按类别划分类(控制器、视图模型、过滤器等)是毫无意义的。

如果你想为你的网站(/)的Home部分写代码，然后创建一个名为Home的文件夹，并在那里放置HomeController, IndexViewModel, AboutViewModel等，以及Home操作使用的所有相关类。

如果你有共享类，比如ApplicationController，你可以把它放在项目的根目录。

为什么要分离相关的东西(HomeController, IndexViewModel)，而把完全没有关系的东西放在一起(HomeController, AccountController) ?

我就这个话题写了一篇博客。

There are no good place to keep your models in. You can keep them in separate assembly if the project is big and there are a lot of ViewModels (Data Transfer Objects). Also you can keep them in separate folder of the site project. For example, in Oxite they are placed in Oxite project which contains a lot of various classes too. Controllers in Oxite are moved to separate project and views are in separate project too. In CodeCampServer ViewModels are named *Form and they are placed in UI project in Models folder. In MvcPress project they are placed in Data project, which also contains all code to work with database and a bit more (but I didn't recommend this approach, it's just for a sample) So you can see there are many point of view. I usually keep my ViewModels (DTO objects) in the site project. But when I have more than 10 models I prefer to move them to separate assembly. Usually in this case I'm moving controllers to separate assembly too. Another question is how to easily map all data from model to your ViewModel. I suggest to have a look at AutoMapper library. I like it very much, it does all dirty work for me. And I also I suggest to look at SharpArchitecture project. It provides very good architecture for projects and it contains a lot of cool frameworks and guidances and great community.

ViewModel类用于将由类实例表示的多个数据封装到一个易于管理的对象中，您可以将该对象传递给View。

将ViewModel类放在自己的文件中，放在自己的目录中是有意义的。在我的项目中，我有一个名为ViewModels的Models文件夹子文件夹。这就是我的ViewModels(例如ProductViewModel.cs)所在的位置。

控制器中的代码:

    [HttpGet]
        public ActionResult EntryEdit(int? entryId)
        {
            ViewData["BodyClass"] = "page-entryEdit";
            EntryEditViewModel viewMode = new EntryEditViewModel(entryId);
            return View(viewMode);
        }

    [HttpPost]
    public ActionResult EntryEdit(Entry entry)
    {
        ViewData["BodyClass"] = "page-entryEdit";            

        #region save

        if (ModelState.IsValid)
        {
            if (EntryManager.Update(entry) == 1)
            {
                return RedirectToAction("EntryEditSuccess", "Dictionary");
            }
            else
            {
                return RedirectToAction("EntryEditFailed", "Dictionary");
            }
        }
        else
        {
            EntryEditViewModel viewModel = new EntryEditViewModel(entry);
            return View(viewModel);
        }

        #endregion
    }

视图模型中的代码:

public class EntryEditViewModel
    {
        #region Private Variables for Properties

        private Entry _entry = new Entry();
        private StatusList _statusList = new StatusList();        

        #endregion

        #region Public Properties

        public Entry Entry
        {
            get { return _entry; }
            set { _entry = value; }
        }

        public StatusList StatusList
        {
            get { return _statusList; }
        }

        #endregion

        #region constructor(s)

        /// <summary>
        /// for Get action
        /// </summary>
        /// <param name="entryId"></param>
        public EntryEditViewModel(int? entryId)
        {
            this.Entry = EntryManager.GetDetail(entryId.Value);                 
        }

        /// <summary>
        /// for Post action
        /// </summary>
        /// <param name="entry"></param>
        public EntryEditViewModel(Entry entry)
        {
            this.Entry = entry;
        }

        #endregion       
    }

项目:

DevJet。Web (ASP。NET MVC web 项目) DevJet.Web.App.Dictionary (a 单独的类库项目) 在这个项目中，我做了一些文件夹，比如: 木豆, BLL, 薄 VM(视图模型文件夹)

在我们的例子中，我们将模型和控制器放在一个独立于视图的项目中。

根据经验，我们已经尝试移动和避免大多数ViewData[”…]的东西，因此我们避免了类型转换和魔术字符串，这是一件好事。

ViewModel还包含一些常见的属性，如列表的分页信息或用于绘制面包屑和标题的页眉信息。在我看来，此时基类包含了太多的信息，我们可以将其分为三部分，基本视图模型中99%的页面的最基本和必要的信息，然后是列表模型和表单模型，其中包含了该场景的特定数据并继承自基类。

最后，我们为每个实体实现一个视图模型来处理特定的信息。

创建一个视图模型基类，它具有通常需要的属性，如操作结果和上下文数据，您还可以放置当前用户数据和角色

class ViewModelBase 
{
  public bool HasError {get;set;} 
  public string ErrorMessage {get;set;}
  public List<string> UserRoles{get;set;}
}

在基控制器类中，有一个类似PopulateViewModelBase()的方法，该方法将填充上下文数据和用户角色。 HasError和ErrorMessage，如果从service/db中提取数据时出现异常，请设置这些属性。在视图上绑定这些属性以显示错误。 用户角色可用于根据角色在视图中显示隐藏部分。

要在不同的get操作中填充视图模型，可以通过具有具有抽象方法FillModel的基本控制器使之一致

class BaseController :BaseController 
{
   public PopulateViewModelBase(ViewModelBase model) 
{
   //fill up common data. 
}
abstract ViewModelBase FillModel();
}

在控制器

class MyController :Controller 
{

 public ActionResult Index() 
{
   return View(FillModel()); 
}

ViewModelBase FillModel() 
{ 
    ViewModelBase  model=;
    string currentAction = HttpContext.Current.Request.RequestContext.RouteData.Values["action"].ToString(); 
 try 
{ 
   switch(currentAction) 
{  
   case "Index": 
   model= GetCustomerData(); 
   break;
   // fill model logic for other actions 
}
}
catch(Exception ex) 
{
   model.HasError=true;
   model.ErrorMessage=ex.Message;
}
//fill common properties 
base.PopulateViewModelBase(model);
return model;
}
}