In the current examples on ASP.NET MVC I see quite basic entities, with simple CRUD methods.
But I'm not sure about what to do with more advanced models. Let me give an example:
We have a garage website. The garage has:
Inventory with carparts
Employees
Customers
Cars that consists of all cars that are/were in the garage
Now lets take the car, the car might have a collection of employees that worked on the car (derived from the original employee class, adds some extra props that tie him to the car), a collection of carparts that have been replaced (also derived, adds for example SerialNr, and ReplacementDate prop) , and of course a customer prop of the customer whoe owns the car.
Now in rest I would like to see the following:
/cars/423 [get] //show car # 423
/cars/423/edit [get] //shows the edit form (ajax enabled, so also shows the other props)
/cars/423/carparts [get] //gets the carparts inside the car
/cars/423/carparts/32/edit [post] //updates that specific carpart inside the specific car
/cars/423/employees [get] //gets the employees who worked on the car
/inventory [get]
/inventory/1234/edit [get] //gets the update form for carpart with ID 1234
/employees [get] //gets all the employees in the company
So how would I build my Controllers? Should I add all those CRUD methods for the child elements inside the CarsController? Doesn't that make a very fat controller because of the tens of methods (this model is overly simplified, the car has many more parent-child relationships)? Or should I create a EmployeesInCar controller (seems bad)...
Thanks a lot.
EDIT:
First of all the example is hypothetical only, just an example.
If I follow the suggestion I would have a CarController which only handles cars. And my PartsController would handle only Parts. But we have two sets of Parts, a general one (for the inventory); and we have a Part inside a car, which derives from the general one, but adds properties such as SerialNumber and ReplacementDate.
So my Parts controller becomes quite fat, for the example let's call the entities: GeneralPart (in inventory) and SpecificPart (the derived class, with the extra properties)
Index -> Returns all GeneralParts in inventory.
IndexByCar(int id) -> Return all SpecificParts by Car.
etc. etc.
Because each action deals with either a General Part or a Specific Part. The same is for employees, we have the base employee class, and a specific one, with extra properties.
If you wish to use the kinds of paths you provided in your example, then it sounds like you should learn how to use the Routing engine in .NET 3.5. You should be able to use the kinds of urls you need, but you will need to create several custom routes and probably a custom route handler to accomplish it. The Routing engine in .NET 3.5 is very flexible and was not designed specifically for MVC...its actually very generic and can provide a very broad, probably unlimited range of url rewriting.
Its a bit late where I live, so the example I was trying to write just isn't forming. :P Suffice to say, a custom route handler and some new route mappings should get you where you need to be.
EDIT: This may be of help:
http://codingcockerel.co.uk/2008/05/26/custom-routing-for-asp-net-mvc/
EDIT 2:
I left out controllers before. The routing just gets you the ability to use the kinds of URLs you want. And thats a good thing...the kinds of URL's you proposed will provide a better SEO experience long-term. As for organizing your controllers, I recommend keeping it simple:
/Controllers/CarsController.cs
/Controllers/PartsController.cs
/Controllers/EmployeesController.cs
/Controllers/InventoryController.cs
Your routes would match your url patterns, and turn them into a proper route to your controller, taking the ID's and matching them to the parameters of your actions.
EDIT 3:
So, now that I understand your question more fully, I hope I can answer it better. Generally speaking, I think controllers should map 1:1 with your entities. If you have a GeneralPart, then you should have a controller dedicated to managing general parts. If you have CarPart, then you should have a controller dedicated to managing car parts. If CarParts are GeneralParts, and they can behave like GeneralParts in some cases, then it is probably best to have those management aspects on the GeneralPartsController, unless that management deals with any special attributes of a CarPart...in which case management should be delegated to CarPartsController. It is kind of elegant how polymorphism plays into controller organization, and how the "is a" relationship allows you to reuse controllers to manage multiple types.
To be honest, you have a fairly complex scenario I havn't directly encountered in my time working with ASP.NET MVC. I try to avoid such scenarios as much as possible, because they give rise to complicated questions like this that tend to be subjective in their answers. Ultimately, you should organize your controllers in a way that makes logical sense to how they are used, how they map to your entities, and how well they organize the behavior you are interested in exposing. Sometimes, it isn't logical to map all of your controllers to a single entity. Sometimes you need to have a "composite" controller that deals with actions that operate on multiple entities at once, or graphs of entities. In these cases, its probably best to have controllers dedicated to those particular behavioral groups, rather than trying to find one entity-specific controller that "sorta fits".
Your design should initially focus on the entities. You have car, employees and inventory. Write a controller for each one of these. That will give you the basic routes:
/cars/{action}/{id}
/employees/{action}/{id}
/inventory/{action}/{id}
From here you should be able to write a new route with a custom route constraint with the following structure:
/cars/{id}/part/{id}/
This is also slightly better than /cars/{id}/carpart/{id} because the word car in carpart is redundant. /car/.../part/ indicates that part is for a car.
Thinking about things like these for URI design is really important early on because changing the URI design later will break search engine indices, bookmarks and links.
Your "CarPart" entity (class CarPartModel) can be in "stock" state (class StockCarPartModel : CarPartModel) or "replaced" state (class ReplacedCarPartModel : CarPartModel). Then:
GET to /carpart/index should return all entities (CarPartModel)
GET to /carpart/replaced should return replaced parts (ReplacedCarPartModel)
GET to /carpart/bycar/{id} should return parts, related to specific car (ReplacedCarPartModel)
GET to /carpart/inventory should return parts in stock (StockCarPartModel)
This all is handled by "Default" route and in your CarPartController your have actions:
public ActionResult Index() { ... }
public ActionResult Replaced() { ... }
public ActionResult ByCar(string carId) { ... }
public ActionResult Inventory() { ... }
I think your controller is not fat. It is normal for controller to deal with Model inheritance. The main complexity will be in your ViewModels and Views
You don't need to have too many things going on in a controller.
Think of a controller as a class that determines what the user wants and what it should see in the end.
What user wants is determined by the ASP.NET MVC engine automatically for you, with the help of the routes you defined in the Global.asax file. Then the controller should decide what view the user should see as a result of the request he/she made.
So for your application things would go something like this :
Global.asax
routes.MapRoute("Cars", "cars/{id}/{action}",
new { controller = "Cars", action = "Details", id = "" } );
//This route should give you the URLs you listed on your question
CarsController.cs
public class CarsController : Controller {
//this action will be called for URLs like "/cars/423"
//or /cars/423/details
public ActionResult Details(int id) {
//You would most likely have a service or business logic layer
//which handles talking to your DAL and returning the objects
//that are needed for presentation and likes
Car car = SomeClass.GetCar(id);
if (car == null) {
//if there's no car with the ID specified in the URL
//then show a view explaining that
return View("CarNotFound");
}
//if there's a car with that ID then return a view that
//takes Car as its model and shows details about it
return View(car);
}
//this action will be called for URLs like /cars/423/edit
[AcceptVerbs(HttpVerbs.Get)]
public ActionResult Edit(int id) {
//again get the car, show an edit form if it's available for editing
}
//this action will be called for URLs like /cars/423/edit too
//but only when the request verb is POST
[AcceptVerbs(HttpVerbs.Post), ActionName("Edit")]
public ActionResult Save(int id) {
//save the changes on the car, again using your business logic layer
//you don't need the actual implementation of saving the car here
//then either show a confirmation message with a view,
//saying the car was saved or redirect the user to the
//Details action with the id of the car they just edited
}
}
So basically this is how you would set up your controller(s) for an application like that.
I would recommend using Restful Routing for ASP .NET MVC:
https://github.com/stevehodgkiss/restful-routing
Related
I have a client who wishes to use a URL naming convention along the lines of:
/{subjectarea}/{subject}/{action}
Which is fine - this works brilliantly, with one controller per subject area, and having the action after the id (subject) is no issue at all.
However, it then gets complicated, as the client then wants to further continue the hierarchy:
/{subjectarea}/{subject}/{action}/{tightlyrelatedsubject}/{tightlyrelatedsubjectvariables}/{tightlyrelatedsubjectaction}
I have a controller for the tightly related subject (its just another subject area) which handles all of the admin side, but the client insists on having the public view hung off of the parent subject rather than its own root.
How can I do this while avoiding breaking the entire principals of MVC, and also avoiding re-implementing a ton of ASP.Net MVC provided functionality in my subject area controller just to be able to handle the related subjects from that same controller?
Is it possible to somehow call the related subjects controller from within the parent subject controller, and return the resulting view (as this would keep the separation of functionality for the subjects to their own controllers)? If that is possible, it would solve a heck of a lot of issues with this.
Here is the solution which solves my given issue - hope it solves someone elses.
As mentioned in my comment to Robert Harvey, all I actually need is another route which doesn't use the first two or three components as the controller, action and id, but instead takes those values from later on - if you hang this off of a static value in the route as well, its much easier to do.
So, here is the url I decided on to simplify the route:
/{subjectarea}/{subject}/related/{tightlyrelatedsubject}/{tightlyrelatedsubjectvariables}/{tightlyrelatedsubjectaction}
The route which satisfies this URL is as follows:
routes.MapRoute(
"RelatedSubjects",
"{parentcontroller}/{parentsubject}/related/{controller}/{id}/{action}",
new { controller = "shoes", action = "view", id = "all" }
);
On the subsequent controller action, I can ask for parameter values for parentcontroller and parentsubject so I can filter out the related item to just be specific to the given parent subject - problem solved!
This route needs to be above the ones which just deal with the first two values, otherwise you run the risk of another route map hijacking the request.
I could do this entirely without the /related/ static portion as the route could easily match on number of values alone, and infact I may indeed do so - however, I consider it better for later administration if there is a static item in there to confirm the use of the route.
I hope this helps someone!
One way you can do it is specify a wildcard route (notice the asterisk):
routes.MapRoute("subjects", "{action}/{*path}",
new { controller = "Subjects", action = "Index" });
This allows the controller to receive the entire path string after the action.
You can then obtain the hierarchy of subjects in the controller method like so:
string[] subjects = path.Split('/');
Once you have that, you can do anything you want, including dispatching different subjects to different handling methods for processing.
I have - I think - a complex URL to deal with in ASP MVC 1.0:
All my actions in most of the controllers require two parameters all the time: Account and Project. This is on top of each Action's requirements. This means a typical URL is like this:
http://abcd.com/myaccount/projects/project_id/sites/edit/12
In this example:
myaccount is the account name. projects can be a controller, others options are like locations, employees. project_id is the id of a project within myaccount, sites could be a controller, other options are like staff or payments. edit is an action and 12 is the id of the site edited.
(hope this is clear enough)
Now one option is to create a route and pass project_id and account into all actions of controllers by adding two extra parameters to all actions. This is not really desired and also I'm not sure the two controllers (projects and sites) are going to work here.
My ideal situation is to use some kind of context that travels with the call to the controller action and store project_id and myaccount in there. The rest of the parameters then can be dealt with in a normal way like:
// sitescontroller
public ActionResult Edit(string id)
{
string account = somecontext["account"];
string project_id = somecontext["project"];
// do stuff
}
Any ideas as to how/where this can happen? Also how is this going to work with ActionLink (i.e. generating correct links based on this context)?
Thanks!
You first need to add the tokens to your routes like {company}/projects/{project}{controller}/{action}/{id}. Then if you wrote your own IControllerFactory then it would be very easy to push the values from the RouteData into the controller via the constructor or however you wanted to do it. Probably the easiest way to get started would be to subclass DefaultControllerFactory and override the CreateController method.
This doesn't quite make sense to me. Why would you have a route that is akin to the following:
{controller}/{id}/{controller}/{id}
?
My question may be obvious but I'd like to build a well-designed web application.
As for any administration area, the admin should be able to list/create/delete/modify users, articles, posts, etc...
I'd like to know what is the best way to design the application.
Should I create a controller for each one of those items (/Users/Create/id or /Posts/Delete/id), or create all of the action in my Administration Controller (/Administration/CreateUser/id or /Administration/DeletePost/id) ?
You should write a separate controller for each entity to keep a clean separation of concerns for your controller classes. If you only have one controller, then you will only have one Views directory with dozens of views, and your controller will contain dozens of methods, and that will soon become unmanageable.
The answer depends on how much functionality will be in the controllers. Just start of with one controller and if it gets too much split it into a few.
The great thing about MVC is where you put things in your controllers doesn't have to have any effect on the URLs. you can very easily map /Users/Create to e.g. UserAdminController class.
I would just build a new MVC website that handles the administration.
You have a more flexible solution as long as you've separated the data and the business logic in different assembly's. You could then publish your website to a subdomain, admin.yoursite.com for example. That way you don't have to mess with your routes and you can keep them in separate views which imho is the most elegant solution.
Pro's and Con's would be nice to hear.
I'm working on a project which will need the same administration site but haven't got that far yet so this question interests me a lot.
I'm currently using ASP.NET for a large client.
The approach I've adopted is to put the functionality of the action into another class.
Example
I am writing an administration section also. There will be one Administration controller (our admin section is small, if it was larger I would change the routing to allow more controllers, for now we are using the out of the box configuration). If I create an "EditUser" view. I will also create an "EditUserAction" class. All EditUser code will go into the class. I construct the EditUserAction class in the Administration controller class in the Edit User method. This removes all the action specific code out of the Controller class. This way all the action specific code is either in the action method or in the action class. Otherwise, the controller would quickly become overrun with code from various actions. The controller class would balloon to an unmanageable mess in short order.
Class examples
public class Administration: Controller
{
public ActionResult EditUser(string userId)
{
EditUserAction action = new EditUserAction();
}
}
public class EditUserAction
{
public void Save(User user)
{
//save code here
}
}
I hope this explanation is clear. If it's not let me know and I'll clarify.
To answer your question I am doing it the latter(/Administration/CreateUser/id or /Administration/DeletePost/id).
You could use DynamicData for this. It isn't MVC but it can be used together with it and it is really easy to setup and use.
Here is another way of asking my question.
A part of my Master Page:
<% if (!String.Equals(ViewContext.RequestContext.RouteData.GetRequiredString("controller"), "Administration")) { %>
<div>
<!-- Some Code -->
</div> <% } %>
As you can see, in my master page, I'd like to display some part of the page, depending on the user working on the administration area or not.
It works pretty well with only the Administration Controller (/Administration/CreateUser/id)... but it becomes a big mess when I use different controller as User or Article (/User/DeleteUser/id or /Article/Details/id).
I'd prefer to use one controller per entity, but I can't find a way to bond this approach with multiple controllers.
I suggest to use this solution.
But I changed definition to this:
public ThemedViewEngine()
{
base.MasterLocationFormats = new string[] {
"~/Views/{1}/{0}.master",
"~/Views/Shared/{0}.master",
"~/Themes/{2}/Views/{1}/{0}.master",
"~/Themes/{2}/Views/Shared/{0}.master",
"~/Themes/Default/Views/{1}/{0}.master",
"~/Themes/Default/Views/Shared/{0}.master"
};
base.ViewLocationFormats = new string[] {
"~/Views/{1}/{0}.aspx",
"~/Views/{1}/{0}.ascx",
"~/Views/Shared/{0}.aspx",
"~/Views/Shared/{0}.ascx",
"~/Themes/{2}/Views/{1}/{0}.aspx",
"~/Themes/{2}/Views/{1}/{0}.ascx",
"~/Themes/{2}/Views/Shared/{0}.aspx",
"~/Themes/{2}/Views/Shared/{0}.ascx",
"~/Themes/Default/Views/{1}/{0}.aspx",
"~/Themes/Default/Views/{1}/{0}.ascx",
"~/Themes/Default/Views/Shared/{0}.aspx",
"~/Themes/Default/Views/Shared/{0}.ascx"
};
base.PartialViewLocationFormats = new string[] {
"~/Views/{1}/{0}.aspx",
"~/Views/{1}/{0}.ascx",
"~/Views/Shared/{0}.aspx",
"~/Views/Shared/{0}.ascx",
"~/Themes/{2}/Views/{1}/{0}.aspx",
"~/Themes/{2}/Views/{1}/{0}.ascx",
"~/Themes/{2}/Views/Shared/{0}.aspx",
"~/Themes/{2}/Views/Shared/{0}.ascx",
"~/Themes/Default/Views/{1}/{0}.aspx",
"~/Themes/Default/Views/{1}/{0}.ascx",
"~/Themes/Default/Views/Shared/{0}.aspx",
"~/Themes/Default/Views/Shared/{0}.ascx"
};
}
Default theme is default so it is required to exists.
Structure of directory will be:
Content
Themes
Default
Content
Views
Home
Blog
Whatever should be skinned
OtherTheme
Content
Views
Home
Blog
Whatever should be skinned
Views
Articles
Posts
Users
Settings
Other Administration stuff
It depends on the scale size of your admin area,
I suggest you consider to do the following (or maybe document it a bit)
Consider how many entities you want to manage independently,
Consider how many actions each one will have,
Check that is there any dependency between your application and admin area (user access, for user friendly URLs )
then you can specify which approach can help you, Having one admin controller, admin actions in entity controllers, or defining a new Admin project in case of large functional applications.
*If the project scale is growing fast and soon it needs large scale, I would choose third one - having a new admin mvc project.
I hope it help you decide.
The 'RenderPartial()' method in ASP.NET MVC offeres a very low level of functionality. It does not provide, nor attempt to provide a true 'sub-controller' model *.
I have an increasing number of controls being rendered via 'RenderPartial()'. They fall into 3 main categories :
1) Controls that are direct
descendants of a specific page that
use that page's model
2) Controls that are direct
descendants of a specific page that
use that page's model with an
additional key of some type.
Think implementation of
'DataRepeater'.
3) Controls that represent unrelated
functionality to the page they appear
on. This could be anything from a
banner rotator, to a feedback form,
store locator, mailing list signup.
The key point being it doesn't care
what page it is put on.
Because of the way the ViewData model works there only exists one model object per request - thats to say anything the subcontrols need must be present in the page model.
Ultimately the MVC team will hopefully come out with a true 'subcontroller' model, but until then I'm just adding anything to the main page model that the child controls also need.
In the case of (3) above this means my model for 'ProductModel' may have to contain a field for 'MailingListSignup' model. Obviously that is not ideal, but i've accepted this at the best compromise with the current framework - and least likely to 'close any doors' to a future subcontroller model.
The controller should be responsible for getting the data for a model because the model should really just be a dumb data structure that doesn't know where it gets its data from. But I don't want the controller to have to create the model in several different places.
What I have begun doing is creating a factory to create me the model. This factory is called by the controller (the model doesn't know about the factory).
public static class JoinMailingListModelFactory {
public static JoinMailingListModel CreateJoinMailingListModel() {
return new JoinMailingListModel()
{
MailingLists = MailingListCache.GetPartnerMailingLists();
};
}
}
So my actual question is how are other people with this same issue actually creating the models. What is going to be the best approach for future compatibility with new MVC features?
NB: There are issues with RenderAction() that I won't go into here - not least that its only in MVCContrib and not going to be in the RTM version of ASP.NET-MVC. Other issues caused sufficent problems that I elected not to use it. So lets pretend for now that only RenderPartial() exists - or at least that thats what I've decided to use.
Instead of adding things like MailingListSignup as a property of your ProductModel, encapsulate both at the same level in a class like ProductViewModel that looks like:
public class ProductViewModel() {
public ProductModel productModel;
public MailingListSignup signup;
}
Then get your View to be strongly-typed to the ProductViewModel class. You can access the ProductModel by calling Model.productModel, and you can access the signup class using Model.signup.
This is a loose interpretation of Fowler's 'Presentation Model' (http://martinfowler.com/eaaDev/PresentationModel.html), but I've seen it used by some Microsoft devs, such as Rob Conery and Stephen Walther.
One approach I've seen for this scenario is to use an action-filter to populate the data for the partial view - i.e. subclass ActionFilterAttribute. In the OnActionExecuting, add the data into the ViewData. Then you just have to decorate the different actions that use that partial view with the filter.
There's a RenderPartial overload I use that let's you specify a new ViewData and Model:
RenderPartial code
If you look at the previous link of the MVC source code, as well as the following (look for RenderPartialInternal method):
RenderPartialInternal code
you can see that if basically copies the viewdata you pass creating a new Dictionary and sets the Model to be used in the control. So the page can have a Model, but then pass a different Model to the sub-control.
If the sub-controls aren't referred directly from the main View Model, you could do the trick Marc Gravell mentions to add your custom logic.
One method I tried was to use a strongly typed partial view with an interface. In most situations an agregated ViewModel is the better way, but I still want to share this.
<%# Control Language="C#" Inherits="System.Web.Mvc.ViewUserControl<IMailingListSignup>" %>
The Viewmodel implements the interface
public class ProductViewModel:IMailingListSignup
Thats not perfect at all but solves some issues: You can still easily map properties from your route to the model. I am not shure if you can have a route parameter map to the properties of MailingListSignup otherwise.
You still have the problem of filling the Model. If its not to late I prefer to do it in OnActionExecuted. I dont see how you can fill a Model in OnActionExecuting.
Imagine a blog engine in ASP.NET MVC where I want strongly-typed Views. I have a Model for BlogPost and a Model for Comments. A typical page would display a blog post and all of that post's comments. What object should I pass to the View()? Should I create a Presentation Model? Maybe I'm using that term incorrectly, but by that I mean should I create a composite model just for a View(), something like BlogAndComments? Should my controller create an instance of BlogAndComments and pas that?
Or, should I somehow pass both a BlogPost and Comments object to the View?
I think you're on the right track with your understanding of Presentation Models. As to when you should create a View Model, the answer is probably 'it depends'. In your example, you can probably get away with passing the BlogPost and Comments in the ViewData object. It's not gorgeous, but hey, it gets the job done.
When and if that starts to feel ugly or unwieldy, then I would start thinking about making a View Model. I usually end up with the notion of some sort of 'Page', which includes the page title, common data, and then specific stuff for a particular page. In your case, that might end up as a BlogViewPage, which includes Title, BlogPost and List comments.
The nice thing about that approach is that you can then test that controller by making a request and testing the BlogViewPage to ensure that it contains the expected data.
In my opinion comments belong to the view as much as the post itself.
Make a BL class for your comments like:
class CommentBO
{
Guid UserID;
string Text;
}
Then you have a BO for your post.
class PostBO
{
Guid UserID;
List<CommentBO> Comments;
}
Then your model can be really simple.
class BlogModel
{
string AuthorName;
string BlogTitle;
List<PostBO> Posts;
}
Pass it to the view and render it.
You may be tempted to omit all BO and just fill the model directly from the database. It is an option, but not exactly the right one. A model is just a package of things for a view to display. These things however should be prepared somewhere else, namely at the business logics level with just a nominal participation of the controller.
I always use strongly typed views and create a presentation model class for each view or view user control. The advantage is that by looking at the presentation model class alone I know exactly what are the values that the view uses. If I were passing domain models then that would not be obvious because domain models may contain many properties that the view does not use. I would have to look at the view's markup to figure out what values it needs. Plus the presentation model usually adds certain properties that are not available in the domain model. It seems a bit tedious but I think it saves time and makes the code better in the long run.
In my opinion if comments belong to a blog post why not create a collection of comments on the blog post model? That makes perfect sense from a domain modeling stand-point and chances are whatever ORM you are using would support lazy-loading that collection which simplifies your controller logic as well.