overlap 字符串重复次数

直接上code:

  public class OverlapCoefficient
    {
        private Collection<string> tokenSet;

        private string[] delimiters;

        public OverlapCoefficient()
        {
            this.tokenSet = new Collection<string>();
            this.delimiters = new string[] { " ", "-", "-", "_" };
        }

        public int GetSimilarity(string first, string second)
        {
            if (first != null && second != null)
            {
                return this.GetOverlapCount(this.Tokenize(first), this.Tokenize(second));
            }

            return 0;
        }


        public int GetOverlapCount(Collection<string> firstTokens, Collection<string> secondTokens)
        {
            this.tokenSet.Clear();
            int firstSetTokenCount = this.CalculateUniqueTokensCount(firstTokens);
            int secondSetTokenCount = this.CalculateUniqueTokensCount(secondTokens);
            this.MergeIntoSet(firstTokens);
            this.MergeIntoSet(secondTokens);

            int overlap = firstSetTokenCount + secondSetTokenCount - this.tokenSet.Count;
            return overlap;
        }

        private void MergeIntoSet(Collection<string> firstTokens)
        {
            foreach (string local in firstTokens)
            {
                if (!this.tokenSet.Contains(local))
                {
                    this.tokenSet.Add(local);
                }
            }
        }

        private Collection<string> Tokenize(string word)
        {
            Collection<string> res = new Collection<string>(word.Split(this.delimiters, StringSplitOptions.RemoveEmptyEntries).ToArray());
            return res;
        }

        private int CalculateUniqueTokensCount(Collection<string> tokenList)
        {
            Collection<string> collection = new Collection<string>();
            foreach (string local in tokenList)
            {
                if (!collection.Contains(local))
                {
                    collection.Add(local);
                }
            }

            return collection.Count;
        }
    }