public void IsUrlAllowed_WildCardAgentWithWhiteSpaceDisallow_ReturnsTrue()
{
string userAgentString = _userAgentString;
_unitUnderTest = new RobotsDotText(_rootUri, @"User-agent: *
Disallow: ");
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri, userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "aa.html", userAgentString));
}
public void IsUrlAllowed_QuerystringMatch_NotSupported_ReturnsFalse()
{
string userAgentString = _userAgentString;
_unitUnderTest = new RobotsDotText(_rootUri, @"User-Agent: *
Disallow: /?category=whatever
Disallow: /?category=another&color=red");
Assert.IsFalse(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "?category=whatever", userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "?category=another&blah=blah", userAgentString));
}
public void IsUrlAllowed_QuerystringMatch_NotSupported_ReturnsTrue()
{
//IF this test starts failing that is a good thing, it means the robots impl now supports querystrings
string userAgentString = _userAgentString;
_unitUnderTest = new RobotsDotText(_rootUri, @"User-Agent: *
Disallow: /?category=whatever
Disallow: /?category=another&color=red");
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "?category=whatever", userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "?category=another&blah=blah", userAgentString));
}
[Test, Ignore]//This is a bug and needs to be fixed
public void IsUrlAllowed_QuerystringOnRoot2_ReturnsTrue()
{
string userAgentString = _userAgentString;
_unitUnderTest = new RobotsDotText(_rootUri, @"User-Agent: *
Disallow: /?/
Disallow: /category/");
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri, userAgentString));
}
public void IsUrlAllowed_QuerystringOnRoot_ReturnsTrue()
{
string userAgentString = _userAgentString;
_unitUnderTest = new RobotsDotText(_rootUri, @"User-Agent: *
Disallow: /?category=whatever
Disallow: /?category=another&color=red");
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri, userAgentString));
}
public void IsUrlAllowed_AllowedPages_ReturnsTrue()
{
string userAgentString = _userAgentString;
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri, userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfolder/aa.html", userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfolder/bb.html", userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfile2", userAgentString));
//User agent "userAgentCrawlDelayIs1" doesn't specify anything to disallow so should allow all ("*" is not inherited)
userAgentString = "userAgentCrawlDelayIs1";
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri, userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "disallowedfile.txt", userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "disallowedfolder", userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "disallowedfolder/", userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "disallowedfolder/subfolder", userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "disallowedfolder/subfolder/", userAgentString));
//Allows all since "userAgentCrawlDelayIs1" does not specify allow or disallow
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri, userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfolder/aa.html", userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfolder/bb.html", userAgentString));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfile2", userAgentString));
}
public void IsUrlAllowed_EmptyRobotsContent_ReturnsTrue()
{
_unitUnderTest = new RobotsDotText(_rootUri, "");
//Should use "*" user agent by default
string userAgent = _userAgentString;
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri, userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfolder/aa.html", userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfolder/bb.html", userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfile2", userAgent));
//User agent "userAgentCrawlDelayIs1" doesn't specify anything to disallow so should allow all ("*" is not inherited)
userAgent = "userAgentCrawlDelayIs1";
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri, userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "disallowedfile.txt", userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "disallowedfolder", userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "disallowedfolder/", userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "disallowedfolder/subfolder", userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "disallowedfolder/subfolder/", userAgent));
//Allows all since "userAgentCrawlDelayIs1" does not specify allow or disallow
userAgent = "userAgentCrawlDelayIs1";
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri, userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfolder/aa.html", userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfolder/bb.html", userAgent));
Assert.IsTrue(_unitUnderTest.IsUrlAllowed(_rootUri.AbsoluteUri + "allowedfile2", userAgent));
}