/// <summary>
/// ExtractSegmentedReversedUrl
/// </summary>
/// <param name="key"></param>
/// <param name="reversed"></param>
/// <param name="srUrl"></param>
/// <returns></returns>
public SegmentedReversedUrl ExtractSegmentedReversedUrl(String reversedKey, bool reversed)
{
SegmentedReversedUrl srUrl = null;
if (!reversed)
{
reversedKey = UrlReverser.Reverse(reversedKey);
}
if (String.IsNullOrEmpty(reversedKey))
{
return srUrl;
}
if (reversedKey.Length < ProtocolPrefixLength)
{
return srUrl;
}
String port = null;
if (reversedKey[0] == SchemeHttpEncodingSymbol)
{
}
else if (reversedKey[0] == schemeHttpsEncodingSymbol)
{
}
else
{
return srUrl;
}
Int32 portEndIndex = reversedKey.IndexOf(PortSpacer);
if (portEndIndex == -1 || portEndIndex > (ProtocolPrefixLength) - 1)
{
portEndIndex = 1 + MaxPortLength;
}
if (portEndIndex > 1)
{
port = reversedKey.Substring(1, portEndIndex - 1);
}
// Start to parse host here
Int32 hostStartIndex = ProtocolPrefixLength;
if (hostStartIndex >= reversedKey.Length)
{
return srUrl;
}
Int32 hostEndIndex = reversedKey.IndexOf(PathDelimeter, hostStartIndex);
if (hostEndIndex == -1)
{
hostEndIndex = reversedKey.Length;
}
if (hostEndIndex == hostStartIndex)
{
return srUrl;
}
String host = reversedKey.Substring(hostStartIndex, hostEndIndex - hostStartIndex);
String[] byDot = host.Split(new char[]{HostDelimeter}, StringSplitOptions.RemoveEmptyEntries);
Boolean isIP = byDot.Length == 4 && IsIPHost(byDot[0], byDot[1], byDot[2], byDot[3]);
int domainEndIndex = hostStartIndex;
if (isIP)
{
domainEndIndex = hostEndIndex;
}
else if (byDot.Length <= 2)
{
domainEndIndex = hostEndIndex;
}
else
{
// Get domain
bool tldfound = false;
for (; domainEndIndex < hostEndIndex; ++domainEndIndex)
{
if (reversedKey[domainEndIndex] != HostDelimeter)
{
continue;
}
else
{
string suspiciousTld = reversedKey.Substring(hostStartIndex, domainEndIndex - hostStartIndex + 1);
if (tlds.ContainsKey(suspiciousTld))
{
tldfound = true;
}
else
{
if (tldfound)
{
// The sub string up to the last HostDelimeter is the tld
// so that the sub string up to here is the domain
++domainEndIndex;
break;
}
}
}
}
// tld prefix not found, so just treat the first two segments as the domain
if (!tldfound)
{
domainEndIndex = hostStartIndex;
domainEndIndex += byDot[0].Length + 1 + byDot[1].Length + 1;
}
}
// Find L1Path
int L1EndIndex = -1;
int L1StartIndex = -1;
int UrlEndIndex = reversedKey.Length;
if (hostEndIndex == reversedKey.Length)
{
L1EndIndex = 0;
L1StartIndex = 0;
UrlEndIndex = 0;
}
else
{
L1StartIndex = hostEndIndex;
L1EndIndex = reversedKey.IndexOfAny(L1PathDelimeters, L1StartIndex + 1);
if (L1EndIndex == -1)
{
L1EndIndex = L1StartIndex;
}
++L1EndIndex;
}
String domain = reversedKey.Substring(0, domainEndIndex );
String hostSuffix = reversedKey.Substring(domainEndIndex, hostEndIndex - domainEndIndex);
String l1PathSuffix = reversedKey.Substring(L1StartIndex, L1EndIndex - L1StartIndex);
String urlSuffix = reversedKey.Substring(L1EndIndex, UrlEndIndex - L1EndIndex);
// Sanity check for segments
// will potentially hurt the performance, so don't use it in production
/*
if (l1PathSuffix.Length > 0)
{
Common.Assert(l1PathSuffix[0] == '/',
"Assertion failed: L1PathSuffix doesn't start with /, L1PathSuffix = {0}, Key = {1}",
l1PathSuffix,
reversedKey);
Common.Assert(l1PathSuffix[l1PathSuffix.Length - 1] == L1PathDelimeters[0] ||
l1PathSuffix[l1PathSuffix.Length - 1] == L1PathDelimeters[1],
"Assertion failed: L1PathSuffix doesn't end with /, L1PathSuffix = {0}, Key = {1}",
l1PathSuffix,
reversedKey);
}
*/
srUrl = new SegmentedReversedUrl(domain,
hostSuffix,
l1PathSuffix,
urlSuffix);
return srUrl;
}
public string MapSegmentedUrlToSite( SegmentedReversedUrl segUrl)
{
String domain, host, l1Path;
domain = segUrl.ReversedDomain;
host = segUrl.ReversedHost;
l1Path = segUrl.ReversedL1Path ;
string site = null;
if( !isDefined )
{
throw new Exception("SitePrefixTable is not defined" );
}
try
{
// same filtering of host/l1path in construction of SitePrefixTable
// normalization of prefixe dupes
{
if( !String.IsNullOrEmpty(l1Path ))
{
l1Path = l1Path.Trim().Replace( " ", "%20" );
if( host == l1Path || l1Path == (host + "/") || l1Path == (host + "//"))
{
l1Path = "";
}
}
if( !String.IsNullOrEmpty(host))
{
if( host == domain || host == (domain + "/") || host == (domain + "//"))
{
host = "";
}
}
}
if(!String.IsNullOrEmpty(l1Path) && table.ContainsKey( l1Path ))
{
site = l1Path;
}
else if( !String.IsNullOrEmpty(host) && table.ContainsKey( host ))
{
site = host;
}
else
{
site = domain;
}
if( !getReversedSite)
{
site = UrlReverser.ReverseBack( site );
}
}
catch(Exception)
{
return null;
}
return site;
}