private Token nextToken(Token reusableToken) throws IOException {
assert reusableToken != null;
// 先使用上次留下来的。
Token nextToken = tokenQueue.poll();
if (nextToken != null) {
return nextToken;
}
/*//在 TokenUtils.nextToken 已经调用了 inc
if(!input.incrementToken()) {
return null;
}*/
/*TermAttribute termAtt = (TermAttribute)input.getAttribute(TermAttribute.class);
OffsetAttribute offsetAtt = (OffsetAttribute)input.getAttribute(OffsetAttribute.class);
TypeAttribute typeAtt = (TypeAttribute)input.getAttribute(TypeAttribute.class);
nextToken = reusableToken.reinit(termAtt.termBuffer(), 0, termAtt.termLength(), offsetAtt.startOffset(), offsetAtt.endOffset(), typeAtt.type());*/
nextToken = TokenUtils.nextToken(input, reusableToken);
if (nextToken != null
&& (Word.TYPE_LETTER_OR_DIGIT.equalsIgnoreCase(nextToken.type())
|| Word.TYPE_DIGIT_OR_LETTER.equalsIgnoreCase(nextToken.type()))) {
final char[] buffer = nextToken.buffer();
final int length = nextToken.length();
byte lastType = (byte) Character.getType(buffer[0]); // 与上次的字符是否同类
int termBufferOffset = 0;
int termBufferLength = 0;
for (int i = 0; i < length; i++) {
byte type = (byte) Character.getType(buffer[i]);
if (type <= Character.MODIFIER_LETTER) {
type = Character.LOWERCASE_LETTER;
}
if (type != lastType) { // 与上一次的不同
addToken(nextToken, termBufferOffset, termBufferLength, lastType);
termBufferOffset += termBufferLength;
termBufferLength = 0;
lastType = type;
}
termBufferLength++;
}
if (termBufferLength > 0) { // 最后一次
addToken(nextToken, termBufferOffset, termBufferLength, lastType);
}
nextToken = tokenQueue.poll();
}
return nextToken;
}
public boolean incrementToken() throws IOException {
clearAttributes();
Token token = nextToken(reusableToken);
if (token != null) {
termAtt.copyBuffer(token.buffer(), 0, token.length());
offsetAtt.setOffset(token.startOffset(), token.endOffset());
typeAtt.setType(token.type());
return true;
} else {
end();
return false;
}
}
/**
* 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;
}
}
public Token next() throws IOException {
Token candidate;
while ((candidate = baseTokeniser.next()) != null) {
try {
Integer integer = Integer.valueOf(candidate.termText());
String valueString = NumericEncoder.encode(integer.intValue());
Token integerToken =
new Token(
valueString, candidate.startOffset(), candidate.startOffset(), candidate.type());
return integerToken;
} catch (NumberFormatException e) {
// just ignore and try the next one
}
}
return null;
}
public final Token next() throws IOException {
Token localToken = this.input.next();
if (localToken == null) return null;
String str1 = localToken.termText();
String str2 = localToken.type();
if ((str2 == APOSTROPHE_TYPE) && ((str1.endsWith("'s")) || (str1.endsWith("'S"))))
return new Token(
str1.substring(0, str1.length() - 2),
localToken.startOffset(),
localToken.endOffset(),
str2);
if (str2 == ACRONYM_TYPE) {
StringBuffer localStringBuffer = new StringBuffer();
for (int i = 0; i < str1.length(); i++) {
char c = str1.charAt(i);
if (c != '.') localStringBuffer.append(c);
}
return new Token(
localStringBuffer.toString(), localToken.startOffset(), localToken.endOffset(), str2);
}
return localToken;
}
/**
* Returns the next token in the stream, or null at EOS.
*
* <p>Removes <tt>'s</tt> from the end of words.
*
* <p>Removes dots from acronyms.
*
* <p>Splits host names ...
*/
public final org.apache.lucene.analysis.Token next() throws java.io.IOException {
if (hostTokens == null) {
org.apache.lucene.analysis.Token t = input.next();
if (t == null) return null;
String text = t.termText();
String type = t.type();
if (type == APOSTROPHE_TYPE
&& // remove 's
(text.endsWith("'s") || text.endsWith("'S"))) {
return new org.apache.lucene.analysis.Token(
text.substring(0, text.length() - 2), t.startOffset(), t.endOffset(), type);
} else if (type == ACRONYM_TYPE) { // remove dots
StringBuffer trimmed = new StringBuffer();
for (int i = 0; i < text.length(); i++) {
char c = text.charAt(i);
if (c != '.') trimmed.append(c);
}
return new org.apache.lucene.analysis.Token(
trimmed.toString(), t.startOffset(), t.endOffset(), type);
} else if (type == HOST_TYPE) {
// <HOST: <ALPHANUM> ("." <ALPHANUM>)+ >
// There must be at least two tokens ....
hostTokens = new LinkedList<org.apache.lucene.analysis.Token>();
StringTokenizer tokeniser = new StringTokenizer(text, ".");
int start = t.startOffset();
int end;
while (tokeniser.hasMoreTokens()) {
String token = tokeniser.nextToken();
end = start + token.length();
hostTokens.offer(new org.apache.lucene.analysis.Token(token, start, end, ALPHANUM_TYPE));
start = end + 1;
}
// check if we have an acronym ..... yes a.b.c ends up here ...
if (text.length() == hostTokens.size() * 2 - 1) {
hostTokens = null;
// acronym
StringBuffer trimmed = new StringBuffer();
for (int i = 0; i < text.length(); i++) {
char c = text.charAt(i);
if (c != '.') trimmed.append(c);
}
return new org.apache.lucene.analysis.Token(
trimmed.toString(), t.startOffset(), t.endOffset(), ALPHANUM_TYPE);
} else {
return hostTokens.remove();
}
} else {
return t;
}
} else {
org.apache.lucene.analysis.Token token = hostTokens.remove();
if (hostTokens.isEmpty()) {
hostTokens = null;
}
return token;
}
}
