Convert TorrentHaven to AngleSharp (#7387)

This commit is contained in:
Diego Heras
2020-02-29 20:42:31 +01:00
committed by GitHub
parent a697aba985
commit dd52a16bff

View File

@@ -6,7 +6,8 @@ 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.Dom;
using AngleSharp.Html.Parser;
using Jackett.Common.Models; using Jackett.Common.Models;
using Jackett.Common.Models.IndexerConfig; using Jackett.Common.Models.IndexerConfig;
using Jackett.Common.Services.Interfaces; using Jackett.Common.Services.Interfaces;
@@ -101,12 +102,13 @@ namespace Jackett.Common.Indexers
public override async Task<ConfigurationData> GetConfigurationForSetup() public override async Task<ConfigurationData> GetConfigurationForSetup()
{ {
var loginPage = await RequestStringWithCookies(IndexUrl, string.Empty); var loginPage = await RequestStringWithCookies(IndexUrl, string.Empty);
CQ dom = loginPage.Content; var parser = new HtmlParser();
var qCaptchaImg = dom.Find("td.tablea > img").First(); var dom = parser.ParseDocument(loginPage.Content);
if (qCaptchaImg.Length == 1) var qCaptchaImg = dom.QuerySelector("td.tablea > img");
if (qCaptchaImg != null)
{ {
var CaptchaUrl = SiteLink + qCaptchaImg.Attr("src"); var captchaUrl = SiteLink + qCaptchaImg.GetAttribute("src");
var captchaImage = await RequestBytesWithCookies(CaptchaUrl, loginPage.Cookies); var captchaImage = await RequestBytesWithCookies(captchaUrl, loginPage.Cookies);
configData.CaptchaImage.Value = captchaImage.Content; configData.CaptchaImage.Value = captchaImage.Content;
} }
else else
@@ -139,10 +141,10 @@ namespace Jackett.Common.Indexers
var result = await RequestLoginAndFollowRedirect(IndexUrl, pairs, configData.CaptchaCookie.Value, true, null, IndexUrl, true); var result = await RequestLoginAndFollowRedirect(IndexUrl, pairs, configData.CaptchaCookie.Value, true, null, IndexUrl, true);
if (result.Content == null || (!result.Content.Contains("login_complete") && !result.Content.Contains("index.php?strWebValue=account&strWebAction=logout"))) if (result.Content == null || (!result.Content.Contains("login_complete") && !result.Content.Contains("index.php?strWebValue=account&strWebAction=logout")))
{ {
CQ dom = result.Content; var parser = new HtmlParser();
var errorMessage = dom["table > tbody > tr > td[valign=top][width=100%]"].Html(); var dom = parser.ParseDocument(result.Content);
if (errorMessage.Length == 0) var errorMessageEl = dom.QuerySelector("table > tbody > tr > td[valign=top][width=100%]");
errorMessage = result.Content; var errorMessage = errorMessageEl != null ? errorMessageEl.InnerHtml : result.Content;
throw new ExceptionWithConfigData(errorMessage, configData); throw new ExceptionWithConfigData(errorMessage, configData);
} }
@@ -190,42 +192,42 @@ namespace Jackett.Common.Indexers
searchUrl += "?" + queryCollection.GetQueryString(); searchUrl += "?" + queryCollection.GetQueryString();
var response = await RequestStringWithCookies(searchUrl); var response = await RequestStringWithCookies(searchUrl);
var results = response.Content;
var TitleRegexp = new Regex(@"^return buildTable\('(.*?)',\s+"); var TitleRegexp = new Regex(@"^return buildTable\('(.*?)',\s+");
try try
{ {
CQ dom = results; var parser = new HtmlParser();
var rows = dom["table.torrenttable > tbody > tr"]; var dom = parser.ParseDocument(response.Content);
var rows = dom.QuerySelectorAll("table.torrenttable > tbody > tr");
foreach (var row in rows.Skip(1)) foreach (var row in rows.Skip(1))
{ {
var release = new ReleaseInfo(); var release = new ReleaseInfo();
release.MinimumRatio = 0.8; release.MinimumRatio = 0.8;
release.MinimumSeedTime = 0; release.MinimumSeedTime = 0;
var qRow = row.Cq();
var qDetailsLink = qRow.Find("a[href^=index.php?strWebValue=torrent&strWebAction=details]").First();
release.Title = TitleRegexp.Match(qDetailsLink.Attr("onmouseover")).Groups[1].Value;
var qCatLink = qRow.Find("a[href^=index.php?strWebValue=torrent&strWebAction=search&dir=]").First(); var qDetailsLink = row.QuerySelector("a[href^=index.php?strWebValue=torrent&strWebAction=details]");
var qDLLink = qRow.Find("a[href^=index.php?strWebValue=torrent&strWebAction=download&id=]").First(); release.Title = TitleRegexp.Match(qDetailsLink.GetAttribute("onmouseover")).Groups[1].Value;
var qSeeders = qRow.Find("td.column1:eq(3)");
var qLeechers = qRow.Find("td.column2:eq(3)");
var qDateStr = qRow.Find("font:has(a)").First();
var qSize = qRow.Find("td.column2[align=center]").First();
var catStr = qCatLink.Attr("href").Split('=')[3].Split('#')[0]; var qCatLink = row.QuerySelector("a[href^=index.php?strWebValue=torrent&strWebAction=search&dir=]");
var qDlLink = row.QuerySelector("a[href^=index.php?strWebValue=torrent&strWebAction=download&id=]");
var qSeeders = row.QuerySelector("td.column1:nth-of-type(4)");
var qLeechers = row.QuerySelector("td.column2:nth-of-type(4)");
var qDateStr = row.QuerySelector("font:has(a)");
var qSize = row.QuerySelector("td.column2[align=center]");
var catStr = qCatLink.GetAttribute("href").Split('=')[3].Split('#')[0];
release.Category = MapTrackerCatToNewznab(catStr); release.Category = MapTrackerCatToNewznab(catStr);
release.Link = new Uri(SiteLink + qDLLink.Attr("href")); release.Link = new Uri(SiteLink + qDlLink.GetAttribute("href"));
release.Comments = new Uri(SiteLink + qDetailsLink.Attr("href")); release.Comments = new Uri(SiteLink + qDetailsLink.GetAttribute("href"));
release.Guid = release.Link; release.Guid = release.Link;
var sizeStr = qSize.Text(); var sizeStr = qSize.Text();
release.Size = ReleaseInfo.GetBytes(sizeStr); release.Size = ReleaseInfo.GetBytes(sizeStr);
release.Seeders = ParseUtil.CoerceInt(qSeeders.Text()); release.Seeders = ParseUtil.CoerceInt(qSeeders.TextContent);
release.Peers = ParseUtil.CoerceInt(qLeechers.Text()) + release.Seeders; release.Peers = ParseUtil.CoerceInt(qLeechers.TextContent) + release.Seeders;
var dateStr = qDateStr.Text().Trim(); var dateStr = qDateStr.Text().Trim();
var dateStrParts = dateStr.Split(); var dateStrParts = dateStr.Split();
@@ -240,12 +242,12 @@ namespace Jackett.Common.Indexers
var pubDateUtc = TimeZoneInfo.ConvertTimeToUtc(dateGerman, germanyTz); var pubDateUtc = TimeZoneInfo.ConvertTimeToUtc(dateGerman, germanyTz);
release.PublishDate = pubDateUtc.ToLocalTime(); release.PublishDate = pubDateUtc.ToLocalTime();
var grabs = qRow.Find("td:nth-child(7)").Text(); var grabs = row.QuerySelector("td:nth-child(7)").TextContent;
release.Grabs = ParseUtil.CoerceInt(grabs); release.Grabs = ParseUtil.CoerceInt(grabs);
if (qRow.Find("img[src=\"themes/images/freeleech.png\"]").Length >= 1) if (row.QuerySelector("img[src=\"themes/images/freeleech.png\"]") != null)
release.DownloadVolumeFactor = 0; release.DownloadVolumeFactor = 0;
else if (qRow.Find("img[src=\"themes/images/DL50.png\"]").Length >= 1) else if (row.QuerySelector("img[src=\"themes/images/DL50.png\"]") != null)
release.DownloadVolumeFactor = 0.5; release.DownloadVolumeFactor = 0.5;
else else
release.DownloadVolumeFactor = 1; release.DownloadVolumeFactor = 1;
@@ -257,7 +259,7 @@ namespace Jackett.Common.Indexers
} }
catch (Exception ex) catch (Exception ex)
{ {
OnParseError(results, ex); OnParseError(response.Content, ex);
} }
return releases; return releases;