System.ComponentModel.DataAnnotations for ASP.NET Web Forms
Although I'm primarily an ASP.NET Web Forms developer, I regularly dabble in new and interesting technologies. I've toyed with other Microsoft technologies such as ASP.NET MVC and Dynamic Data web sites.
ASP.NET MVC offers an interesting mechanism for validating view models called DataAnnotations. Examples can be seen here and here.
Some of the more useful attributes for validation include:
- RequiredAttribute
- RegularExpressionAttribute
- RangeAttribute
- DataTypeAttribute
I recently decided to implement a data validation schema (for lack of a better term) for web forms similar to that of ASP.NET MVC. This would allow us to maintain validation of a model object via attributes at the class level, instead of dispersing these validation rules through the web form code or any of the business logic layers. However, there will be occasions where an object's validity depends upon some state or validity of another object. I've handled this by allowing the validation method to accept any action to be called after the object has been validated.
In order to make a domain object validatable, I'm going to implement the following interface:
IValidatable.cs
namespace Validatable
{
interface IValidatable
{
System.Type EntityType { get; set; }
System.Collections.Generic.List<object> Errors { get; set; }
/// <summary>
/// Determine whether the object is valid. If invalid, errors are added to item.Errors
/// </summary>
/// <returns>true if valid, false if invalid</returns>
bool IsValid();
/// <summary>
/// Validate this item against a supplied action
/// </summary>
/// <param name="action">The action to use for validation</param>
void Validate(System.Action action);
/// <summary>
/// Validate this object against a Repository.
/// <example>
/// Item a = new Item();
/// a.Validate(new ItemRepository());
/// </example>
/// Note: Repository must have a Validate(T item) method
/// </summary>
/// <param name="repository">The Repository to use for validation</param>
void Validate(object repository);
}
}
As you can see, this abstract class adds a property accessor for the entity's type, a list of validation errors, an IsValid method, and two Validate methods.
Validatable<T>
I'll take the code in chunks, since there is a lot more than I usually post.
First of all, the using directives required for this class are pretty sparse:
using System;
using System.Collections.Generic;
using System.ComponentModel.DataAnnotations;
using System.Linq;
using System.Reflection;
The class definition is:
[Serializable]
public abstract class Validatable<T> : IValidatable where T : class, new()
{ /* class */ }
If you're unfamiliar with generic constraints, the where T : class, new() above enforces the T passed as in to be a class and to have a parameterless constructor (new())
The properties can be auto-implemented properties, or they can implement any backing logic you prefer. The constructor as I have it is:
protected Validatable()
{
Errors = new List<object>();
EntityType = typeof(T);
}
Now for the methods and a slight explanation of each:
IsValid()
public virtual bool IsValid()
{
Errors = new List<object>();
PropertyInfo[] props = this.GetType().GetProperties();
foreach (PropertyInfo property in props)
{
foreach (ValidationAttribute va in
property.GetCustomAttributes(true).OfType<ValidationAttribute>())
{
var value = property.GetValue(this, null);
if (!va.IsValid(value))
{
Errors.Add(va.ErrorMessage);
}
}
}
return Errors.Count <= 0;
}
As a side note, all DataAnnotation attributes inherit from ValidationAttribute. You can create your own Validation Attributes by inheriting from this class. This method clears the list of errors and repopulates it by looping over any properties with validation attributes of this object, validates that property's value against the attribute, and adds the ErrorMessage associated with it to the list of Errors.
Validate Methods
public virtual void Validate(Action action)
{
if (IsValid())
{
action();
}
}
public virtual void Validate(object repository)
{
MethodInfo method = repository.GetType().GetMethods()
.Where(x => x.Name.Equals("Validate")).FirstOrDefault();
if (method != null)
{
object[] parameters = new object[] { this };
Validate(() => method.Invoke(repository, parameters));
}
}
These two methods can be changed to throw an exception if the object is invalid. I'll leave that up to you. The way these are set up is to allow you to validate the object and perform either some unknown action or call Validate(T item) against some other object, arbitrarily called a repository. In actuality, you could have a Validate method on some other object that takes an object of this type and essentially chain validations.
The simplicity of the Validate(Action action) method is what makes it beautiful. For instance, you can have an item (a) and cause it save after validation by doing something like:
a.Validate(() => new ItemDAL().Save(a));
As you can see, a Save method in your Item's Data Access Layer taking Item as a parameter can be called only when a.IsValid() is true. Again, there are a number of ways to tweak this and change it to your liking, but I'll leave that up to you.
DataAnnotationValidator
Here is the wonderful part about all of this. You can create a custom validator that validates in a similar way as the Validatable class. I'll post the code and quickly explain what it does. I got the idea from this from another blog and tweaked it a little.
DataAnnotationValidator.cs
// DataAnnotationValidator.cs
namespace Validatable
{
using System;
using System.ComponentModel.DataAnnotations;
using System.Linq;
using System.Reflection;
using System.Web.UI;
using System.Web.UI.WebControls;
[ToolboxData("<{0}:DataAnnotationValidator runat=\"server\" ControlToValidate=\"[Required]\" Display=\"Dynamic\" Text=\"*\" SourceTypeName=\"[FullyQualifiedTypeName]\" PropertyToValidate=\"[PropertyName]\" />")]
public class DataAnnotationValidator : BaseValidator
{
/// <summary>
/// THe Property that should be checked
/// </summary>
public string PropertyToValidate { get; set; }
/// <summary>
/// The object's type
/// </summary>
public string SourceTypeName { get; set; }
protected override bool EvaluateIsValid()
{
Type source = GetValidatedType();
PropertyInfo property = GetValidatedProperty(source);
string value = GetControlValidationValue(ControlToValidate);
foreach (ValidationAttribute va in property
.GetCustomAttributes(typeof(ValidationAttribute), true)
.OfType<ValidationAttribute>())
{
if (!va.IsValid(value))
{
if (string.IsNullOrEmpty(ErrorMessage))
{
this.ErrorMessage = va.ErrorMessage;
}
return false;
}
}
return true;
}
private Type GetValidatedType()
{
if (string.IsNullOrEmpty(SourceTypeName))
{
throw new InvalidOperationException("Null SourceTypeName can't be validated");
}
Type validatedType = Type.GetType(SourceTypeName);
if (validatedType == null)
{
throw new InvalidOperationException(
string.Format("{0}:{1}", "Invalid SourceTypeName", SourceTypeName));
}
return validatedType;
}
private PropertyInfo GetValidatedProperty(Type source)
{
PropertyInfo property = source.GetProperty(PropertyToValidate,
BindingFlags.Public | BindingFlags.Instance);
if (property == null)
{
throw new InvalidOperationException(
string.Format("{0}:{1}", "Validated Property Does Not Exists", PropertyToValidate));
}
return property;
}
}
}
This creates a validator web control which has 'PropertyToValidate' and 'SourceTypeName'. When a Validator web control is added to a page, it must have ControlToValidate, PropertyToValidate, and SourceTypeName specified or an error will be thrown. The validation method verifies the type and the property, gets the value of the property and finally, much like the Validatable abstract class, it loops over all ValidationAttributes and validates the value against that attribute. You can use this with multiple attributes, so it doesn't hurt to add this validator to a property-- that just means you can add an attribute in the future and everything is already wired up! That's pretty cool.
Example
As an example, I'm going to create a single page to **input** a Customer object. The customer object is very simple:
public class Customer : Validatable.Validatable<Customer>
{
[Required(ErrorMessage="UserId is Required")]
public int UserId { get; set; }
[DataType(DataType.PhoneNumber, ErrorMessage="Invalid Phone Number")]
public string PhoneNumber { get; set; }
[RegularExpression(@"(^\d{5}(-\d{4}){0,1}$)", ErrorMessage="Invalid Zip Code")]
public string ZipCode { get; set; }
}
And the Page in full is collapsed below:
<%@ Page Language="C#" AutoEventWireup="true" CodeBehind="Default.aspx.cs" Inherits="Example._Default" %>
<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
<html xmlns="http://www.w3.org/1999/xhtml" >
<head runat="server">
<title></title>
</head>
<body>
<form id="form1" runat="server">
<asp:Label ID="lblStatus" runat="server" />
<asp:ValidationSummary ID="ValidationSummary1" runat="server" />
<div>
<asp:FormView ID="FormView1" runat="server" DataSourceID="srcRepository"
Width="119px" AllowPaging="True">
<EditItemTemplate>
UserId:
<asp:TextBox ID="UserIdTextBox" runat="server" Text='<%# Bind("UserId") %>' />
<example:DataAnnotationValidator ID="DataAnnotationValidator1" runat="server" ControlToValidate="UserIdTextBox" Display="Dynamic"
PropertyToValidate="UserId" ErrorMessage="Invalid User Identification Number"
SourceTypeName="Example.Customer, Example, Version=1.0.0.0, Culture=neutral, PublicKeyToken=null"
Text="*"/>
<br />
PhoneNumber:
<asp:TextBox ID="PhoneNumberTextBox" runat="server"
Text='<%# Bind("PhoneNumber") %>' />
<example:DataAnnotationValidator ID="DataAnnotationValidator2" runat="server" ControlToValidate="PhoneNumberTextBox" Display="Dynamic"
PropertyToValidate="PhoneNumber" OnInit="GetTypeName" SourceTypeName="Example.Customer, Example, Version=1.0.0.0, Culture=neutral, PublicKeyToken=null" Text="*"/>
<br />
ZipCode:
<asp:TextBox ID="ZipCodeTextBox" runat="server" Text='<%# Bind("ZipCode") %>' />
<example:DataAnnotationValidator ID="DataAnnotationValidator3" runat="server" ControlToValidate="ZipCodeTextBox" Display="Dynamic"
PropertyToValidate="ZipCode" SourceTypeName="Example.Customer, Example, Version=1.0.0.0, Culture=neutral, PublicKeyToken=null" Text="*"/>
<br />
EntityType:
<asp:TextBox ID="EntityTypeTextBox" runat="server"
Text='<%# Eval("EntityType") %>' />
<br />
<asp:LinkButton ID="UpdateButton" runat="server" CausesValidation="True"
CommandName="Update" Text="Update" />
<asp:LinkButton ID="UpdateCancelButton" runat="server"
CausesValidation="False" CommandName="Cancel" Text="Cancel" />
</EditItemTemplate>
<InsertItemTemplate>
UserId:
<asp:TextBox ID="UserIdTextBox" runat="server" Text='<%# Bind("UserId") %>' />
<example:DataAnnotationValidator ID="DataAnnotationValidator1" runat="server" ControlToValidate="UserIdTextBox" Display="Dynamic"
PropertyToValidate="UserId" OnInit="GetTypeName" Text="*"/>
<br />
PhoneNumber:
<asp:TextBox ID="PhoneNumberTextBox" runat="server"
Text='<%# Bind("PhoneNumber") %>' />
<example:DataAnnotationValidator ID="DataAnnotationValidator2" runat="server" ControlToValidate="PhoneNumberTextBox" Display="Dynamic"
PropertyToValidate="PhoneNumber" OnInit="GetTypeName" Text="*"/>
<br />
ZipCode:
<asp:TextBox ID="ZipCodeTextBox" runat="server" Text='<%# Bind("ZipCode") %>' />
<example:DataAnnotationValidator ID="DataAnnotationValidator3" runat="server" ControlToValidate="ZipCodeTextBox" Display="Dynamic"
PropertyToValidate="ZipCode" OnInit="GetTypeName" Text="*"/>
<br />
EntityType:
<asp:TextBox ID="EntityTypeTextBox" runat="server"
Text='<%# Eval("EntityType") %>' />
<br />
<asp:LinkButton ID="InsertButton" runat="server" CausesValidation="True"
CommandName="Insert" Text="Insert" />
<asp:LinkButton ID="InsertCancelButton" runat="server"
CausesValidation="False" CommandName="Cancel" Text="Cancel" />
</InsertItemTemplate>
<ItemTemplate>
UserId:
<asp:Label ID="UserIdTextBox" runat="server" Text='<%# Bind("UserId") %>' />
<br />
PhoneNumber:
<asp:Label ID="PhoneNumberTextBox" runat="server"
Text='<%# Bind("PhoneNumber") %>' />
<br />
ZipCode:
<asp:Label ID="ZipCodeTextBox" runat="server" Text='<%# Bind("ZipCode") %>' />
<br />
EntityType:
<asp:Label ID="EntityTypeLabel" runat="server"
Text='<%# Bind("EntityType") %>' />
<br />
<asp:LinkButton ID="EditButton" runat="server" CausesValidation="False"
CommandName="Edit" Text="Edit" />
<asp:LinkButton ID="NewButton" runat="server" CausesValidation="False"
CommandName="New" Text="New" />
</ItemTemplate>
</asp:FormView>
<asp:ObjectDataSource ID="srcRepository" runat="server"
DataObjectTypeName="Example.Customer" InsertMethod="Save"
oninserted="srcRepository_Saved" onupdated="srcRepository_Saved"
SelectMethod="Get" TypeName="Example.CustomerRepository" UpdateMethod="Save"></asp:ObjectDataSource>
</div>
</form>
</body>
</html>
Of particular note here is in the EditTemplate, you'll see
UserId:
<asp:TextBox ID="UserIdTextBox" runat="server" Text='<%# Bind("UserId") %>' />
<example:DataAnnotationValidator ID="DataAnnotationValidator1" runat="server" ControlToValidate="UserIdTextBox" Display="Dynamic"
PropertyToValidate="UserId" ErrorMessage="Invalid User Identification Number"
SourceTypeName="Example.Customer, Example, Version=1.0.0.0, Culture=neutral, PublicKeyToken=null"
Text="*"/>
in which the SourceTypeName must be the full Assembly Qualified Name. This is necessary if the Validatable
UserId:
<asp:TextBox ID="UserIdTextBox" runat="server" Text='<%# Bind("UserId") %>' />
<example:DataAnnotationValidator ID="DataAnnotationValidator1" runat="server" ControlToValidate="UserIdTextBox" Display="Dynamic"
PropertyToValidate="UserId" OnInit="GetTypeName" Text="*"/>
and the code in the code behind is very simple:
protected void GetTypeName(object sender, EventArgs e)
{
DataAnnotationValidator validator = (DataAnnotationValidator)sender;
validator.SourceTypeName = new Customer().EntityType.AssemblyQualifiedName;
}
Also notice that the first validator (the one with the assembly qualified name) has an ErrorMessage property specified. Doing this allows you to override the error message returned from the validated object. We can do this if the model says "UserId" and you want the user to see this property referred to as "User Identification".
Conclusion
Sometimes dabbling in other technologies can open the door for new and simpler ways of doing things. I like this example because it allows you to build web applications quickly and easily, while making them maintainable in the future (new validations only have to be added to the model, not to every control where the model must be validated).
For further reference, this project is linked below. Please download and modify if necessary.
Validatable.zip
NHibernate.Criterion Extensions workaround
I've been working with Fluent NHibernate for the past month or so, and I realized while writing an NHibernate query that the NHibernate Criterion's Restrictions class isn't very refactor-friendly. The refactorability of Fluent NHibernate is ultimately why we've decided to use it. It does take a little longer to set up than some other ORMs, but in the long-run it's nice to be able to change a property name and refactor.
... except the property names are hard-coded in the Criteria!
My workaround for this is to use a bit of static reflection a-la-FNH, and maintain code that is easily refactored.
For instance, I've created a static utility class (Showing the Restrictions.Eq() substitute):
public static class CriterionExtensions
{
public static SimpleExpression Eq<T>(
Expression<Func<T, object>> exp, object value
)
{
var memberExpression = GetMemberExpression(exp);
string propertyName = ((PropertyInfo)memberExpression.Member).Name;
if (!string.IsNullOrEmpty(propertyName))
return Restrictions.Eq(propertyName, value);
else
return null;
}
private static MemberExpression GetMemberExpression<T>(
Expression<Func<T, object>> expression
)
{
MemberExpression memberExpression = null;
if (expression.Body.NodeType == ExpressionType.Convert)
{
var body = (UnaryExpression)expression.Body;
memberExpression = body.Operand as MemberExpression;
}
else if (expression.Body.NodeType == ExpressionType.MemberAccess)
{
memberExpression = expression.Body as MemberExpression;
}
if (memberExpression == null)
{
throw new ArgumentException("Not a member access", "member");
}
return memberExpression;
}
}
And to use this code, you can create an aliased using directive and call it in code:
using Ensure = MyNamespace.Extensions.Criterion.CriterionExtensions;
/* class declarations and whatnot */
internal IEnumerable<Product> GetAll(int shelfNumber)
{
var session = SessionManager.GetCurrentSession();
return session.CreateCriteria<Product>()
.Add(Ensure.Eq<Product>(x => x.ShelfNumber, shelfNumber))
.List<Product>();
}
It's pretty simple to use and is easily refactored. You may be able to add your *extensions* to a namespace: NHibernate.Criterion and call your class Restrictions, but I didn't try this (I don't like mixing namespaces). Let me know what you think!
CopyCat Rails’ Time Extensions
I've become a fan of Ruby's simple syntax. My favorite thing is the ActiveSupport's Numeric Time Extensions. I'm starting a library of useful extensions, and I'm going to be adding a number of shortcuts to mock these extensions.
For those of you that don't want to follow the link, check out the syntax:
# equivalent to Time.now.advance(:months => 1) 1.month.from_now # equivalent to Time.now.advance(:years => 2) 2.years.from_now # equivalent to Time.now.advance(:months => 4, :years => 5) (4.months + 5.years).from_now
Here is what I've gotten so far (it only includes int for now):
public static class DateExtensionsThatMockRuby
{
/// <summary>
/// Retrieve the number of ticks for x Days.
/// <example>long ticks = 5.Days();</example>
/// </summary>
/// <param name="i"></param>
/// <returns>ticks</returns>
public static long Days(this int i)
{
return new TimeSpan(i, 0, 0, 0).Ticks;
}
/// <summary>
/// Retrieve the number of ticks for x Hours.
/// </summary>
/// <example>long ticks = 15.Hours();</example>
/// <param name="i"></param>
/// <returns>ticks</returns>
public static long Hours(this int i)
{
return new TimeSpan(0, i, 0, 0).Ticks;
}
/// <summary>
/// Retrieve the number of ticks for x Minutes
/// </summary>
/// <example>long ticks = 97.Minutes();</example>
/// <param name="i"></param>
/// <returns>ticks</returns>
public static long Minutes(this int i)
{
return new TimeSpan(0, i, 0).Ticks;
}
/// <summary>
/// Retrieve the number of ticks for x Seconds
/// </summary>
/// <example>long ticks = 3000.Seconds();</example>
/// <param name="i"></param>
/// <returns>ticks</returns>
public static long Seconds(this int i)
{
return new TimeSpan(0, 0, i).Ticks;
}
/// <summary>
/// Retrieve the number of ticks for x Milliseconds
/// </summary>
/// <example>long ticks = 3000.Milliseconds();</example>
/// <param name="i"></param>
/// <returns>ticks</returns>
public static long Milliseconds(this int i)
{
return new TimeSpan(0, 0, 0, 0, i).Ticks;
}
/// <summary>
/// Retrieve a DateTime object from ticks
/// <example>DateTime dt = 236423690923466.AsDateTime();</example>
/// /// <example>DateTime dt = 10.Days().AsDateTime();</example>
/// </summary>
/// <param name="ticks"></param>
/// <returns>DateTime</returns>
public static DateTime AsDateTime(this long ticks)
{
return new DateTime(ticks);
}
/// <summary>
/// Retrieve a DateTime object from ticks
/// <example>DateTime dt = 236423690923466.AsDateTime(DateTimeKind.Utc);</example>
/// /// <example>DateTime dt = 10.Days().AsDateTime(DateTimeKind.Utc);</example>
/// </summary>
/// <param name="ticks"></param>
/// <returns>DateTime</returns>
public static DateTime AsDateTime(this long ticks, DateTimeKind kind)
{
return new DateTime(ticks, kind);
}
}
DRY! GenericComparer for sorting Generic Lists
I'm a pretty firm believer in the Ruby/Ruby on Rails idea of DRY ("Don't Repeat Yourself").
That said, I get pretty tired of writing comparers for sorting lists and generic lists. Every one of these comparers is exactly the same: you specify a list of properties related to the object and a sort direction, then call compare on those properties.
This can be changed with a little reflection:
public class GenericComparer<T> : IComparer<T>
{
public string SortExpression { get; set; }
public int SortDirection { get; set; } // 0:Ascending, 1:Descending
public GenericComparer(string sortExpression, int sortDirection)
{
this.SortExpression = sortExpression;
this.SortDirection = sortDirection;
}
public GenericComparer() { }
#region IComparer<T> Members
public int Compare(T x, T y)
{
PropertyInfo propertyInfo = typeof(T).GetProperty(SortExpression);
IComparable obj1 = (IComparable)propertyInfo.GetValue(x, null);
IComparable obj2 = (IComparable)propertyInfo.GetValue(y, null);
if (SortDirection == 0)
{
return obj1.CompareTo(obj2);
}
else return obj2.CompareTo(obj1);
}
#endregion
}
This is a code snippet I love to have in my arsenal.
Here's how you use it:
List<MyObject> objectList = GetObjects(); /* from your repository or whatever */
objectList.Sort(new GenericComparer<MyObject>("ObjectPropertyName", (int)SortDirection.Descending));
dropdown.DataSource = objectList;
dropdown.DataBind();
Note that Sort returns void. This means you can't throw this to the right of a DataSource call. You'd have to call this, then call your DataSource/DataBind.
Quickly create a Dictionary using Linq to Objects
When working with collections or lists, sometimes you just need to get a simple dictionary.
There are a number of ways to do this, but the ToDictionary method seems promising.
For isntance:
IList<Person> people = new List<Person>();
people.Add(new Person
{
FirstName = "Jim",
LastName="Schubert",
Age=29,
Phone="123-333-4444"
});
people.Add(new Person
{
FirstName = "Joe",
LastName="Schubert",
Age=25,
Phone="800-999-1111"
});
Dictionary<int,string> names = people.ToDictionary(k => k.Age, v => v.FirstName);
There may not be many uses for this type of syntax, but simple transformations like this make Linq fun.
Displaying multiple fields in a Dropdownlist’s DataTextField
I've encountered this problem on occasion, where I want to display more than one field in a dropdownlist's DataTextField property. In the past, I've overcome this problem by rewriting a SQL statement, or adding another column in the database itself to accomodate my needs.
In one of my classes (INFO 465: Projects in Information Systems @ VCU), we're working from a database which we're not allowed to change. The reason we can't change it is because the instructor uses the same database for his examples. I could just write another method into my business logic layer, but it would get cluttered pretty quickly.
So, I decided to make use of LINQ and found the following solution:
ddlUsers.DataSource = BLL.Employee.GetEmployees()
.Select(be =>
new {
ID = be.Id,
FullName = String.Format("{0}{1}{2}",
be.LastName,
(!string.IsNullOrEmpty(be.FirstName) ? ", " : string.Empty),
be.FirstName)
}).AsEnumerable();
This takes the List of Business Entity objects and uses the LINQ select statement to generate an implicit/anonymous object from that. The only downfall to this method is that the new object only has local scope. But, since I'm only using this in a dropdown, it's a pretty nifty trick.
T-SQL Multi-Table Delete
I was trying to run a DELETE query on multiple tables and I kept getting this error:
The DELETE statement conflicted with the REFERENCE constraint
I'm not that great with SQL, but I can get by. DELETEs have always been somethng that have given me trouble. So, I thought I'd share this simple fix.
DELETE FROM PRODUCT FROM BOOK WHERE BOOK.ID = PRODUCT.BOOK_ID AND BOOK.ID = @BOOK_ID
You just have to make sure you put the table with the FK as the first table in the query.
I guess I should have paid a little more attention in Database class. We learned Oracle, which is different from Microsoft's SQL in many ways.
