// Searches for the tag in the LexEntries of the vector v
private FreqLexEntry searchTag(String tag, Vector vector) {
for (int i = 0; i < vector.size(); i++) {
FreqLexEntry myEntry = (FreqLexEntry) vector.elementAt(i);
if (myEntry.getTag().equals(tag)) return myEntry;
}
return null;
}
private int getFrequency(String word, HashMap theHash) {
Vector entries = (Vector) theHash.get(word);
if (entries == null) return 0;
int frequency = 0;
for (int i = 0; i < entries.size(); i++) {
FreqLexEntry entry = (FreqLexEntry) entries.elementAt(i);
frequency = frequency + entry.getFrequency();
}
return frequency;
}
private String lookup(HashMap theHash, String word) {
Vector entries = (Vector) theHash.get(word);
String tagStr = null;
if (entries != null) {
for (int i = 0; i < entries.size(); i++) {
FreqLexEntry entry = (FreqLexEntry) entries.elementAt(i);
if (i == 0) tagStr = entry.getTag();
else tagStr = tagStr + "_" + entry.getTag();
}
}
return tagStr;
}
private int getFrequencyWordTag(HashMap theHash, String word, String tag) {
Vector entries = (Vector) theHash.get(word);
if (entries == null) return 0;
int frequency = 0;
for (int i = 0; i < entries.size(); i++) {
FreqLexEntry entry = (FreqLexEntry) entries.elementAt(i);
if (entry.getTag().equals(tag)) {
frequency = entry.getFrequency();
break;
}
}
return frequency;
}
// Builds a hash for tags of rare words used in suffix handling
// A tag is a key, a frequency is a value
private void storeSuffixTags(Vector vector, HashMap suffixTags /*, int howOften*/) {
// The vector stores all the tags and frequencies to add to the hash
for (int i = 0; i < vector.size(); i++) {
FreqLexEntry entry = (FreqLexEntry) vector.elementAt(i); // The entry object
String tag = entry.getTag(); // The tag string
int freq = entry.getFrequency(); // The frequency
Integer intObj = (Integer) suffixTags.get(tag); // Get the integer object
if (intObj == null) // If not found
{
intObj = new Integer(freq);
suffixTags.put(tag, intObj);
} else // else increment the frequency
{
int theFreq = intObj.intValue();
theFreq += freq;
intObj = new Integer(theFreq);
suffixTags.put(tag, intObj);
}
}
}
private void addSuffixEntry(String suffix, Vector vectorToAdd, HashMap theHashMap) {
// We have to make copies of the entries in the vector vectorToAdd and use those entries
// in the hashmap, because else we are manipulating the same objects (FreqLexEntry) as used in
// the word hashmap!
FreqLexEntry entryToAdd;
Vector myVector = (Vector) theHashMap.get(suffix);
if (myVector == null) // If nothing is there yet
{
myVector = new Vector();
for (int i = 0; i < vectorToAdd.size(); i++) {
entryToAdd = (FreqLexEntry) vectorToAdd.elementAt(i);
FreqLexEntry newEntry = new FreqLexEntry(entryToAdd.getTag(), entryToAdd.getFrequency());
myVector.add(newEntry);
}
theHashMap.put(
suffix, myVector); // The suffix is the key, the value is a vector of LexEntries
} else { // The vector is already there - its values now need to be updated
for (int i = 0;
i < vectorToAdd.size();
i++) // Loop through the vector which holds entries to be added
{
entryToAdd = (FreqLexEntry) vectorToAdd.elementAt(i);
String tag = entryToAdd.getTag();
FreqLexEntry myEntry =
searchTag(tag, myVector); // Get my entry which corresponds to this tag
if (myEntry == null) // Then the tag was not found in the vector, add a new entry
{
myEntry = new FreqLexEntry(entryToAdd.getTag(), entryToAdd.getFrequency());
myVector.add(myEntry);
} else { // The tag was found in the vector; increment the tag frequency
myEntry.setFrequency(myEntry.getFrequency() + entryToAdd.getFrequency());
}
}
}
}