public boolean incrementToken() throws IOException {
if (inPhrase) {
inPhrase = false;
clearAttributes();
termAtt.setTermBuffer("phrase2");
offsetAtt.setOffset(savedStart, savedEnd);
return true;
} else
while (input.incrementToken()) {
if (termAtt.term().equals("phrase")) {
inPhrase = true;
savedStart = offsetAtt.startOffset();
savedEnd = offsetAtt.endOffset();
termAtt.setTermBuffer("phrase1");
offsetAtt.setOffset(savedStart, savedEnd);
return true;
} else if (!termAtt.term().equals("stop")) return true;
}
return false;
}
public boolean incrementToken() throws IOException {
if (!input.incrementToken()) // Advance to next token
return false; // When false, end has been reached
String encoded;
encoded = metaphoner.encode(termAttr.term()); // Convert term text to
// Metaphone encoding
termAttr.setTermBuffer(encoded); // Overwrite term text with encoded
// text
typeAttr.setType(METAPHONE); // Set token type
return true;
}
@Override
public boolean incrementToken() throws IOException {
if (!input.incrementToken()) return false;
char[] termBuffer = termAtt.termBuffer();
int termBufferLength = termAtt.termLength();
char[] backup = null;
if (factory.maxWordCount < CapitalizationFilterFactory.DEFAULT_MAX_WORD_COUNT) {
// make a backup in case we exceed the word count
backup = new char[termBufferLength];
System.arraycopy(termBuffer, 0, backup, 0, termBufferLength);
}
if (termBufferLength < factory.maxTokenLength) {
int wordCount = 0;
int lastWordStart = 0;
for (int i = 0; i < termBufferLength; i++) {
char c = termBuffer[i];
if (c <= ' ' || c == '.') {
int len = i - lastWordStart;
if (len > 0) {
factory.processWord(termBuffer, lastWordStart, len, wordCount++);
lastWordStart = i + 1;
i++;
}
}
}
// process the last word
if (lastWordStart < termBufferLength) {
factory.processWord(
termBuffer, lastWordStart, termBufferLength - lastWordStart, wordCount++);
}
if (wordCount > factory.maxWordCount) {
termAtt.setTermBuffer(backup, 0, termBufferLength);
}
}
return true;
}
private void applyToken(Token token) {
termAtt.setTermBuffer(token.termBuffer(), 0, token.termLength());
posAtt.setPositionIncrement(token.getPositionIncrement());
offsetAtt.setOffset(token.startOffset(), token.endOffset());
}
public static void assertTokenStreamContents(
TokenStream ts,
String[] output,
int startOffsets[],
int endOffsets[],
String types[],
int posIncrements[],
Integer finalOffset)
throws IOException {
assertNotNull(output);
CheckClearAttributesAttribute checkClearAtt =
(CheckClearAttributesAttribute) ts.addAttribute(CheckClearAttributesAttribute.class);
assertTrue("has no TermAttribute", ts.hasAttribute(TermAttribute.class));
TermAttribute termAtt = (TermAttribute) ts.getAttribute(TermAttribute.class);
OffsetAttribute offsetAtt = null;
if (startOffsets != null || endOffsets != null || finalOffset != null) {
assertTrue("has no OffsetAttribute", ts.hasAttribute(OffsetAttribute.class));
offsetAtt = (OffsetAttribute) ts.getAttribute(OffsetAttribute.class);
}
TypeAttribute typeAtt = null;
if (types != null) {
assertTrue("has no TypeAttribute", ts.hasAttribute(TypeAttribute.class));
typeAtt = (TypeAttribute) ts.getAttribute(TypeAttribute.class);
}
PositionIncrementAttribute posIncrAtt = null;
if (posIncrements != null) {
assertTrue(
"has no PositionIncrementAttribute", ts.hasAttribute(PositionIncrementAttribute.class));
posIncrAtt = (PositionIncrementAttribute) ts.getAttribute(PositionIncrementAttribute.class);
}
ts.reset();
for (int i = 0; i < output.length; i++) {
// extra safety to enforce, that the state is not preserved and also assign bogus values
ts.clearAttributes();
termAtt.setTermBuffer("bogusTerm");
if (offsetAtt != null) offsetAtt.setOffset(14584724, 24683243);
if (typeAtt != null) typeAtt.setType("bogusType");
if (posIncrAtt != null) posIncrAtt.setPositionIncrement(45987657);
checkClearAtt.getAndResetClearCalled(); // reset it, because we called clearAttribute() before
assertTrue("token " + i + " does not exist", ts.incrementToken());
assertTrue(
"clearAttributes() was not called correctly in TokenStream chain",
checkClearAtt.getAndResetClearCalled());
assertEquals("term " + i, output[i], termAtt.term());
if (startOffsets != null)
assertEquals("startOffset " + i, startOffsets[i], offsetAtt.startOffset());
if (endOffsets != null) assertEquals("endOffset " + i, endOffsets[i], offsetAtt.endOffset());
if (types != null) assertEquals("type " + i, types[i], typeAtt.type());
if (posIncrements != null)
assertEquals("posIncrement " + i, posIncrements[i], posIncrAtt.getPositionIncrement());
}
assertFalse("end of stream", ts.incrementToken());
ts.end();
if (finalOffset != null)
assertEquals("finalOffset ", finalOffset.intValue(), offsetAtt.endOffset());
ts.close();
}
