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

lostfilm: try to prevent parse error #14638

Browse Source 
tested with non-Russian IP for regular TEST
However error only occurs with Russian IP so that is untested as I don't have a VPN to access that region.
This commit is contained in:
Garfield69
2023-08-17 21:10:12 +12:00
parent a8f994871f
commit d6f0592867
1 changed files with 82 additions and 68 deletions
Download Patch File Download Diff File Expand all files Collapse all files

150
src/Jackett.Common/Indexers/LostFilm.cs
View File

@@ -717,89 +717,104 @@ namespace Jackett.Common.Indexers
var document = parser.ParseDocument(results.ContentString); var document = parser.ParseDocument(results.ContentString);
var rows = document.QuerySelectorAll("div.inner-box--item"); var rows = document.QuerySelectorAll("div.inner-box--item");
logger.Debug("> Parsing " + rows.Count().ToString() + " releases"); if (rows.Count() > 0)
var serieTitle = document.QuerySelector("div.inner-box--subtitle").TextContent;
serieTitle = serieTitle.Substring(0, serieTitle.LastIndexOf(','));
var episodeInfo = document.QuerySelector("div.inner-box--text").TextContent;
var episodeName = TrimString(episodeInfo, '(', ')');
foreach (var row in rows)
{ {
try logger.Debug("> Parsing " + rows.Count().ToString() + " releases");
var serieTitle = document.QuerySelector("div.inner-box--subtitle").TextContent;
serieTitle = serieTitle.Substring(0, serieTitle.LastIndexOf(','));
var episodeInfo = document.QuerySelector("div.inner-box--text").TextContent;
var episodeName = TrimString(episodeInfo, '(', ')');
foreach (var row in rows)
{ {
try
var detailsInfo = row.QuerySelector("div.inner-box--desc").TextContent;
var releaseDetails = parseReleaseDetailsRegex.Match(detailsInfo);
// ReSharper states "Expression is always false"
// TODO Refactor to get the intended operation
if (releaseDetails == null)
{ {
throw new FormatException("Failed to map release details string: " + detailsInfo);
}
/* var detailsInfo = row.QuerySelector("div.inner-box--desc").TextContent;
* For supported qualities see: var releaseDetails = parseReleaseDetailsRegex.Match(detailsInfo);
* - TvCategoryParser.cs
* - https://github.com/SickRage/SickRage/wiki/Quality-Settings#quality-names-to-recognize-the-quality-of-a-file
*/
var quality = releaseDetails.Groups["quality"].Value.Trim();
// Adapt shitty quality format for common algorythms
quality = Regex.Replace(quality, "-Rip", "Rip", RegexOptions.IgnoreCase);
quality = Regex.Replace(quality, "WEB-DLRip", "WEBDL", RegexOptions.IgnoreCase);
quality = Regex.Replace(quality, "WEB-DL", "WEBDL", RegexOptions.IgnoreCase);
quality = Regex.Replace(quality, "HDTVRip", "HDTV", RegexOptions.IgnoreCase);
// Fix forgotten p-Progressive suffix in resolution index
quality = Regex.Replace(quality, "1080 ", "1080p ", RegexOptions.IgnoreCase);
quality = Regex.Replace(quality, "720 ", "720p ", RegexOptions.IgnoreCase);
var techComponents = new[] // ReSharper states "Expression is always false"
{ // TODO Refactor to get the intended operation
if (releaseDetails == null)
{
throw new FormatException("Failed to map release details string: " + detailsInfo);
}
/*
* For supported qualities see:
* - TvCategoryParser.cs
* - https://github.com/SickRage/SickRage/wiki/Quality-Settings#quality-names-to-recognize-the-quality-of-a-file
*/
var quality = releaseDetails.Groups["quality"].Value.Trim();
// Adapt shitty quality format for common algorythms
quality = Regex.Replace(quality, "-Rip", "Rip", RegexOptions.IgnoreCase);
quality = Regex.Replace(quality, "WEB-DLRip", "WEBDL", RegexOptions.IgnoreCase);
quality = Regex.Replace(quality, "WEB-DL", "WEBDL", RegexOptions.IgnoreCase);
quality = Regex.Replace(quality, "HDTVRip", "HDTV", RegexOptions.IgnoreCase);
// Fix forgotten p-Progressive suffix in resolution index
quality = Regex.Replace(quality, "1080 ", "1080p ", RegexOptions.IgnoreCase);
quality = Regex.Replace(quality, "720 ", "720p ", RegexOptions.IgnoreCase);
var techComponents = new[]
{
"rus", "rus",
quality, quality,
"(LostFilm)" "(LostFilm)"
}; };
var techInfo = string.Join(" ", techComponents.Where(s => !string.IsNullOrEmpty(s))); var techInfo = string.Join(" ", techComponents.Where(s => !string.IsNullOrEmpty(s)));
// Ru title: downloadLink.TextContent.Replace("\n", ""); // Ru title: downloadLink.TextContent.Replace("\n", "");
// En title should be manually constructed. // En title should be manually constructed.
var titleComponents = new[] { var titleComponents = new[] {
serieTitle, details.GetEpisodeString(), episodeName, techInfo serieTitle, details.GetEpisodeString(), episodeName, techInfo
}; };
var downloadLink = row.QuerySelector("div.inner-box--link > a"); var downloadLink = row.QuerySelector("div.inner-box--link > a");
var sizeString = releaseDetails.Groups["size"].Value.ToUpper(); var sizeString = releaseDetails.Groups["size"].Value.ToUpper();
sizeString = sizeString.Replace("ТБ", "TB"); // untested sizeString = sizeString.Replace("ТБ", "TB"); // untested
sizeString = sizeString.Replace("ГБ", "GB"); sizeString = sizeString.Replace("ГБ", "GB");
sizeString = sizeString.Replace("МБ", "MB"); sizeString = sizeString.Replace("МБ", "MB");
sizeString = sizeString.Replace("КБ", "KB"); // untested sizeString = sizeString.Replace("КБ", "KB"); // untested
var link = new Uri(downloadLink.GetAttribute("href")); var link = new Uri(downloadLink.GetAttribute("href"));
// TODO this feels sparse compared to other trackers. Expand later // TODO this feels sparse compared to other trackers. Expand later
var release = new ReleaseInfo var release = new ReleaseInfo
{
Category = new[] { TorznabCatType.TV.ID },
Title = string.Join(" - ", titleComponents.Where(s => !string.IsNullOrEmpty(s))),
Link = link,
Guid = link,
Size = ParseUtil.GetBytes(sizeString),
// add missing torznab fields not available from results
Seeders = 1,
Peers = 2,
DownloadVolumeFactor = 0,
UploadVolumeFactor = 1,
MinimumRatio = 1,
MinimumSeedTime = 172800 // 48 hours
};
// TODO Other trackers don't have this log line. Remove or add to other trackers?
logger.Debug("> Add: " + release.Title);
releases.Add(release);
}
catch (Exception ex)
{ {
Category = new[] { TorznabCatType.TV.ID }, logger.Error(string.Format("{0}: Error while parsing row '{1}':\n\n{2}", Id, row.OuterHtml, ex));
Title = string.Join(" - ", titleComponents.Where(s => !string.IsNullOrEmpty(s))), }
Link = link,
Guid = link,
Size = ParseUtil.GetBytes(sizeString),
// add missing torznab fields not available from results
Seeders = 1,
Peers = 2,
DownloadVolumeFactor = 0,
UploadVolumeFactor = 1,
MinimumRatio = 1,
MinimumSeedTime = 172800 // 48 hours
};
// TODO Other trackers don't have this log line. Remove or add to other trackers?
logger.Debug("> Add: " + release.Title);
releases.Add(release);
} }
catch (Exception ex) }
else
{
if (results.ContentString.Contains("Контент недоступен на территории Российской Федерации"))
{ {
logger.Error(string.Format("{0}: Error while parsing row '{1}':\n\n{2}", Id, row.OuterHtml, ex)); logger.Debug($"> Content is not available in the Russian Federation");
}
else
{
var message = document.QuerySelector("p").TextContent;
logger.Debug($"> {message}");
} }
} }
} }
@@ -807,7 +822,6 @@ namespace Jackett.Common.Indexers
{ {
OnParseError(results.ContentString, ex); OnParseError(results.ContentString, ex);
} }
return releases; return releases;
} }