因正则表达式搜索总是出现死循环,开始考虑改为其他搜索方式,因为.net自带的IndexOf默认只能找到第一个或最后一个,如果要把全部的匹配项都找出来,还需要自己写循环SubString,所以想找下有没有现成的,就发现了在这个领域里,BM算法是王道,而sunday算法据说是目前最好的改进版,这一点我没有从国外的网站尤其是wiki上找到印证,但中文谈论sunday的文章很多,我就姑且认为它是最好的吧。
int matchPosition = i;
            while (i < text.Length && j < pattern.Length)
            {
                if (text[i] == pattern[j])
                {
                    i++;
                    j++;
                }
                else
                {
                    if(m==text.Length-1)break;
                    int k = pattern.Length - 1;
                    while (k >= 0 && text[m ] != pattern[k])
                    {
                        k--;
                    }
                    int gap = pattern.Length - k;
                    i += gap;
                    m = i + pattern.Length;
                    if (m > text.Length) m = text.Length - 1;
                    matchPosition = i;
                    j = 0;
                }
            }
            if (i <= text.Length)
            {
                return matchPosition;
            }
            return -1;
        }
Stopwatch watch=new Stopwatch();
            //watch.Start();
            //for (int i = 0; i < count; i++)
            //{
            //    int pos= Sunday.GetPositionFirst(context, pattern, true);
            //}
            //watch.Stop();
            //Console.WriteLine(watch.ElapsedMilliseconds);
            watch.Reset();
            watch.Start();
            for (int i = 0; i < count; i++)
            {
                int pos = context.IndexOf(pattern);
            }
            watch.Stop();
            Console.WriteLine(watch.ElapsedMilliseconds);
            watch.Reset();
            watch.Start();
            for (int i = 0; i < count; i++)
            {
                int pos = Sunday.SundaySearch(context, pattern);
            }
            watch.Stop();
            Console.WriteLine(watch.ElapsedMilliseconds);
        }
但千万不要使用substring来实现算法,那样会新生成很多字符串中间变量,算法带来的好处远远不如分配内存复制字符串的消耗大,注释掉的部分就是使用substring实现的,比indexof慢很多。