128 lines
3.6 KiB
C#
128 lines
3.6 KiB
C#
using HtmlAgilityPack;
|
|
using MangaReader.Core.Http;
|
|
using MangaReader.Core.Metadata;
|
|
|
|
namespace MangaReader.Core.Sources.NatoManga.Metadata;
|
|
|
|
public class NatoMangaWebCrawler(IHtmlLoader htmlLoader) : MangaWebCrawler
|
|
{
|
|
public override string SourceId => "NatoManga";
|
|
|
|
public override async Task<SourceManga?> GetMangaAsync(string url, CancellationToken cancellationToken)
|
|
{
|
|
HtmlDocument document = await htmlLoader.GetHtmlDocumentAsync(url, cancellationToken);
|
|
NatoMangaHtmlDocument node = new(document);
|
|
|
|
SourceManga manga = new()
|
|
{
|
|
Title = new()
|
|
{
|
|
Name = node.TitleNode?.InnerText ?? string.Empty,
|
|
Language = Common.Language.Unknown
|
|
},
|
|
Genres = GetGenres(node.GenresNode),
|
|
Chapters = GetChapters(node.ChapterNodes)
|
|
};
|
|
|
|
return manga;
|
|
}
|
|
|
|
private static string[] GetGenres(HtmlNode? node)
|
|
{
|
|
if (node == null)
|
|
return [];
|
|
|
|
HtmlNodeCollection? genreNodes = node.SelectNodes(".//a");
|
|
|
|
if (genreNodes == null)
|
|
return [];
|
|
|
|
return [.. genreNodes.Select(genreNode => genreNode.InnerText.Trim())];
|
|
}
|
|
|
|
private static long GetViews(HtmlNode node)
|
|
{
|
|
string text = node.InnerText.Trim();
|
|
|
|
if (int.TryParse(text, out int number))
|
|
return number;
|
|
|
|
if (double.TryParse(text, out double doubleNumber))
|
|
return (int)doubleNumber;
|
|
|
|
ReadOnlySpan<char> shortText = text.AsSpan(0, text.Length - 1);
|
|
|
|
if (double.TryParse(shortText, out double formattedNumber) == false)
|
|
return 0;
|
|
|
|
char suffix = text[^1];
|
|
|
|
//if (char.GetNumericValue(suffix) > -1)
|
|
// return (int)formattedNumber;
|
|
|
|
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 SourceMangaChapter[] GetChapters(HtmlNodeCollection? chapterNodes)
|
|
{
|
|
if (chapterNodes == null)
|
|
return [];
|
|
|
|
List<SourceMangaChapter> chapters = [];
|
|
|
|
foreach (var node in chapterNodes)
|
|
{
|
|
HtmlNodeCollection? chapterPropertyNodes = node.SelectNodes(".//span");
|
|
|
|
if (chapterPropertyNodes == null || chapterPropertyNodes.Count < 3)
|
|
continue;
|
|
|
|
HtmlNode? chapterNameNode = chapterPropertyNodes[0].SelectSingleNode(".//a");
|
|
HtmlNode chapterViewNode = chapterPropertyNodes[1];
|
|
HtmlNode chapterTimeNode = chapterPropertyNodes[2];
|
|
|
|
if (chapterNameNode == null)
|
|
continue;
|
|
|
|
SourceMangaChapter chapter = new()
|
|
{
|
|
Number = GetChapterNumber(chapterNameNode),
|
|
Title = 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)..].Replace('-', '.');
|
|
|
|
return float.Parse(chapterNumber);
|
|
}
|
|
} |