manga-reader/MangaReader.Core/Sources/MangaNato/Metadata/MangaNatoWebCrawler.cs

using HtmlAgilityPack;
using MangaReader.Core.Common;
using MangaReader.Core.Http;
using MangaReader.Core.Metadata;
using System.Text;
using System.Web;

namespace MangaReader.Core.Sources.MangaNato.Metadata;

public class MangaNatoWebCrawler(IHtmlLoader htmlLoader) : MangaWebCrawler
{
    public override string SourceId => "MangaNato";

    public override async Task<SourceManga?> GetMangaAsync(string url, CancellationToken cancellationToken)
    {
        HtmlDocument document = await htmlLoader.GetHtmlDocumentAsync(url, cancellationToken);
        MangaNatoMangaDocument node = new(document);

        SourceManga manga = new()
        {
            Title = new()
            {
                Name = node.TitleNode?.InnerText ?? string.Empty,
                Language = Language.Unknown
            },
            AlternateTitles = GetAlternateTitles(node.AlternateTitlesNode),
            Contributors = GetContributors(node.AuthorsNode),
            Status = GetStatus(node.StatusNode),
            Genres = GetGenres(node.GenresNode),
            UpdateDate = GetUpdateDate(node.UpdateDateNode),
            RatingPercent = GetRatingPercent(node.AverageRatingNode, node.BestRatingNode),
            Votes = node.VotesNode != null ? int.Parse(node.VotesNode.InnerText) : 0,
            Views = GetViews(node.ViewsNode),
            Description = new()
            {
                Name = GetTextFromNodes(node.StoryDescriptionTextNodes),
                Language = Language.Unknown
            },
            Chapters = GetChapters(node.ChapterNodes)
        };

        return manga;
    }

    private static List<SourceMangaTitle> GetAlternateTitles(HtmlNode? node)
    {
        if (node == null)
            return [];

        List<SourceMangaTitle> sourceMangaTitles = [];

        string[] titles = [.. node.InnerText.Split(';').Select(x => x.Trim())];

        foreach (string title in titles)
        {
            SourceMangaTitle sourceMangaTitle = new()
            {
                Name = title,
                Language = Language.Unknown
            };

            sourceMangaTitles.Add(sourceMangaTitle);
        }

        return [.. sourceMangaTitles];
    }

    private static SourceMangaContributor[] GetContributors(HtmlNode? node)
    {
        if (node == null)
            return [];

        List<SourceMangaContributor> contributors = [];

        string[] names = [.. node.InnerText.Split('-').Select(x => x.Trim())];

        foreach (string name in names)
        {
            SourceMangaContributor contributor = new()
            {
                Name = name,
                Role = SourceMangaContributorRole.Author
            };

            contributors.Add(contributor);
        }

        return [.. contributors];
    }

    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)
    {
        if (node == null)
            return [];

        return [.. node.InnerText.Split('-').Select(x => x.Trim())];
    }

    private static DateTime? GetUpdateDate(HtmlNode? node)
    {
        if (node == null)
            return null;

        List<string> dateAndTime = [.. node.InnerText.Split('-').Select(x => x.Trim())];
        DateOnly date = DateOnly.Parse(dateAndTime[0]);
        TimeOnly time = TimeOnly.Parse(dateAndTime[1]);

        return date.ToDateTime(time);
    }

    private static long GetViews(HtmlNode? node)
    {
        if (node == null)
            return 0;

        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)
    {
        if (averageNode == null || bestNode == null)
            return 0;

        double average = Convert.ToDouble(averageNode.InnerText);
        double best = Convert.ToDouble(bestNode.InnerText);

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

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

        if (chapterNodes == null)
            return 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')]");

            SourceMangaChapter chapter = new()
            {
                Number = GetChapterNumber(chapterNameNode),
                Title = chapterNameNode?.InnerText ?? string.Empty,
                Url = chapterNameNode?.Attributes["href"].Value ?? string.Empty,
                Views = GetViews(chapterViewNode),
                UploadDate = chapterTimeNode != null ? DateTime.Parse(chapterTimeNode.Attributes["title"].Value) : null
            };

            chapters.Add(chapter);
        }

        return chapters;
    }

    private static float GetChapterNumber(HtmlNode? chapterNameNode)
    {
        if (chapterNameNode == null)
            return 0;

        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();
    }
}