mirror of https://github.com/apache/lucene.git
Use Arrays.compareUnsigned instead of iterating compare. (#13252)
This commit is contained in:
parent
1f1181a079
commit
3024e66e4a
|
@ -288,6 +288,8 @@ Improvements
|
|||
Optimizations
|
||||
---------------------
|
||||
|
||||
* GITHUB#13252: Replace handwritten loops compare with Arrays.compareUnsigned in SegmentTermsEnum. (zhouhui)
|
||||
|
||||
* GITHUB#12996: Reduce ArrayUtil#grow in decompress. (Zhang Chao)
|
||||
|
||||
* GITHUB#13115: Short circuit queued flush check when flush on update is disabled (Prabhat Sharma)
|
||||
|
|
|
@ -18,6 +18,7 @@ package org.apache.lucene.codecs.lucene90.blocktree;
|
|||
|
||||
import java.io.IOException;
|
||||
import java.io.PrintStream;
|
||||
import java.util.Arrays;
|
||||
import org.apache.lucene.codecs.BlockTermState;
|
||||
import org.apache.lucene.index.BaseTermsEnum;
|
||||
import org.apache.lucene.index.ImpactsEnum;
|
||||
|
@ -387,31 +388,18 @@ final class SegmentTermsEnum extends BaseTermsEnum {
|
|||
}
|
||||
|
||||
if (cmp == 0) {
|
||||
final int targetUptoMid = targetUpto;
|
||||
|
||||
// Second compare the rest of the term, but
|
||||
// don't save arc/output/frame; we only do this
|
||||
// to find out if the target term is before,
|
||||
// equal or after the current term
|
||||
final int targetLimit2 = Math.min(target.length, term.length());
|
||||
while (targetUpto < targetLimit2) {
|
||||
cmp =
|
||||
(term.byteAt(targetUpto) & 0xFF) - (target.bytes[target.offset + targetUpto] & 0xFF);
|
||||
// if (DEBUG) {
|
||||
// System.out.println(" cycle2 targetUpto=" + targetUpto + " (vs limit=" +
|
||||
// targetLimit + ") cmp=" + cmp + " (targetLabel=" + (char) (target.bytes[target.offset +
|
||||
// targetUpto]) + " vs termLabel=" + (char) (term.bytes[targetUpto]) + ")");
|
||||
// }
|
||||
if (cmp != 0) {
|
||||
break;
|
||||
}
|
||||
targetUpto++;
|
||||
}
|
||||
|
||||
if (cmp == 0) {
|
||||
cmp = term.length() - target.length;
|
||||
}
|
||||
targetUpto = targetUptoMid;
|
||||
cmp =
|
||||
Arrays.compareUnsigned(
|
||||
term.bytes(),
|
||||
targetUpto,
|
||||
term.length(),
|
||||
target.bytes,
|
||||
target.offset + targetUpto,
|
||||
target.offset + target.length);
|
||||
}
|
||||
|
||||
if (cmp < 0) {
|
||||
|
@ -666,28 +654,16 @@ final class SegmentTermsEnum extends BaseTermsEnum {
|
|||
}
|
||||
|
||||
if (cmp == 0) {
|
||||
final int targetUptoMid = targetUpto;
|
||||
// Second compare the rest of the term, but
|
||||
// don't save arc/output/frame:
|
||||
final int targetLimit2 = Math.min(target.length, term.length());
|
||||
while (targetUpto < targetLimit2) {
|
||||
cmp =
|
||||
(term.byteAt(targetUpto) & 0xFF) - (target.bytes[target.offset + targetUpto] & 0xFF);
|
||||
// if (DEBUG) {
|
||||
// System.out.println(" cycle2 targetUpto=" + targetUpto + " (vs limit=" + targetLimit
|
||||
// + ") cmp=" + cmp + " (targetLabel=" + (char) (target.bytes[target.offset + targetUpto])
|
||||
// + " vs termLabel=" + (char) (term.byteAt(targetUpto)) + ")");
|
||||
// }
|
||||
if (cmp != 0) {
|
||||
break;
|
||||
}
|
||||
targetUpto++;
|
||||
}
|
||||
|
||||
if (cmp == 0) {
|
||||
cmp = term.length() - target.length;
|
||||
}
|
||||
targetUpto = targetUptoMid;
|
||||
cmp =
|
||||
Arrays.compareUnsigned(
|
||||
term.bytes(),
|
||||
targetUpto,
|
||||
term.length(),
|
||||
target.bytes,
|
||||
target.offset + targetUpto,
|
||||
target.offset + target.length);
|
||||
}
|
||||
|
||||
if (cmp < 0) {
|
||||
|
|
|
@ -369,6 +369,42 @@ public abstract class BasePostingsFormatTestCase extends BaseIndexFileFormatTest
|
|||
dir.close();
|
||||
}
|
||||
|
||||
// Test seek in disorder.
|
||||
public void testDisorder() throws Exception {
|
||||
Directory dir = newDirectory();
|
||||
|
||||
IndexWriterConfig iwc = newIndexWriterConfig(null);
|
||||
iwc.setCodec(getCodec());
|
||||
iwc.setMergePolicy(newTieredMergePolicy());
|
||||
IndexWriter iw = new IndexWriter(dir, iwc);
|
||||
|
||||
for (int i = 0; i < 10000; i++) {
|
||||
Document document = new Document();
|
||||
document.add(new StringField("id", i + "", Field.Store.NO));
|
||||
iw.addDocument(document);
|
||||
}
|
||||
iw.commit();
|
||||
iw.forceMerge(1);
|
||||
|
||||
DirectoryReader reader = DirectoryReader.open(iw);
|
||||
TermsEnum termsEnum = getOnlyLeafReader(reader).terms("id").iterator();
|
||||
|
||||
for (int i = 0; i < 20000; i++) {
|
||||
int n = random().nextInt(0, 10000);
|
||||
BytesRef target = new BytesRef(n + "");
|
||||
// seekExact.
|
||||
assertTrue(termsEnum.seekExact(target));
|
||||
assertEquals(termsEnum.term(), target);
|
||||
// seekCeil.
|
||||
assertEquals(SeekStatus.FOUND, termsEnum.seekCeil(target));
|
||||
assertEquals(termsEnum.term(), target);
|
||||
}
|
||||
|
||||
reader.close();
|
||||
iw.close();
|
||||
dir.close();
|
||||
}
|
||||
|
||||
protected void subCheckBinarySearch(TermsEnum termsEnum) throws Exception {}
|
||||
|
||||
public void testBinarySearchTermLeaf() throws Exception {
|
||||
|
|
Loading…
Reference in New Issue