文本语句比较算法

左左在右边

已于 2022-03-10 15:37:23 修改

阅读量348

点赞数

分类专栏： C#中高级文章标签：算法 linq c#

于 2022-03-10 15:22:37 首次发布

本文链接：https://blog.csdn.net/weixin_41609327/article/details/123402466

版权

本文介绍了一种C#实现的文本比较算法，通过按标点符号分割文本，对比两个版本之间的变化。算法能识别文本的删除和增加，但无法精确到字的修改。

摘要由CSDN通过智能技术生成

1,算法实现效果

按标点符号进行分割，两段文本上一个文本与当前文本分割为两个list，遍历上一个版本list，与当前版本对比如果每一项出现，表示未修改，如果未出现表示删除

遍历当前版本list与上一个版本对比，如果每一项出现，表示未修改，如果未出现表示增加

注:不能精确定位到字有没有修改

2,算法部分

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Text.RegularExpressions;
using System.Threading.Tasks;

namespace Base
{
    public class CompareModel
    {
        public string currentStr { get; set; }
        public string compareStr { get; set; }
        public List<string> deleteList { get; set; }
        public List<string> insertList { get; set; }
    }
    public class CompareStr
    {
        public CompareModel GetCompareRet(string strcurrent, string strnext)
        {
            if (strcurrent == strnext)
            {
                return new CompareModel { compareStr = strcurrent, currentStr = strcurrent };
            }
            string str1 = strcurrent + "\n";
            string str2 = strnext + "\n";
            DiffTextClass diff = new DiffTextClass();
            DiffTextClass.Item[] items = diff.DiffText(str1, str2);
            Regex reg = new Regex("[\、\,\，\.\。\;\；\?\!\！\：\:\r\n]");

            string str1ret = reg.Replace(str1, "|");
            string[] Lines1 = str1ret.Split('|');
            List<string> note1 = GetNotes(str1);

            string str2ret = reg.Replace(str2, "|");
            string[] Lines2 = str2ret.Split('|');
            List<string> note2 = GetNotes(str2);

            List<string> deleteList = new List<string>();
            List<string> insertList = new List<string>();
            foreach (var item in items)
            {
                string deltestrB = "";
                if (item.deletedA > 0)
                {
                    for (int i = 0; i < item.deletedA; i++)
                    {
                        string deleteStr = Lines1[item.StartA + i];
                        deleteStr = deleteStr.Replace("{tdr}", "");
                        Lines1[item.StartA + i] = "<del>" + deleteStr + "</del>";
                        deleteList.Add(deleteStr);
                        if (item.StartA + i < note1.Count)
                        {
                            deltestrB += "<del>" + deleteStr + "</del>" + note1[item.StartA + i];
                        }

                    }
                }
                if (item.insertedB == 0 && item.deletedA > 0)
                {
                    string insertStr = Lines2[item.StartB];
                    Lines2[item.StartB] = deltestrB + insertStr;
                }

                if (item.insertedB > 0)
                {
                    for (int i = 0; i < item.insertedB; i++)
                    {
                        string insertStr = Lines2[item.StartB + i];
                        if (i != 0)
                        {
                            deltestrB = "";
                        }
                        insertStr = insertStr.Replace("{tdr}", "");
                        Lines2[item.StartB + i] = deltestrB + "<insert>" + insertStr + "</insert>";
                        insertList.Add(insertStr);
                    }
                }

            }

            string str1out = "";
            //第一篇加入标点
            for (int i = 0; i < Lines1.Length - 1; i++)
            {
                if (i >= note1.Count)
                {
                    str1out += Lines1[i] + "。";
                }
                else
                {
                    str1out += Lines1[i] + note1[i];
                }
            }
            string str2out = "";
            //第二篇加入标点
            for (int i = 0; i < Lines2.Length - 1; i++)
            {
                if (i >= note2.Count)
                {
                    str2out += Lines2[i] + "。";
                }
                else
                {
                    str2out += Lines2[i] + note2[i];
                }
            }
            CompareModel model = new CompareModel();
            model.currentStr = str1out;
            model.compareStr = str2out;
            model.deleteList = deleteList;
            model.insertList = insertList;
            return model;

        }
        //是否为标点符号
        public bool IsNote(string str)
        {
            //if (Regex.IsMatch(text[i].ToString(), @"[\u4e00-\u9fbb]+{1}quot;))）
            System.