glax/Tranga
1
0
Fork 0
mirror of https://github.com/C9Glax/tranga.git synced 2025-04-14 20:33:19 +02:00
Code Issues Actions Packages Projects Releases Wiki Activity

Merge pull request #353 from ale-ben/cuttingedge
Some checks failed
Docker Image CI / build (push) Has been cancelled
Details

Browse Source 
Weebcentral: File name also depends on original chapter name
This commit is contained in:
Glax 2025-03-02 16:20:07 +01:00 committed by GitHub
commit 5bb4977876
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 60 additions and 51 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

17
Tranga/Chapter.cs
View File

@ -44,7 +44,7 @@ public readonly struct Chapter : IComparable
if (name is not null && name.Length > 0) if (name is not null && name.Length > 0)
{ {
string chapterName = IllegalStrings.Replace(string.Concat(LegalCharacters.Matches(name)), ""); string chapterName = IllegalStrings.Replace(string.Concat(LegalCharacters.Matches(name)), "");
this.fileName = $"{chapterVolNumStr} - {chapterName}"; this.fileName = chapterName.Length > 0 ? $"{chapterVolNumStr} - {chapterName}" : chapterVolNumStr;
} }
else else
this.fileName = chapterVolNumStr; this.fileName = chapterVolNumStr;
@ -96,17 +96,20 @@ public readonly struct Chapter : IComparable
if(mangaArchive is null) if(mangaArchive is null)
{ {
FileInfo[] archives = new DirectoryInfo(mangaDirectory).GetFiles("*.cbz"); FileInfo[] archives = new DirectoryInfo(mangaDirectory).GetFiles("*.cbz");
Regex volChRex = new(@"(?:Vol(?:ume)?\.([0-9]+)\D*)?Ch(?:apter)?\.([0-9]+(?:\.[0-9]+)*)"); Regex volChRex = new(@"(?:Vol(?:ume)?\.([0-9]+)\D*)?Ch(?:apter)?\.([0-9]+(?:\.[0-9]+)*)(?: - (.*))?.cbz");
Chapter t = this; Chapter t = this;
mangaArchive = archives.FirstOrDefault(archive => mangaArchive = archives.FirstOrDefault(archive =>
{ {
Match m = volChRex.Match(archive.Name); Match m = volChRex.Match(archive.Name);
if (m.Groups[1].Success) /*
return m.Groups[1].Value == t.volumeNumber.ToString(GlobalBase.numberFormatDecimalPoint) && * 1. If the volumeNumber is not present in the filename, it is not checked.
m.Groups[2].Value == t.chapterNumber.ToString(GlobalBase.numberFormatDecimalPoint); * 2. Check the chapterNumber in the chapter against the one in the filename.
else * 3. The chpaterName has to either be absent both in the chapter and the filename or match.
return m.Groups[2].Value == t.chapterNumber.ToString(GlobalBase.numberFormatDecimalPoint); */
return (!m.Groups[1].Success || m.Groups[1].Value == t.volumeNumber.ToString(GlobalBase.numberFormatDecimalPoint)) &&
m.Groups[2].Value == t.chapterNumber.ToString(GlobalBase.numberFormatDecimalPoint) &&
((!m.Groups[3].Success && string.IsNullOrEmpty(t.name)) || m.Groups[3].Value == t.name);
}); });
} }

94
Tranga/MangaConnectors/WeebCentral.cs
View File

@ -1,7 +1,6 @@
using System.Net; using System.Net;
using System.Text.RegularExpressions; using System.Text.RegularExpressions;
using HtmlAgilityPack; using HtmlAgilityPack;
using Soenneker.Utils.String.NeedlemanWunsch;
using Tranga.Jobs; using Tranga.Jobs;
namespace Tranga.MangaConnectors; namespace Tranga.MangaConnectors;
@ -22,10 +21,10 @@ public class Weebcentral : MangaConnector
{ {
Log($"Searching Publications. Term=\"{publicationTitle}\""); Log($"Searching Publications. Term=\"{publicationTitle}\"");
const int limit = 32; //How many values we want returned at once const int limit = 32; //How many values we want returned at once
var offset = 0; //"Page" int offset = 0; //"Page"
var requestUrl = string requestUrl =
$"{_baseUrl}/search/data?limit={limit}&offset={offset}&text={publicationTitle}&sort=Best+Match&order=Ascending&official=Any&display_mode=Minimal%20Display"; $"{_baseUrl}/search/data?limit={limit}&offset={offset}&text={publicationTitle}&sort=Best+Match&order=Ascending&official=Any&display_mode=Minimal%20Display";
var requestResult = RequestResult requestResult =
downloadClient.MakeRequest(requestUrl, RequestType.Default); downloadClient.MakeRequest(requestUrl, RequestType.Default);
if ((int)requestResult.statusCode < 200 || (int)requestResult.statusCode >= 300 || if ((int)requestResult.statusCode < 200 || (int)requestResult.statusCode >= 300 ||
requestResult.htmlDocument == null) requestResult.htmlDocument == null)
@ -34,7 +33,7 @@ public class Weebcentral : MangaConnector
return []; return [];
} }
var publications = ParsePublicationsFromHtml(requestResult.htmlDocument); Manga[] publications = ParsePublicationsFromHtml(requestResult.htmlDocument);
Log($"Retrieved {publications.Length} publications. Term=\"{publicationTitle}\""); Log($"Retrieved {publications.Length} publications. Term=\"{publicationTitle}\"");
return publications; return publications;
@ -45,13 +44,13 @@ public class Weebcentral : MangaConnector
if (document.DocumentNode.SelectNodes("//article") == null) if (document.DocumentNode.SelectNodes("//article") == null)
return []; 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(); .Select(elem => elem.GetAttributeValue("href", "")).ToList();
HashSet<Manga> ret = new(); HashSet<Manga> ret = new();
foreach (var url in urls) foreach (string url in urls)
{ {
var manga = GetMangaFromUrl(url); Manga? manga = GetMangaFromUrl(url);
if (manga is not null) if (manga is not null)
ret.Add((Manga)manga); ret.Add((Manga)manga);
} }
@ -62,9 +61,9 @@ public class Weebcentral : MangaConnector
public override Manga? GetMangaFromUrl(string url) public override Manga? GetMangaFromUrl(string url)
{ {
Regex publicationIdRex = new(@"https:\/\/weebcentral\.com\/series\/(\w*)\/(.*)"); 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 && if ((int)requestResult.statusCode < 300 && (int)requestResult.statusCode >= 200 &&
requestResult.htmlDocument is not null) requestResult.htmlDocument is not null)
return ParseSinglePublicationFromHtml(requestResult.htmlDocument, publicationId, url); return ParseSinglePublicationFromHtml(requestResult.htmlDocument, publicationId, url);
@ -73,26 +72,26 @@ public class Weebcentral : MangaConnector
private Manga ParseSinglePublicationFromHtml(HtmlDocument document, string publicationId, string websiteUrl) private Manga ParseSinglePublicationFromHtml(HtmlDocument document, string publicationId, string websiteUrl)
{ {
var posterNode = HtmlNode? posterNode =
document.DocumentNode.SelectSingleNode("//section[@class='flex items-center justify-center']/picture/img"); document.DocumentNode.SelectSingleNode("//section[@class='flex items-center justify-center']/picture/img");
var posterUrl = posterNode?.GetAttributeValue("src", "") ?? ""; string posterUrl = posterNode?.GetAttributeValue("src", "") ?? "";
var coverFileNameInCache = SaveCoverImageToCache(posterUrl, publicationId, RequestType.MangaCover); string coverFileNameInCache = SaveCoverImageToCache(posterUrl, publicationId, RequestType.MangaCover);
var titleNode = document.DocumentNode.SelectSingleNode("//section/h1"); HtmlNode? titleNode = document.DocumentNode.SelectSingleNode("//section/h1");
var sortName = titleNode?.InnerText ?? "Undefined"; string sortName = titleNode?.InnerText ?? "Undefined";
HtmlNode[] authorsNodes = HtmlNode[] authorsNodes =
document.DocumentNode.SelectNodes("//ul/li[strong/text() = 'Author(s): ']/span")?.ToArray() ?? []; document.DocumentNode.SelectNodes("//ul/li[strong/text() = 'Author(s): ']/span")?.ToArray() ?? [];
var authors = authorsNodes.Select(n => n.InnerText).ToList(); List<string> authors = authorsNodes.Select(n => n.InnerText).ToList();
HtmlNode[] genreNodes = HtmlNode[] genreNodes =
document.DocumentNode.SelectNodes("//ul/li[strong/text() = 'Tags(s): ']/span")?.ToArray() ?? []; document.DocumentNode.SelectNodes("//ul/li[strong/text() = 'Tags(s): ']/span")?.ToArray() ?? [];
HashSet<string> tags = genreNodes.Select(n => n.InnerText).ToHashSet(); HashSet<string> tags = genreNodes.Select(n => n.InnerText).ToHashSet();
var statusNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Status: ']/a"); HtmlNode? statusNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Status: ']/a");
var status = statusNode?.InnerText ?? ""; string status = statusNode?.InnerText ?? "";
Log("unable to parse status"); Log("unable to parse status");
var releaseStatus = Manga.ReleaseStatusByte.Unreleased; Manga.ReleaseStatusByte releaseStatus = Manga.ReleaseStatusByte.Unreleased;
switch (status.ToLower()) switch (status.ToLower())
{ {
case "cancelled": releaseStatus = Manga.ReleaseStatusByte.Cancelled; break; case "cancelled": releaseStatus = Manga.ReleaseStatusByte.Cancelled; break;
@ -101,19 +100,19 @@ public class Weebcentral : MangaConnector
case "ongoing": releaseStatus = Manga.ReleaseStatusByte.Continuing; break; case "ongoing": releaseStatus = Manga.ReleaseStatusByte.Continuing; break;
} }
var yearNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Released: ']/span"); HtmlNode? yearNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Released: ']/span");
var year = Convert.ToInt32(yearNode?.InnerText ?? "0"); int year = Convert.ToInt32(yearNode?.InnerText ?? "0");
var descriptionNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Description']/p"); HtmlNode? descriptionNode = document.DocumentNode.SelectSingleNode("//ul/li[strong/text() = 'Description']/p");
var description = descriptionNode?.InnerText ?? "Undefined"; string description = descriptionNode?.InnerText ?? "Undefined";
HtmlNode[] altTitleNodes = document.DocumentNode HtmlNode[] altTitleNodes = document.DocumentNode
.SelectNodes("//ul/li[strong/text() = 'Associated Name(s)']/ul/li")?.ToArray() ?? []; .SelectNodes("//ul/li[strong/text() = 'Associated Name(s)']/ul/li")?.ToArray() ?? [];
Dictionary<string, string> altTitles = new(), links = new(); Dictionary<string, string> altTitles = new(), links = new();
for (var i = 0; i < altTitleNodes.Length; i++) for (int i = 0; i < altTitleNodes.Length; i++)
altTitles.Add(i.ToString(), altTitleNodes[i].InnerText); altTitles.Add(i.ToString(), altTitleNodes[i].InnerText);
var originalLanguage = ""; string originalLanguage = "";
Manga manga = new(sortName, authors.ToList(), description, altTitles, tags.ToArray(), posterUrl, Manga manga = new(sortName, authors.ToList(), description, altTitles, tags.ToArray(), posterUrl,
coverFileNameInCache, links, coverFileNameInCache, links,
@ -130,8 +129,8 @@ public class Weebcentral : MangaConnector
public override Chapter[] GetChapters(Manga manga, string language = "en") public override Chapter[] GetChapters(Manga manga, string language = "en")
{ {
Log($"Getting chapters {manga}"); Log($"Getting chapters {manga}");
var requestUrl = $"{_baseUrl}/series/{manga.publicationId}/full-chapter-list"; string requestUrl = $"{_baseUrl}/series/{manga.publicationId}/full-chapter-list";
var requestResult = RequestResult requestResult =
downloadClient.MakeRequest(requestUrl, RequestType.Default); downloadClient.MakeRequest(requestUrl, RequestType.Default);
if ((int)requestResult.statusCode < 200 || (int)requestResult.statusCode >= 300) if ((int)requestResult.statusCode < 200 || (int)requestResult.statusCode >= 300)
return []; return [];
@ -139,35 +138,42 @@ public class Weebcentral : MangaConnector
//Return Chapters ordered by Chapter-Number //Return Chapters ordered by Chapter-Number
if (requestResult.htmlDocument is null) if (requestResult.htmlDocument is null)
return []; return [];
var chapters = ParseChaptersFromHtml(manga, requestResult.htmlDocument); List<Chapter> chapters = ParseChaptersFromHtml(manga, requestResult.htmlDocument);
Log($"Got {chapters.Count} chapters. {manga}"); Log($"Got {chapters.Count} chapters. {manga}");
return chapters.Order().ToArray(); return chapters.OrderByDescending(c => c.name).ThenBy(c => c.volumeNumber).ThenBy(c => c.chapterNumber).ToArray();
} }
private List<Chapter> ParseChaptersFromHtml(Manga manga, HtmlDocument document) 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+(?:\.\d+)*)"); Regex chapterRex = new(@"(\d+(?:\.\d+)*)");
Regex chapterNameRex = new(@"(\w* )+");
Regex idRex = new(@"https:\/\/weebcentral\.com\/chapters\/(\w*)"); 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://")) if (!url.StartsWith("https://") && !url.StartsWith("http://"))
return new Chapter(manga, null, null, "-1", "undefined"); return new Chapter(manga, null, null, "-1", "undefined");
var idMatch = idRex.Match(url); Match idMatch = idRex.Match(url);
var id = idMatch.Success ? idMatch.Groups[1].Value : null; string? id = idMatch.Success ? idMatch.Groups[1].Value : null;
var chapterNode = elem.SelectSingleNode("span[@class='grow flex items-center gap-2']/span")?.InnerText ?? string chapterNode = elem.SelectSingleNode("span[@class='grow flex items-center gap-2']/span")?.InnerText ??
"Undefined"; "Undefined";
var chapterNumberMatch = chapterRex.Match(chapterNode); MatchCollection chapterNumberMatch = chapterRex.Matches(chapterNode);
var chapterNumber = chapterNumberMatch.Success ? chapterNumberMatch.Groups[1].Value : "-1"; string chapterNumber = chapterNumberMatch.Count > 0 ? chapterNumberMatch[^1].Groups[1].Value : "-1";
MatchCollection chapterNameMatch = chapterNameRex.Matches(chapterNode);
string chapterName = chapterNameMatch.Count > 0
? string.Join(" - ",
chapterNameMatch.Select(m => m.Groups[1].Value.Trim())
.Where(name => name.Length > 0 && !name.Equals("Chapter", StringComparison.OrdinalIgnoreCase)).ToArray()).Trim()
: "";
return new Chapter(manga, null, null, chapterNumber, url, id); return new Chapter(manga, chapterName != "" ? chapterName : null, null, chapterNumber, url, id);
}).Where(elem => elem.chapterNumber != -1 && elem.url != "undefined").ToList(); }).Where(elem => elem.chapterNumber != -1 && elem.url != "undefined").ToList();
ret.Reverse(); ret.Reverse();
@ -182,7 +188,7 @@ public class Weebcentral : MangaConnector
return HttpStatusCode.RequestTimeout; return HttpStatusCode.RequestTimeout;
} }
var chapterParentManga = chapter.parentManga; Manga chapterParentManga = chapter.parentManga;
if (progressToken?.cancellationRequested ?? false) if (progressToken?.cancellationRequested ?? false)
{ {
progressToken.Cancel(); progressToken.Cancel();
@ -191,18 +197,18 @@ public class Weebcentral : MangaConnector
Log($"Retrieving chapter-info {chapter} {chapterParentManga}"); Log($"Retrieving chapter-info {chapter} {chapterParentManga}");
var requestResult = downloadClient.MakeRequest(chapter.url, RequestType.Default); RequestResult requestResult = downloadClient.MakeRequest(chapter.url, RequestType.Default);
if (requestResult.htmlDocument is null) if (requestResult.htmlDocument is null)
{ {
progressToken?.Cancel(); progressToken?.Cancel();
return HttpStatusCode.RequestTimeout; return HttpStatusCode.RequestTimeout;
} }
var document = requestResult.htmlDocument; HtmlDocument? document = requestResult.htmlDocument;
var imageNodes = HtmlNode[] imageNodes =
document.DocumentNode.SelectNodes($"//section[@hx-get='{chapter.url}/images']/img")?.ToArray() ?? []; 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 DownloadChapterImages(urls, chapter, RequestType.MangaImage, progressToken: progressToken); return DownloadChapterImages(urls, chapter, RequestType.MangaImage, progressToken: progressToken);
} }