/// <summary>
/// Compares the InternetUrl with another InternetUrl object. The comparison
/// is done based on the Url property.
/// </summary>
/// <param name="obj">The InternetUrl object to be compared to this one</param>
/// <returns>An integer indicating difference between the two objects</returns>
public int CompareTo(object obj)
{
if (!(obj is InternetUrl))
{
throw new ArgumentException("InternetUrl can only be compared to InternetUrl objects.");
}
InternetUrl toCompare = (InternetUrl)obj;
return(String.Compare(url.ToString(), toCompare.Url));
}
/// <summary>
/// Returns the host name of an <see cref="InternetUrl"/>
/// </summary>
/// <param name="url">The <see cref="InternetUrl"/> to examine</param>
/// <returns>A string containing the Url's host name or IP Address.</returns>
public static string HostName(InternetUrl url)
{
try
{
Uri uri = new Uri(url.Url);
return(uri.Host);
}
catch
{
return(String.Empty);
}
}
/// <summary>
/// Compares an existing <see cref="InternetUrl"/> instance with the Url of the object for equality.
/// </summary>
/// <param name="obj">The object to compare with the current instance</param>
/// <returns>true if obj represents the same Url as the Url contained in this <see cref="InternetUrl"/> instance; otherwise, false.</returns>
/// <remarks>This method must not throw any exceptions. The method automatically uses
/// the <see cref="Object.GetType"/> method to determine if the runtime types of the
/// objects being compared are equivalent, therefore it is not necessary to check
/// whether the obj parameter is a <see cref="InternetUrl"/> instance.
/// </remarks>
public override bool Equals(object obj)
{
//does not check the ID member
InternetUrl ToCompare = (InternetUrl)obj;
for (int i = 0; i < 16; i++)
{
if ((urlMD5[i] != ToCompare.MD5[i]))
{
return(false);
}
}
return(true);
}
/// <summary>
/// Checks if at least 30 seconds have passed since the last request to a given host
/// was made, in order not to slammer it with simultaneous or frequent requests.
/// </summary>
/// <param name="targetUrl">
/// A <see cref="InternetUrl"/> that is served by a host we wish to check.
/// </param>
/// <returns>
/// An integer containing the number of milliseconds a crawler thread must wait
/// before visiting this host.
/// </returns>
public int FilterUrl(ref InternetUrl targetUrl)
{
string hostName = InternetUtils.HostName(targetUrl);
return FilterHost(ref hostName);
}
/// <summary>
/// Constructs a new <see cref="InternetUrl"/> object from an existing <see cref="InternetUrl"/> object.
/// </summary>
/// <param name="IUrl">The existing <see cref="InternetUrl"/> object</param>
public InternetUrl(InternetUrl IUrl)
{
urlID = IUrl.ID;
url = IUrl.Url;
urlMD5 = IUrl.MD5;
}
/// <summary>
/// Constructs a new <see cref="InternetUrl"/> object from an existing <see cref="InternetUrl"/> object.
/// </summary>
/// <param name="IUrl">The existing <see cref="InternetUrl"/> object</param>
public InternetUrl(InternetUrl IUrl)
{
urlID = IUrl.ID;
url = IUrl.Url;
urlMD5 = IUrl.MD5;
}
/// <summary>
/// Returns the host name of an <see cref="InternetUrl"/>
/// </summary>
/// <param name="url">The <see cref="InternetUrl"/> to examine</param>
/// <returns>A string containing the Url's host name or IP Address.</returns>
public static string HostName(InternetUrl url)
{
try
{
Uri uri = new Uri(url.Url);
return uri.Host;
}
catch
{
return String.Empty;
}
}
