private String wj(Sentence sent, int pos, int i, int j) {
String cp = "wj:" + Integer.toString(i) + ":" + Integer.toString(j) + ":";
if ((pos + i) >= sent.size() || (pos + i) < 0 || (pos + j) < 0 || (pos + j) >= sent.size())
cp = "";
else {
cp += sent.getWordAt(pos + i) + ":" + sent.getWordAt(pos + j);
}
return cp;
}
private String w(Sentence sent, int pos, int i) {
String cp = "w:" + Integer.toString(i) + ":";
if (pos + i == -1) cp += "BS";
else if (pos + i == sent.size()) cp += "ES";
else if (0 <= (pos + i) && (pos + i) < sent.size()) cp += sent.getWordAt(pos + i);
else cp = "";
return cp;
}
private String mk(Sentence sent, int pos, int i) {
String cp;
if (0 <= (pos + i) && (pos + i) < sent.size()) {
cp = "mk:" + Integer.toString(i);
String word = sent.getWordAt(pos + i);
if (!StringUtils.isSign(word)) cp = "";
} else cp = "";
return cp;
}
private String com(Sentence sent, int pos, int i) {
String cp;
if (0 <= (pos + i) && (pos + i) < sent.size()) {
cp = "hyph:" + Integer.toString(i);
String word = sent.getWordAt(pos + i);
if (!word.contains(":")) cp = "";
} else cp = "";
return cp;
}
private String hn(Sentence sent, int pos, int i) {
String cp;
if (0 <= (pos + i) && (pos + i) < sent.size()) {
cp = "hn:" + Integer.toString(i);
String word = sent.getWordAt(pos + i);
if (!StringUtils.containNumber(word)) cp = "";
} else cp = "";
return cp;
}
private String sff(Sentence sent, int pos, int i) {
String cp;
if (0 <= (pos + i) && (pos + i) < sent.size()) {
cp = "sff:" + Integer.toString(i) + ":";
String word = sent.getWordAt(pos + i);
String[] sylls = word.split("_");
if (sylls.length >= 2) {
cp += sylls[sylls.length - 1];
} else cp = "";
} else cp = "";
return cp;
}
private String dict(Sentence sent, int pos, int i) {
String cp = "";
if (0 <= (pos + i) && (pos + i) < sent.size()) {
String word = sent.getWordAt(pos + i);
if (word2dictags.containsKey(word)) {
List tags = (List) word2dictags.get(word);
for (int j = 0; j < tags.size(); ++j) {
cp += "dict:" + Integer.toString(i) + ":" + tags.get(j) + " ";
}
}
}
return cp.trim();
}
private String ac(Sentence sent, int pos, int i) {
String cp;
if (0 <= (pos + i) && (pos + i) < sent.size()) {
cp = "ac:" + Integer.toString(i);
String word = sent.getWordAt(pos + i);
boolean isAllCap = true;
for (int j = 0; j < word.length(); ++j) {
if (word.charAt(j) == '_' || word.charAt(j) == '.') continue;
if (!Character.isUpperCase(word.charAt(j))) {
isAllCap = false;
break;
}
}
if (!isAllCap) cp = "";
} else cp = "";
return cp;
}
private String rr(Sentence sent, int pos, int i) {
String cp = "";
if (0 <= (pos + i) && (pos + i) < sent.size()) {
String word = sent.getWordAt(pos + i);
String[] sylls = word.split("_");
if (sylls.length == 2) { // consider 2-syllable words
VnSyllParser parser1 = new VnSyllParser(sylls[0]);
VnSyllParser parser2 = new VnSyllParser(sylls[1]);
if (parser1.isValidVnSyllable() && parser2.isValidVnSyllable()) {
if (parser1.getNonToneSyll().equalsIgnoreCase(parser2.getNonToneSyll())) {
cp += "fr:" + Integer.toString(i) + " ";
} else if (parser1.getRhyme().equalsIgnoreCase(parser2.getRhyme())) {
cp += "pr:" + Integer.toString(i) + " ";
}
}
}
}
return cp.trim();
}