From 03bbafb0c98896a4da62a0d375350437820af02d Mon Sep 17 00:00:00 2001 From: Cory Date: Fri, 28 Feb 2020 15:40:27 -0600 Subject: [PATCH] Convert DigitalHive to AngleSharp (#7359) #7292 --- src/Jackett.Common/Indexers/DigitalHive.cs | 47 +++++++++++++--------- 1 file changed, 27 insertions(+), 20 deletions(-) diff --git a/src/Jackett.Common/Indexers/DigitalHive.cs b/src/Jackett.Common/Indexers/DigitalHive.cs index ab5fde5ce..c106641c7 100644 --- a/src/Jackett.Common/Indexers/DigitalHive.cs +++ b/src/Jackett.Common/Indexers/DigitalHive.cs @@ -4,7 +4,8 @@ using System.Collections.Specialized; using System.Linq; using System.Text; 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; @@ -90,8 +91,9 @@ namespace Jackett.Common.Indexers public override async Task GetConfigurationForSetup() { var loginPage = await RequestStringWithCookies(LoginUrl, configData.CookieHeader.Value); - CQ cq = loginPage.Content; - var recaptchaSiteKey = cq.Find(".g-recaptcha").Attr("data-sitekey"); + var parser = new HtmlParser(); + var cq = parser.ParseDocument(loginPage.Content); + var recaptchaSiteKey = cq.QuerySelector(".g-recaptcha").GetAttribute("data-sitekey"); if (recaptchaSiteKey != null) { var result = configData; @@ -149,7 +151,8 @@ namespace Jackett.Common.Indexers await ConfigureIfOK(result.Cookies, result.Content.Contains("logout.php"), () => { - CQ errorMessage = result.Content; + var parser = new HtmlParser(); + var errorMessage = parser.ParseDocument(result.Content); throw new ExceptionWithConfigData(errorMessage.Text(), configData); }); @@ -195,43 +198,47 @@ namespace Jackett.Common.Indexers return releases; } - private IEnumerable contentToReleaseInfos(TorznabQuery query, CQ dom) + private IEnumerable contentToReleaseInfos(TorznabQuery query, string content) { + var parser = new HtmlParser(); + var dom = parser.ParseDocument(content); var releases = new List(); // Doesn't handle pagination yet... - var rows = dom["div.panel-body > table.table > tbody > tr"]; + var rows = dom.QuerySelectorAll("div.panel-body > table.table > tbody > tr"); foreach (var row in rows) { var release = new ReleaseInfo(); release.MinimumRatio = 1; release.MinimumSeedTime = 259200; - var qRow = row.Cq(); - release.Title = qRow.Find("td:nth-child(2) > a").First().Text().Trim(); + + release.Title = row.QuerySelector("td:nth-child(2) > a").TextContent.Trim(); if ((query.ImdbID == null || !TorznabCaps.SupportsImdbMovieSearch) && !query.MatchQueryStringAND(release.Title)) continue; - release.Guid = new Uri(SiteLink + qRow.Find("td:nth-child(2) > a").First().Attr("href")); + release.Guid = new Uri(SiteLink + row.QuerySelector("td:nth-child(2) > a").GetAttribute("href")); release.Comments = release.Guid; - release.Link = new Uri(SiteLink + qRow.Find("td:nth-child(3) > a").First().Attr("href")); - var pubDateElement = qRow.Find("td:nth-child(2) > span").First(); - pubDateElement.Find("a").Remove(); // remove snatchinfo links (added after completing a torrent) - var pubDate = pubDateElement.Text().Trim().Replace("Added: ", ""); + release.Link = new Uri(SiteLink + row.QuerySelector("td:nth-child(3) > a").GetAttribute("href")); + var pubDateElement = row.QuerySelector("td:nth-child(2) > span"); + // remove snatchinfo links (added after completing a torrent) + foreach (var element in pubDateElement.QuerySelectorAll("a")) + element.Remove(); + var pubDate = pubDateElement.TextContent.Trim().Replace("Added: ", ""); release.PublishDate = DateTime.Parse(pubDate).ToLocalTime(); - release.Category = MapTrackerCatToNewznab(qRow.Find("td:nth-child(1) > a").First().Attr("href").Split('=')[1]); - release.Size = ReleaseInfo.GetBytes(qRow.Find("td:nth-child(7)").First().Text()); - release.Seeders = ParseUtil.CoerceInt(qRow.Find("td:nth-child(9)").First().Text()); - release.Peers = ParseUtil.CoerceInt(qRow.Find("td:nth-child(10)").First().Text()) + release.Seeders; + release.Category = MapTrackerCatToNewznab(row.QuerySelector("td:nth-child(1) > a").GetAttribute("href").Split('=')[1]); + release.Size = ReleaseInfo.GetBytes(row.QuerySelector("td:nth-child(7)").TextContent); + release.Seeders = ParseUtil.CoerceInt(row.QuerySelector("td:nth-child(9)").TextContent); + release.Peers = ParseUtil.CoerceInt(row.QuerySelector("td:nth-child(10)").TextContent) + release.Seeders; - var files = row.Cq().Find("td:nth-child(5)").Text(); + var files = row.QuerySelector("td:nth-child(5)").TextContent; release.Files = ParseUtil.CoerceInt(files); - var grabs = row.Cq().Find("td:nth-child(8)").Text(); + var grabs = row.QuerySelector("td:nth-child(8)").TextContent; release.Grabs = ParseUtil.CoerceInt(grabs); - if (row.Cq().Find("i.fa-star").Any()) + if (row.QuerySelector("i.fa-star") != null) release.DownloadVolumeFactor = 0; else release.DownloadVolumeFactor = 1;