除了dictionary.add 之外，有没有更快的方法在C# 字典中添加行？

Question

We are using dictionary with key value pair both as string, Dictionary has millions of rows been added in a foreach loop. Profiler results show dictionary.add taking elapsed time in minutes for million plus hits.

Below is the method for reference.

public virtual Dictionary<string, string> CreateDictionaryRow(string[] row)
{
    Dictionary<string, string> dictionaryRow = new Dictionary<string, string>();

    foreach (ColumnMappingBO item in this.Layout.Items)
    {

        string value = string.Empty;
        if (!string.IsNullOrEmpty(item.ColumnPosition))
        {
            if (item.ColumnPosition == "ZZ")
            {
                value = string.Empty;
            }
            else
            {
                if (LayoutPosition.TryGetValue(item.ColumnID, out Ordinal))
                {
                    if (row.Length > Ordinal)
                    {
                        if (row[Ordinal] != null)
                        {
                            value = row[Ordinal];
                        }
                    }
                }
            }
        }
        dictionaryRow.Add(item.ColumnNameID, value);
    }
    return dictionaryRow;
}

All the code lines are fine, performance is impacted when dictionaryRow.Add() is executed. Let know how do we optimise adding rows to dictionary, any external custom library which can be used ?

Answer 1

I would try changing this into a LINQ query.

Maybe something like this:

    public Dictionary<string, string> CreateDictionaryRow(string[] row)
    {
        return this.Layout.Items.ToDictionary(item => item.ColumnNameID, 
                                    item => GetValue(item, row));;
    }

    private string GetValue(ColumnMappingBO item, string[] row)
    {
        string value = string.Empty;
        if (!string.IsNullOrEmpty(item.ColumnPosition))
        {
            if (item.ColumnPosition == "ZZ")
            {
                value = string.Empty;
            }
            else
            {
                if (LayoutPosition.TryGetValue(item.ColumnID, out Ordinal))
                {
                    if (row.Length > Ordinal)
                    {
                        if (row[Ordinal] != null)
                        {
                            value = row[Ordinal];
                        }
                    }
                }
            }
        }

        return value;
    }

Answer 2

One thing to experiment is how many unique hashcodes item.ColumnNameId generates. if there's a lot of collisions you are basically doing the same as iterating over a linkedlist until you reach the end and then add the object at the end.

Here's the code that is called when you call dictionary.Add() taken from http://referencesource.microsoft.com/#q=Dictionary

So with a lot of collisions you start looping and comparing keys a lot and with such a large Dictionary you can expect some collisions.

void Insert(TKey key, TValue value, bool add) {        
            if (buckets == null) Initialize(0);
            int hashCode = comparer.GetHashCode(key) & 0x7FFFFFFF;
            int targetBucket = hashCode % buckets.Length;

            for (int i = buckets[targetBucket]; i >= 0; i = entries[i].next) {
                if (entries[i].hashCode == hashCode && comparer.Equals(entries[i].key, key)) {
                    entries[i].value = value;
                    version++;
                    return;
                } 
}