From 8056e47fefc02680dab0c7d3b41d77cac451aea0 Mon Sep 17 00:00:00 2001 From: Cory Date: Fri, 28 Feb 2020 16:32:02 -0600 Subject: [PATCH] Convert HD-Space from csQuery to AngleSharp (#7352) #7292 --- src/Jackett.Common/Indexers/HDSpace.cs | 146 +++++++++++-------------- 1 file changed, 62 insertions(+), 84 deletions(-) diff --git a/src/Jackett.Common/Indexers/HDSpace.cs b/src/Jackett.Common/Indexers/HDSpace.cs index e240986fa..f871dca63 100644 --- a/src/Jackett.Common/Indexers/HDSpace.cs +++ b/src/Jackett.Common/Indexers/HDSpace.cs @@ -1,12 +1,11 @@ using System; using System.Collections.Generic; using System.Collections.Specialized; -using System.Globalization; -using System.Linq; using System.Text; using System.Text.RegularExpressions; using System.Threading.Tasks; -using CsQuery; +using AngleSharp.Dom; +using AngleSharp.Html.Parser; using Jackett.Common.Models; using Jackett.Common.Models.IndexerConfig; using Jackett.Common.Services.Interfaces; @@ -30,14 +29,14 @@ namespace Jackett.Common.Indexers public HDSpace(IIndexerConfigurationService configService, WebClient wc, Logger l, IProtectionService ps) : base(name: "HD-Space", - description: "Sharing The Universe", - link: "https://hd-space.org/", - caps: TorznabUtil.CreateDefaultTorznabTVCaps(), - configService: configService, - client: wc, - logger: l, - p: ps, - configData: new ConfigurationDataBasicLogin()) + description: "Sharing The Universe", + link: "https://hd-space.org/", + caps: TorznabUtil.CreateDefaultTorznabTVCaps(), + configService: configService, + client: wc, + logger: l, + p: ps, + configData: new ConfigurationDataBasicLogin()) { Encoding = Encoding.UTF8; Language = "en-us"; @@ -45,31 +44,26 @@ namespace Jackett.Common.Indexers AddCategoryMapping(15, TorznabCatType.MoviesBluRay); // Movie / Blu-ray AddMultiCategoryMapping(TorznabCatType.MoviesHD, - 19, // Movie / 1080p - 41, // Movie / 4K UHD - 18, // Movie / 720p - 40, // Movie / Remux - 16 // Movie / HD-DVD - ); - + 19, // Movie / 1080p + 41, // Movie / 4K UHD + 18, // Movie / 720p + 40, // Movie / Remux + 16 // Movie / HD-DVD + ); AddMultiCategoryMapping(TorznabCatType.TVHD, - 21, // TV Show / 720p HDTV - 22 // TV Show / 1080p HDTV - ); - + 21, // TV Show / 720p HDTV + 22 // TV Show / 1080p HDTV + ); AddCategoryMapping(30, TorznabCatType.AudioLossless); // Music / Lossless AddCategoryMapping(31, TorznabCatType.AudioVideo); // Music / Videos - AddMultiCategoryMapping(TorznabCatType.TVDocumentary, - 24, // TV Show / Documentary / 720p - 25 // TV Show / Documentary / 1080p - ); - + 24, // TV Show / Documentary / 720p + 25 // TV Show / Documentary / 1080p + ); AddMultiCategoryMapping(TorznabCatType.XXX, - 33, // XXX / 720p - 34 // XXX / 1080p - ); - + 33, // XXX / 720p + 34 // XXX / 1080p + ); AddCategoryMapping("37", TorznabCatType.PC); AddCategoryMapping("38", TorznabCatType.Other); } @@ -77,21 +71,21 @@ namespace Jackett.Common.Indexers public override async Task ApplyConfiguration(JToken configJson) { LoadValuesFromJson(configJson); - var loginPage = await RequestStringWithCookies(LoginUrl, string.Empty); - - var pairs = new Dictionary { - { "uid", configData.Username.Value }, - { "pwd", configData.Password.Value } + var pairs = new Dictionary + { + {"uid", configData.Username.Value}, + {"pwd", configData.Password.Value} }; // Send Post var response = await RequestLoginAndFollowRedirect(LoginUrl, pairs, loginPage.Cookies, true, null, LoginUrl); - await ConfigureIfOK(response.Cookies, response.Content != null && response.Content.Contains("logout.php"), () => + await ConfigureIfOK(response.Cookies, response.Content?.Contains("logout.php") == true, () => { var errorStr = "You have {0} remaining login attempts"; - var remainingAttemptSpan = new Regex(string.Format(errorStr, "(.*?)")).Match(loginPage.Content).Groups[1].ToString(); + var remainingAttemptSpan = new Regex(string.Format(errorStr, "(.*?)")) + .Match(loginPage.Content).Groups[1].ToString(); var attempts = Regex.Replace(remainingAttemptSpan, "<.*?>", string.Empty); var errorMessage = string.Format(errorStr, attempts); throw new ExceptionWithConfigData(errorMessage, configData); @@ -102,28 +96,25 @@ namespace Jackett.Common.Indexers protected override async Task> PerformQuery(TorznabQuery query) { var releases = new List(); + var queryCollection = new NameValueCollection + { + {"active", "0"}, + {"options", "0"}, + {"category", string.Join(";", MapTorznabCapsToTrackers(query))} + }; var searchString = query.GetQueryString(); - var searchUrl = SearchUrl; - var queryCollection = new NameValueCollection(); - queryCollection.Add("active", "0"); - queryCollection.Add("options", "0"); - queryCollection.Add("category", string.Join(";", MapTorznabCapsToTrackers(query))); - if (!string.IsNullOrWhiteSpace(searchString)) - { queryCollection.Add("search", searchString); - } - searchUrl += queryCollection.GetQueryString(); - - var response = await RequestStringWithCookiesAndRetry(searchUrl); - var results = response.Content; + var response = await RequestStringWithCookiesAndRetry(SearchUrl + queryCollection.GetQueryString()); try { - CQ dom = results; - var rows = dom["table.lista > tbody > tr"]; + var resultParser = new HtmlParser(); + var searchResultDocument = resultParser.ParseDocument(response.Content); + var rows = searchResultDocument.QuerySelectorAll("table.lista > tbody > tr"); + foreach (var row in rows) { // this tracker has horrible markup, find the result rows by looking for the style tag before each one @@ -131,61 +122,48 @@ namespace Jackett.Common.Indexers if (prev == null || prev.NodeName.ToLowerInvariant() != "style") continue; - var qRow = row.Cq(); var release = new ReleaseInfo(); - release.MinimumRatio = 1; release.MinimumSeedTime = 172800; // 48 hours - var qLink = row.ChildElements.ElementAt(1).FirstElementChild.Cq(); - release.Title = qLink.Text().Trim(); - release.Comments = new Uri(SiteLink + qLink.Attr("href")); + var qLink = row.Children[1].FirstElementChild; + release.Title = qLink.TextContent.Trim(); + release.Comments = new Uri(SiteLink + qLink.GetAttribute("href")); release.Guid = release.Comments; - var qDownload = row.ChildElements.ElementAt(3).FirstElementChild.Cq(); - release.Link = new Uri(SiteLink + qDownload.Attr("href")); + var qDownload = row.Children[3].FirstElementChild; + release.Link = new Uri(SiteLink + qDownload.GetAttribute("href")); - //"July 11, 2015, 13:34:09", "Today at 20:04:23" - var dateStr = row.ChildElements.ElementAt(4).Cq().Text().Trim(); - if (dateStr.StartsWith("Today")) - release.PublishDate = DateTime.Today + TimeSpan.ParseExact(dateStr.Replace("Today at ", ""), "hh\\:mm\\:ss", CultureInfo.InvariantCulture); - else if (dateStr.StartsWith("Yesterday")) - release.PublishDate = DateTime.Today - TimeSpan.FromDays(1) + TimeSpan.ParseExact(dateStr.Replace("Yesterday at ", ""), "hh\\:mm\\:ss", CultureInfo.InvariantCulture); - else - release.PublishDate = DateTime.SpecifyKind(DateTime.ParseExact(dateStr, "MMMM dd, yyyy, HH:mm:ss", CultureInfo.InvariantCulture), DateTimeKind.Local); - - var sizeStr = row.ChildElements.ElementAt(5).Cq().Text(); + var dateStr = row.Children[4].TextContent.Trim(); + //"July 11, 2015, 13:34:09", "Today|Yesterday at 20:04:23" + release.PublishDate = DateTimeUtil.FromUnknown(dateStr.Replace("at", string.Empty)); + var sizeStr = row.Children[5].TextContent; release.Size = ReleaseInfo.GetBytes(sizeStr); - - release.Seeders = ParseUtil.CoerceInt(row.ChildElements.ElementAt(7).Cq().Text()); - release.Peers = ParseUtil.CoerceInt(row.ChildElements.ElementAt(8).Cq().Text()) + release.Seeders; - - var grabs = qRow.Find("td:nth-child(10)").Text(); + release.Seeders = ParseUtil.CoerceInt(row.Children[7].TextContent); + release.Peers = ParseUtil.CoerceInt(row.Children[8].TextContent) + release.Seeders; + var grabs = row.QuerySelector("td:nth-child(10)").TextContent; grabs = grabs.Replace("---", "0"); release.Grabs = ParseUtil.CoerceInt(grabs); - - if (qRow.Find("img[title=\"FreeLeech\"]").Length >= 1) + if (row.QuerySelector("img[title=\"FreeLeech\"]") != null) release.DownloadVolumeFactor = 0; - else if (qRow.Find("img[src=\"images/sf.png\"]").Length >= 1) // side freeleech + else if (row.QuerySelector("img[src=\"images/sf.png\"]") != null) // side freeleech release.DownloadVolumeFactor = 0; - else if (qRow.Find("img[title=\"Half FreeLeech\"]").Length >= 1) + else if (row.QuerySelector("img[title=\"Half FreeLeech\"]") != null) release.DownloadVolumeFactor = 0.5; else release.DownloadVolumeFactor = 1; - release.UploadVolumeFactor = 1; - - var qCat = qRow.Find("a[href^=\"index.php?page=torrents&category=\"]"); - var cat = qCat.Attr("href").Split('=')[2]; + var qCat = row.QuerySelector("a[href^=\"index.php?page=torrents&category=\"]"); + var cat = qCat.GetAttribute("href").Split('=')[2]; release.Category = MapTrackerCatToNewznab(cat); - releases.Add(release); } } catch (Exception ex) { - OnParseError(results, ex); + OnParseError(response.Content, ex); } + return releases; } }