Read-Only mode and the other Editing mode.
Let me know if you need any clarification.
ASP.NET MVC does not feature the notion of 'controls', as are found in classic ASP.NET and WPF. The foundational blocks of ASP.NET MVC are HTML elements, like <input>, <button> et cetera. Naturally, these don't offer the functionality you're looking for (i.e. Implementation of the ICommand Interface).
The scenario that you're looking at (i.e. two modes of your form) can be (and arguably should be) dealt with at the View level. You're already facing the right direction: have a 'CanSave' property on your Model, and use this in the View to determine what is generated.
Example:
<% if (Model.CanSave)
{ %>
<p>First Name: <%= Html.TextBox("firstname", Model.firstname) %> </p>
<% }
else
{ %>
<p>First Name: <%=Model.firstname %></p>
<% } %>
You'll probably want to check out the DisplayTemplates and EditorTemplates... very handy for this scenario. Brad Wilson does a good job here.
It will help you move to this:
<%= (Model.CanSave) ? Html.EditorFor(x => x.firstname) : Html.DisplayFor(x => x.firstname) %>
...which makes your View clean and nice.
If you can't get MVC to do this it's relatively worth it to hand-code something like this vb-style pseudocode. This involves...
Subclassing your controls.
Not as much of a pain as it sounds, but, it is a medium sized one. Therefore it is only appropriate for medium-sized to large apps. But worth it for them.
Interface BaseUIControl
Property Enabled as Boolean
Property Visible as Boolean
Property Name as String
Property EntireStateAsXML as string ' You can use this to do EVERYTHING!
Interface UserActionItem
Event Clicked(sender as UserActionItem ... don't pass anything from UI namespaces!)
Class MyButton (or link, etc.) Implement BaseUIControl, UserActionItem Inherits UI.Button
How does this help? You've basically replaced the missing functionality. Your Controller (or even application layer) can be aware of the UI components by interface only, so they won't have to see the UI types.
more...
You can leverage this philosophy to control everything. This has saved me thousands of hours of monkey code.
Interface TextControl
Property Value as text
Interface CheckControl
Property Checked as boolean
The above two are Pretty basic - you inherit MyCheckBox and MyTextBox from the UI versions and implement the appropriate.
Of course you could set up common code to loop thru all controls and auto-validate (or loop thru and get each one's XML to autobind the whole form).
Interface ValidationBase
Property Required as Boolean
If you have a text or numeric-only mask or restricitons built into 2 subclasses...
Interface ValidationNumeric
Property MinVal, MaxVal as double
Interface ValidationText
Property MinLen, MaxLen as double
No, it won't go to the database for you. But this sweeps a ton of crud under the rug.
You can even set these property values in the UI designer - yes, putting BL in bed with UI, BUT, if you only have one UI for the BL, actually works very well.
Now image a UI with a mix of things like listbox/multiselect, double-list picker controls, checked listbox, a groupbox of option buttons/checkboxes ...
Interface Selector
property Items as list (of string)
property SelectedItems as list (of string)
Use what works on the UI - your generic routines can care less what they look like!! The subclassed UI pieces will just implement them to set/get the right values.
In addition ... we added 'validationEquation', ActivatesEquation (gray/ungray), SetValueTriggerEquation (if true, set value to SetValueEquation, otherwise, leave alone), which allowed controls to be set to simple values from other items (basically getting the values from bound objects as if using reflection) via Pascal Gayane's Expression Evaluator (it reads .net types!)
You can also subclass the main form, have it recurse thru all it's subcontrols, put together the XML's for the whole screen, and serialize it like that. You can have your own classes implement these in the non-UI layers and use it to totally (de/)serialize the UI state, and use them to read the UI too, if they relate to a business object, to map to it.
It's unbelievable how much this simplifies a complex app. We have one with 1200+ data entry panels (... pages... ours is a thickclient app) that will fill out 250 different paper forms at 250K LOC. The form definitions contain the 'name' of each control and this is pulled from the XML generated from the screens. We probably saved 500K LOC as many of the screens have no code behind them or only trivial code; all the databinding, validation, etc. is handled by common routines that reference the interfaces.
Like I say, this only works for a big app. Spend at least 2-3 weeks developing 90% of the functionality, though; probably another month throughout the 2 years dev maturing it. I am guessing your apps is big if you're caring about ICommand and its conveniences. I would put the payback at 15-20 moderately complex pages.
If I'm understanding the question correctly, you could write a ControllerCommand class to encapsulate this. Something like this:
public class ControllerCommand
{
public string Action { get; set; }
public string Controller { get; set; }
public object RouteValues { get; set; }
public bool IsEnabled { get; set; }
}
Your Details viewmodel might use it like this:
public class DetailsModel
{
public guid Id { get; set;}
// some other viewmodel properties
public ControllerCommand Edit { get; set; }
}
You could write extension methods on HtmlHelper to replace the built-in ones:
public MvcHtmlString CommandLink(this HtmlHelper html, string linkText, ControllerCommand command, object htmlAttributes)
{
if (command.IsEnabled)
{
return html.ActionLink(linkText, command.Action, command.Controller, command.RouteValues, htmlAttributes);
}
else
{
return MvcHtmlString.Create(linkText);
// perhaps return <span class="disabled-command">linkText</span>
}
}
One of the ways I have found is to use Filter attributes which you can put in your Actions, but that only handles CanExecute on the server side.
For the GUI side, couldnt find better way than putting If statements to check if the user is Priviliged to run particular action (i.e. Edit/Delete buttons)
I have a form with 2 elements that will be submitted and then update part of a user profile.
I don't want to use the entire generated form and have to remove all the fields except for the two I need. I just want to be able to create a quite simple form to do my update.
Is there a way to utilize Symfony's sfValidatorEmail inside the action on the returned value of an email field?
Since the regex is already written in the validator, I would like to reuse it, but I don't know how to use it in the action after the non-symfony form has been submitted.
Two approaches here - you could construct a simple form anyway extending from sfForm/sfFormSymfony (doesn't have to be ORM-based) that just contains the 2 fields you want. That way you can use the existing validation framework, and then use $myForm->getValues() after everything has been validated to get your values for your profile update.
Alternatively, as you've mentioned, you can use the sfValidatorEmail class in your action like so:
$dirtyValue = "broken.email.address"
$v = new sfValidatorEmail();
try
{
$v->clean($dirtyValue);
}
catch (sfValidatorError $e)
{
// Validation failed
}
The latter approach quickly leads to messy code if you have many values that need cleaning, and it's worth putting the logic back into a form to handle this in the usual manner.
If you're submitting a form with 2 elements, it should be a form on the edit and update end, period. Symfony forms are lightweight, there's no performance reason to not use them. Instead, make a custom form for this purpose:
class ProfileUpdateForm extends ProfileForm
{
public function configure()
{
$this->useFields(array('email', 'other_field'));
}
}
I'm trying to build a form using the Zend_Form component, but the number of elements varies. The information for each Zend_Form element is stored in a database (name, options, validators, filters, etc.).
The application I'm working on consists of building surveys which contain a varying number of questions. Each question is associated with different arrays of answers. Ultimately my goal is to build arrays of radio/checkbox buttons, dynamically, server-side.
I'm looking for a pretty way to generate my form, but I'm not sure of the best way to load the model within the form. Should the model be loaded in the controller then passed (somehow, via a parameter?) directly to the form, or is it better to load the model within the Form init() method? Where's the best place to put the logic, should it be within the form class, or within the controller, or within the model?
My idea is to fetch form element properties (name, rules, filters, etc.) in the database, then iterate and finally render the form. What do you think of this approach? Ultimately, elements will be dynamically added (client-side), this time, using AJAX and a JavaScript library (such as jQuery).
Here are a couple useful links I found via Google, but I think they all answer a slightly different question than mine:
On building dynamic forms, server side:
http://framework.zend.com/wiki/display/ZFPROP/Zend_Form+generation+from+models+-+Jani+Hartikainen
http://weierophinney.net/matthew/archives/200-Using-Zend_Form-in-Your-Models.html
http://codeutopia.net/blog/2009/01/07/another-idea-for-using-models-with-forms/
On building dynamic forms, client side, with AJAX processing:
http://www.jeremykendall.net/2009/01/19/dynamically-adding-elements-to-zend-form/
I think I found a possible solution, it involves passing an array of Zend Form elements to the Zend Form::__construct() method. The constructor takes an array of options, one of them is called "elements". Have a look at the source code within the Zend Framework library.
I coded a new private method within the controller, called buildSurveyForm(). Note : the object, passed as a parameter, is built from a huge SQL query with half a dozen JOIN statements, fetching data from a few tables (surveys, questions, answers, etc.) within the database. One of the public attributes for this class consists of an array of questions, stored as objects (with public methods/attributes as well, etc.). Same for answers. The code for building these classes is pretty trivial and out of topic here.
Here's the code within the survey controller. I copy/pasted and edited/dropped a few lines to make it a lot clearer :
private function buildSurveyForm(MyApp_Object_Survey $survey)
{
foreach ($survey->questions as $question)
{
$element = new Zend_Form_Element_MultiCheckbox($question->order);
$element->addMultiOptions($question->getAnswersLabels());
$element->setName($question->order);
$element->setLabel($question->title);
$elements[] = $element;
}
// Here's the trick :
$formOptions = array('elements' => $elements);
$surveyForm = new MyApp_Survey_Form($formOptions);
$urlHelper = $this->_helper->getHelper('url');
$surveyForm->setAction($urlHelper->url(array(
'controller' => 'survey',
'action' => 'vote'),
'default'
));
$surveyForm->setMethod('post');
$this->_forms['survey'] = $surveyForm;
return $this->_forms['survey'];
}
The MyApp Survey Form class only contains a Submit button within the init() method. The dynamically generated elements with the code above are added BEFORE this submit button (which is unexpected, but useful). This class simply extends Zend_Form.
Then, within survey controller / view action :
public function viewAction()
{
$surveyModel = $this->_model['survey'];
$survey = $surveyModel->getFullSurvey($this->_getParam('id'));
$survey = new MyApp_Object_Survey($survey);
// Calls above private method :
$surveyForm = $this->buildSurveyForm($survey);
$this->view->assign(array(
'surveyForm' => $surveyForm,
));
}
Adding filters, validators and decorators to form elements is now trivial. My proposal is a bit dirty, but I think it gets the job done. I will add a new proposal if I find something more elegant. Feel free to post different answers/solutions.
You could extend Zend_Form.
Zend form is not good place for logic, only form representation.
So, Load all needed elements using model in controller and pass them to the form in contructor as parameters.