public int quantifyResultsForUrl(String url_to_test)
{
KernelResults url_results = searchAll(url_to_test);
KernelResults hostname_results = searchHostNameOfUrl(url_to_test);
return(quantifyResultsForUrl(url_to_test, hostname_results, url_results));
}
public KernelResults searchAll(String buf)
{
KernelResults target = new KernelResults();
bad_url_scanner.ScanBuffer(buf, target);
good_url_scanner.ScanBuffer(buf, target);
postbad_url_scanner.ScanBuffer(buf, target);
bad_phrase_scanner.ScanBuffer(buf, target);
return(target);
}
public int quantifyResultsForUrlData(KernelResults text_results, KernelResults link_results)
{
int result = 0;
bool is_good = false;
bool is_bad = false;
bool is_unknown = false;
if (link_results.getTotalBadness() > 0)
{
is_bad = true;
is_good = false;
is_unknown = false;
}
if (link_results.getTotalPostbad() > 0 || link_results.getTotalPostbad() > 0)
{
is_bad = true;
is_good = false;
is_unknown = false;
}
if (text_results.getTotalBadness() > 0)
{
is_bad = true;
is_good = false;
is_unknown = false;
}
if (text_results.getTotalPostbad() > 0)
{
is_bad = true;
is_good = false;
is_unknown = false;
}
if (is_bad)
{
result = -1;
}
if (is_good)
{
result = 1;
}
return(result);
}
public KernelResults searchHostNameOfUrl(String url_string)
{
KernelResults target = new KernelResults();
Uri url = null;
String host = null;
url = new Uri(url_string);
host = url.Host;
bad_url_scanner.ScanBuffer(host, target);
//good_url_scanner.ScanBuffer(host, target);
postbad_url_scanner.ScanBuffer(host, target);
bad_phrase_scanner.ScanBuffer(host, target);
return(target);
}
public int quantifyResultsForUrl(String url, KernelResults hostname_results,
KernelResults url_results)
{
int result = 0;
bool is_good = false;
bool is_bad = false;
bool is_unknown = true;
// if the hostname is known bad then we are bad
if (hostname_results.getTotalBadness() > 0)
{
is_bad = true;
is_good = false;
is_unknown = false;
}
// if the hostname is known good then we are good
if (hostname_results.getTotalGoodness() > 0)
{
is_good = true;
is_bad = false;
is_unknown = false;
}
// if the hostname or the full url is known postbad then we are bad
if (hostname_results.getTotalPostbad() > 0 || url_results.getTotalPostbad() > 0)
{
is_bad = true;
is_good = false;
is_unknown = false;
}
// if the site is not yet known then check the rest of the url
if (is_unknown)
{
// if the rest of the url is known bad then we are bad
if (url_results.getTotalBadness() > 0)
{
is_bad = true;
is_good = false;
is_unknown = false;
}
// if the rest of the url is known postbad then we are bad
if (url_results.getTotalPostbad() > 0)
{
is_bad = true;
is_good = false;
is_unknown = false;
}
if (url_results.getTotalGoodness() > 0)
{
foreach (TreeNode i in url_results.getMatches())
{
TreeFlag f = i.getFlags();
if (f.getGoodness() > 0)
{
String extracted = i.extractWord();
if (url == extracted ||
url == "http://" + extracted ||
url == "http://www." + extracted)
{
is_bad = false;
is_good = true;
is_unknown = false;
break;
}
}
}
}
}
if (is_bad)
{
result = -1;
}
if (is_good)
{
result = 1;
}
return(result);
}
