HTML 테이블에서 데이터 추출 – C#

Install-Package Aspose.HTML

using Aspose.Html;
using System.IO;
using System.Linq;
using System.Text.Json;
using System.Collections.Generic;
...

    // Open the document from which you want to extract table data
    using var document = new HTMLDocument(Path.Combine(DataDir, "chapter-9.htm"));
    {
        // Check if there are any table elements in the document
        var tables = document.GetElementsByTagName("table");

        if (tables.Any())
        {
            var result = new List<Dictionary<string, string>>();
            //var i = 0;
            foreach (var table in tables)
            {
                // extract data from html table
                var tbodies = table.GetElementsByTagName("tbody");

                foreach (var tbody in tbodies)
                {
                    if (tbody.Children.Length > 1)
                    {
                        foreach (var row in tbody.Children)
                        {
                            if (row.HasAttribute("id"))
                            {
                                //test row
                                var data = new Dictionary<string, string>();

                                data["Id"] = row.GetAttribute("id");
                                if (row.Children.Length > 0)
                                {
                                    var td = row.Children[0];
                                    if (td.Children.Length > 0)
                                    {
                                        var element = td.Children[0].TagName == "STRONG"
                                            ? td.Children[0].Children[0]
                                            : td.Children[0];
                                        var href = ((HTMLAnchorElement)element).Href;
                                        data["Href"] = href;
                                        data["TestName"] = Path.GetFileNameWithoutExtension(href);
                                    }
                                }

                                data["TestComment"] = string.Join(" ",
                                    row.Children[3].TextContent
                                        .Split(new char[0], StringSplitOptions.RemoveEmptyEntries).ToList()
                                        .Select(x => x.Trim()));
                                result.Add(data);
                            }
                        }
                        var json = JsonSerializer.Serialize(result);
                        Console.WriteLine(json);
                    }
                }
            }
        }
        else
        {
            // Handle the case where no tables are found
            Console.WriteLine("No tables found in the document.");
        }
    }