mirror of
https://github.com/C9Glax/tranga.git
synced 2025-02-22 23:30:13 +01:00
[postgres-Server-V2] Update Weebcentral with changes from V1
This commit is contained in:
Alessandro Benetton
parent
725813c2f3
commit
ef7ebf022d
@ -17,37 +17,38 @@ public class Weebcentral : MangaConnector
|
||||
downloadClient = new ChromiumDownloadClient();
|
||||
}
|
||||
|
||||
public override (Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?)[] GetManga(string publicationTitle = "")
|
||||
public override (Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?)[] GetManga(
|
||||
string publicationTitle = "")
|
||||
{
|
||||
const int limit = 32; //How many values we want returned at once
|
||||
var offset = 0; //"Page"
|
||||
var requestUrl =
|
||||
int offset = 0; //"Page"
|
||||
string requestUrl =
|
||||
$"{_baseUrl}/search/data?limit={limit}&offset={offset}&text={publicationTitle}&sort=Best+Match&order=Ascending&official=Any&display_mode=Minimal%20Display";
|
||||
var requestResult =
|
||||
RequestResult requestResult =
|
||||
downloadClient.MakeRequest(requestUrl, RequestType.Default);
|
||||
if ((int)requestResult.statusCode < 200 || (int)requestResult.statusCode >= 300 ||
|
||||
requestResult.htmlDocument == null)
|
||||
{
|
||||
return [];
|
||||
}
|
||||
|
||||
var publications = ParsePublicationsFromHtml(requestResult.htmlDocument);
|
||||
(Manga, List<Author>, List<MangaTag>, List<Link>, List<MangaAltTitle>)[] publications =
|
||||
ParsePublicationsFromHtml(requestResult.htmlDocument);
|
||||
|
||||
return publications;
|
||||
}
|
||||
|
||||
private (Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?)[] ParsePublicationsFromHtml(HtmlDocument document)
|
||||
private (Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?)[] ParsePublicationsFromHtml(
|
||||
HtmlDocument document)
|
||||
{
|
||||
if (document.DocumentNode.SelectNodes("//article") == null)
|
||||
return [];
|
||||
|
||||
var urls = document.DocumentNode.SelectNodes("/html/body/article/a[@class='link link-hover']")
|
||||
List<string> urls = document.DocumentNode.SelectNodes("/html/body/article/a[@class='link link-hover']")
|
||||
.Select(elem => elem.GetAttributeValue("href", "")).ToList();
|
||||
|
||||
List<(Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?)> ret = new();
|
||||
foreach (var url in urls)
|
||||
foreach (string url in urls)
|
||||
{
|
||||
var manga = GetMangaFromUrl(url);
|
||||
(Manga, List<Author>, List<MangaTag>, List<Link>, List<MangaAltTitle>)? manga = GetMangaFromUrl(url);
|
||||
if (manga is { } x)
|
||||
ret.Add(x);
|
||||
}
|
||||
@ -55,30 +56,32 @@ public class Weebcentral : MangaConnector
|
||||
return ret.ToArray();
|
||||
}
|
||||
|
||||
public override (Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?)? GetMangaFromUrl(string url)
|
||||
public override (Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?)?
|
||||
GetMangaFromUrl(string url)
|
||||
{
|
||||
Regex publicationIdRex = new(@"https:\/\/weebcentral\.com\/series\/(\w*)\/(.*)");
|
||||
var publicationId = publicationIdRex.Match(url).Groups[1].Value;
|
||||
string publicationId = publicationIdRex.Match(url).Groups[1].Value;
|
||||
|
||||
var requestResult = downloadClient.MakeRequest(url, RequestType.MangaInfo);
|
||||
RequestResult requestResult = downloadClient.MakeRequest(url, RequestType.MangaInfo);
|
||||
if ((int)requestResult.statusCode < 300 && (int)requestResult.statusCode >= 200 &&
|
||||
requestResult.htmlDocument is not null)
|
||||
return ParseSinglePublicationFromHtml(requestResult.htmlDocument, publicationId, url);
|
||||
return null;
|
||||
}
|
||||
|
||||
private (Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?) ParseSinglePublicationFromHtml(HtmlDocument document, string publicationId, string websiteUrl)
|
||||
private (Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?) ParseSinglePublicationFromHtml(
|
||||
HtmlDocument document, string publicationId, string websiteUrl)
|
||||
{
|
||||
var posterNode =
|
||||
HtmlNode? posterNode =
|
||||
document.DocumentNode.SelectSingleNode("//section[@class='flex items-center justify-center']/picture/img");
|
||||
var coverUrl = posterNode?.GetAttributeValue("src", "") ?? "";
|
||||
string coverUrl = posterNode?.GetAttributeValue("src", "") ?? "";
|
||||
|
||||
var titleNode = document.DocumentNode.SelectSingleNode("//section/h1");
|
||||
var sortName = titleNode?.InnerText ?? "Undefined";
|
||||
HtmlNode? titleNode = document.DocumentNode.SelectSingleNode("//section/h1");
|
||||
string sortName = titleNode?.InnerText ?? "Undefined";
|
||||
|
||||
HtmlNode[] authorsNodes =
|
||||
document.DocumentNode.SelectNodes("//ul/li[strong/text() = 'Author(s): ']/span")?.ToArray() ?? [];
|
||||
var authorNames = authorsNodes.Select(n => n.InnerText).ToList();
|
||||
List<string> authorNames = authorsNodes.Select(n => n.InnerText).ToList();
|
||||
List<Author> authors = authorNames.Select(n => new Author(n)).ToList();
|
||||
|
||||
HtmlNode[] genreNodes =
|
||||
@ -86,9 +89,9 @@ public class Weebcentral : MangaConnector
|
||||
HashSet<string> tags = genreNodes.Select(n => n.InnerText).ToHashSet();
|
||||
List<MangaTag> mangaTags = tags.Select(t => new MangaTag(t)).ToList();
|
||||
|
||||
var statusNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Status: ']/a");
|
||||
var status = statusNode?.InnerText ?? "";
|
||||
var releaseStatus = MangaReleaseStatus.Unreleased;
|
||||
HtmlNode? statusNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Status: ']/a");
|
||||
string status = statusNode?.InnerText ?? "";
|
||||
MangaReleaseStatus releaseStatus = MangaReleaseStatus.Unreleased;
|
||||
switch (status.ToLower())
|
||||
{
|
||||
case "cancelled": releaseStatus = MangaReleaseStatus.Cancelled; break;
|
||||
@ -97,33 +100,34 @@ public class Weebcentral : MangaConnector
|
||||
case "ongoing": releaseStatus = MangaReleaseStatus.Continuing; break;
|
||||
}
|
||||
|
||||
var yearNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Released: ']/span");
|
||||
var year = uint.Parse(yearNode?.InnerText ?? "0");
|
||||
HtmlNode? yearNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Released: ']/span");
|
||||
uint year = uint.Parse(yearNode?.InnerText ?? "0");
|
||||
|
||||
var descriptionNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Description']/p");
|
||||
var description = descriptionNode?.InnerText ?? "Undefined";
|
||||
HtmlNode? descriptionNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Description']/p");
|
||||
string description = descriptionNode?.InnerText ?? "Undefined";
|
||||
|
||||
HtmlNode[] altTitleNodes = document.DocumentNode
|
||||
.SelectNodes("//ul/li[strong/text() = 'Associated Name(s)']/ul/li")?.ToArray() ?? [];
|
||||
Dictionary<string, string> altTitlesDict = new(), links = new();
|
||||
for (var i = 0; i < altTitleNodes.Length; i++)
|
||||
for (int i = 0; i < altTitleNodes.Length; i++)
|
||||
altTitlesDict.Add(i.ToString(), altTitleNodes[i].InnerText);
|
||||
List<MangaAltTitle> altTitles = altTitlesDict.Select(a => new MangaAltTitle(a.Key, a.Value)).ToList();
|
||||
|
||||
var originalLanguage = "";
|
||||
string originalLanguage = "";
|
||||
|
||||
Manga manga = new (publicationId, sortName, description, websiteUrl, coverUrl, null, year,
|
||||
Manga manga = new(publicationId, sortName, description, websiteUrl, coverUrl, null, year,
|
||||
originalLanguage, releaseStatus, -1,
|
||||
this,
|
||||
authors,
|
||||
mangaTags,
|
||||
this,
|
||||
authors,
|
||||
mangaTags,
|
||||
[],
|
||||
altTitles);
|
||||
|
||||
|
||||
return (manga, authors, mangaTags, [], altTitles);
|
||||
}
|
||||
|
||||
public override (Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?)? GetMangaFromId(string publicationId)
|
||||
public override (Manga, List<Author>?, List<MangaTag>?, List<Link>?, List<MangaAltTitle>?)? GetMangaFromId(
|
||||
string publicationId)
|
||||
{
|
||||
return GetMangaFromUrl($"https://weebcentral.com/series/{publicationId}");
|
||||
}
|
||||
@ -136,68 +140,67 @@ public class Weebcentral : MangaConnector
|
||||
private SearchResult[] FilteredResults(string publicationTitle, SearchResult[] unfilteredSearchResults)
|
||||
{
|
||||
Dictionary<SearchResult, int> similarity = new();
|
||||
foreach (var sr in unfilteredSearchResults)
|
||||
foreach (SearchResult sr in unfilteredSearchResults)
|
||||
{
|
||||
List<int> scores = new();
|
||||
var filteredPublicationString = ToFilteredString(publicationTitle);
|
||||
var filteredSString = ToFilteredString(sr.s);
|
||||
string filteredPublicationString = ToFilteredString(publicationTitle);
|
||||
string filteredSString = ToFilteredString(sr.s);
|
||||
scores.Add(NeedlemanWunschStringUtil.CalculateSimilarity(filteredSString, filteredPublicationString));
|
||||
foreach (var srA in sr.a)
|
||||
foreach (string srA in sr.a)
|
||||
{
|
||||
var filteredAString = ToFilteredString(srA);
|
||||
string filteredAString = ToFilteredString(srA);
|
||||
scores.Add(NeedlemanWunschStringUtil.CalculateSimilarity(filteredAString, filteredPublicationString));
|
||||
}
|
||||
|
||||
similarity.Add(sr, scores.Sum() / scores.Count);
|
||||
}
|
||||
|
||||
var ret = similarity.OrderBy(s => s.Value).Take(10).Select(s => s.Key).ToList();
|
||||
List<SearchResult> ret = similarity.OrderBy(s => s.Value).Take(10).Select(s => s.Key).ToList();
|
||||
return ret.ToArray();
|
||||
}
|
||||
|
||||
public override Chapter[] GetChapters(Manga manga, string language = "en")
|
||||
{
|
||||
var requestUrl = $"{_baseUrl}/series/{manga.ConnectorId}/full-chapter-list";
|
||||
var requestResult =
|
||||
string requestUrl = $"{_baseUrl}/series/{manga.ConnectorId}/full-chapter-list";
|
||||
RequestResult requestResult =
|
||||
downloadClient.MakeRequest(requestUrl, RequestType.Default);
|
||||
if ((int)requestResult.statusCode < 200 || (int)requestResult.statusCode >= 300)
|
||||
return Array.Empty<Chapter>();
|
||||
return [];
|
||||
|
||||
//Return Chapters ordered by Chapter-Number
|
||||
if (requestResult.htmlDocument is null)
|
||||
return Array.Empty<Chapter>();
|
||||
var chapters = ParseChaptersFromHtml(manga, requestResult.htmlDocument);
|
||||
return [];
|
||||
List<Chapter> chapters = ParseChaptersFromHtml(manga, requestResult.htmlDocument);
|
||||
return chapters.Order().ToArray();
|
||||
}
|
||||
|
||||
private List<Chapter> ParseChaptersFromHtml(Manga manga, HtmlDocument document)
|
||||
{
|
||||
var chaptersWrapper = document.DocumentNode.SelectSingleNode("/html/body");
|
||||
HtmlNode? chaptersWrapper = document.DocumentNode.SelectSingleNode("/html/body");
|
||||
|
||||
Regex chapterRex = new(@".* (\d+)");
|
||||
Regex chapterRex = new(@"(\d+(?:\.\d+)*)");
|
||||
Regex idRex = new(@"https:\/\/weebcentral\.com\/chapters\/(\w*)");
|
||||
|
||||
var ret = chaptersWrapper.Descendants("a").Select(elem =>
|
||||
List<Chapter> ret = chaptersWrapper.Descendants("a").Select(elem =>
|
||||
{
|
||||
var url = elem.GetAttributeValue("href", "") ?? "Undefined";
|
||||
string url = elem.GetAttributeValue("href", "") ?? "Undefined";
|
||||
|
||||
if (!url.StartsWith("https://") && !url.StartsWith("http://"))
|
||||
return new Chapter(manga, "undefined", "-1", null, null);
|
||||
return new Chapter(manga, "undefined", "-1");
|
||||
|
||||
var idMatch = idRex.Match(url);
|
||||
var id = idMatch.Success ? idMatch.Groups[1].Value : null;
|
||||
Match idMatch = idRex.Match(url);
|
||||
string? id = idMatch.Success ? idMatch.Groups[1].Value : null;
|
||||
|
||||
var chapterNode = elem.SelectSingleNode("span[@class='grow flex items-center gap-2']/span")?.InnerText ??
|
||||
"Undefined";
|
||||
string chapterNode = elem.SelectSingleNode("span[@class='grow flex items-center gap-2']/span")?.InnerText ??
|
||||
"Undefined";
|
||||
|
||||
var chapterNumberMatch = chapterRex.Match(chapterNode);
|
||||
Match chapterNumberMatch = chapterRex.Match(chapterNode);
|
||||
|
||||
if(!chapterNumberMatch.Success)
|
||||
return new Chapter(manga, "undefined", "-1", null, null);
|
||||
|
||||
string chapterNumber = new(chapterNumberMatch.Groups[1].Value);
|
||||
var chapter = new Chapter(manga, url, chapterNumber, null, null);
|
||||
return chapter;
|
||||
if (!chapterNumberMatch.Success)
|
||||
return new Chapter(manga, "undefined", "-1");
|
||||
|
||||
string chapterNumber = chapterNumberMatch.Groups[1].Value;
|
||||
return new Chapter(manga, url, chapterNumber);
|
||||
}).Where(elem => elem.ChapterNumber.CompareTo("-1") != 0 && elem.Url != "undefined").ToList();
|
||||
|
||||
ret.Reverse();
|
||||
@ -206,17 +209,15 @@ public class Weebcentral : MangaConnector
|
||||
|
||||
internal override string[] GetChapterImageUrls(Chapter chapter)
|
||||
{
|
||||
var requestResult = downloadClient.MakeRequest(chapter.Url, RequestType.Default);
|
||||
if ((int)requestResult.statusCode < 200 || (int)requestResult.statusCode >= 300 ||requestResult.htmlDocument is null)
|
||||
{
|
||||
return [];
|
||||
}
|
||||
RequestResult requestResult = downloadClient.MakeRequest(chapter.Url, RequestType.Default);
|
||||
if ((int)requestResult.statusCode < 200 || (int)requestResult.statusCode >= 300 ||
|
||||
requestResult.htmlDocument is null) return [];
|
||||
|
||||
var document = requestResult.htmlDocument;
|
||||
HtmlDocument? document = requestResult.htmlDocument;
|
||||
|
||||
var imageNodes =
|
||||
HtmlNode[] imageNodes =
|
||||
document.DocumentNode.SelectNodes($"//section[@hx-get='{chapter.Url}/images']/img")?.ToArray() ?? [];
|
||||
var urls = imageNodes.Select(imgNode => imgNode.GetAttributeValue("src", "")).ToArray();
|
||||
string[] urls = imageNodes.Select(imgNode => imgNode.GetAttributeValue("src", "")).ToArray();
|
||||
return urls;
|
||||
}
|
||||
|
||||
|
||||
Loading…
x
Reference in New Issue
Block a user