public String tokens(String field) {
try {
Field f = doc.getField(field);
if (f == null) fail("No such field " + field);
if (!f.isTokenized()) {
String val = value(field);
Token t = new Token(val, 0, val.length());
return t.getPositionIncrement() + " [" + t.termText() + "]";
}
TokenStream ts = f.tokenStreamValue();
if (ts == null && f.stringValue() != null) ts = analyzer.tokenStream(field, f.stringValue());
if (ts == null && f.readerValue() != null) ts = analyzer.tokenStream(field, f.readerValue());
if (ts == null) fail("No token stream for field " + field);
Token t = null;
StringBuilder sb = new StringBuilder();
while ((t = ts.next()) != null) {
sb.append(t.getPositionIncrement() + " [" + t.termText() + "] ");
}
return sb.toString().trim();
} catch (Exception e) {
e.printStackTrace();
fail(e.getMessage());
return null;
}
}
/**
* Merge two lists of tokens, producing a single list with manipulated positionIncrements so that
* the tokens end up at the same position.
*
* <p>Example: [a b] merged with [c d] produces [a/b c/d] ('/' denotes tokens in the same
* position) Example: [a,5 b,2] merged with [c d,4 e,4] produces [c a,5/d b,2 e,2] (a,n means a
* has posInc=n)
*/
public static List<Token> mergeTokens(List<Token> lst1, List<Token> lst2) {
ArrayList<Token> result = new ArrayList<Token>();
if (lst1 == null || lst2 == null) {
if (lst2 != null) result.addAll(lst2);
if (lst1 != null) result.addAll(lst1);
return result;
}
int pos = 0;
Iterator<Token> iter1 = lst1.iterator();
Iterator<Token> iter2 = lst2.iterator();
Token tok1 = iter1.hasNext() ? iter1.next() : null;
Token tok2 = iter2.hasNext() ? iter2.next() : null;
int pos1 = tok1 != null ? tok1.getPositionIncrement() : 0;
int pos2 = tok2 != null ? tok2.getPositionIncrement() : 0;
while (tok1 != null || tok2 != null) {
while (tok1 != null && (pos1 <= pos2 || tok2 == null)) {
Token tok = new Token(tok1.startOffset(), tok1.endOffset(), tok1.type());
tok.copyBuffer(tok1.buffer(), 0, tok1.length());
tok.setPositionIncrement(pos1 - pos);
result.add(tok);
pos = pos1;
tok1 = iter1.hasNext() ? iter1.next() : null;
pos1 += tok1 != null ? tok1.getPositionIncrement() : 0;
}
while (tok2 != null && (pos2 <= pos1 || tok1 == null)) {
Token tok = new Token(tok2.startOffset(), tok2.endOffset(), tok2.type());
tok.copyBuffer(tok2.buffer(), 0, tok2.length());
tok.setPositionIncrement(pos2 - pos);
result.add(tok);
pos = pos2;
tok2 = iter2.hasNext() ? iter2.next() : null;
pos2 += tok2 != null ? tok2.getPositionIncrement() : 0;
}
}
return result;
}
@Override
public boolean incrementToken() throws IOException {
if (index >= tokens.length) return false;
else {
clearAttributes();
Token token = tokens[index++];
termAtt.setEmpty().append(token);
offsetAtt.setOffset(token.startOffset(), token.endOffset());
posIncAtt.setPositionIncrement(token.getPositionIncrement());
flagsAtt.setFlags(token.getFlags());
typeAtt.setType(token.type());
payloadAtt.setPayload(token.getPayload());
return true;
}
}
@Override
public boolean incrementToken() {
if (upto < tokens.length) {
final Token token = tokens[upto++];
// TODO: can we just capture/restoreState so
// we get all attrs...?
clearAttributes();
termAtt.setEmpty();
termAtt.append(token.toString());
posIncrAtt.setPositionIncrement(token.getPositionIncrement());
posLengthAtt.setPositionLength(token.getPositionLength());
offsetAtt.setOffset(token.startOffset(), token.endOffset());
payloadAtt.setPayload(token.getPayload());
return true;
} else {
return false;
}
}
@Override
public boolean incrementToken() throws IOException {
if (tokens == null) {
fillTokens();
}
// System.out.println("graphTokenizer: incr upto=" + upto + " vs " + tokens.size());
if (upto == tokens.size()) {
// System.out.println(" END @ " + tokens.size());
return false;
}
final Token t = tokens.get(upto++);
// System.out.println(" return token=" + t);
clearAttributes();
termAtt.append(t.toString());
offsetAtt.setOffset(t.startOffset(), t.endOffset());
posIncrAtt.setPositionIncrement(t.getPositionIncrement());
posLengthAtt.setPositionLength(t.getPositionLength());
return true;
}
private void applyToken(Token token) {
termAtt.setTermBuffer(token.termBuffer(), 0, token.termLength());
posAtt.setPositionIncrement(token.getPositionIncrement());
offsetAtt.setOffset(token.startOffset(), token.endOffset());
}