in opennlp-tools/src/main/java/opennlp/tools/tokenize/DefaultTokenContextGenerator.java [121:149]
protected void addCharPreds(String key, char c, List<String> preds) {
preds.add(key + "=" + c);
if (Character.isLetter(c)) {
preds.add(key + "_alpha");
if (Character.isUpperCase(c)) {
preds.add(key + "_caps");
}
}
else if (Character.isDigit(c)) {
preds.add(key + "_num");
}
else if (StringUtil.isWhitespace(c)) {
preds.add(key + "_ws");
}
else {
if (c == '.' || c == '?' || c == '!') {
preds.add(key + "_eos");
}
else if (c == '`' || c == '"' || c == '\'') {
preds.add(key + "_quote");
}
else if (c == '[' || c == '{' || c == '(') {
preds.add(key + "_lp");
}
else if (c == ']' || c == '}' || c == ')') {
preds.add(key + "_rp");
}
}
}