Is using dictionary good practice to keep same classes? - c#

I need to collect some classes and provide it by request to some parts of program. I have following code:
public interface ISameClass
{
int Value { get; set; }
void DoStuff();
}
public class SameClass : ISameClass
{
int Value { get; set; }
void DoStuff()
{
//Do something
}
}
public class SameClassProvider
{
private readonly Dictionary<string, ISameClass> _sameClasses;
public SameClassProvider(string parentDir)
{
_sameClasses = new Dictionary<string, ISameClass>
{
{ "Type1", new SameClass() },
{ "Type2", new SameClass() },
{ "Type3", new SameClass() }
};
}
public bool AddClass(string type, ISameClass class)
{
if (_sameClasses.ContainsKey(type) || class == null)
{
return false;
}
_nodes.Add(type, class);
return true;
}
public ISameClass GetClass(string type)
{
if (_sameClasses.TryGetValue(type, out var someClass))
{
return someClass;
}
}
}
Is using classes like SameClassProvider is good practice? Or i can refactor this something or replace with correct pattern? Tnx

Related

How can I restrict who can make changes using interfaces?

I have a IHasTag interface, a TaggableItem class that implements IHasTag and a TaggingManager class which I want to make it the only one responsible for setting or clearing the Tags but I've been struggling all day trying to implement but with no luck.
How can I make this possible?
public class TaggableItem : TaggingManager.IHasTag
{
public string Tag { get; } // read only
}
public class TaggingManager
{
public interface IHasTag
{
string Tag { get; }
}
private List<IHasTag> _taggedItems = new();
public void TagItem(IHasTag item, string tag)
{
item.Tag = tag; // not working
_taggedItems.Add(item);
}
public void ClearAllTags()
{
foreach (var item in _taggedItems)
{
item.Tag = "": // not working
}
_taggedItems.Clear();
}
}
EDIT
I followed Thomas' suggestion and this is what I end up doing. I know it's not perfect, though. Thank you all for your advices.
public interface ITaggable
{
string? Tag { get; }
}
public interface ISelectable
{
bool IsSelected { get; }
}
public interface IItem : ITaggable, ISelectable
{
}
public class Item : IItem
{
protected Item() { }
public bool IsSelected { get; set; }
public string Tag { get; set; } = string.Empty;
// 'Item' will be created here returning IItem.
// So, unless you 'cast' it, you can't set 'Tag' or 'IsSelected'.
public static IItem CreateItem() => new Item();
}
public class SelectionManager
{
protected List<object> _items = new();
public void Select(ISelectable item)
{
if (item is Item selectable)
{
selectable.IsSelected = true;
_items.Add(item);
}
}
public void Unselect(ISelectable item)
{
if (item is Item selectable)
{
selectable.IsSelected = false;
_items.Remove(item);
}
}
}
public class TaggingManager
{
private List<object> _items = new();
public void Tag(ITaggable item, string tag)
{
if (item is Item taggable)
{
taggable.Tag = tag;
_items.Add(item);
}
}
public void Untag(ITaggable item)
{
if (item is Item taggable)
{
taggable.Tag = string.Empty;
_items.Remove(item);
}
}
}
My suggestion would be to have two interfaces for two purposes: reading and writing.
public interface IHasTag // interface for reading. Maybe IReadTag
{
string Tag { get; }
}
public interface ITagChange // interface for writing. Maybe IWriteTag
{
string Tag { set; }
}
public class TaggableItem : IHasTag, ITagChange // implement both
{
public string Tag { get; set; }
}
// Tagging manager gets write access (ITagChange)
public class TaggingManager
{
private List<ITagChange> _taggedItems = new ();
public void TagItem(ITagChange item, string tag)
{
item.Tag = tag;
_taggedItems.Add(item);
}
public void ClearAllTags()
{
foreach (var item in _taggedItems)
{
item.Tag = "";
}
_taggedItems.Clear();
}
}
// Everyone else has read access only (IHasTag)
class SomeoneElse
{
private List<IHasTag> _taggedItems = new ();
public void DoSomething(IHasTag item)
{
_taggedItems.Add(item);
var tag = item.Tag; // do something with the tag
}
}
class Instantiation
{
public void Main()
{
TaggableItem x = new TaggableItem();
TaggingManager m = new TaggingManager();
m.TagItem(x, "name");
SomeoneElse s = new SomeoneElse();
s.DoSomething(x);
}
}
One possible option is to move the "getter" also to the TagManager. Then the manager is responsible for the tags. The object itself does not even need to know about the tags.
You still can restrict this by exchanging object with an interface.
public class TagManager
{
private Dictionary<object, string> _tagedItems = new Dictionary<object, string>();
public bool HasTag(object item)
{
return _tagedItems.ContainsKey(item);
}
public string GetTag(object item)
{
return _tagedItems[item];
}
public void SetTag(object item, string tag)
{
if(!HasTag(item))
{
_tagedItems.Add(item, tag);
}
else
{
_tagedItems[item] = tag;
}
}
}
I don't think that this is really a answer to the OP, but one possible solution for the underlying problem.

Create game inventory system, without casting to derived

I am trying to implement a high-performance game inventory system. I have This abstract base class to store different type of items in Inventory, for example, Coin, Flashlight, Knife etc..
public abstract class ObtainableItem
{
public string Name { get; private set; }
public ObtainableItem(string name)
{
Name = name;
}
}
For example, I have a DoorKey which opens a door. DoorKey has a property KeyCode which will be used for opening a door.
public class DoorKey : ObtainableItem
{
public int KeyCode { get; private set; }
public DoorKey() : base("key")
{
KeyCode = 1234;
}
}
All ObtainableItem are stored in Inventory
public class Inventory
{
const int slotCount = 2;
ObtainableItem[] slots = new ObtainableItem[slotCount];
public Inventory()
{
slots[0] = new DoorKey();
}
}
Now imagine user drags DoorKey from his Inventory on a Door and triggers Open method
public class Door
{
public void Open(ObtainableItem key)
{
if (key is DoorKey)
{
DoorKey doorKey = (DoorKey)key;
if (doorKey.KeyCode == 1234)
{
// Open door
}
}
else
{
// "can't use this item on a door"
}
}
}
How to avoid cast from ObtainableItem to a DoorKey? I have read that using casting is bad practice and it points at a bad code oop design. Ideally, a Door class should look like this. Is there any pattern I should for my inventory system?
public class Door
{
public void Open(DoorKey key)
{
if (key.KeyCode == 1234)
{
// Open door
}
}
}
There are always exceptions that can be made for ease of implementation and readability. What you describe is common, if not typical.
An alternative would be to have the "control" logic in the class that calls Door.Open. This could be easily achieved with a touch of reflection:
public abstract class ObtainableItem
{
public string Name { get; private set; }
public ObtainableItem(string name)
{
Name = name;
}
}
public abstract class WorldItem
{
}
public interface IActsOn<in TWorldItem>
where TWorldItem : WorldItem
{
void ApplyTo(TWorldItem worldItem);
}
public class World
{
// If profiling shows that this is a performance issue, a cache keyed by tWorldItem, tInvItem
// should fix it. No expiry or invalidation should be needed.
private Action<ObtainableItem, WorldItem> GetApplyTo(Type tWorldItem, Type tInvItem)
{
var tActOn = typeof(IActsOn<>).MakeGenericType(tWorldItem);
if (!tActOn.IsAssignableFrom(tInvItem))
{
return null;
}
var methodInfo = tActOn.GetMethod(nameof(IActsOn<WorldItem>.ApplyTo));
return new Action<ObtainableItem, WorldItem>((invItem, worldItem) =>
{
methodInfo.Invoke(invItem, new object[] { worldItem });
});
}
public bool IsDropTarget(WorldItem worldItem, ObtainableItem item)
=> GetApplyTo(worldItem.GetType(), item.GetType()) != null;
public void ActOn(WorldItem worldItem, ObtainableItem item)
{
var actOn = GetApplyTo(worldItem.GetType(), item.GetType());
if (actOn == null)
{
throw new InvalidOperationException();
}
actOn(item, worldItem);
}
}
While this slightly complicates the implementation of World, it simplifies the implementation of various objects:
class Door : WorldItem
{
public void Unlock(string bitting)
{
if (bitting == "1234")
{
Console.WriteLine("Door Opened");
}
else
{
Console.WriteLine("Door could not unlock");
}
}
}
class DoorKey : ObtainableItem, IActsOn<Door>
{
private readonly string Bitting;
public DoorKey(string bitting)
: base("Key")
{
this.Bitting = bitting;
}
public void ApplyTo(Door worldItem)
{
worldItem.Unlock(this.Bitting);
}
}
class RubberChicken : ObtainableItem
{
public RubberChicken()
: base("Rubber chicken")
{
}
}
Example usage:
class Program
{
static void Main(string[] args)
{
var key1 = new DoorKey("1234");
var key2 = new DoorKey("4321");
var rubberChicken = new RubberChicken();
var door = new Door();
var world = new World();
Debug.Assert(!world.IsDropTarget(door, rubberChicken));
Debug.Assert(world.IsDropTarget(door, key1));
world.ActOn(door, key2);
world.ActOn(door, key1);
Console.ReadLine();
}
}

Avoid having to cast with generics and a list of custom data

I have the following code for supporting a list of different types :
public enum eType
{
tInt,
tString,
tDateTime
}
public interface ICustomType<out T>
{
T Value { get; }
}
public abstract class DifferentType
{
protected DifferentType(eType type, string mnemonic)
{
Type = type;
Mnemonic = mnemonic;
}
public string Mnemonic { get; private set; }
public eType Type { get; private set; }
}
public class DateTimeType : DifferentType, ICustomType<DateTime>
{
public DateTimeType(DateTime value, string mnemonic)
: base(eType.tDateTime, mnemonic)
{
Value = value;
}
public DateTime Value { get; private set; }
}
public class IntType : DifferentType, ICustomType<int>
{
public IntType(int value, string mnemonic)
: base(eType.tInt, mnemonic)
{
Value = value;
}
public int Value { get; private set; }
}
public class StringType : DifferentType, ICustomType<string>
{
public StringType(string value, string mnemonic)
: base(eType.tString, mnemonic)
{
Value = value;
}
public string Value { get; private set; }
}
public static class UtilValue
{
public static T GetValue<T>(DifferentType customType)
{
return ((ICustomType<T>)customType).Value;
}
}
public class testTypes2
{
public testTypes2()
{
var values = new List<DifferentType> { GetInt(), GetString(), GetDate() };
foreach (var i in values)
{
switch (i.Type)
{
case eType.tInt:
int resInt = UtilValue.GetValue<int>(i);
break;
case eType.tString:
string resString = UtilValue.GetValue<string>(i);
break;
case eType.tDateTime:
DateTime resDateTime = UtilValue.GetValue<DateTime>(i);
break;
}
}
}
private DateTimeType GetDate()
{
return new DateTimeType(new DateTime(2000, 1, 1), "MnemonicDate");
}
private IntType GetInt()
{
return new IntType(5, "MnemonicInt");
}
private StringType GetString()
{
return new StringType("ok", "MnemonicString");
}
}
and would like to avoid the cast at line return ((ICustomType<T>)customType).Value; in the UtilValue class, any idea how I can get rid of that while still keeping the design?
I am not even sure if this cast is expensive to do? My guess is most certainly.
Visitor-pattern example:
interface IDifferentTypeVisitor
{
void Visit(DateTimeType dt);
void Visit(StringType st);
}
class DifferentType
{
public abstract void Accept(IDifferentTypeVisitor visitor);
}
class DateTimeType : DifferentType
{
public void Accept(IDifferentTypeVisitor visitor)
{
visitor.Visit(this);
}
}
class StringType : DifferentType
{
public void Accept(IDifferentTypeVisitor visitor)
{
visitor.Visit(this);
}
}
class SomeVisitor : IDifferentTypeVisitor
{
public void Visit(DateTimeType dt)
{
//DateTime resDateTime = dt.Value; Or similar
}
public void Visit(StringType st)
{
//string resString = st.Value; Or similar
}
}
public class testTypes2
{
public testTypes2()
{
var values = new List<DifferentType> { /* Content */ };
var visitor = new SomeVisitor();
foreach (var i in values)
{
i.Accept(visitor);
}
}
}
In C# 4 with dynamic it's possible to save some code by adding this to DifferentType:
public void Accept(IDifferentTypeVisitor visitor)
{
visitor.Visit((dynamic)this);
}
and then delete all other Accept methods. It hurts performance but it looks better ;-)

Stack overflow error in singleton pattern

I have implemented Single Pattern. Here is my code i am getting the an error when i call the Test.BuildData() function. Please help
public class WordDataItem
{
public string Word { get; set; }
public string Definition { get; set; }
public int WordGroupKey { get; set; }
}
public class WordDataGroup
{
public List<WordDataItem> listItem = new List<WordDataItem>();
public int GroupKey { get; set; }
}
public sealed class WordDataSource
{
private static WordDataSource _dataSoruce;
private List<WordDataGroup> listGroup = new List<WordDataGroup>();
public List<WordDataGroup> ListGroup
{
get { return listGroup; }
set { listGroup = value; }
}
private WordDataSource() { }
public static WordDataSource Instance
{
get
{
if (Instance == null)
{
_dataSoruce = new WordDataSource();
}
return _dataSoruce;
}
}
}
public static class Test
{
public static void BuildData()
{
WordDataSource.Instance.ListGroup.Add(new WordDataGroup() { GroupKey = 8, listItem = new List<WordDataItem>() { new WordDataItem() {Word = "Hello", Definition="Greetings", WordGroupKey = 8}} });
}
}
I get an error of stack over flow when i call the Test.BuildData() function.
Your Instance property is recursively calling into itself when you check if it is null.
Try this:
public static WordDataSource Instance
{
get
{
if (_dataSoruce == null)
{
_dataSoruce = new WordDataSource();
}
return _dataSoruce;
}
}

Recursive generic types

Is it possible to define a generic type in C# that references itself?
E.g. I want to define a Dictionary<> that holds its type as TValue (for a hierarchy).
Dictionary<string, Dictionary<string, Dictionary<string, [...]>>>
Try:
class StringToDictionary : Dictionary<string, StringToDictionary> { }
Then you can write:
var stuff = new StringToDictionary
{
{ "Fruit", new StringToDictionary
{
{ "Apple", null },
{ "Banana", null },
{ "Lemon", new StringToDictionary { { "Sharp", null } } }
}
},
};
General principle for recursion: find some way to give a name to the recursive pattern, so it can refer to itself by name.
Another example would be generic tree
public class Tree<TDerived> where TDerived : Tree<TDerived>
{
public TDerived Parent { get; private set; }
public List<TDerived> Children { get; private set; }
public Tree(TDerived parent)
{
this.Parent = parent;
this.Children = new List<TDerived>();
if(parent!=null) { parent.Children.Add(this); }
}
public bool IsRoot { get { return Parent == null; } }
public bool IsLeaf { get { return Children.Count==0; } }
}
Now to use it
public class CoordSys : Tree<CoordSys>
{
CoordSys() : base(null) { }
CoordSys(CoordSys parent) : base(parent) { }
public double LocalPosition { get; set; }
public double GlobalPosition { get { return IsRoot?LocalPosition:Parent.GlobalPosition+LocalPosition; } }
public static CoordSys NewRootCoordinate() { return new CoordSys(); }
public CoordSys NewChildCoordinate(double localPos)
{
return new CoordSys(this) { LocalPosition = localPos };
}
}
static void Main()
{
// Make a coordinate tree:
//
// +--[C:50]
// [A:0]---[B:100]--+
// +--[D:80]
//
var A=CoordSys.NewRootCoordinate();
var B=A.NewChildCoordinate(100);
var C=B.NewChildCoordinate(50);
var D=B.NewChildCoordinate(80);
Debug.WriteLine(C.GlobalPosition); // 100+50 = 150
Debug.WriteLine(D.GlobalPosition); // 100+80 = 180
}
Note that you cannot directly instantiate Tree<TDerived>. It has to be a base class to the node class in the tree. Think class Node : Tree<Node> { }.

Categories