shpaq/Jackett-Jackett
1
0
Fork 0
mirror of https://github.com/Jackett/Jackett.git synced 2025-09-17 17:34:09 +02:00
Code Issues Packages Projects Releases Wiki Activity

Convert NCore to AngleSharp (#7378)

Browse Source
This commit is contained in:
Cory
2020-02-29 13:03:51 -06:00
committed by GitHub
parent 65fc651d88
commit e3bb63aac4
1 changed files with 26 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files

49
src/Jackett.Common/Indexers/NCore.cs
View File

@@ -5,7 +5,7 @@ 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.Html.Parser;
using Jackett.Common.Models; using Jackett.Common.Models;
using Jackett.Common.Models.IndexerConfig.Bespoke; using Jackett.Common.Models.IndexerConfig.Bespoke;
using Jackett.Common.Services.Interfaces; using Jackett.Common.Services.Interfaces;
@@ -107,10 +107,11 @@ namespace Jackett.Common.Indexers
var result = await RequestLoginAndFollowRedirect(LoginUrl, pairs, loginPage.Cookies, true, referer: SiteLink); var result = await RequestLoginAndFollowRedirect(LoginUrl, pairs, loginPage.Cookies, true, referer: SiteLink);
await ConfigureIfOK(result.Cookies, result.Content != null && result.Content.Contains("profile.php"), () => await ConfigureIfOK(result.Cookies, result.Content != null && result.Content.Contains("profile.php"), () =>
{ {
CQ dom = result.Content; var parser = new HtmlParser();
var messageEl = dom["#hibauzenet table tbody tr"]; var dom = parser.ParseDocument(result.Content);
var msgContainer = messageEl.Get(0).ChildElements.ElementAt(1); var messageEl = dom.QuerySelector("#hibauzenet table tbody tr");
var errorMessage = msgContainer != null ? msgContainer.InnerText : "Error while trying to login."; var msgContainer = messageEl.Children[1];
var errorMessage = msgContainer != null ? msgContainer.TextContent : "Error while trying to login.";
throw new ExceptionWithConfigData(errorMessage, configData); throw new ExceptionWithConfigData(errorMessage, configData);
}); });
@@ -122,24 +123,25 @@ namespace Jackett.Common.Indexers
var releases = new List<ReleaseInfo>(); var releases = new List<ReleaseInfo>();
try try
{ {
CQ dom = results.Content; var parser = new HtmlParser();
var dom = parser.ParseDocument(results.Content);
ReleaseInfo release; ReleaseInfo release;
var rows = dom[".box_torrent_all"].Find(".box_torrent"); var rows = dom.QuerySelector(".box_torrent_all").QuerySelectorAll(".box_torrent");
// Check torrents only till we reach the query Limit // Check torrents only till we reach the query Limit
for (var i = previously_parsed_on_page; (i < rows.Length && ((already_founded + releases.Count) < limit)); i++) for (var i = previously_parsed_on_page; (i < rows.Length && ((already_founded + releases.Count) < limit)); i++)
{ {
try try
{ {
var qRow = rows[i].Cq(); var row = rows[i];
var key = dom["link[rel=alternate]"].First().Attr("href").Split('=').Last(); var key = dom.QuerySelector("link[rel=alternate]").GetAttribute("href").Split('=').Last();
release = new ReleaseInfo(); release = new ReleaseInfo();
var torrentTxt = qRow.Find(".torrent_txt, .torrent_txt2").Find("a").Get(0); var torrentTxt = row.QuerySelector(".torrent_txt, .torrent_txt2").QuerySelector("a");
//if (torrentTxt == null) continue; //if (torrentTxt == null) continue;
release.Title = torrentTxt.GetAttribute("title"); release.Title = torrentTxt.GetAttribute("title");
release.Description = qRow.Find("span").Get(0).GetAttribute("title") + " " + qRow.Find("a.infolink").Text(); release.Description = row.QuerySelector("span").GetAttribute("title") + " " + row.QuerySelector("a.infolink").TextContent;
release.MinimumRatio = 1; release.MinimumRatio = 1;
release.MinimumSeedTime = 172800; // 48 hours release.MinimumSeedTime = 172800; // 48 hours
@@ -153,11 +155,11 @@ namespace Jackett.Common.Indexers
release.Comments = new Uri(SiteLink.ToString() + "torrents.php?action=details&id=" + downloadId); release.Comments = new Uri(SiteLink.ToString() + "torrents.php?action=details&id=" + downloadId);
release.Guid = new Uri(release.Comments.ToString() + "#comments"); release.Guid = new Uri(release.Comments.ToString() + "#comments");
; ;
release.Seeders = ParseUtil.CoerceInt(qRow.Find(".box_s2").Find("a").First().Text()); release.Seeders = ParseUtil.CoerceInt(row.QuerySelector(".box_s2").QuerySelector("a").TextContent);
release.Peers = ParseUtil.CoerceInt(qRow.Find(".box_l2").Find("a").First().Text()) + release.Seeders; release.Peers = ParseUtil.CoerceInt(row.QuerySelector(".box_l2").QuerySelector("a").TextContent) + release.Seeders;
var imdblink = qRow.Find("a[href*=\".imdb.com/title\"]").Attr("href"); var imdblink = row.QuerySelector("a[href*=\".imdb.com/title\"]").GetAttribute("href");
release.Imdb = ParseUtil.GetLongFromString(imdblink); release.Imdb = ParseUtil.GetLongFromString(imdblink);
var banner = qRow.Find("img.infobar_ico").Attr("onmouseover"); var banner = row.QuerySelector("img.infobar_ico").GetAttribute("onmouseover");
if (banner != null) if (banner != null)
{ {
var BannerRegEx = new Regex(@"mutat\('(.*?)', '", RegexOptions.Compiled); var BannerRegEx = new Regex(@"mutat\('(.*?)', '", RegexOptions.Compiled);
@@ -165,10 +167,10 @@ namespace Jackett.Common.Indexers
var bannerurl = BannerMatch.Groups[1].Value; var bannerurl = BannerMatch.Groups[1].Value;
release.BannerUrl = new Uri(bannerurl); release.BannerUrl = new Uri(bannerurl);
} }
release.PublishDate = DateTime.Parse(qRow.Find(".box_feltoltve2").Get(0).InnerHTML.Replace("<br />", " "), CultureInfo.InvariantCulture); release.PublishDate = DateTime.Parse(row.QuerySelector(".box_feltoltve2").InnerHtml.Replace("<br />", " "), CultureInfo.InvariantCulture);
var sizeSplit = qRow.Find(".box_meret2").Get(0).InnerText.Split(' '); var sizeSplit = row.QuerySelector(".box_meret2").TextContent.Split(' ');
release.Size = ReleaseInfo.GetBytes(sizeSplit[1].ToLower(), ParseUtil.CoerceFloat(sizeSplit[0])); release.Size = ReleaseInfo.GetBytes(sizeSplit[1].ToLower(), ParseUtil.CoerceFloat(sizeSplit[0]));
var catlink = qRow.Find("a:has(img[class='categ_link'])").First().Attr("href"); var catlink = row.QuerySelector("a:has(img[class='categ_link'])").GetAttribute("href");
var cat = ParseUtil.GetArgumentFromQueryString(catlink, "tipus"); var cat = ParseUtil.GetArgumentFromQueryString(catlink, "tipus");
release.Category = MapTrackerCatToNewznab(cat); release.Category = MapTrackerCatToNewznab(cat);
@@ -217,7 +219,7 @@ namespace Jackett.Common.Indexers
} }
catch (FormatException ex) catch (FormatException ex)
{ {
logger.Error("Problem of parsing Torrent:" + rows[i].InnerHTML); logger.Error("Problem of parsing Torrent:" + rows[i].InnerHtml);
logger.Error("Exception was the following:" + ex); logger.Error("Exception was the following:" + ex);
} }
} }
@@ -270,11 +272,12 @@ namespace Jackett.Common.Indexers
var results = await PostDataWithCookiesAndRetry(SearchUrl, pairs); var results = await PostDataWithCookiesAndRetry(SearchUrl, pairs);
CQ dom = results.Content; var parser = new HtmlParser();
var dom = parser.ParseDocument(results.Content);
var numVal = 0; var numVal = 0;
// find number of torrents / page // find number of torrents / page
var torrent_per_page = dom[".box_torrent_all"].Find(".box_torrent").Length; var torrent_per_page = dom.QuerySelector(".box_torrent_all").QuerySelectorAll(".box_torrent").Length;
if (torrent_per_page == 0) if (torrent_per_page == 0)
return releases; return releases;
var start_page = (query.Offset / torrent_per_page) + 1; var start_page = (query.Offset / torrent_per_page) + 1;
@@ -283,13 +286,13 @@ namespace Jackett.Common.Indexers
previously_parsed_on_page = query.Offset; previously_parsed_on_page = query.Offset;
// find pagelinks in the bottom // find pagelinks in the bottom
var pagelinks = dom["div[id=pager_bottom]"].Find("a"); var pagelinks = dom.QuerySelector("div[id=pager_bottom]").QuerySelectorAll("a");
if (pagelinks.Length > 0) if (pagelinks.Length > 0)
{ {
// If there are several pages find the link for the latest one // If there are several pages find the link for the latest one
for (var i = pagelinks.Length - 1; i > 0; i--) for (var i = pagelinks.Length - 1; i > 0; i--)
{ {
var last_page_link = (pagelinks[i].Cq().Attr("href")).Trim(); var last_page_link = (pagelinks[i].GetAttribute("href")).Trim();
if (last_page_link.Contains("oldal")) if (last_page_link.Contains("oldal"))
{ {
var match = Regex.Match(last_page_link, @"(?<=[\?,&]oldal=)(\d+)(?=&)"); var match = Regex.Match(last_page_link, @"(?<=[\?,&]oldal=)(\d+)(?=&)");