Convert HD-Space from csQuery to AngleSharp (#7352) #7292

This commit is contained in:
Cory 2020-02-28 16:32:02 -06:00 committed by GitHub
parent 63a3465aa6
commit 8056e47fef
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 62 additions and 84 deletions

View File

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