trieTree.xiaoshi.sz.com/testtrieTree/Program.cs

using IPRS.xiaoshi.sz.com;
using Microsoft.VisualBasic;
using System;
using System.IO;
using System.Reflection.Metadata;
using System.Runtime.CompilerServices;
using System.Text.Json;
using testtrieTree;
using trieTree.xiaoshi.sz.com;
using static System.Runtime.InteropServices.JavaScript.JSType;


// See https://aka.ms/new-console-template for more information
string? key = null;
if (args.Length > 0)
{
    key = args[0];
}
else
{

    Console.WriteLine("请选择：");
    Console.WriteLine("1、手动输入地址解析；");
    Console.WriteLine("2、从文件中读取地址批量解析；");
    Console.WriteLine("3、测试IPRS检索；");
    Console.WriteLine("4、测试IPRS获取中国专利全文文本；");
    Console.WriteLine("5、测试IPRS获取中国专利PDF；");
    Console.WriteLine("6、测试IPRS获取中国专利著录项目；");

    key = Console.ReadLine();
}

switch (key)
{
    
    case "4":
        Console.Write("请输入专利申请号：");
        string strAppNo = Console.ReadLine();
        IPRSSearcher searcher = new IPRSSearcher();
        var ret =searcher.GetPatentFullTxtInfo(strAppNo);
        Console.WriteLine(ret.ToString());
        
        Parser paser = new Parser();
        var p = paser.ParserFulltext(ret.ToString());
        break;
    case "5":
        Console.Write("请输入专利申请号：");
        strAppNo = Console.ReadLine();
        searcher = new IPRSSearcher();
        ret = searcher.GetPatentCNPdf(strAppNo);
        Console.WriteLine(ret.ToString());
        break;
    case "6":
        Console.Write("请输入专利申请号：");
        strAppNo = Console.ReadLine();
        searcher = new IPRSSearcher();
        ret = searcher.GetPatentCNBiblio(strAppNo);
        Console.WriteLine(ret.ToString());
        break;
    case "1":
        Console.WriteLine(DateTime.Now.ToString());
        AddressUtility parser = new AddressUtility();

        Console.WriteLine(DateTime.Now.ToString());

        while (true)
        {
            Console.Write("请输入解析的地址：");
            string strAddress = Console.ReadLine();

            if (strAddress != "exit")
            {
                List<string>? result = parser.Paser(strAddress);

                if (result != null)
                {
                    foreach (string s in result)
                    {
                        Console.Write($"{s}\t");
                    }
                    Console.WriteLine();
                }

                Console.WriteLine(DateTime.Now);
            }
            else
            {
                Console.WriteLine("bye!");
                break;
            }

        }
        break;
    case "2":
        Console.WriteLine(DateTime.Now.ToString());
        parser = new AddressUtility();

        Console.WriteLine(DateTime.Now.ToString());
        string csvFile = "c:/temp/地址.txt";
        List<string> strings = new List<string>();
        using (StreamReader sr = new StreamReader(csvFile))
        {
            while (!sr.EndOfStream)
            {
                string line = sr.ReadLine();

                List<string> retList = parser.Paser(line);
                string str = "\"" + line.Trim() + "\"";
                string ssqx = "";
                string sszq = "";
                int i = 0;
                foreach (var s in retList)
                {
                    if (i < 2)
                    {
                        if (s != "市辖区" && s != "省直辖县级行政区划")
                            sszq += s;
                    }

                    if (i < 3)
                    {
                        if (s != "市辖区" && s != "省直辖县级行政区划")
                            ssqx += s;

                    }
                    str += "," + s;
                    i++;
                }

                for (int j = i; j < 5; j++)
                {
                    str += ",";
                }
                str = str + "," + line.Contains(ssqx).ToString() + "," + line.Contains(sszq).ToString();
                Console.WriteLine(str);
                strings.Add(str);
            }
        }

        string strPath = $"c:/temp/address_{DateTime.Now.ToString("hhmmss")}.csv";
        using (StreamWriter streamWriter = new StreamWriter(strPath, true, System.Text.Encoding.UTF8))
        {
            streamWriter.WriteLine("地址,省,市,区县,乡镇,村级,区县正确,市正确,省正确");
            foreach (var s in strings)
            {
                streamWriter.WriteLine(s);
            }
        }

        Console.WriteLine(DateTime.Now);
        break;
    case "3":
        string savePath = "e:/CNPatents/";
        Console.Write("请输入开始日期(yyyy-mm-dd):");
        string? strStart = Console.ReadLine();

        DateTime start;
        if(!DateTime.TryParse(strStart, out start))
        {
            Console.WriteLine("输入日期格式错误，系统退出！");
        } //DateTime.Parse("2024-11-01");

        Console.Write("请输入结束日期(yyyy-mm-dd):");
        string? strEnd = Console.ReadLine();

        DateTime end;
        if (!DateTime.TryParse(strEnd, out end))
        {
            Console.WriteLine("输入日期格式错误，系统退出！");
        } //DateTime.Parse("2024-11-01");
        //DateTime end = DateTime.Parse("2024-11-30");
        while(start.CompareTo(end) <= 0)
        {
            //Console.WriteLine("请输入下载的日期(yyyymmdd):");
            string strPubDate = start.ToString("yyyyMMdd");
            string strCondition = $"F XX (({strPubDate}/GD*3/PT)+({strPubDate}/GD*2/PT)+({strPubDate}/GD*9/PT)+({strPubDate}/PD*1/PT)+({strPubDate}/PD*8/PT))";

            string strRet = new IPRSSearcher().GetPatents(strCondition, 1, 1);
            int TotalPatents = 0;
            int RetiredPatents =0;
            using (JsonDocument document = JsonDocument.Parse(strRet))
            {
                var retCode = document.RootElement.GetProperty("Ret").GetInt32();
                TotalPatents = document.RootElement.GetProperty("Data").GetProperty("HitCount").GetInt32();
            }


            List<string> list = new List<string>();
            string temSavePath = $"{savePath}{strPubDate}";
            if (TotalPatents > 0)
            {
                
                if (!System.IO.Directory.Exists(temSavePath))
                {
                    System.IO.Directory.CreateDirectory(temSavePath);
                }

                if (TotalPatents > 10000)
                {
                    list = new IPRSSearcher().splitCondition(strCondition,DateTime.Now.AddYears(-15),DateTime.Now);
                }
                else
                {
                    list.Add(strCondition);
                }
            }
            
            foreach (string c in list)
            {
                string strRet1 = new IPRSSearcher().GetPatents(c, 1, 1);
                int TotalPatents1 = 0;
                using (JsonDocument document = JsonDocument.Parse(strRet1))
                {
                    var retCode1 = document.RootElement.GetProperty("Ret").GetInt32();
                    TotalPatents1 = document.RootElement.GetProperty("Data").GetProperty("HitCount").GetInt32();
                }

                if (TotalPatents1 > 0)
                {
                    int totalPages = (TotalPatents1 % 50 == 0) ? TotalPatents1 / 50 : TotalPatents1 / 50 + 1;

                    for (int page = 1; page <= totalPages; page++)
                    {
                        int iListTry = 0;
                        tryListAgain:
                        strRet1 = new IPRSSearcher().GetPatents(c, page, 50);

                        if (string.IsNullOrEmpty(strRet1))
                        {
                            iListTry++;
                            if (iListTry < 3)
                            {
                                System.Threading.Thread.Sleep(1000);
                                goto tryListAgain;
                            }
                        }
                        using (JsonDocument document = JsonDocument.Parse(strRet1))
                        {
                            var retCode = document.RootElement.GetProperty("Ret").GetInt32();
                            if(retCode == 500)
                            {
                                iListTry++;
                                if (iListTry < 3)
                                {
                                    goto tryListAgain;
                                }
                            }
                            //TotalPatents = document.RootElement.GetProperty("Data").GetProperty("HitCount").GetInt32();
                            var Datas = document.RootElement.GetProperty("Data").GetProperty("List");

                            for (int i = 0; i < Datas.GetArrayLength(); i++)
                            {
                                string appNo = Datas[i].GetProperty("AN").GetString().Replace(".", "");

                                string strType = appNo.Length > 9 ? appNo.Substring(6, 1) : appNo.Substring(4, 1);

                                if (!System.IO.File.Exists($"{temSavePath}/{appNo}.json"))
                                {
                                    RetiredPatents++;
                                    using (StreamWriter writer = new StreamWriter($"{temSavePath}/{appNo}.json"))
                                    {
                                        writer.WriteLine(Datas[i].ToString());
                                    }
                                    var searcher1 = new IPRSSearcher();
                                    int iGetMPic = 0;
                                TryGetMainPic:
                                    try
                                    {
                                        string mpicUrl = searcher1.GetPatentCnMainPic(appNo);
                                        byte[] data = new HttpClient().GetByteArrayAsync(mpicUrl).Result;
                                        using (FileStream fileStream = new FileStream($"{temSavePath}/{appNo}_abs.jpeg", FileMode.Create))
                                        {
                                            fileStream.Write(data, 0, data.Length);
                                        }
                                    }
                                    catch (Exception ex)
                                    {
                                        iGetMPic++;
                                        if (iGetMPic < 3)
                                        {
                                            goto TryGetMainPic;
                                        }
                                    }

                                    if (strType == "3")
                                    {
                                        Console.WriteLine($"{(page - 1) * 50 + i + 1}/{TotalPatents1}-{RetiredPatents}/{TotalPatents}\t{appNo}\t{DateTime.Now}\t{start.ToString("yyyMMdd")}");
                                        continue;
                                        string strWGImagesUrl = searcher1.GetPatentCNWGImage(appNo);

                                        if (!string.IsNullOrEmpty(strWGImagesUrl))
                                        {
                                            using (StreamWriter writer = new StreamWriter($"{temSavePath}/{appNo}_ImageUrls.txt"))
                                            {
                                                writer.WriteLine(strWGImagesUrl);
                                            }
                                            #region 
                                            //break;

                                            //string[] img_urls = strWGImagesUrl.Split('|');

                                            //for (int idx = 0; idx < img_urls.Length; idx++)
                                            //{
                                            //    int iTryGetPic = 0;
                                            //TryGetPic:
                                            //    try
                                            //    {
                                            //        byte[] data = new HttpClient().GetByteArrayAsync(img_urls[idx]).Result;
                                            //        using (FileStream fileStream = new FileStream($"{temSavePath}/{appNo}_img{idx + 1}.jpeg", FileMode.Create))
                                            //        {
                                            //            fileStream.Write(data, 0, data.Length);
                                            //        }
                                            //    }
                                            //    catch
                                            //    {
                                            //        iTryGetPic++;

                                            //        if (iTryGetPic < 3)
                                            //        {
                                            //            goto TryGetPic;
                                            //        }
                                            //        else
                                            //        {
                                            //            Console.WriteLine($"{appNo}图{idx}获取错误！");
                                            //        }
                                            //    }

                                            //}
                                            #endregion

                                        }
                                    }
                                    else
                                    {
                                        string strFullText = searcher1.GetPatentFullTxtInfo(appNo);
                                        using (StreamWriter writer = new StreamWriter($"{temSavePath}/{appNo}_fulltext.xml"))
                                        {
                                            writer.WriteLine(strFullText);
                                        }
                                    }

                                }
                                
                                Console.WriteLine($"{(page - 1) * 50 + i + 1}/{TotalPatents1}-{RetiredPatents}/{TotalPatents}\t{appNo}\t{DateTime.Now}\t{start.ToString("yyyMMdd")}");

                            }

                        }


                    }
                }
            }
            start = start.AddDays(1);
            while ((start.DayOfWeek != DayOfWeek.Friday ) && (start.DayOfWeek != DayOfWeek.Tuesday))
            {
                start = start.AddDays(1);
            }
                
        }
        

            //return;
        break;
    default:
        break;
}