在区分大小写的字典 C# 上进行不区分大小写的搜索

Question

I have a case Sensitive dictionary,

 Dictionary<string, uint> itemNames = new Dictionary<string, uint>(StringComparer.Ordinal);

So I can have case sensitive keys in this dictionary.

For example I can have below key value pairs,

1. { test, 10 }
2. { TEST, 20 }
3. { test1, 30 }
4. { test2, 40 }

...

When someone passes key, I want to retrieve the value. The retrieval should be partially case insensitive which means, If exact case is matched then return the case sensitive result, if case sensitive key doesn't exists then retrieve case insensitive key value.

For example, with the above values inserted in the dictionary

If user passes key as "TEST" I need to return 20.

If user passes key as "TEST1" , the case sensitive key is not found so I need to return 30.

How to achieve this in C# ?

Answer 1

You should first use TryGetValue to check if there is an item. If not, select the first match:

string key = "test1";
int val;
if (!itemNames.TryGetValue(key, out val))
{
    val = itemNames.FirstOrDefault
                    (k => string.Equals(k.Key, key, StringComparison.OrdinalIgnoreCase)
                    )?.Value ?? 0;
}

Be aware for the performance of this code though. If you have a large dictionary with a lot of misses on the first attempt, a second (case-insensitive) dictionary would be better.

Answer 2

Performance wise, the only way to get O(1) average lookup time is to have two dictionaries, where one will use a default comparer, and the other one a case-insensitive one:

// map each string to single item
Dictionary<string, int> CaseSensitive;
    
// map each string to multiple items, case insensitive
Dictionary<string, List<KeyValuePair<string, int>>> CaseInsensitive;

or even

// map each string to single item
Dictionary<string, int> CaseSensitive;
    
// map each string to multiple items, case insensitive, with O(1) lookup time
Dictionary<string, HashSet<KeyValuePair<string, int>>> CaseInsensitive;

Your case insensitive search can then properly return all matches, not just the most recently inserted one.