.NET dictionary with two keys and one value

As you wish your value to be “findable” from either key, I would just use two dictionaries like you are doing now. However I would wrap this up in a class, with methods names like FindByXXX and FindByYYY.

The much harder question is how do you do a delete, as you need to know both keys at the time of the delete. Maybe your value stores both keys so you can pass the value into your delete method. Maybe you never need to remove items from the dictionaries. Or the code that needs to remove items knows both keys.

Hence there is no standard dictionary to do this, as the requirements are different between each user.

(Note you don’t want a dictionary with a composite key, as that would require you to know both keys whenever you wished to look up an item.)

Maybe, something like this:

public class TwoKeyDictionary<Tkey1, Tkey2, TValue> { private object m_data_lock = new object(); private Dictionary<Tkey1, Tkey2> m_dic1 = new Dictionary<Tkey1, Tkey2>(); private Dictionary<Tkey2, TValue> m_dic2 = new Dictionary<Tkey2, TValue>(); public void AddValue(Tkey1 key1, Tkey2 key2, TValue value) { lock(m_data_lock) { m_dic1[key1] = key2; m_dic2[key2] = value; } } public TValue getByKey1(Tkey1 key1) { lock(m_data_lock) return m_dic2[m_dic1[key1]]; } public TValue getByKey2(Tkey2 key2) { lock(m_data_lock) return m_dic2[key2]; } public void removeByKey1(Tkey1 key1) { lock(m_data_lock) { Tkey2 tmp_key2 = m_dic1[key1]; m_dic1.Remove(key1); m_dic2.Remove(tmp_key2); } } public void removeByKey2(Tkey2 key2) { lock(m_data_lock) { Tkey1 tmp_key1 = m_dic1.First((kvp) => kvp.Value.Equals(key2)).Key; m_dic1.Remove(tmp_key1); m_dic2.Remove(key2); } } } 

I can offer a second solution, but it seems more slow and ugly vs. the first.

public class TwoKeysDictionary<K1, K2, V> { private class TwoKeysValue<K1, K2, V> { public K1 Key1 { get; set; } public K2 Key2 { get; set; } public V Value { get; set; } } private List<TwoKeysValue<K1, K2, V>> m_list = new List<TwoKeysValue<K1, K2, V>>(); public void Add(K1 key1, K2 key2, V value) { lock (m_list) m_list.Add(new TwoKeysValue<K1, K2, V>() { Key1 = key1, Key2 = key2, Value = value }); } public V getByKey1(K1 key1) { lock (m_list) return m_list.First((tkv) => tkv.Key1.Equals(key1)).Value; } public V getByKey2(K2 key2) { lock (m_list) return m_list.First((tkv) => tkv.Key2.Equals(key2)).Value; } public void removeByKey1(K1 key1) { lock (m_list) m_list.Remove(m_list.First((tkv) => tkv.Key1.Equals(key1))); } public void removeByKey2(K2 key2) { lock (m_list) m_list.Remove(m_list.First((tkv) => tkv.Key2.Equals(key2))); } } 

In very bad case, when Keys are a big structures (i.e. big value-types) and Keys are equals by size, and values are small value-types (for instance, a byte), with first solution you had: one set of Key1 , two sets of Key2, one set of values = 3 sets of big objects and 1 set of small values. With second solution you had: one set of Key1 , one set of Key2, one set of values = 2 sets of big objects and small set with values. I.e. with using of first solution you need by 50% (or by lower) more memory space vs. second, but a second solution is a very, very slow vs. first.

Your solution has a big impact on the memory footprint of your application. As the dictionary grows it will take at least double the amount memory (for value types) required to store the actual data.

You could probably approach this from a different angle. Have two dictionaries :

var lookupDictionary = new Dictionary<string, string>(); var valuesDictionary = new Dictionary<string, [YourValueType]>(); 

From here on in its pretty simple.

// Add a new entry into the values dictionary and give it a unique key valuesDictionary.Add("FooBar", "FUBAR VALUE"); // Add any number of lookup keys with the same value key lookupDictionary.Add("Foo", "FooBar"); lookupDictionary.Add("Bar", "FooBar"); lookupDictionary.Add("Rab", "FooBar"); lookupDictionary.Add("Oof", "FooBar"); 

When you need to find something from valuesDictionary you hit lookupDictionary first. This will give you the key of the value you are looking for in the valuesDictionary.

EDIT

I haven't addressed the deletion issue in my answer so here it goes :D

You would hit lookupDictionary to find the value key and then delete all entries from lookupDictionary that have that value.

Should be simple enough and safe since the valuesDictionary is guaranteed to have a unique key hence you will not accidentally delete a lookup key for some other value.

However, as Ian Ringrose pointed out in a comment, you are going to do a full scan on the lookupDictionary to delete. This may have an undesirable impact on performance in tight loops etc.

I can't really think of a good way to solve this issue at the moment. Perhaps someone else might have some ideas on how this could be improved.

I hope this helps.

You can't do it just with a single Dictionary without losing look up speed. The reason is that if you were to create a composite key there is no meaningful value you can return when you override GetHashCode. This means an equality comparison would need to be done against every key until a dictionary entry is found. You would also have a potential problem with a composite key in this case: because your Equals method would check whether one property or the other are equal, the following keys would essentially be duplicate keys { Id=1, Name="Bob" } { Id=1, Name="Anna" }, which doesn't give me a warm fuzzy feeling.

This leaves you with wrapping a dictionary, or pair of dictionaries with your own class.

interesting question, here's one solution. You have to add an indexer for every key type you want to support though.

public class NewDic<T> { public void Add(string key1, long key2, T value) { mDic.Add(key1, value); mDic.Add(key2, value); } public T this[string s] { get { return mDic[s]; } } public T this[long l] { get { return mDic[l]; } } Dictionary<object, T> mDic = new Dictionary<object, T>(); } NewDic<long> dic = new NewDic<long>(); dic.Add("abc", 20, 10); Console.WriteLine(dic["abc"]); Console.WriteLine(dic[20]); 

This is NOT a proper dictionary, but can be used for simple dictionary-like add remove functionalities.

This can be made generic as well, with proper implementation of IComparable in the keys types, and changing the dictionary code accordingly. (Note, default values of keys are not allowed to manage ambiguity!)

internal class KeyValueSet //this dictionary item is tailor made for this example { public string KeyStr { get; set; } public int KeyInt { get; set; } public int Value { get; set; } public KeyValueSet() { } public KeyValueSet(string keyStr, int keyInt, int value) { KeyStr = keyStr; KeyInt = keyInt; Value = value; } } public class DoubleKeyDictionary { List<KeyValueSet> _list = new List<KeyValueSet>(); private void Add(KeyValueSet set) { if (set == null) throw new InvalidOperationException("Cannot add null"); if (string.IsNullOrEmpty(set.KeyStr) && set.KeyInt == 0) throw new InvalidOperationException("Invalid key"); if (!string.IsNullOrEmpty(set.KeyStr) && _list.Any(l => l.KeyStr.Equals(set.KeyStr)) || set.KeyInt != 0 && _list.Any(l => l.KeyInt == set.KeyInt)) throw new InvalidOperationException("Either of keys exists"); _list.Add(set); } public void Add(string keyStr, int keyInt, int value) { Add(new KeyValueSet { KeyInt = keyInt, KeyStr = keyStr, Value = value }); } public void Add(string key, int value) { Add(new KeyValueSet { KeyInt = 0, KeyStr = key, Value = value }); } public void Add(int key, int value) { Add(new KeyValueSet { KeyInt = key, KeyStr = string.Empty, Value = value }); } public void Remove(int key) { if (key == 0) throw new InvalidDataException("Key not found"); var val = _list.First(l => l.KeyInt == key); _list.Remove(val); } public void Remove(string key) { if (string.IsNullOrEmpty(key)) throw new InvalidDataException("Key not found"); var val = _list.First(l => l.KeyStr == key); _list.Remove(val); } public void Remove(KeyValueSet item) { _list.Remove(item); } public int this[int index] { get { if (index != 0 && _list.Any(l => l.KeyInt == index)) return _list.First(l => l.KeyInt == index).Value; throw new InvalidDataException("Key not found"); } set { Add(index, value); } } public int this[string key] { get { if (!string.IsNullOrEmpty(key) && _list.Any(l => l.KeyStr == key)) return _list.First(l => l.KeyStr == key).Value; throw new InvalidDataException("Key not found"); } set { Add(key, value); } } } 

Testing the DoubleKeyDictionary

var dict = new DoubleKeyDictionary(); dict.Add(123, 1); dict.Add(234, 2); dict.Add("k1", 3); dict.Add("k2", 4); dict[456] = 5; dict["k3"] = 6; dict.Add("k4", 567, 7); dict.Remove(123); Console.WriteLine(dict[234]); //2 Console.WriteLine(dict["k2"]); //4 Console.WriteLine(dict[456]); //5 Console.WriteLine(dict[567]); //7 Console.WriteLine(dict["k4"]); //7 Console.WriteLine(dict[123]); //exception 

As a local solution I use the easy approach:

Imagine I have a collection of products identified by a string and a form with buttons for each one of the products.

When managing the state of the buttons I need to find buttons by string key. When handling the clicks I need to find product IDs by button instance.

Instead of maintaining two separate dictionaries I do the following:

public class SPurchaseOption { public Button Button; public string ProductID; public string SomeOtherAssociatedData; } Dictionary<object, SPurchaseOption> purchaseOptions; 

When the buttons are initialized I append two entries into the Dictionary i.e.

Key: ProductID, Value: "SPurchaseOption" Key: Button, Value: "SPurchaseOption" 

For a more general approach and if you need a commonly used component you will have to build a wrap around two dictionaries i.e:

public class DoubleKeyedDictionary<TKey1, TKey2, TValue> { class SItem { public TKey1 key1; public TKey2 key2; public TValue value; } Dictionary<TKey1, SItem> dic1; Dictionary<TKey2, SItem> dic2; } 

this will give access to both the value and alternative key by any of the keys.

As suggested in a comment to your question you could simply use an Object key for your Dictionary:

Dictionary<Object, long> dict = new Dictionary<Object, long>(); dict.Add("abc", 111); dict.Add(345, 111); 

To get a cleaner solution you could wrap this dictionary in a custom class and create your version of Add method:

public void Add(ISet<Object> keys, T value){ foreach(Object k in keys) { _privateDict.Add(k, value); } } 

How about a Dictionary<Tuple<string, long>, long>? Tuples are compared by value, so it should index uniquely in the expected manner. Plus, now you won't have to pack the long value in two places (and deal with the wonderful pain of synchronizing the values everywhere).

How about this approach? Basically, still use a dictionary-based strategy, but facade it through a class with overloaded indexer properties. So it looks like a dictionary, feels like a dictionary, but supports multiple keys (not like a dictionary, LOL).

public class MultiKeyDictionary<TFirstKey, TSecondKey, TValue> { private readonly Dictionary<TFirstKey, TValue> firstKeyDictionary = new Dictionary<TFirstKey, TValue>(); private readonly Dictionary<TSecondKey, TFirstKey> secondKeyDictionary = new Dictionary<TSecondKey, TFirstKey>(); public TValue this[TFirstKey idx] { get { return firstKeyDictionary[idx]; } set { firstKeyDictionary[idx] = value; } } public TValue this[TSecondKey idx] { get { var firstKey = secondKeyDictionary[idx]; return firstKeyDictionary[firstKey]; } set { var firstKey = secondKeyDictionary[idx]; firstKeyDictionary[firstKey] = value; } } public IEnumerable<KeyValuePair<TFirstKey, TValue>> GetKeyValuePairsOfFirstKey() { return firstKeyDictionary.ToList(); } public IEnumerable<KeyValuePair<TSecondKey, TValue>> GetKeyValuePairsOfSecondKey() { var r = from s in secondKeyDictionary join f in firstKeyDictionary on s.Value equals f.Key select new KeyValuePair<TSecondKey, TValue>(s.Key, f.Value); return r.ToList(); } public void Add(TFirstKey firstKey, TSecondKey secondKey, TValue value) { firstKeyDictionary.Add(firstKey, value); secondKeyDictionary.Add(secondKey, firstKey); } public bool Remove(TFirstKey firstKey) { if (!secondKeyDictionary.Any(f => f.Value.Equals(firstKey))) return false; var secondKeyToDelete = secondKeyDictionary.First(f => f.Value.Equals(firstKey)); secondKeyDictionary.Remove(secondKeyToDelete.Key); firstKeyDictionary.Remove(firstKey); return true; } public bool Remove(TSecondKey secondKey) { if (!secondKeyDictionary.ContainsKey(secondKey)) return false; var firstKey = secondKeyDictionary[secondKey]; secondKeyDictionary.Remove(secondKey); firstKeyDictionary.Remove(firstKey); return true; } } 

Test the code...

 static void Main(string[] args) { var dict = new MultiKeyDictionary<string, long, long>(); dict.Add("abc", 111, 1234); dict.Add("def", 222, 7890); dict.Add("hij", 333, 9090); Console.WriteLine(dict["abc"]); // expect 1234 Console.WriteLine(dict["def"]); // expect 7890 Console.WriteLine(dict[333]); // expect 9090 Console.WriteLine(); Console.WriteLine("removing def"); dict.Remove("def"); Console.WriteLine(); Console.WriteLine("now we have:"); foreach (var d in dict.GetKeyValuePairsOfFirstKey()) { Console.WriteLine($"{d.Key} : {d.Value}"); } Console.WriteLine(); Console.WriteLine("removing 333"); dict.Remove(333); Console.WriteLine(); Console.WriteLine("now we have:"); foreach (var d in dict.GetKeyValuePairsOfSecondKey()) { Console.WriteLine($"{d.Key} : {d.Value}"); } Console.ReadLine(); } 

At first I thought I could create a class that implmented IDictionary<TKey1, TValue> and IDictionary<TKey2, TValue>, and just have a single Dictionary as a field and delegate most methods to the single dictionary with minimal logic.

The problem with this approach is that TKey1 and TKey2 could be of the same type, which is a problem because this new class would be implementing the same interface twice. Which method should the runtime invoke when TKey1 is a string and TKey2 is also a string?

As others above have suggested, it is best to create your own data structure that utilizes one or two dictionaries behind the scenes. For example, if you knew ahead of time that you wanted to use a string and an int as your keys, you could use this approach:

public class StringIntDictionary<TValue> : IDictionary<string, TValue>, IDictionary<int, TValue> { private IDictionary<object, TValue> _dictionary = new Dictionary<object, TValue>(); // implement interface below, delegate to _dictionary } 

That would allow you to look use both string and int keys:

var dict = StringIntDictionary<bool>(); dict["abc"] = true; dict[123] = true; 

A very crude way that may suffice until I find a better one.

class MyClass { public string StringKey = ""; public int IntKey = 0; public override Equals(object obj) { // Code to return true if all fields are equal } } Dictionary <MyClass, string> MyDict; MyClass myClass; MyDict[MyDict.Keys.FirstOrDefault(x => x.Equals(MyClass))]; 

For my money, the answer saying to use tuples is the right one. Unfortunately, my NuGet is too old to get the ValueTuple package I'd want to use so my fields aren't 'item1', 'item2' etc. That would be more confusing than what I've done here. When I change VS/NuGet versions, it's ValueTuples all the way for this kind of situation. Second time this week I've encountered the need!

Here's something better in terms of efficiency.

public class MultiKeyDictionary<TKeyType1, TKeyType2, TValueType> { private readonly object threadLock = new object(); private readonly Dictionary<TKeyType1, TValueType> _dictionary1 = new Dictionary<TKeyType1, TValueType>(); private readonly Dictionary<TKeyType2, TValueType> _dictionary2 = new Dictionary<TKeyType2, TValueType>(); private readonly Dictionary<TKeyType1, TKeyType2> _Key1Key2Map = new Dictionary<TKeyType1, TKeyType2>(); private readonly Dictionary<TKeyType2, TKeyType1> _Key2Key1Map = new Dictionary<TKeyType2, TKeyType1>(); public bool Add(TKeyType1 key1, TKeyType2 key2, TValueType v) { if (ContainsKey1(key1) || ContainsKey2(key2)) return false; _dictionary1.Add(key1, v); _dictionary2.Add(key2, v); _Key1Key2Map.Add(key1, key2); _Key2Key1Map.Add(key2, key1); return true; } public bool ContainsKey1(TKeyType1 key) { return _dictionary1.ContainsKey(key); } public bool ContainsKey2(TKeyType2 key) { return _dictionary2.ContainsKey(key); } //Note if TKeyType1 and TKeyType2 are the same then we are forced to use GetBy functions public TValueType GetByKey1(TKeyType1 key) { return _dictionary1[key]; } public TValueType GetByKey2(TKeyType2 key) { return _dictionary2[key]; } public bool SetByKey1(TKeyType1 key, TValueType val) { if (ContainsKey1(key)) return false; lock (threadLock) { var key2 = _Key1Key2Map[key]; _dictionary1[key] = val; _dictionary2[key2] = val; } return true; } public bool SetByKey2(TKeyType2 key, TValueType val) { if (ContainsKey2(key)) return false; lock (threadLock) { var key1 = _Key2Key1Map[key]; _dictionary1[key1] = val; _dictionary2[key] = val; } return true; } public void RemoveUsingKey1(TKeyType1 key) { lock (threadLock) { var key2 = _Key1Key2Map[key]; _dictionary1.Remove(key); _dictionary2.Remove(key2); _Key1Key2Map.Remove(key); _Key2Key1Map.Remove(key2); } } public void RemoveUsingKey2(TKeyType2 key) { lock (threadLock) { var key1 = _Key2Key1Map[key]; _dictionary1.Remove(key1); _dictionary2.Remove(key); _Key1Key2Map.Remove(key1); _Key2Key1Map.Remove(key); } } public bool Contains(TKeyType1 key) { return _dictionary1.ContainsKey(key); } public bool Contains(TKeyType2 key) { return _dictionary2.ContainsKey(key); } public TValueType this[TKeyType1 key] { get => GetByKey1(key); set => SetByKey1(key, value); } public TValueType this[TKeyType2 key] { get => GetByKey2(key); set => SetByKey2(key, value); } public void Remove(TKeyType1 key) { RemoveUsingKey1(key); } public void Remove(TKeyType2 key) { RemoveUsingKey2(key); } public int Count => _dictionary1.Count; public Dictionary<TKeyType1, TValueType>.KeyCollection Key1s => _dictionary1.Keys; public Dictionary<TKeyType2, TValueType>.KeyCollection Key2s => _dictionary2.Keys; public Dictionary<TKeyType1, TValueType>.ValueCollection Values => _dictionary1.Values; public void Clear() { lock (threadLock) { _dictionary1.Clear(); _dictionary2.Clear(); _Key1Key2Map.Clear(); _Key2Key1Map.Clear(); } } //Map between Keys public TKeyType2 Key2(TKeyType1 key) { return _Key1Key2Map[key]; } public TKeyType1 Key1(TKeyType2 key) { return _Key2Key1Map[key]; } } 

I have created my own version of it. It is a bit more sophisticated. The documentation should explain some of the functionality. In essence it allows a reasonable way to handle 2 keys for the same value, auto-merges entries, is presumably thread-safe (untested), allows mapping keys together, and handles deleting entries, all the while having the functionality of dictionaries at its base. When adding an entry, but one of its keys already exists, it will just add the key and overwrite the value. It's quite distinct logically from other forms of collections, so these are the most I was able to implement.

Some structure with key pairs didn't seem fitting, given that I need this to arbitrarily add a second key as needed to existing entries, and given the merging functionality. I also took regard for the situation where one uses the same types for both keys, but also for situations where they don't.

/// <summary> A collection that internally uses a list (which in turn internally uses an array), and two dictionaries for the index. /// This allows operating it based on two keys and provides means to (automatically) map keys to each other. /// The indexing of the internal list is treated specially. In order to not infringe on the validity of the dictionaries' references to the indexes, /// they are kept identical. Removing is handled by setting the entries to 'null', and once a new item is added, they are overwritten. </summary> /// <typeparam name="TKey1"> The first key. </typeparam> /// <typeparam name="TKey2"> The second key. </typeparam> /// <typeparam name="T"> The stored value type. </typeparam> public class TwoKeyDictionary<TKey1, TKey2, T> : IEnumerable<TwoKeyDictionaryEntry<TKey1, TKey2, T>>, IReadOnlyCollection<TwoKeyDictionaryEntry<TKey1, TKey2, T>> { private readonly Dictionary<TKey1, int> _keys01 = new Dictionary<TKey1, int> (); private readonly Dictionary<TKey2, int> _keys02 = new Dictionary<TKey2, int> (); private readonly List<TwoKeyDictionaryEntry<TKey1, TKey2, T>> _items = new List<TwoKeyDictionaryEntry<TKey1, TKey2, T>> (); private int _freeIndex = 0; // The index of the first free slot. private int _freeCount = 0; // Free before the last value. private readonly object _lock = new object (); public TwoKeyDictionary () { } /// <summary> Adds an item. </summary> public bool Add (TKey1 key, T value) { return AddByKey1 (key, value); } /// <summary> Adds an item. </summary> public bool Add (TKey2 key, T value) { return AddByKey2 (key, value); } /// <summary> Adds an item. </summary> public bool AddByKey1 (TKey1 key, T value) { lock (_lock) { return AddByKey1Internal (key, value); } } /// <summary> Adds an item. </summary> public bool AddByKey2 (TKey2 key, T value) { lock (_lock) { return AddByKey2Internal (key, value); } } /// <summary> Adds an item with two keys. If either key already exists, it will map the other key to it. The value will only be overwritten if it's 'null'. </summary> public bool Add (TKey1 key1, TKey2 key2, T value) { return Add (key1, key2, value, false); } /// <summary> Adds an item with two keys. If either key already exists, it will map the other key to it. The value will only be overwritten if it's 'null'. /// This may also define how the key is mapped, if occurring. </summary> public bool Add (TKey1 key1, TKey2 key2, T value, bool mapToKey2) { lock (_lock) { return AddInternal (key1, key2, value, mapToKey2); } } /// <summary> Maps both keys together. If either key exists, it will add the other one to it. If both exist, it will merge the entries and delete the other. /// By default this will map to key1. </summary> public bool Map (TKey1 key1, TKey2 key2) { return MapToKey1 (key1, key2); } /// <summary> Maps both keys together. If either key exists, it will add the other one to it. If both exist, it will merge the entries and delete the one with key2. </summary> public bool MapToKey1 (TKey1 key1, TKey2 key2) { lock (_lock) { return MapToKey1Internal (key1, key2); } } /// <summary> Maps both keys together. If either key exists, it will add the other one to it. If both exist, it will merge the entries and delete the one with key1. </summary> public bool MapToKey2 (TKey1 key1, TKey2 key2) { lock (_lock) { return MapToKey2Internal (key1, key2); } } /// <summary> Removes an entry based on key1. If there is a key2 mapped to it, it will be removed as well. </summary> public bool Remove (TKey1 key) { return RemoveByKey1 (key); } /// <summary> Removes an entry based on key2. If there is a key1 mapped to it, it will be removed as well. </summary> public bool Remove (TKey2 key) { return RemoveByKey2 (key); } /// <summary> Removes an entry based on key1. If there is a key2 mapped to it, it will be removed as well. </summary> public bool RemoveByKey1 (TKey1 key) { lock (_lock) { return RemoveByKey1Internal (key); } } /// <summary> Removes an entry based on key2. If there is a key1 mapped to it, it will be removed as well. </summary> public bool RemoveByKey2 (TKey2 key) { lock (_lock) { return RemoveByKey2Internal (key); } } /// <summary> Removes an entry based on both, key1 and key2. Any entries related to either keys will be removed. </summary> public bool Remove (TKey1 key1, TKey2 key2) { lock (_lock) { return RemoveByKey1Internal (key1) | RemoveByKey2Internal (key2); } } /// <summary> Tries to return a value based on key1. </summary> public bool TryGetValue (TKey1 key, out T value) { return TryGetValueByKey1 (key, out value); } /// <summary> Tries to return a value based on key2. </summary> public bool TryGetValue (TKey2 key, out T value) { return TryGetValueByKey2 (key, out value); } /// <summary> Tries to return a value based on key1. </summary> public bool TryGetValueByKey1 (TKey1 key, out T value) { if (key == null) { value = default; return false; } if (_keys01.TryGetValue (key, out int index)) { TwoKeyDictionaryEntry<TKey1, TKey2, T> entry = _items[index]; if (entry != null) { value = entry.Value; return true; } } value = default; return false; } /// <summary> Tries to return a value based on key2. </summary> public bool TryGetValueByKey2 (TKey2 key, out T value) { if (key == null) { value = default; return false; } if (_keys02.TryGetValue (key, out int index)) { TwoKeyDictionaryEntry<TKey1, TKey2, T> entry = _items[index]; if (entry != null) { value = entry.Value; return true; } } value = default; return false; } /// <summary> Tries to return a value based on key1 or key2. Prioritizes key1. </summary> public bool TryGetValue (TKey1 key1, TKey2 key2, out T value) { return TryGetValue (key1, key2, false, out value); } /// <summary> Tries to return a value based on key1 or key2. </summary> public bool TryGetValue (TKey1 key1, TKey2 key2, bool prioritizeKey2, out T value) { return prioritizeKey2 ? TryGetValue (key1, out value) || TryGetValue (key2, out value) : TryGetValue (key2, out value) || TryGetValue (key1, out value); } /// <summary> Returns 'true' if they key and the entry still exists. The stored value itself may still be 'null' regardless. </summary> public bool ContainsKey (TKey1 key) { return ContainsKey1 (key); } /// <summary> Returns 'true' if they key and the entry still exists. The stored value itself may still be 'null' regardless. </summary> public bool ContainsKey (TKey2 key) { return ContainsKey2 (key); } /// <summary> Returns 'true' if they key and the entry still exists. The stored value itself may still be 'null' regardless. </summary> public bool ContainsKey1 (TKey1 key) { if (key == null) return false; if (_keys01.TryGetValue (key, out int index)) return _items[index] != null; else return false; } /// <summary> Returns 'true' if they key and the entry still exists. The stored value itself may still be 'null' regardless. </summary> public bool ContainsKey2 (TKey2 key) { if (key == null) return false; if (_keys02.TryGetValue (key, out int index)) return _items[index] != null; else return false; } /// <summary> Returns 'true' if they key and the entry still exists. The stored value itself may still be 'null' regardless. </summary> public bool ContainsKey (TKey1 key1, TKey2 key2) { return ContainsKey1 (key1) || ContainsKey2 (key2); } #region Internal // Returns true if this wasn't the last position. private bool GetFreeIndex (bool apply, out int index) { if (_freeCount == 0) { index = _items.Count; return false; } else { index = _freeIndex; if (apply) { // We must find the next free slot. int freeIndex = _freeIndex + 1; int count = _items.Count; while (freeIndex < count && _items[freeIndex] != null) { freeIndex++; } if (freeIndex == count) _freeCount = 0; else Interlocked.Decrement (ref _freeCount); _freeIndex = freeIndex; } return true; } } private bool MapToKey1Internal (TKey1 key1, TKey2 key2) { if (key1 == null || key2 == null) return false; bool s1 = _keys01.TryGetValue (key1, out int index1); bool s2 = _keys02.TryGetValue (key2, out int index2); if (s1 && s2) { TwoKeyDictionaryEntry<TKey1, TKey2, T> e1 = _items[index1]; TwoKeyDictionaryEntry<TKey1, TKey2, T> e2 = _items[index2]; RemoveByKey2Internal (key2); e1.Key2 = key2; if (e1.Value == null) e1.Value = e2.Value; return true; } else if (s1) { _items[index1].Key2 = key2; _keys02.Add (key2, index1); return true; } else if (s2) { _items[index2].Key1 = key1; _keys01.Add (key1, index2); return true; } else return false; } private bool MapToKey2Internal (TKey1 key1, TKey2 key2) { if (key1 == null || key2 == null) return false; bool s1 = _keys01.TryGetValue (key1, out int index1); bool s2 = _keys02.TryGetValue (key2, out int index2); if (s1 && s2) { TwoKeyDictionaryEntry<TKey1, TKey2, T> e1 = _items[index1]; TwoKeyDictionaryEntry<TKey1, TKey2, T> e2 = _items[index2]; RemoveByKey1Internal (key1); e2.Key1 = key1; if (e2.Value == null) e2.Value = e1.Value; return true; } else if (s1) { _items[index1].Key2 = key2; return true; } else if (s2) { _items[index2].Key1 = key1; return true; } else return false; } private bool AddByKey1Internal (TKey1 key, T value) { if (key == null) return false; if (_keys01.TryGetValue (key, out int index)) { TwoKeyDictionaryEntry<TKey1, TKey2, T> entry = _items[index]; if (entry != null) { entry.Value = value; return true; } else { _keys01.Remove (key); return AddByKey1Internal (key, value); } } else { TwoKeyDictionaryEntry<TKey1, TKey2, T> item = new TwoKeyDictionaryEntry<TKey1, TKey2, T> (key, default, value); if (GetFreeIndex (true, out int freeIndex)) { _items[freeIndex] = item; } else { _items.Add (item); } _keys01.Add (key, freeIndex); return true; } } private bool AddByKey2Internal (TKey2 key, T value) { if (key == null) return false; if (_keys02.TryGetValue (key, out int index)) { TwoKeyDictionaryEntry<TKey1, TKey2, T> entry = _items[index]; if (entry != null) { entry.Value = value; return true; } else { _keys02.Remove (key); return AddByKey2Internal (key, value); } } else { TwoKeyDictionaryEntry<TKey1, TKey2, T> item = new TwoKeyDictionaryEntry<TKey1, TKey2, T> (default, key, value); if (GetFreeIndex (true, out int freeIndex)) { _items[freeIndex] = item; } else { _items.Add (item); } _keys02.Add (key, freeIndex); return true; } } private bool AddInternal (TKey1 key1, TKey2 key2, T value, bool mapToKey2) { if (key1 == null) return AddByKey2Internal (key2, value); else if (key2 == null) return AddByKey1Internal (key1, value); bool hasKey1 = _keys01.TryGetValue (key1, out int index1); bool hasKey2 = _keys02.TryGetValue (key2, out int index2); if (hasKey1 && hasKey2) { // We have 2 different entries' keys that point to the same value. Merge them to one key, remove the other. if (mapToKey2) { if (MapToKey2Internal (key1, key2)) { _items[index2].Value = value; } } else { if (MapToKey1Internal (key1, key2)) { _items[index1].Value = value; } } } else if (hasKey1) { TwoKeyDictionaryEntry<TKey1, TKey2, T> entry = _items[index1]; entry.Key2 = key2; entry.Value = value; } else if (hasKey2) { TwoKeyDictionaryEntry<TKey1, TKey2, T> entry = _items[index2]; entry.Key1 = key1; entry.Value = value; } else { _items.Add (new TwoKeyDictionaryEntry<TKey1, TKey2, T> (key1, key2, value)); } return true; } private bool RemoveByKey1Internal (TKey1 key) { if (key == null) return false; if (_keys01.TryGetValue (key, out int index)) { TwoKeyDictionaryEntry<TKey1, TKey2, T> entry = _items[index]; if (entry != null) { _keys01.Remove (key); if (entry.Key2 != null) _keys02.Remove (entry.Key2); if (index == _items.Count - 1) { _items.RemoveAt (index); } else { _items[index] = null; _freeIndex = _freeCount > 0 ? Math.Min (_freeIndex, index) : index; Interlocked.Increment (ref _freeCount); } return true; } else { _keys01.Remove (key); } } return false; } private bool RemoveByKey2Internal (TKey2 key) { if (key == null) return false; if (_keys02.TryGetValue (key, out int index)) { TwoKeyDictionaryEntry<TKey1, TKey2, T> entry = _items[index]; if (entry != null) { _keys02.Remove (key); if (entry.Key1 != null) _keys01.Remove (entry.Key1); if (index == _items.Count - 1) { _items.RemoveAt (index); } else { _items[index] = null; _freeIndex = _freeCount > 0 ? Math.Min (_freeIndex, index) : index; Interlocked.Increment (ref _freeCount); } return true; } else { _keys02.Remove (key); } } return false; } #endregion #region Interface Implementations public int Count => _items.Count (j => j != null); public IEnumerator<TwoKeyDictionaryEntry<TKey1, TKey2, T>> GetEnumerator () { return _items.Where (j => j != null).GetEnumerator (); } IEnumerator IEnumerable.GetEnumerator () { return _items.Where (j => j != null).GetEnumerator (); } #endregion } /// <summary> The entry class of <see cref="TwoKeyDictionary{TKey1, TKey2, T}"/>, which grants references to the keys in both dictionaries used. </summary> /// <typeparam name="TKey1"> The first key. </typeparam> /// <typeparam name="TKey2"> The second key. </typeparam> /// <typeparam name="T"> The stored value type. </typeparam> public class TwoKeyDictionaryEntry<TKey1, TKey2, T> { public TKey1 Key1 { get; internal set; } public TKey2 Key2 { get; internal set; } public T Value { get; internal set; } internal TwoKeyDictionaryEntry () { } internal TwoKeyDictionaryEntry (TKey1 key1, TKey2 key2, T value) { Key1 = key1; Key2 = key2; Value = value; } public override string ToString () { return $"{Key1?.ToString () ?? "---"} | {Key2?.ToString () ?? "---"} | {Value}"; } } 

What you're asking for is more the function of a DataTable:

var myDataTable = new DataTable(); myDataTable.Columns.Add("Key1", typeof(long)); myDataTable.Columns.Add("Key2", typeof(string)); myDataTable.Columns.Add("Value", typeof(long)); myDataTable.Rows.Add(345, "abc", 111); // DataRow selectedRow = myDataTable.Select("Key1 = 345").FirstOrDefault(); DataRow sameRow = myDataTable.Select("Key2 = 'abc'").FirstOrDefault(); var selectedValue = (long)selectedRow["Value"]; // //Delete sameRow?.Delete(); 

Hope this helps (ten years later)!