public void matchFound(Scanner s, TreeNode match_item)
{
// add the match item to our matches list
this.matches.AddLast(match_item);
TreeFlag f = match_item.getFlags();
this.totalBadness += f.getBadness();
this.totalGoodness += f.getGoodness();
this.totalPostbad += f.getPostbadness();
// if the match is marked bad, add the category to our bad category list
if (f.getBadness() > 0)
{
this.badCategories.AddLast(f.getCategory());
}
// if the match is marked good, add the category to our good category list
if (f.getGoodness() > 0)
{
this.goodCategories.AddLast(f.getCategory());
}
// if the match is marked postbad, add the category to our postbad category list
if (f.getPostbadness() > 0)
{
this.postbadCategories.AddLast(f.getCategory());
}
}
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);
}
