manga-reader/MangaReader.Core/WebCrawlers/MangaNato/MangaNatoWebCrawler.cs

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

namespace MangaReader.Core.WebCrawlers.MangaNato;

public class MangaNatoWebCrawler : MangaWebCrawler
{
    public override MangaDTO GetManga(string url)
    {
        HtmlDocument document = GetHtmlDocument(url);
        MangaNatoMangaDocument node = new(document);

        MangaDTO manga = new()
        {
            Title = node.TitleNode.InnerText,
            AlternateTitles = GetAlternateTitles(node.AlternateTitlesNode),
            Authors = GetAuthors(node.AuthorsNode),
            Status = GetStatus(node.StatusNode),
            Genres = GetGenres(node.GenresNode),
            UpdateDate = GetUpdateDate(node.UpdateDateNode),
            RatingPercent = GetRatingPercent(node.AverageRatingNode, node.BestRatingNode),
            Votes = int.Parse(node.VotesNode.InnerText),
            Views = GetViews(node.ViewsNode),
            Description = GetTextFromNodes(node.StoryDescriptionTextNodes),
            Chapters = GetChapters(node.ChapterNodes)
        };

        return manga;
    }

    private static List<string> GetAlternateTitles(HtmlNode node)
    {
        return node.InnerText.Split(';').Select(x => x.Trim()).ToList();
    }

    private static List<string> GetAuthors(HtmlNode node)
    {
        return node.InnerText.Split('-').Select(x => x.Trim()).ToList();
    }

    private static MangaStatus GetStatus(HtmlNode node)
    {
        return node.InnerText switch
        {
            "Ongoing" => MangaStatus.Ongoing,
            "Completed" => MangaStatus.Complete,
            _ => MangaStatus.Unknown,
        };
    }

    private static List<string> GetGenres(HtmlNode node)
    {
        return node.InnerText.Split('-').Select(x => x.Trim()).ToList();
    }

    private static DateTime GetUpdateDate(HtmlNode node)
    {
        List<string> dateAndTime = node.InnerText.Split('-').Select(x => x.Trim()).ToList();
        DateOnly date = DateOnly.Parse(dateAndTime[0]);
        TimeOnly time = TimeOnly.Parse(dateAndTime[1]);

        return date.ToDateTime(time);
    }

    private static long GetViews(HtmlNode node)
    {
        string text = node.InnerText;

        if (int.TryParse(text, out int number))
            return number;

        ReadOnlySpan<char> shortText = text.AsSpan(0, text.Length - 1);

        if (double.TryParse(shortText, out double formattedNumber) == false)
            return 0;

        char suffix = text[^1];
        long multiplier = GetMultiplier(suffix);

        return (int)(formattedNumber * multiplier);
    }

    private static long GetMultiplier(char c)
    {
        return c switch
        {
            'K' => 1_000,
            'M' => 1_000_000,
            'B' => 1_000_000_000,
            'T' => 1_000_000_000_000,
            _ => 0,
        };
    }

    private static int GetRatingPercent(HtmlNode averageNode, HtmlNode bestNode)
    {
        double average = Convert.ToDouble(averageNode.InnerText);
        double best = Convert.ToDouble(bestNode.InnerText);

        return (int)Math.Round(average / best * 100);
    }

    private static List<MangaChapterDTO> GetChapters(HtmlNodeCollection chapterNodes)
    {
        List<MangaChapterDTO> chapters = [];

        foreach (var node in chapterNodes)
        {
            HtmlNode chapterNameNode = node.SelectSingleNode(".//a[contains(@class, 'chapter-name')]");
            HtmlNode chapterViewNode = node.SelectSingleNode(".//span[contains(@class, 'chapter-view')]");
            HtmlNode chapterTimeNode = node.SelectSingleNode(".//span[contains(@class, 'chapter-time')]");

            MangaChapterDTO chapter = new()
            {
                Number = GetChapterNumber(chapterNameNode),
                Name = chapterNameNode.InnerText,
                Url = chapterNameNode.Attributes["href"].Value,
                Views = GetViews(chapterViewNode),
                UploadDate = DateTime.Parse(chapterTimeNode.Attributes["title"].Value)
            };

            chapters.Add(chapter);
        }

        return chapters;
    }

    private static float GetChapterNumber(HtmlNode chapterNameNode)
    {
        string url = chapterNameNode.Attributes["href"].Value;
        int index = url.IndexOf("/chapter-");

        if (index == -1)
            return 0;

        string chapterNumber = url[(index + "/chapter-".Length)..];

        return float.Parse(chapterNumber);
    }

    private static string GetTextFromNodes(List<HtmlNode> nodes)
    {
        StringBuilder stringBuilder = new();

        foreach (HtmlNode node in nodes)
        {
            if (node.Name == "br")
            {
                stringBuilder.AppendLine();
            }
            else
            {
                stringBuilder.Append(HttpUtility.HtmlDecode(node.InnerText).Replace("\r\n", "").Trim());
            }
        }

        return stringBuilder.ToString();
    }
}