موارد مهم این نکات رو اگر تبدیل به یک متد کمکی کنیم، کلاس زیر بدست خواهد آمد:
با این تستها جهت بررسی آن:
using System;
using System.Web;
namespace UrlNormalizationTest
{
public static class UrlNormalization
{
public static bool AreTheSameUrls(this string url1, string url2)
{
url1 = url1.NormalizeUrl();
url2 = url2.NormalizeUrl();
return url1.Equals(url2);
}
public static bool AreTheSameUrls(this Uri uri1, Uri uri2)
{
var url1 = uri1.NormalizeUrl();
var url2 = uri2.NormalizeUrl();
return url1.Equals(url2);
}
public static string[] DefaultDirectoryIndexes = new[]
{
"default.asp",
"default.aspx",
"index.htm",
"index.html",
"index.php"
};
public static string NormalizeUrl(this Uri uri)
{
var url = urlToLower(uri);
url = limitProtocols(url);
url = removeDefaultDirectoryIndexes(url);
url = removeTheFragment(url);
url = removeDuplicateSlashes(url);
url = addWww(url);
url = removeFeedburnerPart(url);
return removeTrailingSlashAndEmptyQuery(url);
}
public static string NormalizeUrl(this string url)
{
return NormalizeUrl(new Uri(url));
}
private static string removeFeedburnerPart(string url)
{
var idx = url.IndexOf("utm_source=", StringComparison.Ordinal);
return idx == -1 ? url : url.Substring(0, idx - 1);
}
private static string addWww(string url)
{
if (new Uri(url).Host.Split('.').Length == 2 && !url.Contains("://www."))
{
return url.Replace("://", "://www.");
}
return url;
}
private static string removeDuplicateSlashes(string url)
{
var path = new Uri(url).AbsolutePath;
return path.Contains("//") ? url.Replace(path, path.Replace("//", "/")) : url;
}
private static string limitProtocols(string url)
{
return new Uri(url).Scheme == "https" ? url.Replace("https://", "http://") : url;
}
private static string removeTheFragment(string url)
{
var fragment = new Uri(url).Fragment;
return string.IsNullOrWhiteSpace(fragment) ? url : url.Replace(fragment, string.Empty);
}
private static string urlToLower(Uri uri)
{
return HttpUtility.UrlDecode(uri.AbsoluteUri.ToLowerInvariant());
}
private static string removeTrailingSlashAndEmptyQuery(string url)
{
return url
.TrimEnd(new[] { '?' })
.TrimEnd(new[] { '/' });
}
private static string removeDefaultDirectoryIndexes(string url)
{
foreach (var index in DefaultDirectoryIndexes)
{
if (url.EndsWith(index))
{
url = url.TrimEnd(index.ToCharArray());
break;
}
}
return url;
}
}
}using NUnit.Framework;
using UrlNormalizationTest;
namespace UrlNormalization.Tests
{
[TestFixture]
public class UnitTests
{
[Test]
public void Test1ConvertingTheSchemeAndHostToLowercase()
{
var url1 = "HTTP://www.Example.com/".NormalizeUrl();
var url2 = "http://www.example.com/".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test2CapitalizingLettersInEscapeSequences()
{
var url1 = "http://www.example.com/a%c2%b1b".NormalizeUrl();
var url2 = "http://www.example.com/a%C2%B1b".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test3DecodingPercentEncodedOctetsOfUnreservedCharacters()
{
var url1 = "http://www.example.com/%7Eusername/".NormalizeUrl();
var url2 = "http://www.example.com/~username/".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test4RemovingTheDefaultPort()
{
var url1 = "http://www.example.com:80/bar.html".NormalizeUrl();
var url2 = "http://www.example.com/bar.html".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test5AddingTrailing()
{
var url1 = "http://www.example.com/alice".NormalizeUrl();
var url2 = "http://www.example.com/alice/?".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test6RemovingDotSegments()
{
var url1 = "http://www.example.com/../a/b/../c/./d.html".NormalizeUrl();
var url2 = "http://www.example.com/a/c/d.html".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test7RemovingDirectoryIndex1()
{
var url1 = "http://www.example.com/default.asp".NormalizeUrl();
var url2 = "http://www.example.com/".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test7RemovingDirectoryIndex2()
{
var url1 = "http://www.example.com/default.asp?id=1".NormalizeUrl();
var url2 = "http://www.example.com/default.asp?id=1".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test7RemovingDirectoryIndex3()
{
var url1 = "http://www.example.com/a/index.html".NormalizeUrl();
var url2 = "http://www.example.com/a/".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test8RemovingTheFragment()
{
var url1 = "http://www.example.com/bar.html#section1".NormalizeUrl();
var url2 = "http://www.example.com/bar.html".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test9LimitingProtocols()
{
var url1 = "https://www.example.com/".NormalizeUrl();
var url2 = "http://www.example.com/".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test10RemovingDuplicateSlashes()
{
var url1 = "http://www.example.com/foo//bar.html".NormalizeUrl();
var url2 = "http://www.example.com/foo/bar.html".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test11AddWww()
{
var url1 = "http://example.com/".NormalizeUrl();
var url2 = "http://www.example.com".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
[Test]
public void Test12RemoveFeedburnerPart()
{
var url1 = "http://site.net/2013/02/firefox-19-released/?utm_source=rss&utm_medium=rss&utm_campaign=firefox-19-released".NormalizeUrl();
var url2 = "http://site.net/2013/02/firefox-19-released".NormalizeUrl();
Assert.AreEqual(url1, url2);
}
}
}
