Skip to content

Apache Lucene open-source search software

Use ArrayDeque for JapaneseCompletionFilter output buffering (#16416)

Luiz Anselmo52bbea2e0bf0
diff --git a/lucene/CHANGES.txt b/lucene/CHANGES.txt
index 422332fcda5..3a3fdbf3ec7 100644
--- a/lucene/CHANGES.txt
+++ b/lucene/CHANGES.txt
@@ -392,6 +392,9 @@ Improvements
Optimizations
---------------------
+* GITHUB#16416: Use ArrayDeque instead of ArrayList for FIFO output buffering in
+ JapaneseCompletionFilter, avoiding repeated element shifting on removal. (Luiz Lima)
+
* GITHUB#16394: Add DocIdSetIterator#intoArray, which DisjunctionDISIApproximation implements by
loading a window of doc IDs through #intoBitSet rather than one nextDoc() call per match.
ConstantScoreScorer#nextDocsAndScores now builds on it, which avoids a priority-queue update per
diff --git a/lucene/analysis/kuromoji/src/java/org/apache/lucene/analysis/ja/JapaneseCompletionFilter.java b/lucene/analysis/kuromoji/src/java/org/apache/lucene/analysis/ja/JapaneseCompletionFilter.java
index 310a143a52e..116be917514 100644
--- a/lucene/analysis/kuromoji/src/java/org/apache/lucene/analysis/ja/JapaneseCompletionFilter.java
+++ b/lucene/analysis/kuromoji/src/java/org/apache/lucene/analysis/ja/JapaneseCompletionFilter.java
@@ -17,7 +17,8 @@
package org.apache.lucene.analysis.ja;
import java.io.IOException;
-import java.util.ArrayList;
+import java.util.ArrayDeque;
+import java.util.Deque;
import java.util.Iterator;
import java.util.List;
import org.apache.lucene.analysis.TokenFilter;
@@ -145,7 +146,7 @@ public final class JapaneseCompletionFilter extends TokenFilter {
private final Mode mode;
- private List<CompletionToken> outputs;
+ private final Deque<CompletionToken> outputs;
private CharsRefBuilder pdgSurface;
private CharsRefBuilder pdgReading;
@@ -154,7 +155,7 @@ public final class JapaneseCompletionFilter extends TokenFilter {
CompletionTokenGenerator(Mode mode) {
this.mode = mode;
- outputs = new ArrayList<>();
+ outputs = new ArrayDeque<>();
}
public void reset() {
@@ -164,12 +165,12 @@ public final class JapaneseCompletionFilter extends TokenFilter {
@Override
public boolean hasNext() {
- return outputs.size() > 0;
+ return !outputs.isEmpty();
}
@Override
public CompletionToken next() {
- return outputs.remove(0);
+ return outputs.removeFirst();
}
void addToken(String surface, String reading, int startOffset, int endOffset) {
@@ -215,7 +216,8 @@ public final class JapaneseCompletionFilter extends TokenFilter {
private void generateOutputs() {
// preserve original surface form as an output.
- outputs.add(new CompletionToken(pdgSurface.toString(), true, pdgStartOffset, pdgEndOffset));
+ outputs.addLast(
+ new CompletionToken(pdgSurface.toString(), true, pdgStartOffset, pdgEndOffset));
// skip readings that cannot be translated to romaji.
if (pdgReading == null
|| pdgReading.length() == 0
@@ -226,7 +228,7 @@ public final class JapaneseCompletionFilter extends TokenFilter {
List<CharsRef> romaji = KatakanaRomanizer.getInstance().romanize(pdgReading.get());
for (CharsRef ref : romaji) {
// set the same start/end offset as the original surface form for romanized tokens.
- outputs.add(new CompletionToken(ref.toString(), false, pdgStartOffset, pdgEndOffset));
+ outputs.addLast(new CompletionToken(ref.toString(), false, pdgStartOffset, pdgEndOffset));
}
}