😺

C#/ASP.NET Core Advanced features

2024/04/30に公開

Abstruct

This article provides the insights regarding features of C# or ASP.NET Core to embody the code with clean architecture and in high performance and security.

Clean Code

Interface and Abstruct class

Both abstruct and encupslate their implementation and what the classes should provide to external components

Interface

Defines what members the class should have
- method, property, indexer, event
Need to implement the details of the members
Useful for:
- DI
- Mock a class for unit test
- Implementation of Strategy Pattern to reduce unnecessary conditional branchs.

interface IVehecleOperation
{
    void Run()
}

class Car : IVehecleOperation
{
    // Explicit interface member implementation:
    void Run()
    {
        // Method implementation.
    }
}

Abstruct class / method

A super class doesn't include its implementation
Cannot instanciate it.
Useful for:
- provide common functionality and interface to its subclass

abstract class Vehecle
{
    public abstract void Run();
}

class Car : Vehecle
{
    public override void Run()
    {
        // Method implementation.
    }
}

Comparison

	Abstruct Class	Interface
Inheritance	single	multiple
access modifier	public, protected	public
method definition	Can implement	Cannnot implement
member variable	static/instance fields	static fields

Generics

Functionality for a code independent on types

Useful for:
- Reuse a code in a variety of Type
- Implementation for collection
- Keeping type safe in compile

// T: Type
// where T: condition of the type
static T Max<T>(T a, T b)
  where T : IComparable
{
  // a and b are secured to have IComparable interface
  // Thus a and b has CompareTo() 
  return a.CompareTo(b) > 0 ? a : b;
}

Conditions pettern

condition	meaning	note
where T : struct	Value Type
where T : class	Reference Type
where T : notnull	Not Null
where T : Enum	Enum	can use with "struct"
where T : Delegate	Delegate
where T : [base class]	Inherites base class
where T : [interface]	Has interface
where T : new()	Have constructer without arguments	Should place the last

Can use multiple condtitions separated commas(,)
Cannot specify duplicated or contradicted conditions
- e.g) where T: struct, class

Access modifiers

Encupslate class/members and protect its internal

modifiier	scope	usage
public	anyware OK	interfaces for external access
internal	only in the assembly(project)	as default usage
protected	the class and its subclass	inheritance
private	only in the class	encupslation

Default is private

in/readonly/required modifier

They are useful for prevention from invalid value in the class.

modifier	target	effect
in	argument	prohibit reassignment to an argument
readonly	variable	prohibit reassignment after initialisation
required	parameter	set the pamameter required to initialise

ref/out

Both are the function to pass argument to output result.

modifier	initialization in caller	assign in the callee
ref	Nessesary	Unnecessary
out	Unnessesary	Nessesary

Nullable reference type

NOT Nullable as default in reference time
- But not throw error, just alert it on IDE
Need to add ? modifier to its type to set it Nullable
- e.g) string? descripton;

Delegation

Functionality to pass a method to another method.

// Define delegate type
delegate void SingleDelegate(int a);
delegate void MultiDelegate(string b);

class DelegateTest
{
  static void Main()
  {
    // assign a method
    SingleDelegate single = new SingleDelegate(SayAge);
    single(18); // call the method via delegate

    // Multi deligation
    MultiDelegate multi = new MultiDelegate(SayName);
    multi += new MultiDelegate(SayHello);
    // sequential excecution SayName -> SayHello 
    multi("John"); 
  }

  static void SayAge(int age)
  {
    Console.WriteLine($"I'm {age} years old.");
  }
  static void SayName(string name)
  {
    Console.WriteLine($"My name is {name}.");
  }
  static void SayHello(string name)
  {
    Console.WriteLine($"Hello {name}.");
  }
}

Can delegate both static/instance methods.
Can delegate multiple methods
- execute sequentially by FIFO order
Userful for:
- Event handler
- Reuse code by replacing only method

record

Something like immutable class. Generates such as Equal or ToString automatically after compile.

Useful for:
- Value Object in DDD

Comparison: class/struct/record

	type	usage
class	reference type	complex and mutable structure which needs modification
struct	value type	simple and mutable structure which doesn't needs modification
record	value type	the same as class but immutable

Index and range

Convenient index tips to express more consice.

	meaning	note
^i	i th index from the last	zero start
i..j	i th to j-1 th indices	include i but exclude j
i..	i th index to the last
..j	0 th index to j-1 th

DI

Needs interface to inject
Configure in Program.cs
Inject to constructor
Useful for:
- Mock injection in unit test
- Loose coupling
- Reusability

Program.cs

var builder = WebApplication.CreateBuilder(args);

builder.Services.AddTransient<IUserSessionService, UserSessionService>();
builder.Services.AddScoped<ICartService, CartService>();
builder.Services.AddSingleton<IGlobalSettingService, GlobalSettingService>();

Each methods provides with diffirent lifetime of instance

method	from		to
AddTransient	Every access of the service	->	End of the instance, becoming the target of GC
AddScoped	Start of HTTP request	->	End of HTTP request
AddSingleton	Start of application	->	End of application

Constructor

public class ECommerse {
  ECommerse(IUserSessionService userSessionService,
      ICartService cartService,
      IGlobalSettingService globalSettingService) {
        ....
    }
}

Reflection

Functionality to fetch or modify metadata of class in runtime.

Type type = typeof(DateTime); // Get type name from Class
Type type = obj.GetType(); // Get type name from Object
PropertyInfo prop = type.GetProperty("MyProperty"); // Get property from type
prop.SetValue(obj, 987); // Set value to property

*Useful for:
* Debug/Unit test
* The development of generic libraries or frameworks

High Performance

Boxing and Unboxing

Conversion between Value type and Reference type.
Could cause unnecessary overhead and shouldn't use without reasons.

	type	memory		type	memory
Boxing	Value type	stack	->	Reference type	heap
Unoxing	Reference type	heap	->	Value type	stack

Boxing

int intValue = 42;
object objValue = intValue; // Boxing

Unboxing

object objValue = 42;
int intValue = (int)objValue; // Unboxing

Finalizer and Dispose pattern

Code pattern for your unique class to open in using() variables.
Implement IDisposable interface.

Managed resource: Managed by .NET Framework
- e.g.) System.IO.File class
- Should dispose when Dispose() is called
Unmanaged resource:
- e.g.) Windows API to operate more detailed functionality for files
- Should dispose when Finalizer is called
- High cost to dispose resource here
  - Don't use it as much as possible

Implementation

// Implement IDisposable interface
class Resource : IDisposable
{
    public void Dispose()
    {
        Dispose(true);
        GC.SuppressFinalize(this);
    }

    protected virtual void Dispose(bool disposing)
    {
        if (disposing)
        {
            // Dispose managed resource 
        }
        // Dispose unmanaged resource
    }

    // finalizer
    ~Resource()
    {
        Dispose(false);
    }
}

Usage

using(var resource = new Resource()){
  // process resource
}

Loop

Use StringBuilder instead of string to concatnate strings.
Use Dictionary instead of List when you need fast search in loop
- Time complexity(Dictionary) -> O(1)
- Time complexity(List) -> O(n)
Use Parallel.For or Parallel.ForEach when you need to precess nested loop
- It compute each row parallelly

Parallel.For(0, digits.Count, (i) => {
    for (int j = 0; j < digits.Count; j++) {
        var a = j;
    }
});

unsafe

A keyword to use pointer. Usually no need to use but could be the last option for performance tuning.

Useful for:
- To interporate with pointer language like C
- Fine-tuning for performance

// unsafe method
unsafe struct X
{
  byte x;
  int y;
}
unsafe void UnsafeMethod()
{
  // can use pointer
  var p = new Point();
  var pp = &p;
  int* pi = (int*)pp;
  *pi = 0x00010002;

  // keep array on stack
  int* x = stackalloc int[N];

  // fetch the size of struct
  Console.WriteLine(sizeof(X));

  // fix address not to move in GC
  var array = new[] { 1, 2, 3, 4, 5 };
  fixed (int* px = array)
  {
    for (int* p = px; p != px + array.Length; ++p)
      *p = (*p) * (*p);
  }
}

void SafeMethod()
{
  // cannot use pointer

  // unsafe block
  unsafe
  {
    // need unsafe block if you use pointer or call unsafe mehtod
    UnsafeMethod()
  }
}

.csproj

<Project Sdk="Microsoft.NET.Sdk">
 
  <PropertyGroup>
    <TargetFramework>net5.0</TargetFramework>
    <!-- enable unsafe feature -->
    <AllowUnsafeBlocks>true</AllowUnsafeBlocks>
  </PropertyGroup>
 
</Project>

Span<T>

Functionality to read/write sequential data without allocating memory.

Sequential data
- Array
- String
- Binary
Useful for:
- Efficient memory performance without copy or memory allocation
- To create library which operates pointer but to call without unsafe statement.
- Avoiding GC by the combination with stackalloc.

// creation
var span = new Span<int>(array, 2, 3);
var span = array.AsSpan().Slice(2, 3);

// Avoid unsafe block for stackalloc
// No need to use unsafe block
Span<byte> buffer = stackalloc byte[BufferSize];

// callee
// using Span<> in argument
static void Clear(Span<byte> span)
{
    unsafe
    {
        fixed (byte* pin = span)
        {
            // process
        }
    }
}

// caller
static void Main()
{
    var array = new byte[256];
    // No need to use unsafe block
    Clear(array);
}

Generics

You can generate new instance of the generic time.
But it's so slow due to using runtime type information.
Better to avoid when you have severe performance requirement.
It's possibly a better option to receive generater function.

Slow

static T[] Array<T>(int n)
    where T : new()
{
    var array = new T[n];
    for (int i = 0; i < array.Length; i++)
        array[i] = new T(); // 10 times slower than usual new()
    return array;
}

Fast

static T[] Array<T>(int n, Func<T> generator)
{
  var array = new T[n];
  for (int i = 0; i < array.Length; i++)
    array[i] = generator(); // call generator
  return array;
}

// Pass generator from caller
Array(1000, () => new MyClass());

Async programming

Background processing

a.k.a Non-blocking processing. Asyncronously process between UI and Time-consuming process.

Use sugarcoat syntax of async / await
- Makes the steps serial and flat as sync processing
Useful for:
- Securing responsiveness on UI thread from time-consuming tasks
  - I/O processing
  - Network processing
  - DB access

public void DoSomething(string uri) {
  var url = new Uri(uri);
  // add await to async method
  await ProcessAsync(url);
}

// add async keyword
// return type should be Task, Task<T> or void
// append Async suffix for async method
private async Task<boolean> ProcessAsync(Uri resource)
{
    var client = new WebClient();
    // add await to async method
    return await client.DownloadAsync(resource);
}

Parallel processing

Data parallelism

Execute the same processing to the diferrent dataset in multi threads

Suitable for the independent dataset
Unsuitable when the dataset needs:
- To be reordered after all
- To use the result of other threads

Parallel

// single thread
for (var i = 0; i < source.Length; i++)
{
    result[i] = selector(source[i]);
}

// multi threads
Parallel.For(0, source.Length, i =>
{
    result[i] = selector(source[i]);
});

LINQ

// single thread
var result = source.Select(selector);

// multi threads
var result = source.AsParallel().Select(selector);

Task parallelism

Exchange data among independent tasks simaltaniously.
The most simple way is using TPL(Task Parallel Library) to implement.

// create tasks
Task task1 = Task.Run(() =>
{
    // procces in task1
    return 1;
});
Task task2 = Task.Run(() =>
{
    // procces in task2
    return 2;
});

// wait for all task to be completed
Task.WhenAll(task1, task2);

// Aggregate the results of tasks
var res1 = task1.Result
var res2 = task2.Result;
var totalRes = res1 + res2;

Security

SQL Injection

An attack to inject malicious query through user input mechanism.

Solution

API

Entity Framework
- Uses parameterised query inside
- Could be solution for SQL Injection if you can use it.
Parameterised query

string query = "SELECT * FROM Users WHERE Username = @Username AND Password = @Password";
SqlCommand cmd = new SqlCommand(query, connection);
cmd.Parameters.AddWithValue("@Username", username);
cmd.Parameters.AddWithValue("@Password", password);

XSS

An attack to inject malicious script into application.
Usually Javascript is used for script via <script> tag in html.

Solution

API

Input Validation
- Verify the input value and pass only genuine value

UI

HTMLEncode
- Encode HTML string
  - e.g) "<\"123\">" -> <"123">
  - Use System.Text.Encodings.Web.HtmlEncoder
Razor
- Razor engine used in MVC automatically encodes all output sourced from variables

CSRF

An malicious attack to deceive the users to send malicious requests via their fraud web pages and operate the resources illigally.

Solution

API

Add X-CSRF-TOKEN to HTTP Header which set on its Cookie
Same-origin policy
- But it disturbs interaction between your own system
- You may adjust acceptable origins by CORS mechanism

UI

Avoid using GET
Use AntiForgeryToken

Others

.NET Framework vs .NET Core

	.NET Framework	.NET Core
Use case	Existing Enterprize system, Desktop application for Windows	Cloud Native application, Containerlized application
Platform	Only Windows	Cross platform (Windows、Linux、macOS)
Life Cycle	Stable	Frequently updated
Open Source	Partially	Fully

Discussion

Okamasa

thank you