720-pier: Update parsing. Resolves #7877 resolves #7190 (#7895)

This commit is contained in:
Cory
2020-03-29 10:21:09 -05:00
committed by GitHub
parent 729edfa262
commit 2608a4cd44

View File

@@ -1,8 +1,8 @@
using System; using System;
using System.Collections.Generic; using System.Collections.Generic;
using System.Collections.Specialized; using System.Collections.Specialized;
using System.Linq;
using System.Text; using System.Text;
using System.Text.RegularExpressions;
using System.Threading.Tasks; using System.Threading.Tasks;
using AngleSharp.Html.Parser; using AngleSharp.Html.Parser;
using Jackett.Common.Models; using Jackett.Common.Models;
@@ -17,34 +17,20 @@ namespace Jackett.Common.Indexers
{ {
public class Pier720 : BaseWebIndexer public class Pier720 : BaseWebIndexer
{ {
private string LoginUrl => SiteLink + "ucp.php?mode=login"; public Pier720(IIndexerConfigurationService configService, WebClient wc, Logger l, IProtectionService ps) :
private string SearchUrl => SiteLink + "search.php"; base("720pier",
description: "720pier is a RUSSIAN Private Torrent Tracker for HD SPORTS",
public override string[] LegacySiteLinks { get; protected set; } = { link: "https://720pier.ru/",
"http://720pier.ru/", caps: TorznabUtil.CreateDefaultTorznabTVCaps(),
}; configService: configService,
client: wc,
private new ConfigurationDataBasicLoginWithRSSAndDisplay configData logger: l,
{ p: ps,
get => (ConfigurationDataBasicLoginWithRSSAndDisplay)base.configData; configData: new ConfigurationDataBasicLoginWithRSSAndDisplay())
set => base.configData = value;
}
public Pier720(IIndexerConfigurationService configService, WebClient wc, Logger l, IProtectionService ps)
: base(name: "720pier",
description: "720pier is a RUSSIAN Private Torrent Tracker for HD SPORTS",
link: "https://720pier.ru/",
caps: TorznabUtil.CreateDefaultTorznabTVCaps(),
configService: configService,
client: wc,
logger: l,
p: ps,
configData: new ConfigurationDataBasicLoginWithRSSAndDisplay())
{ {
Encoding = Encoding.UTF8; Encoding = Encoding.UTF8;
Language = "ru-ru"; Language = "ru-ru";
Type = "private"; Type = "private";
AddCategoryMapping(32, TorznabCatType.TVSport, "Basketball"); AddCategoryMapping(32, TorznabCatType.TVSport, "Basketball");
AddCategoryMapping(34, TorznabCatType.TVSport, "Basketball - NBA"); AddCategoryMapping(34, TorznabCatType.TVSport, "Basketball - NBA");
AddCategoryMapping(87, TorznabCatType.TVSport, "Basketball - NBA Playoffs"); AddCategoryMapping(87, TorznabCatType.TVSport, "Basketball - NBA Playoffs");
@@ -58,7 +44,6 @@ namespace Jackett.Common.Indexers
AddCategoryMapping(51, TorznabCatType.TVSport, "Basketball - Reviews and highlights"); AddCategoryMapping(51, TorznabCatType.TVSport, "Basketball - Reviews and highlights");
AddCategoryMapping(41, TorznabCatType.TVSport, "Basketball - Other"); AddCategoryMapping(41, TorznabCatType.TVSport, "Basketball - Other");
AddCategoryMapping(38, TorznabCatType.TVSport, "Basketball - Olympic Games"); AddCategoryMapping(38, TorznabCatType.TVSport, "Basketball - Olympic Games");
AddCategoryMapping(42, TorznabCatType.TVSport, "Football"); AddCategoryMapping(42, TorznabCatType.TVSport, "Football");
AddCategoryMapping(43, TorznabCatType.TVSport, "Football - NFL"); AddCategoryMapping(43, TorznabCatType.TVSport, "Football - NFL");
AddCategoryMapping(66, TorznabCatType.TVSport, "Football - Super Bowls"); AddCategoryMapping(66, TorznabCatType.TVSport, "Football - Super Bowls");
@@ -68,7 +53,6 @@ namespace Jackett.Common.Indexers
AddCategoryMapping(54, TorznabCatType.TVSport, "Football - Reviews and highlights"); AddCategoryMapping(54, TorznabCatType.TVSport, "Football - Reviews and highlights");
AddCategoryMapping(97, TorznabCatType.TVSport, "Football - Documentaries"); AddCategoryMapping(97, TorznabCatType.TVSport, "Football - Documentaries");
AddCategoryMapping(44, TorznabCatType.TVSport, "Football - Other"); AddCategoryMapping(44, TorznabCatType.TVSport, "Football - Other");
AddCategoryMapping(46, TorznabCatType.TVSport, "Hockey"); AddCategoryMapping(46, TorznabCatType.TVSport, "Hockey");
AddCategoryMapping(48, TorznabCatType.TVSport, "Hockey - NHL"); AddCategoryMapping(48, TorznabCatType.TVSport, "Hockey - NHL");
AddCategoryMapping(88, TorznabCatType.TVSport, "Hockey - NHL Playoffs"); AddCategoryMapping(88, TorznabCatType.TVSport, "Hockey - NHL Playoffs");
@@ -82,12 +66,10 @@ namespace Jackett.Common.Indexers
AddCategoryMapping(68, TorznabCatType.TVSport, "Hockey - Documentaries"); AddCategoryMapping(68, TorznabCatType.TVSport, "Hockey - Documentaries");
AddCategoryMapping(64, TorznabCatType.TVSport, "Hockey - Reviews and highlights"); AddCategoryMapping(64, TorznabCatType.TVSport, "Hockey - Reviews and highlights");
AddCategoryMapping(50, TorznabCatType.TVSport, "Hockey - Other"); AddCategoryMapping(50, TorznabCatType.TVSport, "Hockey - Other");
AddCategoryMapping(55, TorznabCatType.TVSport, "Baseball"); AddCategoryMapping(55, TorznabCatType.TVSport, "Baseball");
AddCategoryMapping(71, TorznabCatType.TVSport, "Baseball - MLB"); AddCategoryMapping(71, TorznabCatType.TVSport, "Baseball - MLB");
AddCategoryMapping(72, TorznabCatType.TVSport, "Baseball - Other"); AddCategoryMapping(72, TorznabCatType.TVSport, "Baseball - Other");
AddCategoryMapping(85, TorznabCatType.TVSport, "Baseball - Reviews, highlights, documentaries"); AddCategoryMapping(85, TorznabCatType.TVSport, "Baseball - Reviews, highlights, documentaries");
AddCategoryMapping(59, TorznabCatType.TVSport, "Soccer"); AddCategoryMapping(59, TorznabCatType.TVSport, "Soccer");
AddCategoryMapping(61, TorznabCatType.TVSport, "Soccer - English soccer"); AddCategoryMapping(61, TorznabCatType.TVSport, "Soccer - English soccer");
AddCategoryMapping(86, TorznabCatType.TVSport, "Soccer - UEFA"); AddCategoryMapping(86, TorznabCatType.TVSport, "Soccer - UEFA");
@@ -95,7 +77,6 @@ namespace Jackett.Common.Indexers
AddCategoryMapping(62, TorznabCatType.TVSport, "Soccer - Other tournaments, championships"); AddCategoryMapping(62, TorznabCatType.TVSport, "Soccer - Other tournaments, championships");
AddCategoryMapping(63, TorznabCatType.TVSport, "Soccer - World Championships"); AddCategoryMapping(63, TorznabCatType.TVSport, "Soccer - World Championships");
AddCategoryMapping(98, TorznabCatType.TVSport, "Soccer - FIFA World Cup"); AddCategoryMapping(98, TorznabCatType.TVSport, "Soccer - FIFA World Cup");
AddCategoryMapping(45, TorznabCatType.TVSport, "Other sports"); AddCategoryMapping(45, TorznabCatType.TVSport, "Other sports");
AddCategoryMapping(79, TorznabCatType.TVSport, "Other sports - Rugby"); AddCategoryMapping(79, TorznabCatType.TVSport, "Other sports - Rugby");
AddCategoryMapping(78, TorznabCatType.TVSport, "Other sports - Lacrosse"); AddCategoryMapping(78, TorznabCatType.TVSport, "Other sports - Lacrosse");
@@ -106,72 +87,60 @@ namespace Jackett.Common.Indexers
AddCategoryMapping(73, TorznabCatType.TVSport, "Other sports - Auto, moto racing"); AddCategoryMapping(73, TorznabCatType.TVSport, "Other sports - Auto, moto racing");
AddCategoryMapping(91, TorznabCatType.TVSport, "Other sports - Olympic Games"); AddCategoryMapping(91, TorznabCatType.TVSport, "Other sports - Olympic Games");
AddCategoryMapping(94, TorznabCatType.TVSport, "Other sports - Misc"); AddCategoryMapping(94, TorznabCatType.TVSport, "Other sports - Misc");
AddCategoryMapping(56, TorznabCatType.TVSport, "Sports on tv"); AddCategoryMapping(56, TorznabCatType.TVSport, "Sports on tv");
AddCategoryMapping(30, TorznabCatType.TVSport, "Sports"); AddCategoryMapping(30, TorznabCatType.TVSport, "Sports");
} }
private new ConfigurationDataBasicLoginWithRSSAndDisplay configData => (ConfigurationDataBasicLoginWithRSSAndDisplay)base.configData;
public override string[] LegacySiteLinks { get; protected set; } =
{
"http://720pier.ru/"
};
private string LoginUrl => SiteLink + "ucp.php?mode=login";
private string SearchUrl => SiteLink + "search.php";
public override async Task<IndexerConfigurationStatus> ApplyConfiguration(JToken configJson) public override async Task<IndexerConfigurationStatus> ApplyConfiguration(JToken configJson)
{ {
LoadValuesFromJson(configJson); LoadValuesFromJson(configJson);
var pairs = new Dictionary<string, string> var pairs = new Dictionary<string, string>
{ {
{"username", configData.Username.Value}, { "username", configData.Username.Value },
{"password", configData.Password.Value}, { "password", configData.Password.Value },
{"redirect", "/"}, { "redirect", "/" },
{"login", "Login"}, { "login", "Login" },
{"autologin", "on"} { "autologin", "on" }
}; };
var htmlParser = new HtmlParser(); var htmlParser = new HtmlParser();
var loginDocument = htmlParser.ParseDocument((await RequestStringWithCookies(LoginUrl)).Content); var loginDocument = htmlParser.ParseDocument((await RequestStringWithCookies(LoginUrl)).Content);
pairs["creation_time"] = loginDocument.GetElementsByName("creation_time")[0].GetAttribute("value"); pairs["creation_time"] = loginDocument.GetElementsByName("creation_time")[0].GetAttribute("value");
pairs["form_token"] = loginDocument.GetElementsByName("form_token")[0].GetAttribute("value"); pairs["form_token"] = loginDocument.GetElementsByName("form_token")[0].GetAttribute("value");
pairs["sid"] = loginDocument.GetElementsByName("sid")[0].GetAttribute("value"); pairs["sid"] = loginDocument.GetElementsByName("sid")[0].GetAttribute("value");
var result = await RequestLoginAndFollowRedirect(LoginUrl, pairs, null, true, null, LoginUrl, true); var result = await RequestLoginAndFollowRedirect(LoginUrl, pairs, null, true, null, LoginUrl, true);
await ConfigureIfOK(result.Cookies, result.Content?.Contains("ucp.php?mode=logout&") == true, () => await ConfigureIfOK(
{ result.Cookies, result.Content?.Contains("ucp.php?mode=logout&") == true,
var errorMessage = result.Content; () => throw new ExceptionWithConfigData(result.Content, configData));
throw new ExceptionWithConfigData(errorMessage, configData);
});
return IndexerConfigurationStatus.RequiresTesting; return IndexerConfigurationStatus.RequiresTesting;
} }
protected override async Task<IEnumerable<ReleaseInfo>> PerformQuery(TorznabQuery query) protected override async Task<IEnumerable<ReleaseInfo>> PerformQuery(TorznabQuery query)
{ {
var releases = new List<ReleaseInfo>();
var searchString = query.GetQueryString(); var searchString = query.GetQueryString();
var keywordSearch = !string.IsNullOrWhiteSpace(searchString);
var queryCollection = new NameValueCollection var releases = new List<ReleaseInfo>();
{ var queryCollection = !keywordSearch
{"st", "0"}, ? new NameValueCollection
{"sd", "d"}, {
{"sk", "t"}, { "search_id", "active_topics" }
{"tracker_search", "torrent"}, }
{"t", "0"}, : new NameValueCollection
{"submit", "Search"}, {
{"sr", "topics"}, { "sr", "posts" }, //Search all posts
{"ot", "1" } { "ot", "1" }, //Search only in forums trackers (checked)
}; { "keywords", searchString },
{ "sf", "titleonly" }
//queryCollection.Add("sr", "posts"); };
//queryCollection.Add("ch", "99999");
// if the search string is empty use the getnew view
if (string.IsNullOrWhiteSpace(searchString))
{
queryCollection.Add("search_id", "active_topics");
}
else // use the normal search
{
searchString = searchString.Replace("-", " ");
queryCollection.Add("keywords", searchString);
queryCollection.Add("sf", "titleonly");
queryCollection.Add("sr", "topics");
queryCollection.Add("pt", "t");
}
var searchUrl = SearchUrl + "?" + queryCollection.GetQueryString(); var searchUrl = SearchUrl + "?" + queryCollection.GetQueryString();
var results = await RequestStringWithCookies(searchUrl); var results = await RequestStringWithCookies(searchUrl);
if (!results.Content.Contains("ucp.php?mode=logout")) if (!results.Content.Contains("ucp.php?mode=logout"))
@@ -179,68 +148,53 @@ namespace Jackett.Common.Indexers
await ApplyConfiguration(null); await ApplyConfiguration(null);
results = await RequestStringWithCookies(searchUrl); results = await RequestStringWithCookies(searchUrl);
} }
try try
{ {
const string rowsSelector = "ul.topics > li.row";
var resultParser = new HtmlParser(); var resultParser = new HtmlParser();
var searchResultDocument = resultParser.ParseDocument(results.Content); var searchResultDocument = resultParser.ParseDocument(results.Content);
var rows = searchResultDocument.QuerySelectorAll(rowsSelector); var rowSelector = keywordSearch
foreach (var row in rows) ? "div.search div.postbody > h3 > a"
: "ul.topics > li.row:has(i.fa-paperclip) a.topictitle"; // Torrent lines have paperclip icon. Chat topics don't
var rows = searchResultDocument.QuerySelectorAll(rowSelector);
foreach (var rowLink in rows)
{ {
try var detailLink = SiteLink + rowLink.GetAttribute("href");
var detailsResult = await RequestStringWithCookies(detailLink);
var detailsDocument = resultParser.ParseDocument(detailsResult.Content);
var detailRow = detailsDocument.QuerySelector("table.table2 > tbody > tr");
if (detailRow == null)
continue; //No torrents in result
var qDownloadLink = detailRow.QuerySelector("a[href^=\"/download/torrent\"]");
var link = new Uri(SiteLink + qDownloadLink.GetAttribute("href").TrimStart('/'));
var timestr = detailRow.Children[0].QuerySelector("ul.dropdown-contents span.my_tt").TextContent;
var publishDate = DateTimeUtil.FromUnknown(timestr, "UK");
var forumId = detailsDocument.QuerySelector("li.breadcrumbs").LastElementChild
.GetAttribute("data-forum-id");
var sizeString = detailRow.Children[4].QuerySelector("span.my_tt").GetAttribute("title");
var size = ParseUtil.CoerceLong(Regex.Replace(sizeString, @"[^0-9]", string.Empty));
var comments = new Uri(detailLink);
var grabs = ParseUtil.CoerceInt(detailRow.Children[0].QuerySelector("span.complet").TextContent);
var seeders = ParseUtil.CoerceInt(detailRow.Children[2].QuerySelector("span.seed").TextContent);
var leechers = ParseUtil.CoerceInt(detailRow.Children[3].QuerySelector("span.leech").TextContent);
var release = new ReleaseInfo
{ {
var seeders = ParseUtil.CoerceInt(row.QuerySelector("span.seed").TextContent); MinimumRatio = 1,
var grabs = ParseUtil.CoerceLong(row.QuerySelector("span.complet").TextContent); MinimumSeedTime = 0,
var qDetailsLink = row.QuerySelector("a.topictitle"); DownloadVolumeFactor = 1,
var detailsResult = await RequestStringWithCookies(SiteLink + qDetailsLink.GetAttribute("href")); UploadVolumeFactor = 1,
var detailsResultDocument = resultParser.ParseDocument(detailsResult.Content); Seeders = seeders,
var qDownloadLink = detailsResultDocument.QuerySelector("table.table2 > tbody > tr > td > a[href^=\"/download/torrent\"]"); Grabs = grabs,
var author = row.QuerySelector("dd.lastpost > span"); Peers = leechers + seeders,
var timestr = author.TextContent.Split('\n') Title = rowLink.TextContent,
.Where(str => !string.IsNullOrWhiteSpace(str)) //Filter blank lines Comments = comments,
.Skip(1) //Skip author name Guid = comments,
.FirstOrDefault() Link = link,
.Trim(); PublishDate = publishDate,
Category = MapTrackerCatToNewznab(forumId),
var forum = row.QuerySelector("a[href^=\"./viewforum.php?f=\"]"); Size = size,
var forumid = forum.GetAttribute("href").Split('=')[1]; };
var sizeString = row.QuerySelector("dl.row-item > dt > div.list-inner > div[style^=\"float:right\"]") releases.Add(release);
.TextContent
.Replace("GiB", "GB")
.Replace("MiB", "MB")
.Replace("KiB", "KB")
.Replace("ГБ", "GB")
.Replace("МБ", "MB")
.Replace("КБ", "KB");
var comments = new Uri(SiteLink + qDetailsLink.GetAttribute("href"));
var leechers = ParseUtil.CoerceInt(row.QuerySelector("span.leech").TextContent);
var link = new Uri(SiteLink + qDownloadLink.GetAttribute("href").TrimStart('/'));
var publishDate = DateTimeUtil.FromUnknown(timestr, "UK");
var size = ReleaseInfo.GetBytes(sizeString);
var release = new ReleaseInfo
{
MinimumRatio = 1,
MinimumSeedTime = 0,
DownloadVolumeFactor = 1,
UploadVolumeFactor = 1,
Seeders = seeders,
Grabs = grabs,
Peers = leechers + seeders,
Title = qDetailsLink.TextContent,
Comments = comments,
Guid = comments,
Link = link,
PublishDate = publishDate,
Category = MapTrackerCatToNewznab(forumid),
Size = size,
};
releases.Add(release);
}
catch (Exception ex)
{
logger.Error($"{ID}: Error while parsing row '{row.OuterHtml}':\n\n{ex}");
}
} }
} }
catch (Exception ex) catch (Exception ex)