mirror/Jackett
1
0
Fork 0
mirror of https://github.com/Jackett/Jackett synced 2024-12-28 02:39:26 +00:00
Code Issues Releases Wiki Activity

bithdtv: improve imdb search. resolves #7596 (#7629)

Browse source
This commit is contained in:
Cory 2020-03-14 13:16:03 -05:00 committed by GitHub
parent 8ebeb2bb17
commit 24f7fa461d
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
1 changed files with 72 additions and 79 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

179
src/Jackett.Common/Indexers/BitHdtv.cs
View file

@ -5,6 +5,7 @@ using System.Globalization;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using System.Web;
using AngleSharp.Html.Parser;
using Jackett.Common.Models;
using Jackett.Common.Models.IndexerConfig;
@ -20,8 +21,7 @@ namespace Jackett.Common.Indexers
{
private string LoginUrl => SiteLink + "login.php";
private string TakeLoginUrl => SiteLink + "takelogin.php";
private string SearchUrl => SiteLink + "torrents.php?";
private string DownloadUrl => SiteLink + "download.php?id={0}";
private string SearchUrl => SiteLink + "torrents.php";
private new ConfigurationDataRecaptchaLogin configData
{
@ -29,8 +29,9 @@ namespace Jackett.Common.Indexers
set => base.configData = value;
}
public BitHdtv(IIndexerConfigurationService configService, WebClient w, Logger l, IProtectionService ps)
: base(name: "BIT-HDTV",
public BitHdtv(IIndexerConfigurationService configService, WebClient w, Logger l, IProtectionService ps) :
base(
name: "BIT-HDTV",
description: "BIT-HDTV - Home of High Definition",
link: "https://www.bit-hdtv.com/",
caps: new TorznabCapabilities(),
@ -43,9 +44,7 @@ namespace Jackett.Common.Indexers
Encoding = Encoding.GetEncoding("iso-8859-1");
Language = "en-us";
Type = "private";
TorznabCaps.SupportsImdbMovieSearch = true;
AddCategoryMapping(1, TorznabCatType.TVAnime); // Anime
AddCategoryMapping(2, TorznabCatType.MoviesBluRay); // Blu-ray
AddCategoryMapping(4, TorznabCatType.TVDocumentary); // Documentaries
@ -76,13 +75,12 @@ namespace Jackett.Common.Indexers
public override async Task<IndexerConfigurationStatus> ApplyConfiguration(JToken configJson)
{
LoadValuesFromJson(configJson);
var pairs = new Dictionary<string, string> {
var pairs = new Dictionary<string, string>
{
{"username", configData.Username.Value},
{"password", configData.Password.Value},
{ "g-recaptcha-response", configData.Captcha.Value },
{"g-recaptcha-response", configData.Captcha.Value}
};
if (!string.IsNullOrWhiteSpace(configData.Captcha.Cookie))
{
// Cookie was manually supplied
@ -92,7 +90,6 @@ namespace Jackett.Common.Indexers
var results = await PerformQuery(new TorznabQuery());
if (!results.Any())
throw new Exception("Your cookie did not work");
IsConfigured = true;
SaveConfig();
return IndexerConfigurationStatus.Completed;
@ -104,17 +101,12 @@ namespace Jackett.Common.Indexers
}
}
var response = await RequestLoginAndFollowRedirect(TakeLoginUrl, pairs, null, true, null, SiteLink);
await ConfigureIfOK(response.Cookies, response.Content != null && response.Content.Contains("logout.php"), () =>
var response = await RequestLoginAndFollowRedirect(TakeLoginUrl, pairs, null, true, referer: SiteLink);
await ConfigureIfOK(response.Cookies, response.Content?.Contains("logout.php") == true, () =>
{
var parser = new HtmlParser();
var dom = parser.ParseDocument(response.Content);
var messageEl = dom.QuerySelectorAll("table.detail td.text").Last();
foreach (var child in messageEl.QuerySelectorAll("a"))
child.Remove();
foreach (var child in messageEl.QuerySelectorAll("style"))
child.Remove();
var errorMessage = messageEl.TextContent.Trim();
var errorMessage = dom.QuerySelector("table.detail td.text").FirstChild.TextContent.Trim();
throw new ExceptionWithConfigData(errorMessage, configData);
});
return IndexerConfigurationStatus.RequiresTesting;
@ -123,94 +115,95 @@ namespace Jackett.Common.Indexers
protected override async Task<IEnumerable<ReleaseInfo>> PerformQuery(TorznabQuery query)
{
var releases = new List<ReleaseInfo>();
var qc = new NameValueCollection();
if (!string.IsNullOrWhiteSpace(query.ImdbID))
var results = new List<WebClientStringResult>();
var search = new UriBuilder(SearchUrl);
if (query.IsImdbQuery)
{
qc.Add("search", query.ImdbID);
qc.Add("options", "4");
}
else if (!string.IsNullOrWhiteSpace(query.GetQueryString()))
{
qc.Add("search", query.GetQueryString());
qc.Add("options", "0");
}
var searchUrl = SearchUrl + qc.GetQueryString();
var trackerCats = MapTorznabCapsToTrackers(query, mapChildrenCatsToParent: true);
var results = await RequestStringWithCookiesAndRetry(searchUrl);
try
{
var parser = new HtmlParser();
var dom = parser.ParseDocument(results.Content);
foreach (var child in dom.QuerySelectorAll("#needseed"))
child.Remove();
foreach (var table in dom.QuerySelectorAll("table[align=center] + br + table > tbody"))
{
var rows = table.Children;
foreach (var row in rows.Skip(1))
{
var release = new ReleaseInfo();
var qLink = row.Children[2].QuerySelector("a");
release.MinimumRatio = 1;
release.MinimumSeedTime = 172800; // 48 hours
release.Title = qLink.GetAttribute("title");
if (!query.MatchQueryStringAND(release.Title))
continue;
release.Files = ParseUtil.CoerceLong(row.QuerySelector("td:nth-child(4)").TextContent);
release.Grabs = ParseUtil.CoerceLong(row.QuerySelector("td:nth-child(8)").TextContent);
release.Guid = new Uri(qLink.GetAttribute("href"));
release.Comments = release.Guid;
release.Link = new Uri(string.Format(DownloadUrl, qLink.GetAttribute("href").Split('=')[1]));
var catUrl = row.Children[1].FirstElementChild.GetAttribute("href");
var catNum = catUrl.Split('=', '&')[1];
release.Category = MapTrackerCatToNewznab(catNum);
// This tracker cannot search multiple cats at a time, so search all cats then filter out results from different cats
if (trackerCats.Count > 0 && !trackerCats.Contains(catNum))
continue;
var dateString = row.Children[5].TextContent.Trim();
var pubDate = DateTime.ParseExact(dateString, "yyyy-MM-ddHH:mm:ss", CultureInfo.InvariantCulture);
release.PublishDate = DateTime.SpecifyKind(pubDate, DateTimeKind.Local);
var sizeStr = row.Children[6].TextContent;
release.Size = ReleaseInfo.GetBytes(sizeStr);
release.Seeders = ParseUtil.CoerceInt(row.Children[8].TextContent.Trim());
release.Peers = ParseUtil.CoerceInt(row.Children[9].TextContent.Trim()) + release.Seeders;
var bgcolor = row.GetAttribute("bgcolor");
if (bgcolor == "#DDDDDD")
{
release.DownloadVolumeFactor = 1;
release.UploadVolumeFactor = 2;
}
else if (bgcolor == "#FFFF99")
{
release.DownloadVolumeFactor = 0;
release.UploadVolumeFactor = 1;
}
else if (bgcolor == "#CCFF99")
{
release.DownloadVolumeFactor = 0;
release.UploadVolumeFactor = 2;
qc.Add("options", "4"); //Search URL field for IMDB link
search.Query = qc.GetQueryString();
results.Add(await RequestStringWithCookiesAndRetry(search.ToString()));
qc["Options"] = "1"; //Search Title and Description
search.Query = qc.GetQueryString();
results.Add(await RequestStringWithCookiesAndRetry(search.ToString()));
}
else
{
//Site handles empty string on search param. No need to check for IsNullOrEmpty()
qc.Add("search", query.GetQueryString());
qc.Add("options", "0"); //Search Title Only
search.Query = qc.GetQueryString();
results.Add(await RequestStringWithCookiesAndRetry(search.ToString()));
}
var trackerCats = MapTorznabCapsToTrackers(query, true);
var parser = new HtmlParser();
foreach (var result in results)
try
{
var dom = parser.ParseDocument(result.Content);
foreach (var child in dom.QuerySelectorAll("#needseed"))
child.Remove();
var table = dom.QuerySelector("table[align=center] + br + table > tbody");
foreach (var row in table.Children.Skip(1))
{
var release = new ReleaseInfo();
var qLink = row.Children[2].QuerySelector("a");
release.MinimumRatio = 1;
release.MinimumSeedTime = 172800; // 48 hours
release.Title = qLink.GetAttribute("title");
var detailsLink = new Uri(qLink.GetAttribute("href"));
//Skip irrelevant and duplicate entries
if (!query.MatchQueryStringAND(release.Title) || releases.Any(r => r.Guid == detailsLink))
continue;
release.Files = ParseUtil.CoerceLong(row.Children[3].TextContent);
release.Grabs = ParseUtil.CoerceLong(row.Children[7].TextContent);
release.Guid = detailsLink;
release.Comments = release.Guid;
release.Link = new Uri(SiteLink + row.QuerySelector("a[href^=\"download.php\"]").GetAttribute("href"));
var catUrl = new Uri(SiteLink + row.Children[1].FirstElementChild.GetAttribute("href"));
var catQuery = HttpUtility.ParseQueryString(catUrl.Query);
var catNum = catQuery["cat"];
release.Category = MapTrackerCatToNewznab(catNum);
// This tracker cannot search multiple cats at a time
// so search all cats then filter out results from different cats
if (trackerCats.Any() && !trackerCats.Contains(catNum))
continue;
var dateString = row.Children[5].TextContent.Trim();
var pubDate = DateTime.ParseExact(dateString, "yyyy-MM-ddHH:mm:ss", CultureInfo.InvariantCulture);
release.PublishDate = DateTime.SpecifyKind(pubDate, DateTimeKind.Local);
var sizeStr = row.Children[6].TextContent;
release.Size = ReleaseInfo.GetBytes(sizeStr);
release.Seeders = ParseUtil.CoerceInt(row.Children[8].TextContent.Trim());
release.Peers = ParseUtil.CoerceInt(row.Children[9].TextContent.Trim()) + release.Seeders;
switch (row.GetAttribute("bgcolor"))
{
case "#DDDDDD":
release.DownloadVolumeFactor = 1;
release.UploadVolumeFactor = 2;
break;
case "#FFFF99":
release.DownloadVolumeFactor = 0;
release.UploadVolumeFactor = 1;
break;
case "#CCFF99":
release.DownloadVolumeFactor = 0;
release.UploadVolumeFactor = 2;
break;
default:
release.DownloadVolumeFactor = 1;
release.UploadVolumeFactor = 1;
break;
}
releases.Add(release);
}
}
}
catch (Exception ex)
{
OnParseError(results.Content, ex);
OnParseError(result.Content, ex);
}
return releases;