Radarr/NzbDrone.Core/Indexers/BasicRssParser.cs

175 lines
5.7 KiB
C#
Raw Normal View History

2013-04-07 07:30:37 +00:00
using System;
using System.Collections.Generic;
2013-04-07 22:40:13 +00:00
using System.Globalization;
2013-04-07 07:30:37 +00:00
using System.IO;
2013-08-06 05:30:56 +00:00
using System.Linq;
2013-04-07 22:40:13 +00:00
using System.Text.RegularExpressions;
2013-08-06 02:45:57 +00:00
using System.Xml.Linq;
2013-04-07 07:30:37 +00:00
using NLog;
using NzbDrone.Core.Parser.Model;
2013-04-07 07:30:37 +00:00
namespace NzbDrone.Core.Indexers
{
public interface IParseFeed
{
IEnumerable<ReportInfo> Process(Stream source);
2013-04-07 07:30:37 +00:00
}
public class BasicRssParser : IParseFeed
{
private readonly Logger _logger;
public BasicRssParser()
{
_logger = LogManager.GetCurrentClassLogger();
}
public IEnumerable<ReportInfo> Process(Stream source)
2013-04-07 07:30:37 +00:00
{
2013-08-06 02:45:57 +00:00
var xdoc = XDocument.Load(source);
var items = xdoc.Descendants("item");
2013-04-07 07:30:37 +00:00
var result = new List<ReportInfo>();
2013-04-07 07:30:37 +00:00
2013-08-06 02:45:57 +00:00
foreach (var item in items)
2013-04-07 07:30:37 +00:00
{
try
{
2013-08-06 02:45:57 +00:00
var reportInfo = ParseFeedItem(item);
if (reportInfo != null)
2013-04-07 07:30:37 +00:00
{
2013-08-06 02:45:57 +00:00
reportInfo.NzbUrl = GetNzbUrl(item);
reportInfo.NzbInfoUrl = GetNzbInfoUrl(item);
result.Add(reportInfo);
2013-04-07 07:30:37 +00:00
}
}
catch (Exception itemEx)
{
2013-08-06 02:45:57 +00:00
itemEx.Data.Add("Item", item.Title());
2013-04-07 07:30:37 +00:00
_logger.ErrorException("An error occurred while processing feed item", itemEx);
}
}
return result;
}
2013-08-06 02:45:57 +00:00
protected virtual string GetTitle(XElement item)
2013-04-07 07:30:37 +00:00
{
2013-08-06 02:45:57 +00:00
return item.Title();
2013-04-07 07:30:37 +00:00
}
2013-08-06 02:45:57 +00:00
protected virtual string GetNzbUrl(XElement item)
2013-04-07 07:30:37 +00:00
{
2013-08-06 05:30:56 +00:00
return item.Links().First();
2013-04-07 07:30:37 +00:00
}
2013-08-06 02:45:57 +00:00
protected virtual string GetNzbInfoUrl(XElement item)
2013-04-07 07:30:37 +00:00
{
2013-04-07 22:40:13 +00:00
return String.Empty;
2013-04-07 07:30:37 +00:00
}
2013-08-06 02:45:57 +00:00
protected virtual ReportInfo PostProcessor(XElement item, ReportInfo currentResult)
2013-04-07 07:30:37 +00:00
{
return currentResult;
}
2013-08-06 02:45:57 +00:00
private ReportInfo ParseFeedItem(XElement item)
2013-04-07 07:30:37 +00:00
{
var title = GetTitle(item);
var reportInfo = new ReportInfo();
reportInfo.Title = title;
2013-08-06 02:45:57 +00:00
reportInfo.Age = DateTime.Now.Date.Subtract(item.PublishDate().Date).Days;
reportInfo.ReleaseGroup = ParseReleaseGroup(title);
2013-04-07 07:30:37 +00:00
2013-08-06 02:45:57 +00:00
_logger.Trace("Parsed: {0} from: {1}", reportInfo, item.Title());
2013-04-07 07:30:37 +00:00
return PostProcessor(item, reportInfo);
2013-04-07 07:30:37 +00:00
}
2013-04-07 22:40:13 +00:00
public static string ParseReleaseGroup(string title)
{
title = title.Trim();
var index = title.LastIndexOf('-');
if (index < 0)
index = title.LastIndexOf(' ');
if (index < 0)
return String.Empty;
var group = title.Substring(index + 1);
2013-04-07 22:40:13 +00:00
if (@group.Length == title.Length)
return String.Empty;
2013-04-07 22:40:13 +00:00
return @group;
}
private static readonly Regex[] HeaderRegex = new[]
{
new Regex(@"(?:\[.+\]\-\[.+\]\-\[.+\]\-\[)(?<nzbTitle>.+)(?:\]\-.+)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
new Regex(@"(?:\[.+\]\W+\[.+\]\W+\[.+\]\W+\"")(?<nzbTitle>.+)(?:\"".+)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
new Regex(@"(?:\[)(?<nzbTitle>.+)(?:\]\-.+)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
};
public static string ParseHeader(string header)
{
foreach (var regex in HeaderRegex)
{
var match = regex.Matches(header);
if (match.Count != 0)
return match[0].Groups["nzbTitle"].Value.Trim();
}
return header;
}
2013-07-17 07:33:54 +00:00
private static readonly Regex ReportSizeRegex = new Regex(@"(?<value>\d+\.\d{1,2}|\d+\,\d+\.\d{1,2}|\d+)\W?(?<unit>GB|MB|GiB|MiB)",
2013-04-07 22:40:13 +00:00
RegexOptions.IgnoreCase | RegexOptions.Compiled);
public static long GetReportSize(string sizeString)
{
var match = ReportSizeRegex.Matches(sizeString);
if (match.Count != 0)
{
var cultureInfo = new CultureInfo("en-US");
var value = Decimal.Parse(Regex.Replace(match[0].Groups["value"].Value, "\\,", ""), cultureInfo);
var unit = match[0].Groups["unit"].Value;
if (unit.Equals("MB", StringComparison.InvariantCultureIgnoreCase) ||
unit.Equals("MiB", StringComparison.InvariantCultureIgnoreCase))
{
2013-07-17 07:33:54 +00:00
return ConvertToBytes(Convert.ToDouble(value), 2);
}
2013-04-07 22:40:13 +00:00
if (unit.Equals("GB", StringComparison.InvariantCultureIgnoreCase) ||
unit.Equals("GiB", StringComparison.InvariantCultureIgnoreCase))
{
2013-07-17 07:33:54 +00:00
return ConvertToBytes(Convert.ToDouble(value), 3);
}
2013-04-07 22:40:13 +00:00
}
return 0;
}
2013-07-17 07:33:54 +00:00
private static long ConvertToBytes(double value, int power)
{
var multiplier = Math.Pow(1024, power);
var result = value*multiplier;
return Convert.ToInt64(result);
}
2013-04-07 07:30:37 +00:00
}
}