norbits: refactor parsing (#13955)

This commit is contained in:
Bogdan
2023-02-01 10:06:23 +02:00
committed by GitHub
parent 40fcb1e43b
commit 0ba4d305b0

View File

@@ -27,8 +27,6 @@ namespace Jackett.Common.Indexers
private string LoginUrl => SiteLink + "login.php"; private string LoginUrl => SiteLink + "login.php";
private string LoginCheckUrl => SiteLink + "takelogin.php"; private string LoginCheckUrl => SiteLink + "takelogin.php";
private string SearchUrl => SiteLink + "browse.php"; private string SearchUrl => SiteLink + "browse.php";
private string TorrentDetailsUrl => SiteLink + "details.php?id={id}";
private string TorrentDownloadUrl => SiteLink + "download.php?id={id}&passkey={passkey}";
private ConfigurationDataNorbits ConfigData => (ConfigurationDataNorbits)configData; private ConfigurationDataNorbits ConfigData => (ConfigurationDataNorbits)configData;
@@ -218,14 +216,14 @@ namespace Jackett.Common.Indexers
// Check login before performing a query // Check login before performing a query
await CheckLoginAsync(); await CheckLoginAsync();
var SearchTerms = new List<string> { exactSearchTerm }; var searchTerms = new List<string> { exactSearchTerm };
// duplicate search without diacritics // duplicate search without diacritics
var baseSearchTerm = StringUtil.RemoveDiacritics(exactSearchTerm); var baseSearchTerm = StringUtil.RemoveDiacritics(exactSearchTerm);
if (baseSearchTerm != exactSearchTerm) if (baseSearchTerm != exactSearchTerm)
SearchTerms.Add(baseSearchTerm); searchTerms.Add(baseSearchTerm);
foreach (var searchTerm in SearchTerms) foreach (var searchTerm in searchTerms)
{ {
// Build our query // Build our query
var request = BuildQuery(searchTerm, query, searchUrl); var request = BuildQuery(searchTerm, query, searchUrl);
@@ -259,49 +257,38 @@ namespace Jackett.Common.Indexers
logger.Info("\nNorBits - Found " + nbResults + " result(s) (+/- " + firstPageRows.Length + ") in " + pageLinkCount + " page(s) for this query !"); logger.Info("\nNorBits - Found " + nbResults + " result(s) (+/- " + firstPageRows.Length + ") in " + pageLinkCount + " page(s) for this query !");
logger.Info("\nNorBits - There are " + firstPageRows.Length + " results on the first page !"); logger.Info("\nNorBits - There are " + firstPageRows.Length + " results on the first page !");
// Loop on results
foreach (var row in firstPageRows) foreach (var row in firstPageRows)
{ {
var id = row.QuerySelector("td:nth-of-type(2) > a:nth-of-type(1)").GetAttribute("href").Split('=').Last(); // ID var link = new Uri(SiteLink + row.QuerySelector("td:nth-of-type(2) > a[href*=\"download.php?id=\"]")?.GetAttribute("href").TrimStart('/'));
var name = row.QuerySelector("td:nth-of-type(2) > a:nth-of-type(1)").GetAttribute("title"); // Release Name var qDetails = row.QuerySelector("td:nth-of-type(2) > a[href*=\"details.php?id=\"]");
var categoryName = row.QuerySelector("td:nth-of-type(1) > div > a:nth-of-type(1)").GetAttribute("title"); // Category
var mainCat = row.QuerySelector("td:nth-of-type(1) > div > a:nth-of-type(1)").GetAttribute("href").Split('?').Last(); var title = qDetails?.GetAttribute("title").Trim();
var qSubCat2 = row.QuerySelector("td:nth-of-type(1) > div > a[href^=\"/browse.php?sub2_cat[]=\"]"); var details = new Uri(SiteLink + qDetails?.GetAttribute("href").TrimStart('/'));
var cat = mainCat;
if (qSubCat2 != null) var mainCategory = row.QuerySelector("td:nth-of-type(1) > div > a[href*=\"main_cat[]\"]")?.GetAttribute("href")?.Split('?').Last();
cat += '&' + qSubCat2.GetAttribute("href").Split('?').Last(); var secondCategory = row.QuerySelector("td:nth-of-type(1) > div > a[href*=\"sub2_cat[]\"]")?.GetAttribute("href")?.Split('?').Last();
var seeders = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(9)").TextContent); // Seeders
var leechers = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(10)").TextContent); // Leechers var categoryList = new[] { mainCategory, secondCategory };
var regexObj = new Regex(@"[^\d]"); // Completed var cat = string.Join("&", categoryList.Where(c => !string.IsNullOrWhiteSpace(c)));
var completed2 = row.QuerySelector("td:nth-of-type(8)").TextContent;
var completed = ParseUtil.CoerceLong(regexObj.Replace(completed2, "")); var seeders = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(9)").TextContent);
var qFiles = row.QuerySelector("td:nth-of-type(3) > a"); // Files var leechers = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(10)").TextContent);
var files = qFiles != null ? ParseUtil.CoerceInt(Regex.Match(qFiles.TextContent, @"\d+").Value) : 1;
var humanSize = row.QuerySelector("td:nth-of-type(7)").TextContent.ToLowerInvariant(); // Size
var size = ReleaseInfo.GetBytes(humanSize); // Date
var dateTimeOrig = row.QuerySelector("td:nth-of-type(5)").TextContent;
var dateTime = Regex.Replace(dateTimeOrig, @"<[^>]+>|&nbsp;", "").Trim();
var date = DateTime.ParseExact(dateTime, "yyyy-MM-ddHH:mm:ss", CultureInfo.InvariantCulture, DateTimeStyles.AssumeUniversal).ToLocalTime();
var details = new Uri(TorrentDetailsUrl.Replace("{id}", id.ToString())); // Description Link
var passkey = row.QuerySelector("td:nth-of-type(2) > a:nth-of-type(2)").GetAttribute("href"); // Download Link
var key = Regex.Match(passkey, "(?<=passkey\\=)([a-zA-z0-9]*)");
var downloadLink = new Uri(TorrentDownloadUrl.Replace("{id}", id.ToString()).Replace("{passkey}", key.ToString()));
// Building release infos
var release = new ReleaseInfo var release = new ReleaseInfo
{ {
Category = MapTrackerCatToNewznab(cat),
Title = name,
Seeders = seeders,
Peers = seeders + leechers,
PublishDate = date,
Size = size,
Files = files,
Grabs = completed,
Guid = details, Guid = details,
Details = details, Details = details,
Link = downloadLink, Link = link,
Title = title,
Category = MapTrackerCatToNewznab(cat),
Size = ReleaseInfo.GetBytes(row.QuerySelector("td:nth-of-type(7)").TextContent),
Files = ParseUtil.CoerceInt(row.QuerySelector("td:nth-of-type(3) > a")?.TextContent.Trim()),
Grabs = ParseUtil.CoerceLong(row.QuerySelector("td:nth-of-type(8)")?.FirstChild?.TextContent.Trim()),
Seeders = seeders,
Peers = seeders + leechers,
PublishDate = DateTime.ParseExact(row.QuerySelector("td:nth-of-type(5)")?.TextContent.Trim(), "yyyy-MM-ddHH:mm:ss", CultureInfo.InvariantCulture),
DownloadVolumeFactor = 1,
UploadVolumeFactor = 1,
MinimumRatio = 1, MinimumRatio = 1,
MinimumSeedTime = 172800 // 48 hours MinimumSeedTime = 172800 // 48 hours
}; };
@@ -311,8 +298,7 @@ namespace Jackett.Common.Indexers
{ {
genres = genres.Trim().Replace("\xA0", " ").Replace("(", "").Replace(")", "").Replace(" | ", ","); genres = genres.Trim().Replace("\xA0", " ").Replace("(", "").Replace(")", "").Replace(" | ", ",");
release.Description = genres; release.Description = genres;
if (release.Genres == null) release.Genres ??= new List<string>();
release.Genres = new List<string>();
release.Genres = release.Genres.Union(genres.Split(',')).ToList(); release.Genres = release.Genres.Union(genres.Split(',')).ToList();
} }
@@ -326,10 +312,6 @@ namespace Jackett.Common.Indexers
release.DownloadVolumeFactor = 0.5; release.DownloadVolumeFactor = 0.5;
else if (row.QuerySelector("img[title=\"90% Freeleech\"]") != null) else if (row.QuerySelector("img[title=\"90% Freeleech\"]") != null)
release.DownloadVolumeFactor = 0.1; release.DownloadVolumeFactor = 0.1;
else
release.DownloadVolumeFactor = 1;
release.UploadVolumeFactor = 1;
releases.Add(release); releases.Add(release);
} }
@@ -348,29 +330,26 @@ namespace Jackett.Common.Indexers
/// </summary> /// </summary>
/// <param name="term">Term to search</param> /// <param name="term">Term to search</param>
/// <param name="query">Torznab Query for categories mapping</param> /// <param name="query">Torznab Query for categories mapping</param>
/// <param name="url">Search url for provider</param> /// <param name="searchUrl">Search url for provider</param>
/// <param name="page">Page number to request</param> /// <param name="page">Page number to request</param>
/// <returns>URL to query for parsing and processing results</returns> /// <returns>URL to query for parsing and processing results</returns>
private string BuildQuery(string term, TorznabQuery query, string url, int page = 0) private string BuildQuery(string term, TorznabQuery query, string searchUrl, int page = 0)
{ {
var parameters = new NameValueCollection();
var categoriesList = MapTorznabCapsToTrackers(query);
var searchterm = term; var searchterm = term;
// Building our tracker query // Building our tracker query
parameters.Add("incldead", "1"); var parameters = new NameValueCollection
parameters.Add("fullsearch", ConfigData.UseFullSearch.Value ? "1" : "0"); {
parameters.Add("scenerelease", "0"); { "incldead", "1" },
{ "fullsearch", ConfigData.UseFullSearch.Value ? "1" : "0" },
{ "scenerelease", "0" }
};
// If search term provided // If search term provided
if (!string.IsNullOrWhiteSpace(query.ImdbID)) if (!string.IsNullOrWhiteSpace(query.ImdbID))
{
searchterm = "imdbsearch=" + query.ImdbID; searchterm = "imdbsearch=" + query.ImdbID;
}
else if (!string.IsNullOrWhiteSpace(term)) else if (!string.IsNullOrWhiteSpace(term))
{
searchterm = "search=" + WebUtilityHelpers.UrlEncode(term, Encoding.GetEncoding(28591)); searchterm = "search=" + WebUtilityHelpers.UrlEncode(term, Encoding.GetEncoding(28591));
}
else else
{ {
// Showing all torrents (just for output function) // Showing all torrents (just for output function)
@@ -378,17 +357,16 @@ namespace Jackett.Common.Indexers
term = "all"; term = "all";
} }
var CatQryStr = "";
foreach (var cat in categoriesList)
CatQryStr += "&" + cat;
// Building our query // Building our query
url += "?" + searchterm + "&" + parameters.GetQueryString() + "&" + CatQryStr; searchUrl += "?" + searchterm + "&" + parameters.GetQueryString();
logger.Info("\nBuilded query for \"" + term + "\"... " + url); var categoriesList = MapTorznabCapsToTrackers(query);
if (categoriesList.Any())
searchUrl += "&" + string.Join("&", categoriesList);
// Return our search url logger.Info("\nBuilded query for \"" + term + "\"... " + searchUrl);
return url;
return searchUrl;
} }
/// <summary> /// <summary>