We Keep Coding

What is the difference between declaring an explicit generic parameter and use a constraint?

Which is the difference between doing this:
public static bool IsNullOrEmpty<TKey, TValue>
(this Dictionary<TKey, List<TValue>> dictionary, TKey key)
{
return !dictionary.ContainsKey(key) ||
dictionary.ContainsKey(key) && dictionary[key].Count == 0;
}
And this:
public static bool IsNullOrEmpty<TKey, TValue>
(this Dictionary<TKey, TValue> dictionary, TKey key)
where TValue : List<TValue>
{
return !dictionary.ContainsKey(key) ||
dictionary.ContainsKey(key) && dictionary[key].Count == 0;
}
As far as I can notice, the compiler doesn't tell me that there isn't anything wrong. But, is this any approach better than the other? Will it have any different return value (because I don't realize about this yet)?

First off: your best bet for answering this question yourself would have been try it. You would have soon seen that when you attempted to call the function in its second form, it does not work.
That said, let's dig into this a bit. We have
public static bool IsNullOrEmpty<TKey, TValue>(
this Dictionary<TKey, List<TValue>> dictionary, TKey key)
versus
public static bool IsNullOrEmpty<TKey, TValue>(
this Dictionary<TKey, TValue> dictionary, TKey key)
where TValue : List<TValue>
Why is the second one wrong? Well, what type argument would you like to pass for TValue? Suppose we have a Dictionary<string, List<int>> in hand. What is the TValue we can use? It's not int, because that doesn't meet the constraint: int does not derive from List<int>. But it is not List<int> either because List<int>> does not derive from List<TValue>, which is List<List<int>>.
So, now we know why the second one is wrong. Let's now answer more questions:
Under what circumstances does this sort of "recursive" constraint make sense?
Suppose we are trying to find the maximum key in a dictionary:
public static TKey MaxKey<TKey, TValue>(
this Dictionary<TKey, TValue> dictionary)
where TKey : IComparable<TKey>
{
if (dictionary.Count == 0) throw ...
TKey best = default(TKey);
bool first = true;
foreach(TKey k in dictionary.Keys)
{
if (first || best.CompareTo(k) < 0)
best = k;
first = false;
}
return best;
}
Here it totally makes sense to constrain TKey to IComparable<TKey>; we're going to compare keys.
What are some other ways that people use and misuse this pattern?
See https://blogs.msdn.microsoft.com/ericlippert/2011/02/03/curiouser-and-curiouser/ for many examples.
Why is the first method less than perfect?
Because it does not handle these two cases:
First, suppose we have a multidictionary but it is not from keys to lists:
Dictionary<string, int[]>
or
Dictionary<string, Stack<int>>
or
Dictionary<string, IEnumerable<int>>
or whatever.
Second, it also does not handle the case
class MyList : List<int> {}
...
Dictionary<string, MyList>
though that case is pretty rare; you ought not to be extending List<T> normally.
What is the right way to implement my dictionary method for maximum generality?
Here's one way:
public static bool IsEmpty(this IEnumerable items)
{
// EXERCISE: Why is this implementation bad?
// EXERCISE: Can you improve it?
foreach(var item in items)
return false;
return true;
}
public static bool IsNullOrEmpty<TKey, TValue>(
this Dictionary<TKey, TValue> dictionary, TKey key)
where TValue : IEnumerable
{
return !dictionary.ContainsKey(key) || dictionary[key].IsEmpty();
}
For even more generality you might use IDictionary instead of Dictionary.

Extension method for dictionary with value implementing interface

I'm trying to implement an extension method for all dictionaries whose value is of a type that implement a certain interface.
In this case I would like to have a ToListSortedByValue() method that returns
List<KeyValuePair<string, IComparable>>
for any dictionary of type
Dictionary<string, IComparable>
that would be cool because it would allow me to use dictionaries instead of lists, but to be able to have them sorted when needed (for example when printing in files or at console).
This is what I tried, but it doesn't work, any idea why?
public static List<KeyValuePair<string, IComparable>> ToListSortedByValue(this Dictionary<string, IComparable> Dic)
{
return Dic.OrderBy(x => x.Value).ToList();
}
EDIT:
it's solved already, but for completeness sake this is the problem I got:
when trying to use the method I got the an error as if such method didn't exist. If instead of IComparable I use an actual comparable type, let's say int or a class implementing IComparable, than it would work.

Basically you need to make the method generic on the value type and then constrain that type to be IComparable<T>.
public static List<KeyValuePair<string, T>> ToListSortedByValue<T>(
this Dictionary<string, T> Dic) where T : IComparable<T>
{
return Dic.OrderBy(x => x.Value).ToList();
}
This has the added bonus of returning the values as there passed in type. You might even want to make the key type generic too so it's not limited to just string
public static List<KeyValuePair<TKey, TValue>> ToListSortedByValue<TKey, TValue>(
this Dictionary<TKey, TValue> Dic) where TValue : IComparable<TValue>
{
return Dic.OrderBy(x => x.Value).ToList();
}

You need to make your method generic, so that it extends your actual type instead of just IComparable:
public static List<KeyValuePair<string, T>> ToListSortedByValue<T>(this Dictionary<string, T> Dic) where T : IComparable<T>

Case Insensitive Dictionary with Tuple Key

I have a dictionary where the key is a Tuple where the first item is a Date and the second item is a string. I would like the dictionary to be case insensitive.
I know that if the key was just a string I could pass StringComparer.OrdinalIgnoreCase as a parameter when declaring the dictionary, but this does not seem to work when the key is a Tuple.
Is there some way to specify the StringComparer to use on the second item of the Tuple?
Thanks

Use this overload of the Dictionary constructor, which allows you to specify a custom comparer for the keys. You would accompany this with creating a class that implements
IEqualityComparer<Tuple<string, DateTime>>
Which might look like this:
class CustomEqualityComparer : IEqualityComparer<Tuple<string, DateTime>>
{
public bool Equals(Tuple<string, DateTime> lhs, Tuple<string, DateTime> rhs)
{
return
StringComparer.CurrentCultureIgnoreCase.Equals(lhs.Item1, rhs.Item1)
&& lhs.Item2 == rhs.Item2;
}
public int GetHashCode(Tuple<string, DateTime> tuple)
{
return StringComparer.CurrentCultureIgnoreCase.GetHashCode(tuple.Item1)
^ tuple.Item2.GetHashCode();
}
}
There are no argument checks here, so please don't treat this as production code. Also, care needs to be taken so that the Equals and GetHashCode implementations satisfy the all-important condition that if two tuples compare equal, they must have the same hash code. When dealing with custom text comparisons it is easy to introduce bugs if not extra careful: for example, using ToLowerInvariant instead of ToLower above would be a bug (albeit one that might not surface for some time).

I needed this in a Dictionary<Tuple<>> wrapper, so I used #Jon 's code to create a generic version
public class TupleEqualityComparer<T1, T2> : IEqualityComparer<Tuple<T1, T2>>
{
private IEqualityComparer<T1> comparer1;
private IEqualityComparer<T2> comparer2;
public TupleEqualityComparer(IEqualityComparer<T1> comparer1, IEqualityComparer<T2> comparer2)
{
this.comparer1 = comparer1 ?? EqualityComparer<T1>.Default;
this.comparer2 = comparer2 ?? EqualityComparer<T2>.Default;
}
public bool Equals(Tuple<T1, T2> lhs, Tuple<T1, T2> rhs)
{
return comparer1.Equals(lhs.Item1, rhs.Item1) && comparer2.Equals(lhs.Item2, rhs.Item2);
}
public int GetHashCode(Tuple<T1, T2> tuple)
{
return comparer1.GetHashCode(tuple.Item1) ^ comparer2.GetHashCode(tuple.Item2);
}
}
public class Dictionary<TKey1, TKey2, TValue> : Dictionary<Tuple<TKey1, TKey2>, TValue>()
{
public Dictionary() : base() { }
public Dictionary(IEqualityComparer<TKey1> comparer1, IEqualityComparer<TKey2> comparer2) : base(new TupleEqualityComparer<TKey1, Tkey2>(comparer1, comparer2) { }
public TValue this[TKey1 key1, TKey2 key2]
{
get { return base[Tuple.Create(key1, key2)]; }
}
public void Add(TKey1 key1, TKey2 key2, TValue value)
{
base.Add(Tuple.Create(key1, key2), value);
}
public bool ContainsKey(TKey1 key1, TKey2 key2)
{
return base.ContainsKey(Tuple.Create(key1, key2));
}
public bool TryGetValue(TKey1 key1, TKey2 key2, out TValue value)
{
return base.TryGetValue(Tuple.Create(key1, key2), out value);
}
}
Usage
var dict = new Dictionary<string, DateTime, int>(
StringComparer.OrdinalIgnoreCase, null);
dict.Add("value1", DateTime.Now, 123);
Assert.IsTrue(dict.ContainsKey("VALUe1"));

Since the comparisons are going to be case-insensitive, you could use the toLower/toUpper method in the string side when making the tuples, and then always lower or upper the strings you'll have in the tuples used to retrive/compare entries in the dictionary.

Building a generic collection class

I am building the following class to manage a dictionary.
public class EnumDictionary<TKey, TValue>
{
private Dictionary<TKey, TValue> _Dict;
public EnumDictionary(Dictionary<TKey, TValue> Dict)
{
this._Dict = Dict;
}
public TKey GetValue(TValue value)
{
foreach (KeyValuePair<TKey, TValue> kvp in _Dict)
{
if (kvp.Value == value)
return kvp.Key;
}
throw new Exception("Undefined data type: " + value);
}
}
But I am getting an error "Operator '==' cannot be applied to operands of type 'TValue' and 'TValue'".
BTW, I am making this custom collection is because my dictionary has unique value, but I can't get key by value from a dictionary.
Any help is appreciated. Thank you.

Did you try using the Equals method?
if (kvp.Value.Equals(value))
I think this restriction is due to the fact that the == operator can't be used with all types. Take the following for instance:
struct Test
{
public int Value;
}
Given the above struct, the following code will not compile:
Test a, b;
a = b = new Test();
bool areEqual = a == b; // Operator '==' cannot be applied to
// operands of type 'Test' and 'Test'
However, all types have the Equals method, so calling that will work:
Test a, b;
a = b = new Test();
bool areEqual = a.Equals(b);

Fredrik is right; you need to use Equals as you can't presume to be able to use == for all types, since the operator isn't defined for every type.
Depending on your scenario, it might also make sense to add
where TValue : IEquatable<TValue>
as a generic type constraint on your class. The reason for this is that object.Equals accepts another object as a parameter, which means that if TValue is a value type it will be boxed. If it can be known to implement IEquatable<TValue>, on the other hand, then Equals can be resolved to IEquatable<TValue>.Equals*, which takes a TValue as a parameter and therefore won't require value types to be boxed.
I might also recommend that you rethink the internal structure of this class. As it currently stands, there's no reason you need this class at all, as you could easily add an extension method to IDictionary<TKey, TValue> to find a key by value via enumeration over the values. What I would do instead is store two dictionaries: a Dictionary<TKey, TValue> and a Dictionary<TValue, TKey>, so that two-way lookup is possible in O(1).
*By the way, in case you're curious, the reason you can't use IEquatable<T> (or any interface for that matter) to ensure that a type has implemented the == operator is that operators are static, and interfaces cannot provide static methods (and thus can't provide operators).

When you use generic comparsions I think you should implement a (x)CompareTo(Y) or a comparable class. Please correct me if im wrong.

you can use if (kvp.Value.Equals(value)) instead of ==.

Use the "where" condition on your generic types
class Dictionary<TKey,TVal>
where TKey: IComparable, IEnumerable
where TVal: MyI
{
public void Add(TKey key, TVal val)
{
}
}
from http://msdn.microsoft.com/en-us/library/6b0scde8%28VS.80%29.aspx

Don't create a new class. Create a extension method:
public static class DictionaryHelper
{
public static TKey GetKeyFromValue<TKey, TValue>(this IDictionary<TKey, TValue> instance, TValue value)
{
foreach (var kvp in instance)
{
if (kvp.Value.Equals(value))
return kvp.Key;
}
return default(TKey);
}
}
public class Example
{
public static void Main(string[] argv)
{
Dictionary<string, string> test = new Dictionary<string, string> { { "Mykey", "MyValue" }, { "Key1", "Value2" } };
string key = test.GetKeyFromValue("MyValue");
}
}

If you want this to be general purpose, then you will want the definition of equality to be configurable, just as it is in the dictionary for keys.
Have a property of type IEqualityComparer<TValue>, which is set in the constructor.
Then have a version of the constructor that makes the default EqualityComparer<TValue>.Default. This will work by calling Equals on the type in question.
public class EnumDictionary<TKey, TValue>
{
private Dictionary<TKey, TValue> _Dict;
private readonly IEqualityComparer<TValue> _cmp;
public EnumDictionary(Dictionary<TKey, TValue> Dict, IEqualityComparer<TValue> cmp)
{
this._Dict = Dict;
_cmp = cmp;
}
public EnumDictionary(Dictionary<TKey, TValue> Dict)
:this(Dict, IEqualityComparer<TValue>.Default){}
public TKey GetValue(TValue value)
{
foreach (KeyValuePair<TKey, TValue> kvp in _Dict)
{
if (cmp.Equals(kvp.Value, value))
return kvp.Key;
}
throw new Exception("Undefined data type: " + value);
}
}

Is there an IDictionary implementation that, on missing key, returns the default value instead of throwing?

The indexer into Dictionary throws an exception if the key is missing. Is there an implementation of IDictionary that instead will return default(T)?
I know about the TryGetValue() method, but that's impossible to use with LINQ.
Would this efficiently do what I need?:
myDict.FirstOrDefault(a => a.Key == someKeyKalue);
I don't think it will as I think it will iterate the keys instead of using a Hash lookup.

Indeed, that won't be efficient at all.
As per comments, in .Net Core 2+ / NetStandard 2.1+ / Net 5, MS added the extension method GetValueOrDefault()
For earlier versions you can write the extension method yourself:
public static TValue GetValueOrDefault<TKey,TValue>
(this IDictionary<TKey, TValue> dictionary, TKey key)
{
TValue ret;
// Ignore return value
dictionary.TryGetValue(key, out ret);
return ret;
}
Or with C# 7.1:
public static TValue GetValueOrDefault<TKey,TValue>
(this IDictionary<TKey, TValue> dictionary, TKey key) =>
dictionary.TryGetValue(key, out var ret) ? ret : default;
That uses:
An expression-bodied method (C# 6)
An out variable (C# 7.0)
A default literal (C# 7.1)

If you're using .NET Core 2 or above (C# 7.x), the CollectionExtensions class is introduced and you can use the GetValueOrDefault method to get default value if key is not there in a dictionary.
Dictionary<string, string> colorData = new Dictionary<string, string>();
string color = colorData.GetValueOrDefault("colorId", string.Empty);

Carrying these extension methods can help..
public static V GetValueOrDefault<K, V>(this IDictionary<K, V> dict, K key)
{
return dict.GetValueOrDefault(key, default(V));
}
public static V GetValueOrDefault<K, V>(this IDictionary<K, V> dict, K key, V defVal)
{
return dict.GetValueOrDefault(key, () => defVal);
}
public static V GetValueOrDefault<K, V>(this IDictionary<K, V> dict, K key, Func<V> defValSelector)
{
V value;
return dict.TryGetValue(key, out value) ? value : defValSelector();
}

Collections.Specialized.StringDictionary provides a non-exception result when looking up a missing key's value. It is also case-insensitive by default.
Caveats
It is only valid for its specialized uses, and — being designed before generics — it doesn't have a very good enumerator if you need to review the whole collection.

If you're using .Net Core, you can use the CollectionExtensions.GetValueOrDefault method. This is the same as the implementation provided in the accepted answer.
public static TValue GetValueOrDefault<TKey,TValue> (
this System.Collections.Generic.IReadOnlyDictionary<TKey,TValue> dictionary,
TKey key);

public class DefaultIndexerDictionary<TKey, TValue> : IDictionary<TKey, TValue>
{
private IDictionary<TKey, TValue> _dict = new Dictionary<TKey, TValue>();
public TValue this[TKey key]
{
get
{
TValue val;
if (!TryGetValue(key, out val))
return default(TValue);
return val;
}
set { _dict[key] = value; }
}
public ICollection<TKey> Keys => _dict.Keys;
public ICollection<TValue> Values => _dict.Values;
public int Count => _dict.Count;
public bool IsReadOnly => _dict.IsReadOnly;
public void Add(TKey key, TValue value)
{
_dict.Add(key, value);
}
public void Add(KeyValuePair<TKey, TValue> item)
{
_dict.Add(item);
}
public void Clear()
{
_dict.Clear();
}
public bool Contains(KeyValuePair<TKey, TValue> item)
{
return _dict.Contains(item);
}
public bool ContainsKey(TKey key)
{
return _dict.ContainsKey(key);
}
public void CopyTo(KeyValuePair<TKey, TValue>[] array, int arrayIndex)
{
_dict.CopyTo(array, arrayIndex);
}
public IEnumerator<KeyValuePair<TKey, TValue>> GetEnumerator()
{
return _dict.GetEnumerator();
}
public bool Remove(TKey key)
{
return _dict.Remove(key);
}
public bool Remove(KeyValuePair<TKey, TValue> item)
{
return _dict.Remove(item);
}
public bool TryGetValue(TKey key, out TValue value)
{
return _dict.TryGetValue(key, out value);
}
IEnumerator IEnumerable.GetEnumerator()
{
return _dict.GetEnumerator();
}
}

One could define an interface for the key-lookup function of a dictionary. I'd probably define it as something like:
Interface IKeyLookup(Of Out TValue)
Function Contains(Key As Object)
Function GetValueIfExists(Key As Object) As TValue
Function GetValueIfExists(Key As Object, ByRef Succeeded As Boolean) As TValue
End Interface
Interface IKeyLookup(Of In TKey, Out TValue)
Inherits IKeyLookup(Of Out TValue)
Function Contains(Key As TKey)
Function GetValue(Key As TKey) As TValue
Function GetValueIfExists(Key As TKey) As TValue
Function GetValueIfExists(Key As TKey, ByRef Succeeded As Boolean) As TValue
End Interface
The version with non-generic keys would allow code that was using code using non-structure key types to allow for arbitrary key variance, which would not be possible with a generic type parameter. One should not be allowed to use a mutable Dictionary(Of Cat, String) as a mutable Dictionary(Of Animal, String) since the latter would allow SomeDictionaryOfCat.Add(FionaTheFish, "Fiona"). But there's nothing wrong with using a mutable Dictionary(Of Cat, String) as an immutable Dictionary(Of Animal, String), since SomeDictionaryOfCat.Contains(FionaTheFish) should be considered a perfectly well-formed expression (it should return false, without having to search the dictionary, for anything that isn't of type Cat).
Unfortunately, the only way one will be able to actually use such an interface is if one wraps a Dictionary object in a class which implements the interface. Depending upon what you're doing, however, such an interface and the variance it allows might make it worth the effort.

If you are using ASP.NET MVC, you could leverage the RouteValueDictionary class that do the job.
public object this[string key]
{
get
{
object obj;
this.TryGetValue(key, out obj);
return obj;
}
set
{
this._dictionary[key] = value;
}
}

I used encapsulation to create an IDictionary with behavior very similar to an STL map, for those of you who are familiar with c++. For those who aren't:
indexer get {} in SafeDictionary below returns the default value if a key is not present, and adds that key to the dictionary with a default value. This is often the desired behavior, as you're looking up items that will appear eventually or have a good chance of appearing.
method Add(TK key, TV val) behaves as an AddOrUpdate method, replacing the value present if it exists instead of throwing. I don't see why m$ doesn't have an AddOrUpdate method and thinks throwing errors in very common scenarios is a good idea.
TL/DR - SafeDictionary is written so as to never throw exceptions under any circumstances, other than perverse scenarios, such as the computer being out of memory (or on fire). It does this by replacing Add with AddOrUpdate behavior and returning default instead of throwing NotFoundException from the indexer.
Here's the code:
using System;
using System.Collections;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
public class SafeDictionary<TK, TD>: IDictionary<TK, TD> {
Dictionary<TK, TD> _underlying = new Dictionary<TK, TD>();
public ICollection<TK> Keys => _underlying.Keys;
public ICollection<TD> Values => _underlying.Values;
public int Count => _underlying.Count;
public bool IsReadOnly => false;
public TD this[TK index] {
get {
TD data;
if (_underlying.TryGetValue(index, out data)) {
return data;
}
_underlying[index] = default(TD);
return default(TD);
}
set {
_underlying[index] = value;
}
}
public void CopyTo(KeyValuePair<TK, TD>[] array, int arrayIndex) {
Array.Copy(_underlying.ToArray(), 0, array, arrayIndex,
Math.Min(array.Length - arrayIndex, _underlying.Count));
}
public void Add(TK key, TD value) {
_underlying[key] = value;
}
public void Add(KeyValuePair<TK, TD> item) {
_underlying[item.Key] = item.Value;
}
public void Clear() {
_underlying.Clear();
}
public bool Contains(KeyValuePair<TK, TD> item) {
return _underlying.Contains(item);
}
public bool ContainsKey(TK key) {
return _underlying.ContainsKey(key);
}
public IEnumerator<KeyValuePair<TK, TD>> GetEnumerator() {
return _underlying.GetEnumerator();
}
public bool Remove(TK key) {
return _underlying.Remove(key);
}
public bool Remove(KeyValuePair<TK, TD> item) {
return _underlying.Remove(item.Key);
}
public bool TryGetValue(TK key, out TD value) {
return _underlying.TryGetValue(key, out value);
}
IEnumerator IEnumerable.GetEnumerator() {
return _underlying.GetEnumerator();
}
}

It could be a one-liner to check TryGetValue and return default value if it is false.
Dictionary<string, int> myDic = new Dictionary<string, int>() { { "One", 1 }, { "Four", 4} };
string myKey = "One"
int value = myDic.TryGetValue(myKey, out value) ? value : 100;
myKey = "One" => value = 1
myKey = "two" => value = 100
myKey = "Four" => value = 4
Try it online

Since .NET core 2.0 you can use:
myDict.GetValueOrDefault(someKeyKalue)

What about this one-liner that checks whether a key is present using ContainsKey and then returns either the normally retreived value or the default value using the conditional operator?
var myValue = myDictionary.ContainsKey(myKey) ? myDictionary[myKey] : myDefaultValue;
No need to implement a new Dictionary class that supports default values, simply replace your lookup statements with the short line above.

This question helped to confirm that the TryGetValue plays the FirstOrDefault role here.
One interesting C# 7 feature I would like to mention is the out variables feature, and if you add the null-conditional operator from C# 6 to the equation your code could be much more simple with no need of extra extension methods.
var dic = new Dictionary<string, MyClass>();
dic.TryGetValue("Test", out var item);
item?.DoSomething();
The downside of this is that you can't do everything inline like this;
dic.TryGetValue("Test", out var item)?.DoSomething();
If we'd need/want to do this we should code one extension method like Jon's.

Here is a version of #JonSkeet's for the world of C# 7.1 that also allows for an optional default to be passed in:
public static TV GetValueOrDefault<TK, TV>(this IDictionary<TK, TV> dict, TK key, TV defaultValue = default) => dict.TryGetValue(key, out TV value) ? value : defaultValue;
It may be more efficient to have two functions to optimize the case where you want to return default(TV):
public static TV GetValueOrDefault<TK, TV>(this IDictionary<TK, TV> dict, TK key, TV defaultValue) => dict.TryGetValue(key, out TV value) ? value : defaultValue;
public static TV GetValueOrDefault2<TK, TV>(this IDictionary<TK, TV> dict, TK key) {
dict.TryGetValue(key, out TV value);
return value;
}
Unfortunately C# doesn't (yet?) have a comma operator (or the C# 6 proposed semicolon operator) so you have to have an actual function body (gasp!) for one of the overloads.

Modern Answer
As of .NET Core 2.0, there is a built-in extension method with 2 overloads:
TValue GetValueOrDefault<TKey,TValue>(TKey)
TValue GetValueOrDefault<TKey,TValue>(TKey, TValue)
Usage:
var dict = new Dictionary<string, int>();
dict.GetValueOrDefault("foo"); // 0: the datatype's default
dict.GetValueOrDefault("foo", 2); // 2: the specified default
The first version returns null for nullable types, of course.
See the documentation for more detail.

In general I would support the answer from Jon Skeet, however I prefer an implementation where I can give the default value as parameter:
public static TValue GetValueOrDefault<TKey, TValue> (this IDictionary<TKey, TValue> dictionary, TKey key, TValue defaultValue)
{
if (dictionary.ContainsKey(key))
return dictionary[key];
else
return defaultValue;
}

No, because otherwise how would you know the difference when the key exists but stored a null value? That could be significant.