Jackett/src/Jackett.Common/Indexers/Newpct.cs

711 lines
28 KiB
C#
Raw Normal View History

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 AngleSharp.Parser.Html;
using Jackett.Common.Models;
using Jackett.Common.Models.IndexerConfig;
using Jackett.Common.Services.Interfaces;
using Jackett.Common.Utils;
using Jackett.Common.Utils.Clients;
using Newtonsoft.Json.Linq;
using NLog;
using static Jackett.Common.Models.IndexerConfig.ConfigurationData;
namespace Jackett.Common.Indexers
{
public class Newpct : BaseCachingWebIndexer
{
enum ReleaseType
{
TV,
Movie,
}
class NewpctRelease : ReleaseInfo
{
public ReleaseType NewpctReleaseType;
public string SeriesName;
public int? Season;
public int? Episode;
public int? EpisodeTo;
public int Score;
public NewpctRelease()
{
}
public NewpctRelease(NewpctRelease copyFrom) :
base(copyFrom)
{
NewpctReleaseType = copyFrom.NewpctReleaseType;
SeriesName = copyFrom.SeriesName;
Season = copyFrom.Season;
Episode = copyFrom.Episode;
EpisodeTo = copyFrom.EpisodeTo;
Score = copyFrom.Score;
}
public override object Clone()
{
return new NewpctRelease(this);
}
}
2018-12-14 21:56:33 +00:00
private static Uri[] SiteLinkUris = new Uri[]
{
new Uri("http://descargas2020.com/"),
new Uri("http://torrentrapid.com/"),
new Uri("http://torrentlocura.com/"),
new Uri("http://tumejortorrent.com/"),
new Uri("http://www.tvsinpagar.com/"),
};
private NewpctRelease _mostRecentRelease;
2018-12-14 21:56:33 +00:00
private char[] _wordSeparators = new char[] { ' ', '.', ',', ';', '(', ')', '[', ']', '-', '_' };
private int _wordNotFoundScore = 100000;
private Regex _searchStringRegex = new Regex(@"(.+?)S0?(\d+)(E0?(\d+))?$", RegexOptions.IgnoreCase);
private Regex _titleListRegex = new Regex(@"Serie( *Descargar)?(.+?)(Temporada(.+?)(\d+)(.+?))?Capitulos?(.+?)(\d+)((.+?)(\d+))?(.+?)-(.+?)Calidad(.*)", RegexOptions.IgnoreCase);
private Regex _titleClassicRegex = new Regex(@"(\[[^\]]*\])?\[Cap\.(\d{1,2})(\d{2})([_-](\d{1,2})(\d{2}))?\]", RegexOptions.IgnoreCase);
private Regex _titleClassicTvQualityRegex = new Regex(@"\[([^\]]*HDTV[^\]]*)", RegexOptions.IgnoreCase);
private int _maxDailyPages = 7;
2018-12-14 21:56:33 +00:00
private int _maxMoviesPages = 30;
private int _maxEpisodesListPages = 100;
private int[] _allTvCategories = TorznabCatType.TV.SubCategories.Select(c => c.ID).ToArray();
private int[] _allMoviesCategories = TorznabCatType.Movies.SubCategories.Select(c => c.ID).ToArray();
private bool _includeVo;
private bool _filterMovies;
private DateTime _dailyNow;
private int _dailyResultIdx;
private string _searchUrl = "/buscar";
private string _dailyUrl = "/ultimas-descargas/pg/{0}";
private string[] _seriesLetterUrls = new string[] { "/series/letter/{0}", "/series-hd/letter/{0}" };
private string[] _seriesVOLetterUrls = new string[] { "/series-vo/letter/{0}" };
private string _seriesUrl = "{0}/pg/{1}";
private string[] _voUrls = new string[] { "serie-vo", "serievo" };
public Newpct(IIndexerConfigurationService configService, WebClient wc, Logger l, IProtectionService ps)
: base(name: "Newpct",
description: "Newpct - descargar torrent peliculas, series",
2018-12-14 21:56:33 +00:00
link: SiteLinkUris[0].AbsoluteUri,
caps: new TorznabCapabilities(TorznabCatType.TV,
TorznabCatType.TVSD,
TorznabCatType.TVHD,
TorznabCatType.Movies),
configService: configService,
client: wc,
logger: l,
p: ps,
configData: new ConfigurationData())
{
Encoding = Encoding.GetEncoding("windows-1252");
Language = "es-es";
Type = "public";
var voItem = new BoolItem() { Name = "Include original versions in search results", Value = false };
configData.AddDynamic("IncludeVo", voItem);
var filterMoviesItem = new BoolItem() { Name = "Only full match movies", Value = true };
configData.AddDynamic("FilterMovies", filterMoviesItem);
}
public override async Task<IndexerConfigurationStatus> ApplyConfiguration(JToken configJson)
{
configData.LoadValuesFromJson(configJson);
var releases = await PerformQuery(new TorznabQuery());
await ConfigureIfOK(string.Empty, releases.Count() > 0, () =>
{
throw new Exception("Could not find releases from this URL");
});
return IndexerConfigurationStatus.Completed;
}
protected override async Task<IEnumerable<ReleaseInfo>> PerformQuery(TorznabQuery query)
{
2018-12-14 21:56:33 +00:00
Uri link = new Uri(configData.SiteLink.Value);
lock (cache)
{
CleanCache();
}
return await PerformQuery(link, query, 0);
}
2018-12-14 21:56:33 +00:00
public override async Task<byte[]> Download(Uri linkParam)
{
2018-12-14 21:56:33 +00:00
List<string> links = new List<string>();
links.Add(linkParam.AbsoluteUri);
2018-12-14 21:56:33 +00:00
foreach (Uri extraSiteUri in SiteLinkUris)
{
UriBuilder ub = new UriBuilder(linkParam);
ub.Host = extraSiteUri.Host;
links.Add(ub.Uri.AbsoluteUri);
}
links = links.Distinct().ToList();
foreach (string link in links)
{
byte[] result = null;
2018-12-14 21:56:33 +00:00
try
{
var results = await RequestStringWithCookies(link);
var content = results.Content;
Regex regex = new Regex("[^\"]*/descargar-torrent/\\d+_[^\"]*");
Match match = regex.Match(content);
if (match.Success)
result = await base.Download(new Uri(match.Groups[0].Value));
}
catch
{
}
if (result != null)
return result;
else
this.logger.Warn("Newpct - download link not found in " + link);
}
return null;
}
2018-12-14 21:56:33 +00:00
private async Task<IEnumerable<ReleaseInfo>> PerformQuery(Uri siteLink, TorznabQuery query, int attempts)
{
var releases = new List<ReleaseInfo>();
_includeVo = ((BoolItem)configData.GetDynamic("IncludeVo")).Value;
_filterMovies = ((BoolItem)configData.GetDynamic("FilterMovies")).Value;
_dailyNow = DateTime.Now;
_dailyResultIdx = 0;
bool rssMode = string.IsNullOrEmpty(query.SanitizedSearchTerm);
if (rssMode)
{
int pg = 1;
while (pg <= _maxDailyPages)
{
2018-12-14 21:56:33 +00:00
Uri url = new Uri(siteLink, string.Format(_dailyUrl, pg));
var results = await RequestStringWithCookies(url.AbsoluteUri);
var items = ParseDailyContent(results.Content);
if (items == null || !items.Any())
break;
releases.AddRange(items);
//Check if we need to go to next page
bool recentFound = _mostRecentRelease != null &&
items.Any(r => r.Title == _mostRecentRelease.Title && r.Link.AbsoluteUri == _mostRecentRelease.Link.AbsoluteUri);
if (pg == 1)
_mostRecentRelease = (NewpctRelease)items.First().Clone();
if (recentFound)
break;
pg++;
}
}
else
{
bool isTvSearch = query.Categories == null || query.Categories.Length == 0 ||
query.Categories.Any(c => _allTvCategories.Contains(c));
if (isTvSearch)
{
2018-12-14 21:56:33 +00:00
releases.AddRange(await TvSearch(siteLink, query));
}
bool isMovieSearch = query.Categories == null || query.Categories.Length == 0 ||
query.Categories.Any(c => _allMoviesCategories.Contains(c));
if (isMovieSearch)
{
2018-12-14 21:56:33 +00:00
releases.AddRange(await MovieSearch(siteLink, query));
}
}
return releases;
}
2018-12-14 21:56:33 +00:00
private async Task<IEnumerable<ReleaseInfo>> TvSearch(Uri siteLink, TorznabQuery query)
{
List<ReleaseInfo> newpctReleases = null;
string seriesName = query.SanitizedSearchTerm;
int? season = query.Season > 0 ? (int?)query.Season : null;
int? episode = null;
if (!string.IsNullOrWhiteSpace(query.Episode) && int.TryParse(query.Episode, out int episodeTemp))
episode = episodeTemp;
//If query has no season/episode info, try to parse title
if (season == null && episode == null)
{
Match searchMatch = _searchStringRegex.Match(query.SanitizedSearchTerm);
if (searchMatch.Success)
{
seriesName = searchMatch.Groups[1].Value.Trim();
season = int.Parse(searchMatch.Groups[2].Value);
episode = searchMatch.Groups[4].Success ? (int?)int.Parse(searchMatch.Groups[4].Value) : null;
}
}
//Try to reuse cache
lock (cache)
{
var cachedResult = cache.FirstOrDefault(i => i.Query == seriesName.ToLower());
if (cachedResult != null)
newpctReleases = cachedResult.Results.Select(r => (ReleaseInfo)r.Clone()).ToList();
}
if (newpctReleases == null)
{
newpctReleases = new List<ReleaseInfo>();
//Search series url
2018-12-14 21:56:33 +00:00
foreach (Uri seriesListUrl in SeriesListUris(siteLink, seriesName))
{
newpctReleases.AddRange(await GetReleasesFromUri(seriesListUrl, seriesName));
}
//Sonarr removes "the" from shows. If there is nothing try prepending "the"
if (newpctReleases.Count == 0 && !(seriesName.ToLower().StartsWith("the")))
{
seriesName = "The " + seriesName;
2018-12-14 21:56:33 +00:00
foreach (Uri seriesListUrl in SeriesListUris(siteLink, seriesName))
{
newpctReleases.AddRange(await GetReleasesFromUri(seriesListUrl, seriesName));
}
}
//Cache ALL episodes
lock (cache)
{
cache.Add(new CachedQueryResult(seriesName.ToLower(), newpctReleases));
}
}
//Filter only episodes needed
return newpctReleases.Where(r =>
{
NewpctRelease nr = r as NewpctRelease;
return (
nr.Season.HasValue != season.HasValue || //Can't determine if same season
nr.Season.HasValue && season.Value == nr.Season.Value && //Same season and ...
(
nr.Episode.HasValue != episode.HasValue || //Can't determine if same episode
nr.Episode.HasValue &&
(
nr.Episode.Value == episode.Value || //Same episode
nr.EpisodeTo.HasValue && episode.Value >= nr.Episode.Value && episode.Value <= nr.EpisodeTo.Value //Episode in interval
)
)
);
});
}
private async Task<IEnumerable<ReleaseInfo>> GetReleasesFromUri(Uri uri, string seriesName)
{
var newpctReleases = new List<ReleaseInfo>();
var results = await RequestStringWithCookies(uri.AbsoluteUri);
//Episodes list
string seriesEpisodesUrl = ParseSeriesListContent(results.Content, seriesName);
if (!string.IsNullOrEmpty(seriesEpisodesUrl))
{
int pg = 1;
while (pg < _maxEpisodesListPages)
{
Uri episodesListUrl = new Uri(string.Format(_seriesUrl, seriesEpisodesUrl, pg));
results = await RequestStringWithCookies(episodesListUrl.AbsoluteUri);
var items = ParseEpisodesListContent(results.Content);
if (items == null || !items.Any())
break;
newpctReleases.AddRange(items);
pg++;
}
}
return newpctReleases;
}
2018-12-14 21:56:33 +00:00
private IEnumerable<Uri> SeriesListUris(Uri siteLink, string seriesName)
{
IEnumerable<string> lettersUrl;
if (!_includeVo)
{
lettersUrl = _seriesLetterUrls;
}
else
{
lettersUrl = _seriesLetterUrls.Concat(_seriesVOLetterUrls);
}
string seriesLetter = !char.IsDigit(seriesName[0]) ? seriesName[0].ToString() : "0-9";
return lettersUrl.Select(urlFormat =>
{
2018-12-14 21:56:33 +00:00
return new Uri(siteLink, string.Format(urlFormat, seriesLetter.ToLower()));
});
}
private IEnumerable<NewpctRelease> ParseDailyContent(string content)
{
var SearchResultParser = new HtmlParser();
var doc = SearchResultParser.Parse(content);
List<NewpctRelease> releases = new List<NewpctRelease>();
try
{
var rows = doc.QuerySelectorAll(".content .info");
foreach (var row in rows)
{
var anchor = row.QuerySelector("a");
var title = Regex.Replace(anchor.TextContent, @"\s+", " ").Trim();
var title2 = Regex.Replace(anchor.GetAttribute("title"), @"\s+", " ").Trim();
if (title2.Length >= title.Length)
title = title2;
var detailsUrl = anchor.GetAttribute("href");
if (!_includeVo && _voUrls.Any(vo => detailsUrl.ToLower().Contains(vo.ToLower())))
continue;
var span = row.QuerySelector("span");
var quality = span.ChildNodes[0].TextContent.Trim();
ReleaseType releaseType = ReleaseTypeFromQuality(quality);
var sizeText = span.ChildNodes[1].TextContent.Replace("Tama\u00F1o", "").Trim();
var div = row.QuerySelector("div");
var language = div.ChildNodes[1].TextContent.Trim();
_dailyResultIdx++;
NewpctRelease newpctRelease;
if (releaseType == ReleaseType.TV)
newpctRelease = GetReleaseFromData(releaseType,
string.Format("Serie {0} - {1} Calidad [{2}]", title, language, quality),
detailsUrl, quality, language, ReleaseInfo.GetBytes(sizeText), _dailyNow - TimeSpan.FromMilliseconds(_dailyResultIdx));
else
newpctRelease = GetReleaseFromData(releaseType,
string.Format("{0} [{1}][{2}]", title, quality, language),
detailsUrl, quality, language, ReleaseInfo.GetBytes(sizeText), _dailyNow - TimeSpan.FromMilliseconds(_dailyResultIdx));
releases.Add(newpctRelease);
}
}
catch (Exception ex)
{
OnParseError(content, ex);
}
return releases;
}
private string ParseSeriesListContent(string content, string title)
{
var SearchResultParser = new HtmlParser();
var doc = SearchResultParser.Parse(content);
Dictionary<string, string> results = new Dictionary<string, string>();
try
{
var rows = doc.QuerySelectorAll(".pelilist li a");
foreach (var anchor in rows)
{
var h2 = anchor.QuerySelector("h2");
if (h2.TextContent.Trim().ToLower() == title.Trim().ToLower())
return anchor.GetAttribute("href");
}
}
catch (Exception ex)
{
OnParseError(content, ex);
}
return null;
}
private IEnumerable<NewpctRelease> ParseEpisodesListContent(string content)
{
var SearchResultParser = new HtmlParser();
var doc = SearchResultParser.Parse(content);
List<NewpctRelease> releases = new List<NewpctRelease>();
try
{
var rows = doc.QuerySelectorAll(".content .info");
foreach (var row in rows)
{
var anchor = row.QuerySelector("a");
var title = anchor.TextContent.Replace("\t", "").Trim();
var detailsUrl = anchor.GetAttribute("href");
var span = row.QuerySelector("span");
var pubDateText = row.ChildNodes[3].TextContent.Trim();
var sizeText = row.ChildNodes[5].TextContent.Trim();
long size = ReleaseInfo.GetBytes(sizeText);
DateTime publishDate = DateTime.ParseExact(pubDateText, "dd-MM-yyyy", null);
NewpctRelease newpctRelease = GetReleaseFromData(ReleaseType.TV, title, detailsUrl, null, null, size, publishDate);
releases.Add(newpctRelease);
}
}
catch (Exception ex)
{
OnParseError(content, ex);
}
return releases;
}
2018-12-14 21:56:33 +00:00
private async Task<IEnumerable<ReleaseInfo>> MovieSearch(Uri siteLink, TorznabQuery query)
{
var releases = new List<NewpctRelease>();
string searchStr = query.SanitizedSearchTerm;
int pg = 1;
while (pg <= _maxMoviesPages)
{
var queryCollection = new Dictionary<string, string>();
queryCollection.Add("q", searchStr);
queryCollection.Add("pg", pg.ToString());
2018-12-14 21:56:33 +00:00
Uri url = new Uri(siteLink, string.Format(_searchUrl, pg));
var results = await PostDataWithCookies(url.AbsoluteUri, queryCollection);
var items = ParseSearchContent(results.Content);
2018-12-14 21:56:33 +00:00
if (items == null)
break;
releases.AddRange(items);
pg++;
}
ScoreReleases(releases, searchStr);
if (_filterMovies)
releases = releases.Where(r => r.Score < _wordNotFoundScore).ToList();
return releases;
}
private IEnumerable<NewpctRelease> ParseSearchContent(string content)
{
var SearchResultParser = new HtmlParser();
var doc = SearchResultParser.Parse(content);
List<NewpctRelease> releases = new List<NewpctRelease>();
try
{
var rows = doc.QuerySelectorAll(".content .info");
2018-12-14 21:56:33 +00:00
if (rows == null || !rows.Any())
return null;
foreach (var row in rows)
{
var anchor = row.QuerySelector("a");
var h2 = anchor.QuerySelector("h2");
var title = Regex.Replace(h2.TextContent, @"\s+", " ").Trim();
var detailsUrl = anchor.GetAttribute("href");
bool isSeries = h2.QuerySelector("span") != null && h2.TextContent.ToLower().Contains("calidad");
bool isGame = title.ToLower().Contains("pcdvd");
if (isSeries || isGame)
continue;
var span = row.QuerySelectorAll("span");
var pubDateText = span[1].TextContent.Trim();
var sizeText = span[2].TextContent.Trim();
long size = ReleaseInfo.GetBytes(sizeText);
DateTime publishDate = DateTime.ParseExact(pubDateText, "dd-MM-yyyy", null);
var div = row.QuerySelector("div");
NewpctRelease newpctRelease;
newpctRelease = GetReleaseFromData(ReleaseType.Movie, title, detailsUrl, null, null, size, publishDate);
releases.Add(newpctRelease);
}
}
catch (Exception ex)
{
OnParseError(content, ex);
}
return releases;
}
private void ScoreReleases(IEnumerable<NewpctRelease> releases, string searchTerm)
{
string[] searchWords = searchTerm.ToLower().Split(_wordSeparators, StringSplitOptions.None).
Select(s => s.Trim()).
Where(s => !string.IsNullOrEmpty(s)).ToArray();
foreach (NewpctRelease release in releases)
{
release.Score = 0;
string[] releaseWords = release.Title.ToLower().Split(_wordSeparators, StringSplitOptions.None).
Select(s => s.Trim()).
Where(s => !string.IsNullOrEmpty(s)).ToArray();
foreach (string search in searchWords)
{
int index = Array.IndexOf(releaseWords, search);
if (index >= 0)
{
release.Score += index;
releaseWords[index] = null;
}
else
{
release.Score += _wordNotFoundScore;
}
}
}
}
ReleaseType ReleaseTypeFromQuality(string quality)
{
if (quality.Trim().ToLower().StartsWith("hdtv"))
return ReleaseType.TV;
else
return ReleaseType.Movie;
}
NewpctRelease GetReleaseFromData(ReleaseType releaseType, string title, string detailsUrl, string quality, string language, long size, DateTime publishDate)
{
NewpctRelease result = new NewpctRelease();
result.NewpctReleaseType = releaseType;
//Sanitize
title = title.Replace("\t", "").Replace("\x2013", "-");
Match match = _titleListRegex.Match(title);
if (match.Success)
{
result.SeriesName = match.Groups[2].Value.Trim(' ', '-');
result.Season = int.Parse(match.Groups[5].Success ? match.Groups[5].Value.Trim() : "1");
result.Episode = int.Parse(match.Groups[8].Value.Trim().PadLeft(2, '0'));
result.EpisodeTo = match.Groups[11].Success ? (int?)int.Parse(match.Groups[11].Value.Trim()) : null;
string audioQuality = match.Groups[13].Value.Trim(' ', '[', ']');
if (string.IsNullOrEmpty(language))
language = audioQuality;
quality = match.Groups[14].Value.Trim(' ', '[', ']');
string seasonText = result.Season.ToString();
string episodeText = seasonText + result.Episode.ToString().PadLeft(2, '0');
string episodeToText = result.EpisodeTo.HasValue ? "_" + seasonText + result.EpisodeTo.ToString().PadLeft(2, '0') : "";
result.Title = string.Format("{0} - Temporada {1} [{2}][Cap.{3}{4}][{5}]",
result.SeriesName, seasonText, quality, episodeText, episodeToText, audioQuality);
}
else
{
Match matchClassic = _titleClassicRegex.Match(title);
if (matchClassic.Success)
{
result.Season = matchClassic.Groups[2].Success ? (int?)int.Parse(matchClassic.Groups[2].Value) : null;
result.Episode = matchClassic.Groups[3].Success ? (int?)int.Parse(matchClassic.Groups[3].Value) : null;
result.EpisodeTo = matchClassic.Groups[6].Success ? (int?)int.Parse(matchClassic.Groups[6].Value) : null;
if (matchClassic.Groups[1].Success)
quality = matchClassic.Groups[1].Value;
}
result.Title = title;
}
if (releaseType == ReleaseType.TV)
{
if (!string.IsNullOrWhiteSpace(quality) && (quality.Contains("720") || quality.Contains("1080")))
result.Category = new List<int> { TorznabCatType.TVHD.ID };
else
result.Category = new List<int> { TorznabCatType.TV.ID };
}
else
{
result.Title = title;
result.Category = new List<int> { TorznabCatType.Movies.ID };
}
result.Size = size;
result.Link = new Uri(detailsUrl);
result.Guid = result.Link;
result.PublishDate = publishDate;
result.Seeders = 1;
result.Peers = 1;
result.Title = FixedTitle(result, quality, language);
return result;
}
private string FixedTitle(NewpctRelease release, string quality, string language)
{
if (String.IsNullOrEmpty(release.SeriesName))
{
release.SeriesName = release.Title;
if (release.NewpctReleaseType == ReleaseType.TV && release.SeriesName.Contains("-"))
release.SeriesName = release.Title.Substring(0, release.SeriesName.IndexOf('-') - 1);
}
var titleParts = new List<string>();
titleParts.Add(release.SeriesName);
if (release.NewpctReleaseType == ReleaseType.TV)
{
if (String.IsNullOrEmpty(quality))
quality = "HDTV";
var seasonAndEpisode = "S" + release.Season.ToString().PadLeft(2, '0');
seasonAndEpisode += "E" + release.Episode.ToString().PadLeft(2, '0');
if (release.EpisodeTo != release.Episode && release.EpisodeTo != null && release.EpisodeTo != 0)
{
seasonAndEpisode += "-" + release.EpisodeTo.ToString().PadLeft(2, '0');
}
titleParts.Add(seasonAndEpisode);
}
if (!string.IsNullOrEmpty(quality) && !release.SeriesName.Contains(quality))
{
titleParts.Add(quality);
}
if (!string.IsNullOrWhiteSpace(language) && !release.SeriesName.Contains(language))
{
titleParts.Add(language);
}
if (release.Title.ToLower().Contains("espa\u00F1ol") ||
release.Title.ToLower().Contains("espanol") ||
release.Title.ToLower().Contains("castellano") ||
release.Title.ToLower().EndsWith("espa"))
{
titleParts.Add("Spanish");
}
string result = String.Join(".", titleParts);
result = Regex.Replace(result, @"[\[\]]+", ".");
result = Regex.Replace(result, @"\.[ \.]*\.", ".");
return result;
}
}
}