Tranga/Jobs/JobBoss.cs

@@ -1,5 +1,4 @@
-using System.Diagnostics;
+using System.Runtime.InteropServices;
-using System.Runtime.InteropServices;
 using System.Text.RegularExpressions;
 using Newtonsoft.Json;
 using Tranga.MangaConnectors;
@@ -146,24 +145,26 @@ public class JobBoss : GlobalBase
 
     private void LoadJobsList(HashSet<MangaConnector> connectors)
     {
+        Directory.CreateDirectory(TrangaSettings.jobsFolderPath);
+        if(RuntimeInformation.IsOSPlatform(OSPlatform.Linux))
+            File.SetUnixFileMode(TrangaSettings.jobsFolderPath, UserRead | UserWrite | UserExecute | GroupRead | OtherRead);
         if (!Directory.Exists(TrangaSettings.jobsFolderPath)) //No jobs to load
-        {
-            Directory.CreateDirectory(TrangaSettings.jobsFolderPath);
-            if(RuntimeInformation.IsOSPlatform(OSPlatform.Linux))
-                File.SetUnixFileMode(TrangaSettings.jobsFolderPath, UserRead | UserWrite | UserExecute | GroupRead | OtherRead);
             return;
-        }
 
         //Load json-job-files
         foreach (FileInfo file in Directory.GetFiles(TrangaSettings.jobsFolderPath, "*.json").Select(f => new FileInfo(f)))
         {
             Log($"Adding {file.Name}");
-            try
+            Job? job = JsonConvert.DeserializeObject<Job>(File.ReadAllText(file.FullName),
+                new JobJsonConverter(this, new MangaConnectorJsonConverter(this, connectors)));
+            if (job is null)
+            {
+                string newName = file.FullName + ".failed";
+                Log($"Failed loading file {file.Name}.\nMoving to {newName}");
+                File.Move(file.FullName, newName);
+            }
+            else
             {
-                Job? job = JsonConvert.DeserializeObject<Job>(File.ReadAllText(file.FullName),
-                    new JobJsonConverter(this, new MangaConnectorJsonConverter(this, connectors)));
-                if (job is null) throw new NullReferenceException();
-                
                 Log($"Adding Job {job}");
                 if (!AddJob(job, file.FullName)) //If we detect a duplicate, delete the file.
                 {
@@ -172,17 +173,6 @@ public class JobBoss : GlobalBase
                     Log($"Duplicate detected or otherwise not able to add job to list.\nMoved job {job} to {path}");
                 }
             }
-            catch (Exception e)
-            {
-                if (e is not UnreachableException or NullReferenceException)
-                    throw;
-                Log(e.Message);
-                string newName = file.FullName + ".failed";
-                Log($"Failed loading file {file.Name}.\nMoving to {newName}.\n" +
-                    $"If you think this is a bug, upload contents of the file to the Bugreport!");
-                File.Move(file.FullName, newName);
-                continue;
-            }
         }
 
         //Connect jobs to parent-jobs and add Publications to cache

Tranga/MangaConnectors/ChromiumDownloadClient.cs

@@ -10,6 +10,7 @@ namespace Tranga.MangaConnectors;
 internal class ChromiumDownloadClient : DownloadClient
 {
     private static IBrowser? _browser;
+    private const int StartTimeoutMs = 10000;
     private readonly HttpDownloadClient _httpDownloadClient;
     
     private static async Task<IBrowser> StartBrowser(Logging.Logger? logger = null)
@@ -23,7 +24,7 @@ internal class ChromiumDownloadClient : DownloadClient
                 "--disable-dev-shm-usage",
                 "--disable-setuid-sandbox",
                 "--no-sandbox"},
-            Timeout = TrangaSettings.ChromiumStartupTimeoutMs
+            Timeout = StartTimeoutMs
         }, new LoggerFactory([new LogProvider(logger)]));
     }
 
@@ -69,19 +70,17 @@ internal class ChromiumDownloadClient : DownloadClient
 
     private RequestResult MakeRequestBrowser(string url, string? referrer = null, string? clickButton = null)
     {
-        if (_browser is null)
-            return new RequestResult(HttpStatusCode.InternalServerError, null, Stream.Null);
         IPage page = _browser.NewPageAsync().Result;
-        page.DefaultTimeout = TrangaSettings.ChromiumPageTimeoutMs;
+        page.DefaultTimeout = 10000;
         IResponse response;
         try
         {
             response = page.GoToAsync(url, WaitUntilNavigation.Networkidle0).Result;
-            Log($"Page loaded. {url}");
+            Log("Page loaded.");
         }
         catch (Exception e)
         {
-            Log($"Could not load Page {url}\n{e.Message}");
+            Log($"Could not load Page:\n{e.Message}");
             page.CloseAsync();
             return new RequestResult(HttpStatusCode.InternalServerError, null, Stream.Null);
         }

Tranga/MangaConnectors/MangaConnectorJsonConverter.cs

@@ -32,6 +32,7 @@ public class MangaConnectorJsonConverter : JsonConverter
             "MangaDex" => this._connectors.First(c => c is MangaDex),
             "Manganato" => this._connectors.First(c => c is Manganato),
             "MangaKatana" => this._connectors.First(c => c is MangaKatana),
+            "Mangasee" => this._connectors.First(c => c is Mangasee),
             "Mangaworld" => this._connectors.First(c => c is Mangaworld),
             "Bato" => this._connectors.First(c => c is Bato),
             "Manga4Life" => this._connectors.First(c => c is MangaLife),

Tranga/MangaConnectors/Mangasee.cs

@@ -0,0 +1,234 @@
+using System.Data;
+using System.Net;
+using System.Text.RegularExpressions;
+using System.Xml.Linq;
+using HtmlAgilityPack;
+using Newtonsoft.Json;
+using Soenneker.Utils.String.NeedlemanWunsch;
+using Tranga.Jobs;
+
+namespace Tranga.MangaConnectors;
+
+public class Mangasee : MangaConnector
+{
+    public Mangasee(GlobalBase clone) : base(clone, "Mangasee", ["en"])
+    {
+        this.downloadClient = new ChromiumDownloadClient(clone);
+    }
+
+    private struct SearchResult
+    {
+        public string i { get; set; }
+        public string s { get; set; }
+        public string[] a { get; set; }
+    }
+
+    public override Manga[] GetManga(string publicationTitle = "")
+    {
+        Log($"Searching Publications. Term=\"{publicationTitle}\"");
+        string requestUrl = "https://mangasee123.com/_search.php";
+        RequestResult requestResult =
+            downloadClient.MakeRequest(requestUrl, RequestType.Default);
+        if ((int)requestResult.statusCode < 200 || (int)requestResult.statusCode >= 300)
+        {
+            Log($"Failed to retrieve search: {requestResult.statusCode}");
+            return Array.Empty<Manga>();
+        }
+        
+        try
+        {
+            SearchResult[] searchResults = JsonConvert.DeserializeObject<SearchResult[]>(requestResult.htmlDocument!.DocumentNode.InnerText) ??
+                                           throw new NoNullAllowedException();
+            SearchResult[] filteredResults = FilteredResults(publicationTitle, searchResults);
+            Log($"Total available manga: {searchResults.Length} Filtered down to: {filteredResults.Length}");
+            
+
+            string[] urls = filteredResults.Select(result => $"https://mangasee123.com/manga/{result.i}").ToArray();
+            List<Manga> searchResultManga = new();
+            foreach (string url in urls)
+            {
+                Manga? newManga = GetMangaFromUrl(url);
+                if(newManga is { } manga)
+                    searchResultManga.Add(manga);
+            }
+            Log($"Retrieved {searchResultManga.Count} publications. Term=\"{publicationTitle}\"");
+            return searchResultManga.ToArray();
+        }
+        catch (NoNullAllowedException)
+        {
+            Log("Failed to retrieve search");
+            return Array.Empty<Manga>();
+        }
+    }
+
+    private readonly string[] _filterWords = {"a", "the", "of", "as", "to", "no", "for", "on", "with", "be", "and", "in", "wa", "at", "be", "ni"};
+    private string ToFilteredString(string input) => string.Join(' ', input.ToLower().Split(' ').Where(word => _filterWords.Contains(word) == false));
+    private SearchResult[] FilteredResults(string publicationTitle, SearchResult[] unfilteredSearchResults)
+    {
+        Dictionary<SearchResult, int> similarity = new();
+        foreach (SearchResult sr in unfilteredSearchResults)
+        {
+            List<int> scores = new();
+            string filteredPublicationString = ToFilteredString(publicationTitle);
+            string filteredSString = ToFilteredString(sr.s);
+            scores.Add(NeedlemanWunschStringUtil.CalculateSimilarity(filteredSString, filteredPublicationString));
+            foreach (string srA in sr.a)
+            {
+                string filteredAString = ToFilteredString(srA);
+                scores.Add(NeedlemanWunschStringUtil.CalculateSimilarity(filteredAString, filteredPublicationString));
+            }
+            similarity.Add(sr, scores.Sum() / scores.Count);
+        }
+
+        List<SearchResult> ret = similarity.OrderBy(s => s.Value).Take(10).Select(s => s.Key).ToList();
+        return ret.ToArray();
+    }
+
+    public override Manga? GetMangaFromId(string publicationId)
+    {
+        return GetMangaFromUrl($"https://mangasee123.com/manga/{publicationId}");
+    }
+
+    public override Manga? GetMangaFromUrl(string url)
+    {
+        Regex publicationIdRex = new(@"https:\/\/mangasee123.com\/manga\/(.*)(\/.*)*");
+        string publicationId = publicationIdRex.Match(url).Groups[1].Value;
+
+        RequestResult requestResult = this.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 ParseSinglePublicationFromHtml(HtmlDocument document, string publicationId, string websiteUrl)
+    {
+        string originalLanguage = "", status = "";
+        Dictionary<string, string> altTitles = new(), links = new();
+        HashSet<string> tags = new();
+        Manga.ReleaseStatusByte releaseStatus = Manga.ReleaseStatusByte.Unreleased;
+
+        HtmlNode posterNode = document.DocumentNode.SelectSingleNode("//div[@class='BoxBody']//div[@class='row']//img");
+        string posterUrl = posterNode.GetAttributeValue("src", "");
+        string coverFileNameInCache = SaveCoverImageToCache(posterUrl, publicationId, RequestType.MangaCover);
+
+        HtmlNode titleNode = document.DocumentNode.SelectSingleNode("//div[@class='BoxBody']//div[@class='row']//h1");
+        string sortName = titleNode.InnerText;
+
+        HtmlNode[] authorsNodes = document.DocumentNode
+            .SelectNodes("//div[@class='BoxBody']//div[@class='row']//span[text()='Author(s):']/..").Descendants("a")
+            .ToArray();
+        List<string> authors = new();
+        foreach (HtmlNode authorNode in authorsNodes)
+            authors.Add(authorNode.InnerText);
+
+        HtmlNode[] genreNodes = document.DocumentNode
+            .SelectNodes("//div[@class='BoxBody']//div[@class='row']//span[text()='Genre(s):']/..").Descendants("a")
+            .ToArray();
+        foreach (HtmlNode genreNode in genreNodes)
+            tags.Add(genreNode.InnerText);
+
+        HtmlNode yearNode = document.DocumentNode
+            .SelectNodes("//div[@class='BoxBody']//div[@class='row']//span[text()='Released:']/..").Descendants("a")
+            .First();
+        int year = Convert.ToInt32(yearNode.InnerText);
+
+        HtmlNode[] statusNodes = document.DocumentNode
+            .SelectNodes("//div[@class='BoxBody']//div[@class='row']//span[text()='Status:']/..").Descendants("a")
+            .ToArray();
+        foreach (HtmlNode statusNode in statusNodes)
+            if (statusNode.InnerText.Contains("publish", StringComparison.CurrentCultureIgnoreCase))
+                status = statusNode.InnerText.Split(' ')[0];
+        switch (status.ToLower())
+        {
+            case "cancelled": releaseStatus = Manga.ReleaseStatusByte.Cancelled; break;
+            case "hiatus": releaseStatus = Manga.ReleaseStatusByte.OnHiatus; break;
+            case "discontinued": releaseStatus = Manga.ReleaseStatusByte.Cancelled; break;
+            case "complete": releaseStatus = Manga.ReleaseStatusByte.Completed; break;
+            case "ongoing": releaseStatus = Manga.ReleaseStatusByte.Continuing; break;
+        }
+
+        HtmlNode descriptionNode = document.DocumentNode
+            .SelectNodes("//div[@class='BoxBody']//div[@class='row']//span[text()='Description:']/..")
+            .Descendants("div").First();
+        string description = descriptionNode.InnerText;
+
+        Manga manga = new(sortName, authors.ToList(), description, altTitles, tags.ToArray(), posterUrl,
+            coverFileNameInCache, links,
+            year, originalLanguage, publicationId, releaseStatus, websiteUrl: websiteUrl);
+        AddMangaToCache(manga);
+        return manga;
+    }
+
+    public override Chapter[] GetChapters(Manga manga, string language="en")
+    {
+        Log($"Getting chapters {manga}");
+        try
+        {
+            XDocument doc = XDocument.Load($"https://mangasee123.com/rss/{manga.publicationId}.xml");
+            XElement[] chapterItems = doc.Descendants("item").ToArray();
+            List<Chapter> chapters = new();
+            Regex chVolRex = new(@".*chapter-([0-9\.]+)(?:-index-([0-9\.]+))?.*");
+            foreach (XElement chapter in chapterItems)
+            {
+                string url = chapter.Descendants("link").First().Value;
+                Match m = chVolRex.Match(url);
+                string? volumeNumber = m.Groups[2].Success ? m.Groups[2].Value : "1";
+                string chapterNumber = m.Groups[1].Value;
+
+                string chapterUrl = Regex.Replace(url, @"-page-[0-9]+(\.html)", ".html");
+                try
+                {
+                    chapters.Add(new Chapter(manga, "", volumeNumber, chapterNumber, chapterUrl));
+                }
+                catch (Exception e)
+                {
+                    Log($"Failed to load chapter {chapterNumber}: {e.Message}");
+                }
+            }
+
+            //Return Chapters ordered by Chapter-Number
+            Log($"Got {chapters.Count} chapters. {manga}");
+            return chapters.Order().ToArray();
+        }
+        catch (HttpRequestException e)
+        {
+            Log($"Failed to load https://mangasee123.com/rss/{manga.publicationId}.xml \n\r{e}");
+            return Array.Empty<Chapter>();
+        }
+    }
+
+    public override HttpStatusCode DownloadChapter(Chapter chapter, ProgressToken? progressToken = null)
+    {
+        if (progressToken?.cancellationRequested ?? false)
+        {
+            progressToken.Cancel();
+            return HttpStatusCode.RequestTimeout;
+        }
+
+        Manga chapterParentManga = chapter.parentManga;
+        if (progressToken?.cancellationRequested ?? false)
+        {
+            progressToken.Cancel();
+            return HttpStatusCode.RequestTimeout;
+        }
+
+        Log($"Retrieving chapter-info {chapter} {chapterParentManga}");
+
+        RequestResult requestResult = this.downloadClient.MakeRequest(chapter.url, RequestType.Default);
+        if (requestResult.htmlDocument is null)
+        {
+            progressToken?.Cancel();
+            return HttpStatusCode.RequestTimeout;
+        }
+
+        HtmlDocument document = requestResult.htmlDocument;
+        
+        HtmlNode gallery = document.DocumentNode.Descendants("div").First(div => div.HasClass("ImageGallery"));
+        HtmlNode[] images = gallery.Descendants("img").Where(img => img.HasClass("img-fluid")).ToArray();
+        List<string> urls = new();
+        foreach(HtmlNode galleryImage in images)
+            urls.Add(galleryImage.GetAttributeValue("src", ""));
+        
+        return DownloadChapterImages(urls.ToArray(), chapter, RequestType.MangaImage, progressToken:progressToken);
+    }
+}

Tranga/Tranga.cs

@@ -18,6 +18,7 @@ public partial class Tranga : GlobalBase
         _connectors = new HashSet<MangaConnector>()
         {  
             new Manganato(this),
+            new Mangasee(this),
             new MangaDex(this),
             new MangaKatana(this),
             new Mangaworld(this),

Tranga/TrangaSettings.cs

@@ -35,8 +35,6 @@ public static class TrangaSettings
     };
 
     public static Dictionary<RequestType, int> requestLimits { get; set; } = DefaultRequestLimits;
-    public static int ChromiumStartupTimeoutMs { get; set; } = 30000;
-    public static int ChromiumPageTimeoutMs { get; set; } = 30000;
 
     public static void LoadFromWorkingDirectory(string directory)
     {
@@ -169,8 +167,6 @@ public static class TrangaSettings
         jobj.Add("requestLimits", JToken.FromObject(requestLimits));
         jobj.Add("bufferLibraryUpdates", JToken.FromObject(bufferLibraryUpdates));
         jobj.Add("bufferNotifications", JToken.FromObject(bufferNotifications));
-        jobj.Add("chromiumStartTimeout", JToken.FromObject(ChromiumStartupTimeoutMs));
-        jobj.Add("chromiumPageTimeout", JToken.FromObject(ChromiumPageTimeoutMs));
         return jobj;
     }
 
@@ -195,9 +191,5 @@ public static class TrangaSettings
             bufferLibraryUpdates = blu.Value<bool>()!;
         if (jobj.TryGetValue("bufferNotifications", out JToken? bn))
             bufferNotifications = bn.Value<bool>()!;
-        if (jobj.TryGetValue("chromiumStartTimeout", out JToken? cst))
-            ChromiumStartupTimeoutMs = cst.Value<int>();
-        if (jobj.TryGetValue("chromiumPageTimeout", out JToken? cpt))
-            ChromiumPageTimeoutMs = cpt.Value<int>();
     }
 }