Fixed: (norbits) Refactor parsing

pull/1386/head
Bogdan 2 years ago
parent ec389987df
commit b6018a4cd7

@ -5,11 +5,9 @@ using System.Globalization;
using System.Linq;
using System.Net.Http;
using System.Text;
using System.Text.RegularExpressions;
using System.Threading.Tasks;
using AngleSharp.Dom;
using AngleSharp.Html.Parser;
using FluentValidation;
using NLog;
using NzbDrone.Common.Extensions;
using NzbDrone.Common.Http;
@ -21,15 +19,13 @@ using NzbDrone.Core.IndexerSearch.Definitions;
using NzbDrone.Core.Messaging.Events;
using NzbDrone.Core.Parser;
using NzbDrone.Core.Parser.Model;
using NzbDrone.Core.ThingiProvider;
using NzbDrone.Core.Validation;
namespace NzbDrone.Core.Indexers.Definitions
{
namespace NzbDrone.Core.Indexers.Definitions;
public class NorBits : TorrentIndexerBase<NorBitsSettings>
{
public override string Name => "NorBits";
public override string[] IndexerUrls => new string[] { "https://norbits.net/" };
public override string[] IndexerUrls => new[] { "https://norbits.net/" };
public override string Description => "NorBits is a Norwegian Private site for MOVIES / TV / GENERAL";
public override string Language => "nb-NO";
public override Encoding Encoding => Encoding.GetEncoding("iso-8859-1");
@ -44,7 +40,7 @@ namespace NzbDrone.Core.Indexers.Definitions
public override IIndexerRequestGenerator GetRequestGenerator()
{
return new NorBitsRequestGenerator() { Settings = Settings, Capabilities = Capabilities };
return new NorBitsRequestGenerator(Settings, Capabilities);
}
public override IParseIndexerResponse GetParser()
@ -62,7 +58,7 @@ namespace NzbDrone.Core.Indexers.Definitions
var indexPage = await ExecuteAuth(requestBuilder.Build());
var loginUrl = string.Format("{0}/{1}", Settings.BaseUrl.TrimEnd('/'), "login.php");
var loginUrl = $"{Settings.BaseUrl.TrimEnd('/')}/login.php";
var requestBuilder2 = new HttpRequestBuilder(loginUrl)
{
@ -77,7 +73,7 @@ namespace NzbDrone.Core.Indexers.Definitions
// Get login page -- (not used, but simulation needed by tracker security's checks)
await ExecuteAuth(authLoginRequest);
var requestBuilder3 = new HttpRequestBuilder(string.Format("{0}/{1}", Settings.BaseUrl.TrimEnd('/'), "takelogin.php"))
var requestBuilder3 = new HttpRequestBuilder($"{Settings.BaseUrl.TrimEnd('/')}/takelogin.php")
{
LogResponseContent = true,
AllowAutoRedirect = true,
@ -95,29 +91,18 @@ namespace NzbDrone.Core.Indexers.Definitions
if (!loginResponse.GetCookies().ContainsKey("uid"))
{
// Default error message
var message = "Error during attempt !";
// Oops, unable to login
_logger.Info("NorBits - Login failed: " + message, "error");
throw new IndexerAuthException(message);
throw new IndexerAuthException("Login failed");
}
var cookies = loginResponse.GetCookies();
UpdateCookies(cookies, DateTime.Now + TimeSpan.FromDays(30));
_logger.Debug("NorBits authentication succeeded.");
_logger.Debug("Authentication succeeded.");
}
protected override bool CheckIfLoginNeeded(HttpResponse httpResponse)
{
if (!httpResponse.Content.Contains("logout.php"))
{
return true;
}
return false;
return !httpResponse.Content.Contains("logout.php");
}
private IndexerCapabilities SetCapabilities()
@ -170,51 +155,45 @@ namespace NzbDrone.Core.Indexers.Definitions
public class NorBitsRequestGenerator : IIndexerRequestGenerator
{
public NorBitsSettings Settings { get; set; }
public IndexerCapabilities Capabilities { get; set; }
private readonly NorBitsSettings _settings;
private readonly IndexerCapabilities _capabilities;
public NorBitsRequestGenerator()
public NorBitsRequestGenerator(NorBitsSettings settings, IndexerCapabilities capabilities)
{
_settings = settings;
_capabilities = capabilities;
}
private IEnumerable<IndexerRequest> GetPagedRequests(string term, int[] categories, string imdbId = null)
{
var searchUrl = string.Format("{0}/browse.php", Settings.BaseUrl.TrimEnd('/'));
var searchUrl = $"{_settings.BaseUrl.TrimEnd('/')}/browse.php";
var parameters = new NameValueCollection();
var categoriesList = Capabilities.Categories.MapTorznabCapsToTrackers(categories);
var searchterm = term;
var parameters = new NameValueCollection
{
{ "incldead", "1" },
{ "fullsearch", _settings.UseFullSearch ? "1" : "0" },
{ "scenerelease", "0" }
};
// Building our tracker query
parameters.Add("incldead", "1");
parameters.Add("fullsearch", Settings.UseFullSearch ? "1" : "0");
parameters.Add("scenerelease", "0");
var searchTerm = "search=";
// If search term provided
if (!string.IsNullOrWhiteSpace(imdbId))
{
searchterm = "imdbsearch=" + imdbId;
searchTerm = "imdbsearch=" + imdbId;
}
else if (!string.IsNullOrWhiteSpace(term))
{
searchterm = "search=" + term.UrlEncode(Encoding.GetEncoding(28591));
}
else
{
// Showing all torrents (just for output function)
searchterm = "search=";
searchTerm = "search=" + term.UrlEncode(Encoding.GetEncoding(28591));
}
var catQryStr = "";
searchUrl += "?" + searchTerm + "&" + parameters.GetQueryString();
foreach (var cat in categoriesList)
var categoriesList = _capabilities.Categories.MapTorznabCapsToTrackers(categories);
if (categoriesList.Any())
{
catQryStr += "&" + cat;
searchUrl += "&" + string.Join("&", categoriesList);
}
// Building our query
searchUrl += "?" + searchterm + "&" + parameters.GetQueryString() + "&" + catQryStr;
var request = new IndexerRequest(searchUrl, HttpAccept.Html);
yield return request;
@ -224,7 +203,7 @@ namespace NzbDrone.Core.Indexers.Definitions
{
var pageableRequests = new IndexerPageableRequestChain();
pageableRequests.Add(GetPagedRequests(string.Format("{0}", searchCriteria.SanitizedSearchTerm), searchCriteria.Categories, searchCriteria.FullImdbId));
pageableRequests.Add(GetPagedRequests($"{searchCriteria.SanitizedSearchTerm}", searchCriteria.Categories, searchCriteria.FullImdbId));
return pageableRequests;
}
@ -233,7 +212,7 @@ namespace NzbDrone.Core.Indexers.Definitions
{
var pageableRequests = new IndexerPageableRequestChain();
pageableRequests.Add(GetPagedRequests(string.Format("{0}", searchCriteria.SanitizedSearchTerm), searchCriteria.Categories));
pageableRequests.Add(GetPagedRequests($"{searchCriteria.SanitizedSearchTerm}", searchCriteria.Categories));
return pageableRequests;
}
@ -242,7 +221,7 @@ namespace NzbDrone.Core.Indexers.Definitions
{
var pageableRequests = new IndexerPageableRequestChain();
pageableRequests.Add(GetPagedRequests(string.Format("{0}", searchCriteria.SanitizedTvSearchString), searchCriteria.Categories, searchCriteria.FullImdbId));
pageableRequests.Add(GetPagedRequests($"{searchCriteria.SanitizedTvSearchString}", searchCriteria.Categories, searchCriteria.FullImdbId));
return pageableRequests;
}
@ -251,7 +230,7 @@ namespace NzbDrone.Core.Indexers.Definitions
{
var pageableRequests = new IndexerPageableRequestChain();
pageableRequests.Add(GetPagedRequests(string.Format("{0}", searchCriteria.SanitizedSearchTerm), searchCriteria.Categories));
pageableRequests.Add(GetPagedRequests($"{searchCriteria.SanitizedSearchTerm}", searchCriteria.Categories));
return pageableRequests;
}
@ -260,7 +239,7 @@ namespace NzbDrone.Core.Indexers.Definitions
{
var pageableRequests = new IndexerPageableRequestChain();
pageableRequests.Add(GetPagedRequests(string.Format("{0}", searchCriteria.SanitizedSearchTerm), searchCriteria.Categories));
pageableRequests.Add(GetPagedRequests($"{searchCriteria.SanitizedSearchTerm}", searchCriteria.Categories));
return pageableRequests;
}
@ -282,76 +261,45 @@ namespace NzbDrone.Core.Indexers.Definitions
public IList<ReleaseInfo> ParseResponse(IndexerResponse indexerResponse)
{
var torrentInfos = new List<ReleaseInfo>();
var releaseInfos = new List<ReleaseInfo>();
var parser = new HtmlParser();
var dom = parser.ParseDocument(indexerResponse.Content);
var firstPageRows = dom.QuerySelectorAll("#torrentTable > tbody > tr").Skip(1).ToCollection();
var rows = dom.QuerySelectorAll("#torrentTable > tbody > tr").Skip(1).ToCollection();
// If pagination available
int nbResults;
// Check if we have a minimum of one result
if (firstPageRows?.Length >= 1)
{
// Retrieve total count on our alone page
nbResults = firstPageRows.Count();
}
else
foreach (var row in rows)
{
// No result found for this query
return torrentInfos;
}
var link = _settings.BaseUrl + row.QuerySelector("td:nth-of-type(2) > a[href*=\"download.php?id=\"]")?.GetAttribute("href").TrimStart('/');
var qDetails = row.QuerySelector("td:nth-of-type(2) > a[href*=\"details.php?id=\"]");
var torrentDetailsUrl = _settings.BaseUrl + "details.php?id={id}";
var torrentDownloadUrl = _settings.BaseUrl + "download.php?id={id}&passkey={passkey}";
var title = qDetails?.GetAttribute("title").Trim();
var details = _settings.BaseUrl + qDetails?.GetAttribute("href").TrimStart('/');
// Loop on results
foreach (var row in firstPageRows)
{
var id = row.QuerySelector("td:nth-of-type(2) > a:nth-of-type(1)").GetAttribute("href").Split('=').Last(); // ID
var name = row.QuerySelector("td:nth-of-type(2) > a:nth-of-type(1)").GetAttribute("title"); // Release Name
var categoryName = row.QuerySelector("td:nth-of-type(1) > div > a:nth-of-type(1)").GetAttribute("title"); // Category
var mainCat = row.QuerySelector("td:nth-of-type(1) > div > a:nth-of-type(1)").GetAttribute("href").Split('?').Last();
var qSubCat2 = row.QuerySelector("td:nth-of-type(1) > div > a[href^=\"/browse.php?sub2_cat[]=\"]");
var cat = mainCat;
if (qSubCat2 != null)
{
cat += '&' + qSubCat2.GetAttribute("href").Split('?').Last();
}
var mainCategory = row.QuerySelector("td:nth-of-type(1) > div > a[href*=\"main_cat[]\"]")?.GetAttribute("href")?.Split('?').Last();
var secondCategory = row.QuerySelector("td:nth-of-type(1) > div > a[href*=\"sub2_cat[]\"]")?.GetAttribute("href")?.Split('?').Last();
var categoryList = new[] { mainCategory, secondCategory };
var cat = string.Join("&", categoryList.Where(c => !string.IsNullOrWhiteSpace(c)));
var seeders = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(9)").TextContent);
var leechers = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(10)").TextContent);
var seeders = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(9)").TextContent); // Seeders
var leechers = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(10)").TextContent); // Leechers
var regexObj = new Regex(@"[^\d]"); // Completed
var completed2 = row.QuerySelector("td:nth-of-type(8)").TextContent;
var completed = ParseUtil.CoerceInt(regexObj.Replace(completed2, ""));
var qFiles = row.QuerySelector("td:nth-of-type(3) > a"); // Files
var files = qFiles != null ? ParseUtil.CoerceInt(Regex.Match(qFiles.TextContent, @"\d+").Value) : 1;
var humanSize = row.QuerySelector("td:nth-of-type(7)").TextContent.ToLowerInvariant(); // Size
var size = ParseUtil.GetBytes(humanSize); // Date
var dateTimeOrig = row.QuerySelector("td:nth-of-type(5)").TextContent;
var dateTime = Regex.Replace(dateTimeOrig, @"<[^>]+>|&nbsp;", "").Trim();
var date = DateTime.ParseExact(dateTime, "yyyy-MM-ddHH:mm:ss", CultureInfo.InvariantCulture, DateTimeStyles.AssumeUniversal).ToLocalTime();
var details = new Uri(torrentDetailsUrl.Replace("{id}", id.ToString())); // Description Link
var passkey = row.QuerySelector("td:nth-of-type(2) > a:nth-of-type(2)").GetAttribute("href"); // Download Link
var key = Regex.Match(passkey, "(?<=passkey\\=)([a-zA-z0-9]*)");
var downloadLink = new Uri(torrentDownloadUrl.Replace("{id}", id.ToString()).Replace("{passkey}", key.ToString()));
// Building release infos
var release = new TorrentInfo
{
Guid = details,
InfoUrl = details,
DownloadUrl = link,
Title = title,
Categories = _categories.MapTrackerCatToNewznab(cat),
Title = name,
Size = ParseUtil.GetBytes(row.QuerySelector("td:nth-of-type(7)")?.TextContent),
Files = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(3) > a")?.TextContent.Trim()),
Grabs = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(8)")?.FirstChild?.TextContent.Trim()),
Seeders = seeders,
Peers = seeders + leechers,
PublishDate = date,
Size = size,
Files = files,
Grabs = completed,
Guid = details.AbsoluteUri,
InfoUrl = details.AbsoluteUri,
DownloadUrl = downloadLink.AbsoluteUri,
PublishDate = DateTime.ParseExact(row.QuerySelector("td:nth-of-type(5)")?.TextContent.Trim(), "yyyy-MM-ddHH:mm:ss", CultureInfo.InvariantCulture),
DownloadVolumeFactor = 1,
UploadVolumeFactor = 1,
MinimumRatio = 1,
MinimumSeedTime = 172800 // 48 hours
};
@ -359,10 +307,11 @@ namespace NzbDrone.Core.Indexers.Definitions
var genres = row.QuerySelector("span.genres")?.TextContent;
if (!string.IsNullOrEmpty(genres))
{
genres = genres.Trim().Replace("\xA0", " ").Replace("(", "").Replace(")", "").Replace(" | ", ",");
release.Description = genres;
release.Genres = genres.Split(',', StringSplitOptions.TrimEntries | StringSplitOptions.RemoveEmptyEntries).ToList();
}
// IMDB
var imdbLink = row.QuerySelector("a[href*=\"imdb.com/title/tt\"]")?.GetAttribute("href");
release.ImdbId = ParseUtil.GetImdbID(imdbLink) ?? 0;
@ -378,17 +327,11 @@ namespace NzbDrone.Core.Indexers.Definitions
{
release.DownloadVolumeFactor = 0.1;
}
else
{
release.DownloadVolumeFactor = 1;
}
release.UploadVolumeFactor = 1;
torrentInfos.Add(release);
releaseInfos.Add(release);
}
return torrentInfos.ToArray();
return releaseInfos.ToArray();
}
public Action<IDictionary<string, string>, DateTime?> CookiesUpdater { get; set; }
@ -398,9 +341,9 @@ namespace NzbDrone.Core.Indexers.Definitions
{
public NorBitsSettings()
{
UseFullSearch = false;
}
[FieldDefinition(4, Label = "Use Full Search", HelpText = "Use Full Search from Site", Type = FieldType.Checkbox)]
public bool UseFullSearch { get; set; }
}
}

Loading…
Cancel
Save