public boolean addWord(String str, long fileid, long pos) {
/*
* Filter out certain strings. e.g. words that are either very rare (e.g.
* spelling mistakes) or extremely common "stop words".
* http://en.wikipedia.org/wiki/Stop_words
*
* Our definition of stop words is completely arbitrary. It has more to do
* with what kind of computation we can do on a single appengine node.
*
* Ideally we would not ignore any words, but that will require sharding the
* word index across a number of serving machines.
*/
if (str.length() < 3) return false;
int freq = dict.getFrequency(str);
if (freq == 0 || freq > 90000) return false;
ArrayList<FilePositions> fplist = idx.get(str);
if (fplist == null) {
fplist = new ArrayList<FilePositions>();
idx.put(str, fplist);
}
if (fplist.size() == 0 || fplist.get(fplist.size() - 1).fileid != fileid) {
FilePositions fp = new FilePositions();
fp.fileid = fileid;
fp.poslist = new ArrayList<Long>();
fplist.add(fp);
}
fplist.get(fplist.size() - 1).poslist.add(pos);
return true;
}
public static int getMaxFrequency(
final ConcurrentHashMap<String, Dictionary> dictionaries, CharSequence word) {
if (TextUtils.isEmpty(word)) {
return Dictionary.NOT_A_PROBABILITY;
}
int maxFreq = -1;
for (final String key : dictionaries.keySet()) {
final Dictionary dictionary = dictionaries.get(key);
if (null == dictionary) continue;
final int tempFreq = dictionary.getFrequency(word);
if (tempFreq >= maxFreq) {
maxFreq = tempFreq;
}
}
return maxFreq;
}
