jsmr/JSMR.Infrastructure/Scanning/ScannerUtilities.cs

using HtmlAgilityPack;
using System.Text.Json;
using System.Web;

namespace JSMR.Infrastructure.Scanning;

public static class ScannerUtilities
{
    public static List<string> GetStringListFromNodes(HtmlNode[] nodes)
    {
        return [.. nodes
            .Where(node => string.IsNullOrEmpty(node.InnerHtml) == false)
            .Select(node => HttpUtility.HtmlDecode(node.InnerHtml))];
    }

    public static string GetDecodedText(HtmlNode? node)
    {
        if (node == null)
            return string.Empty;

        if (string.IsNullOrWhiteSpace(node.InnerHtml))
            return string.Empty;

        return HttpUtility.HtmlDecode(node.InnerHtml.Replace("\n", "")).Trim();
    }

    public static string GetTextBetween(string text, string startText, string endText)
    {
        int startIndex = text.IndexOf(startText) + startText.Length;
        int endIndex = text.IndexOf(endText);

        int length = endIndex - startIndex;

        if (length <= 0)
            return "";

        return text.Substring(startIndex, length);
    }

    public static string GetImageSource(HtmlNode imageNode)
    {
        string imageSource = imageNode.GetAttributeValue("src", "");

        if (string.IsNullOrEmpty(imageSource))
            imageSource = imageNode.GetAttributeValue("data-src", "");

        return imageSource;
    }

    public static string[] ParseJavaScriptArray(string value)
    {
        try
        {
            string json = NormalizeJavaScriptArray(value);

            return JsonSerializer.Deserialize<string[]>(json) ?? [];
        }
        catch
        {
            return [.. value
                .Trim('[', ']')
                .Split(',', StringSplitOptions.RemoveEmptyEntries)
                .Select(x => x.Trim().Trim('\'', '"'))];
        }
    }

    private static string NormalizeJavaScriptArray(string input)
    {
        return input.Trim().Replace('\'', '"');
    }
}