TEAMModelOS/TEAMModelOS.Service/Evaluation/Implements/HtmlAnalyzeService.cs

using HtmlAgilityPack;
using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Text.RegularExpressions;
using System.Threading.Tasks;
using TEAMModelOS.Model.Core.Dtos;
using TEAMModelOS.Model.Evaluation.Dtos;
using TEAMModelOS.Model.Evaluation.Dtos.Own;
using TEAMModelOS.SDK.Helper.Common.CollectionHelper;
using TEAMModelOS.SDK.Helper.Common.StringHelper;
using TEAMModelOS.SDK.Helper.Security.ShaHash;
using TEAMModelOS.Service.Core.Implements;
using TEAMModelOS.Service.Evaluation.Interfaces;

namespace TEAMModelOS.Service.Evaluation.Implements
{
    class HtmlAnalyzeService: BaseService, IHtmlAnalyzeService
    {
        private static string SummaryTag = "【题文】";
        private static string AnswerTag = "【答案】";
        private static string AnalysisTag = "【解析】";
        private static string EndedTag = "【结束】";
        private static string Options = "ABCDEFGHIJ";
        private static string CompleteStart = "【";
        private static string CompleteEnd = "】";
        private static string ComposeStart = "【综合题】";
        private static string ComposeEnd = "【综合题-题干】";
        private static string ComposeTag = "【综合题-";
        private static Dictionary<string, string> TestType = new Dictionary<string, string> {
            { "Single", "【单选题】|【结束】" }, { "Multiple", "【多选题】|【结束】" },
            { "Judge", "【判断题】|【结束】" }, { "Complete", "【填空题】|【结束】" },
            { "Subjective", "【问答题】|【结束】" } , { "Compose", "【综合题】|【完结】" }};
        public  List<ExerciseDto> AnalyzeWordAsync(string html, string Lang)
        {
            //去除class 以及span标签"
            string classpattern = "class=\"([^\"]*)\"";
            html = Regex.Replace(html, classpattern, "");
            string pattern = "<span([^>]{0,})>";
            html = Regex.Replace(html, pattern, "");
            html = html.Replace("\t", " ").Replace("<span>", "").Replace("</span>", "").Replace("dir=\"ltr\"","");
            Dictionary<string, List<string>> TestInType=  ConvertTest(html);
            List<ExerciseDto> tests = new List<ExerciseDto>();
            foreach (string key in TestInType.Keys)
            {
                switch (key)
                {
                    case "Single":
                        List<ExerciseDto> exercisesSingle = SingleConvert(key, TestInType[key]);
                        exercisesSingle.ForEach(x => { x.PShaCode = x.ShaCode; });
                        tests.AddRange(exercisesSingle); break;
                    case "Multiple":
                        List<ExerciseDto> exercisesMultiple = MultipleConvert(key, TestInType[key]);
                        exercisesMultiple.ForEach(x => { x.PShaCode = x.ShaCode; });
                        tests.AddRange(exercisesMultiple); break;
                    case "Judge":
                        List<ExerciseDto> exercisesJudge = JudgeConvert(key, TestInType[key]);
                        exercisesJudge.ForEach(x => { x.PShaCode = x.ShaCode; });
                        tests.AddRange(exercisesJudge); break;
                    case "Complete":
                        List<ExerciseDto> exercisesComplete = CompleteConvert(key, TestInType[key]);
                        exercisesComplete.ForEach(x => { x.PShaCode = x.ShaCode; });
                        tests.AddRange(exercisesComplete); break;
                    case "Subjective":
                        List<ExerciseDto> exercisesSubjective = SubjectiveConvert(key, TestInType[key]);
                        exercisesSubjective.ForEach(x => { x.PShaCode = x.ShaCode; });
                        tests.AddRange(exercisesSubjective); break;
                    case "Compose":
                        List<ExerciseDto> exercisesCompose = ComposeConvert(key, TestInType[key],Lang);
                        exercisesCompose.ForEach(x => { x.PShaCode = x.ShaCode; });
                        tests.AddRange(exercisesCompose);
                        break;
                    default: break;
                }
            }
            return tests; 
        }

        private List<ExerciseDto> SingleConvert(string TypeKey, List<string> list)
        {
            List<ExerciseDto> testInfos = OptionProcess(TypeKey, list);
            return testInfos;
        }

        private List<ExerciseDto> MultipleConvert(string TypeKey, List<string> list)
        {
            List<ExerciseDto> testInfos = OptionProcess(TypeKey, list);
            return testInfos;
        }

        private List<ExerciseDto> JudgeConvert(string TypeKey, List<string> list)
        {
            List<ExerciseDto> testInfos = OptionProcess(TypeKey, list);
            return testInfos;
        }

       
        private List<ExerciseDto> CompleteConvert(string TypeKey, List<string> list)
        {
            List<ExerciseDto> testInfos = CompleteProcess(TypeKey, list);
            return testInfos;
        }

       

        private List<ExerciseDto> SubjectiveConvert(string TypeKey, List<string> tests)
        {
           // List<string> tests = ConvertTest(testHtml);
            List<ExerciseDto> testInfos = ConvertTestInfo(tests, TypeKey);

            foreach (ExerciseDto testInfo in testInfos)
            {
                testInfo.Question = testInfo.Question.Replace(AnalysisTag, "").Replace(SummaryTag, "").Replace(AnswerTag, "");
                testInfo.Question = HtmlHelper.DoUselessTag(testInfo.Question);
                StringBuilder textImg = new StringBuilder(HtmlHelper.DoTextImg(testInfo.Question));
                testInfo.ShaCode = ShaHashHelper.GetSHA1(textImg.ToString());
                for (int i = 0; i < testInfo.Answer.Count; i++)
                {
                    testInfo.Answer[i] = testInfo.Answer[i].Replace(AnswerTag, "").Replace(AnalysisTag, "");
                    testInfo.Answer[i] = HtmlHelper.DoUselessTag(testInfo.Answer[i]);
                }
                testInfo.Explain = testInfo.Explain.Replace(AnalysisTag, "").Replace(EndedTag, "");
                testInfo.Explain = HtmlHelper.DoUselessTag(testInfo.Explain);
            }
            return testInfos;
        }

        private List<ExerciseDto> ComposeConvert(string TypeKey, List<string> list ,string Lang)
        {
            List<ExerciseDto> exerciseDtos = new List<ExerciseDto>();
            foreach (string html in list) {
                ExerciseDto exercise = new ExerciseDto() { Type=TypeKey };
                string RegexStr = ComposeStart + "([\\s\\S]*?)" + ComposeEnd;
                Match mt = Regex.Match(html, RegexStr);
                exercise.Question= HtmlHelper.DoUselessTag(mt.Value.Replace(ComposeStart,"").Replace(ComposeEnd,""));
                string testinfo = Regex.Replace(html, RegexStr, "").Replace(ComposeTag,CompleteStart);
                //获取综合题的材料加每个小题的sha1Code
                string testQs= HtmlHelper.DoTextImg(exercise.Question);
                List<ExerciseDto> dtos =  AnalyzeWordAsync(testinfo, Lang);
                if (dtos.IsNotEmpty()) {
                    dtos.ForEach(x => { testQs = testQs + x.ShaCode; });
                    exercise.ShaCode=ShaHashHelper.GetSHA1(testQs);
                    dtos.ForEach(x => { x.PShaCode = exercise.ShaCode; });
                    exercise.Children.AddRange(dtos);
                }
                exerciseDtos.Add(exercise);
            }
            return exerciseDtos; 
        }

        private List<ExerciseDto> OptionProcess(string typeKey, List<string> list)
        {
            string[] optionsKeys = Options.Select(s => s.ToString()).ToArray();
            List<ExerciseDto> testInfos = ConvertTestInfo(list, typeKey);
            foreach (ExerciseDto testInfo in testInfos)
            {
                string optsRgex = optionsKeys[0] + "(\\．|\\.|\\、|\\:|\\：)([\\s\\S]*?)" + AnswerTag;
                string optsHtml = Regex.Match(testInfo.Question, optsRgex).Value;
                //HtmlDocument doc = new HtmlDocument();
                //doc.LoadHtml(optsHtml);
                //optsHtml = doc.DocumentNode.InnerText;
                //处理选项
                StringBuilder textImg = new StringBuilder();
                for (int i = 0; i < optionsKeys.Length - 1; i++)
                {
                    string optRgex = optionsKeys[i] + "(\\．|\\.|\\、|\\:|\\：)([\\s\\S]*?)" + optionsKeys[i + 1] + "(\\．|\\.|\\、|\\:|\\：)";
                    string optHtml = Regex.Match(optsHtml, optRgex).Value;
                    if (!string.IsNullOrEmpty(optHtml))
                    {
                        optHtml = optHtml.Substring(2, optHtml.Length - 4);
                        optHtml = HtmlHelper.DoUselessTag(optHtml);
                        textImg.Append(HtmlHelper.DoTextImg(optHtml));
                        testInfo.Option.Add(new CodeValue { Code = optionsKeys[i], Value = optHtml });
                        //testInfo.Option.Add(new Dictionary<string, string> { { "code", optionsKeys[i] },{ "value", optHtml } });
                        //testInfo.Option.TryAdd(optionsKeys[i], optHtml);
                    }
                    else
                    {
                        optRgex = optionsKeys[i] + "(\\．|\\.|\\、|\\:|\\：)([\\s\\S]*?)" + AnswerTag;
                        optHtml = Regex.Match(optsHtml, optRgex).Value;
                        if (!string.IsNullOrEmpty(optHtml))
                        {
                            optHtml = optHtml.Substring(2, optHtml.Length - 6);
                            optHtml = HtmlHelper.DoUselessTag(optHtml);
                            textImg.Append(HtmlHelper.DoTextImg(optHtml));
                            testInfo.Option.Add(new CodeValue { Code = optionsKeys[i], Value = optHtml });
                            //testInfo.Option.Add(new Dictionary<string, string> { { "code", optionsKeys[i] }, { "value", optHtml } });
                            //testInfo.Option.TryAdd(optionsKeys[i], optHtml);
                        }
                    }
                }
                //处理题干
                testInfo.Question = testInfo.Question.Replace(optsHtml, "").Replace(SummaryTag, "").Replace(AnswerTag, "");
                testInfo.Question = HtmlHelper.DoUselessTag(testInfo.Question);
                textImg.Append(HtmlHelper.DoTextImg(testInfo.Question));
                testInfo.ShaCode = ShaHashHelper.GetSHA1(textImg.ToString());
                List<string> answers = testInfo.Answer;
                HashSet<string> ans = new HashSet<string>();
                //处理答案
                for (int i = 0; i < answers.Count; i++)
                {
                    string Answer = answers[i].Replace(AnswerTag, "").Replace(AnalysisTag, "").TrimStart().TrimEnd();
                    Answer.Select(s => s.ToString()).ToList().ForEach(x =>
                    {
                        ans.Add(x);
                    });
                }
                testInfo.Answer = ans.ToList();
                //处理解析
                testInfo.Explain = testInfo.Explain.Replace(AnalysisTag, "").Replace(EndedTag, "");
                testInfo.Explain = HtmlHelper.DoUselessTag(testInfo.Explain);
            }
            return testInfos;
        }


        public static List<ExerciseDto> ConvertTestInfo(List<string> tests, string TypeKey) {
            List<ExerciseDto> testInfos = new List<ExerciseDto>();
            foreach (string html in tests)
            {
                Dictionary<string, string> regex = new Dictionary<string, string>();
                Dictionary<string, string> question = new Dictionary<string, string> { { "Summary", TestType[TypeKey].Split("|")[0] + "|" + AnswerTag }, { "Answer", AnswerTag + "|" + AnalysisTag }, { "Analysis", AnalysisTag + "|" + EndedTag } };
                Dictionary<string, string> compquestion = new Dictionary<string, string> { { "Summary", TestType[TypeKey].Split("|")[0] + "|" + AnalysisTag }, { "Analysis", AnalysisTag + "|" + EndedTag } };
                ExerciseDto test = new ExerciseDto();
                test.Type = TypeKey;
                List<string> keys = new List<string>();
                if (TypeKey.Equals("Complete"))
                {
                    keys = compquestion.Keys.ToList();
                    regex = compquestion;
                }
                else
                {
                    keys = question.Keys.ToList();
                    regex = question;
                }
                foreach (string key in keys)
                {
                    string[] tags = regex[key].Split("|");
                    string RegexStr = tags[0] + "([\\s\\S]*?)" + tags[1];
                    Match mt = Regex.Match(html, RegexStr);
                    switch (key)
                    {
                        case "Summary":
                            test.Question = mt.Value.Replace(TestType[TypeKey].Split("|")[0],""); break;
                        case "Answer":
                            string Answer = mt.Value;
                            ///单选或多选,判断答案 脱html标签
                            if (TypeKey.Equals("Single") || TypeKey.Equals("Multiple") || TypeKey.Equals("Judge"))
                            {
                                HtmlDocument doc = new HtmlDocument();
                                doc.LoadHtml(mt.Value);
                                Answer = doc.DocumentNode.InnerText;
                            }
                            test.Answer = new List<string>() { Answer }; break;
                        case "Analysis":
                            test.Explain = mt.Value; break;
                        default: break;
                    }
                }
                testInfos.Add(test);
            }
            return testInfos; 
        }
        private List<ExerciseDto> CompleteProcess(string TypeKey, List<string> tests)
        {
            //List<string> tests = ConvertTest(testHtml);
            List<ExerciseDto> testInfos = ConvertTestInfo(tests, TypeKey);
            HtmlDocument doc = new HtmlDocument();
            foreach (ExerciseDto testInfo in testInfos)
            {
                List<string> ans = new List<string>();
                testInfo.Question = testInfo.Question.Replace(AnalysisTag, "").Replace(SummaryTag, "").Replace(AnswerTag, "");
                string regRex = CompleteStart + "([\\s\\S]*?)" + CompleteEnd;
                List<ReplaceDto> replaces = new List<ReplaceDto>();
                var m = Regex.Match(testInfo.Question, regRex);
                int index = 1;
                while (m.Success)
                {
                    string an = m.Groups[1].ToString();
                    doc.LoadHtml(an);
                    string anstr = doc.DocumentNode.InnerText;
                    string nbsp = "";
                    int length = System.Text.Encoding.Default.GetBytes(anstr).Length;
                    for (int i = 0; i < length * 3; i++)
                    {
                        nbsp += "&nbsp;";
                    }
                    ReplaceDto replaceDto = new ReplaceDto { oldstr = CompleteStart + an + CompleteEnd, newstr = "<underline data=\"" + index + "\"><u>" + nbsp + "</u></underline>" };
                    replaces.Add(replaceDto);
                    ans.Add(an);
                    m = m.NextMatch();
                    index++;
                }
                string textImg = testInfo.Question;
                //消除答案
                foreach (ReplaceDto replace in replaces)
                {
                    testInfo.Question = testInfo.Question.Replace(replace.oldstr, replace.newstr);
                    testInfo.Question = HtmlHelper.DoUselessTag(testInfo.Question);
                    //只要题干文字和图片 
                    //不加underline标记
                    textImg = testInfo.Question.Replace(replace.oldstr, "");

                }
                textImg = HtmlHelper.DoTextImg(textImg);
                testInfo.ShaCode = ShaHashHelper.GetSHA1(textImg);
                //处理解析
                testInfo.Explain = testInfo.Explain.Replace(AnalysisTag, "").Replace(EndedTag, "");
                testInfo.Explain = HtmlHelper.DoUselessTag(testInfo.Explain);
                testInfo.Answer.AddRange(ans);
            }
            return testInfos;
        }
        /// <summary>
        /// 解析题型
        /// </summary>
        /// <param name="testHtml"></param>
        /// <returns></returns>
        public static Dictionary<string, List<string>> ConvertTest(string testHtml)
        {
            string start = SummaryTag;
            string end = EndedTag;
            Dictionary<string, List<string>> TestInType = new Dictionary<string, List<string>>();
            foreach (string key in TestType.Keys) {
                string[] tags = TestType[key].Split("|");
                string regRex = tags[0] + "([\\s\\S]*?)" + tags[1];
                var m = Regex.Match(testHtml, regRex);
                //int index = 1;
                List<string> tests = new List<string>();
                while (m.Success)
                {
                    string testInfo = tags[0]+m.Groups[1].ToString()+tags[1];
                    tests.Add(testInfo);
                    m = m.NextMatch();
                }
                TestInType.Add(key, tests);
            }
            return TestInType;
        }
    }
}