大数据量下的高性能字符串处理 string.Substring优化

有一个txt文本,我需要对他每一行进行遍历,并在匹配的地方做文本截取,

常规情况下肯定就直接用string.Substring,但一测试,耗时直接干到60000ms,也就是60s.

如下面代码中被注释掉的部分,无论是用Substring还是Split都在60s左右.(还不是每一行都要截取)

最后使用了StringBuilder.CopyTo(),直接从60000ms降到25ms

代码中其他的代码作用不用管,测试过,那些逻辑耗时不到5ms,

修改前:

            for (var i = 0; i < lines.Count; i++)
            {
                var instance = lines[i];
                if (i > 0) offset += 2;//+2是换行符
                offsetsDic.Add(i, new KeyValuePair(offset, offset += instance.LineText.Length));
                var LineText = instance.LineText.Trim();
                if (instance.LineText.Length > 2 && LineText[0] == '[' && LineText[1] == '@')
                {
                    //var index = LineText.IndexOf(']');
                    //var funName = index > -1 ? LineText.Substring(0, index) : LineText;
                    //var funName = instance.LineText.Split(new char[] { ' ', 't' }, StringSplitOptions.RemoveEmptyEntries).FirstOrDefault().ToUpperInvariant();
                    fieldLineoffset.Add(i, funName);
                }
                else
                {
                    if (fieldLineoffset.Any())
                        fieldLineoffset.Add(i, fieldLineoffset[i - 1]);
                }
            }

修改后: text 就是 lines的原文本

var sb = new StringBuilder(text);
            for (var i = 0; i < lines.Count; i++)
            {
                var instance = lines[i];
                if (i > 0) offset += 2;//+2是换行符
                offsetsDic.Add(i, new KeyValuePair(offset, offset + instance.LineText.Length));
                var LineText = instance.LineText.Trim();
                if (instance.LineText.Length > 2 && LineText[0] == '[' && LineText[1] == '@')
                {
                    var funName = LineText;
                    var index = LineText.IndexOf(']');
                    if (index > -1)
                    {
                        var chars = new char[instance.LineText.Length - LineText.Length + index];
                        sb.CopyTo(offset + (instance.LineText.Length - LineText.Length), chars, 0, chars.Length);
                        funName = new string(chars);
                    }
                    fieldLineoffset.Add(i, funName);
                }
                else
                {
                    if (fieldLineoffset.Any())
                        fieldLineoffset.Add(i, fieldLineoffset[i - 1]);
                }
                offset += instance.LineText.Length;
            }