Added the rest of collocabilities
This commit is contained in:
@@ -52,10 +52,53 @@ public class Ngrams {
|
||||
// generate proper MultipleHMKeys depending on filter data
|
||||
String key = wordToString(ngramCandidate, stats.getFilter().getCalculateFor(), stats.getFilter().getWordParts());
|
||||
|
||||
if(key.length() < stats.getFilter().getPrefixLength() + stats.getFilter().getSuffixLength()){
|
||||
if(stats.getFilter().getPrefixLength() != null && stats.getFilter().getSuffixLength() != null &&
|
||||
key.length() < stats.getFilter().getPrefixLength() + stats.getFilter().getSuffixLength()){
|
||||
continue;
|
||||
}
|
||||
|
||||
if(stats.getFilter().getPrefixList() != null && stats.getFilter().getSuffixList() != null &&
|
||||
(stats.getFilter().getPrefixList().size() > 0 || stats.getFilter().getSuffixList().size() > 0)){
|
||||
|
||||
String correctPrefix = "";
|
||||
// go over all prefixes in PrefixList and look for them in words
|
||||
for(String pf : stats.getFilter().getPrefixList()){
|
||||
if (pf.length() <= key.length() && pf.equals(key.substring(0, pf.length()))){
|
||||
correctPrefix = pf;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
String correctSuffix = "";
|
||||
// go over all prefixes in SuffixList and look for them in words
|
||||
for(String sf : stats.getFilter().getSuffixList()){
|
||||
if (sf.length() <= key.length() && sf.equals(key.substring(key.length() - sf.length()))){
|
||||
correctSuffix = sf;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
// boolean a = (correctPrefix.equals("") && !correctSuffix.equals(""));
|
||||
// boolean b = (!correctPrefix.equals("") && correctSuffix.equals(""));
|
||||
// boolean c = (!correctPrefix.equals("") && !correctSuffix.equals("") && correctPrefix.length() + correctSuffix.length() <= key.length());
|
||||
// boolean d = !((correctPrefix.equals("") && !correctSuffix.equals("")) ||
|
||||
// (!correctPrefix.equals("") && correctSuffix.equals("")) ||
|
||||
// (!correctPrefix.equals("") && !correctSuffix.equals("") && correctPrefix.length() + correctSuffix.length() <= key.length()));
|
||||
|
||||
if(!((stats.getFilter().getPrefixList().size() == 0 && !correctSuffix.equals("")) ||
|
||||
(!correctPrefix.equals("") && stats.getFilter().getSuffixList().size() == 0) ||
|
||||
(!correctPrefix.equals("") && !correctSuffix.equals("") && correctPrefix.length() + correctSuffix.length() <= key.length()))){
|
||||
continue;
|
||||
}
|
||||
|
||||
// if(!((correctPrefix.equals("") && !correctSuffix.equals("")) ||
|
||||
// (!correctPrefix.equals("") && correctSuffix.equals("")) ||
|
||||
// (!correctPrefix.equals("") && !correctSuffix.equals("") && correctPrefix.length() + correctSuffix.length() <= key.length()))){
|
||||
// continue;
|
||||
// }
|
||||
|
||||
}
|
||||
|
||||
// if last letter is ',' erase it
|
||||
|
||||
// if (key.equals("")){
|
||||
|
||||
Reference in New Issue
Block a user