Skip to content

Apache Lucene open-source search software

Not maintain docBufferUpTo when only docs needed (#14164)

gf212152d380948827
diff --git a/lucene/core/src/java/org/apache/lucene/codecs/lucene101/Lucene101PostingsReader.java b/lucene/core/src/java/org/apache/lucene/codecs/lucene101/Lucene101PostingsReader.java
index b4ccff69fed..04b39c23974 100644
--- a/lucene/core/src/java/org/apache/lucene/codecs/lucene101/Lucene101PostingsReader.java
+++ b/lucene/core/src/java/org/apache/lucene/codecs/lucene101/Lucene101PostingsReader.java
@@ -615,19 +615,21 @@ public final class Lucene101PostingsReader extends PostingsReaderBase {
numLongs = -bitsPerValue;
docIn.readLongs(docBitSet.getBits(), 0, numLongs);
}
- // Note: we know that BLOCK_SIZE bits are set, so no need to compute the cumulative pop
- // count at the last index, it will be BLOCK_SIZE.
- // Note: this for loop auto-vectorizes
- for (int i = 0; i < numLongs - 1; ++i) {
- docCumulativeWordPopCounts[i] = Long.bitCount(docBitSet.getBits()[i]);
- }
- for (int i = 1; i < numLongs - 1; ++i) {
- docCumulativeWordPopCounts[i] += docCumulativeWordPopCounts[i - 1];
+ if (needsFreq) {
+ // Note: we know that BLOCK_SIZE bits are set, so no need to compute the cumulative pop
+ // count at the last index, it will be BLOCK_SIZE.
+ // Note: this for loop auto-vectorizes
+ for (int i = 0; i < numLongs - 1; ++i) {
+ docCumulativeWordPopCounts[i] = Long.bitCount(docBitSet.getBits()[i]);
+ }
+ for (int i = 1; i < numLongs - 1; ++i) {
+ docCumulativeWordPopCounts[i] += docCumulativeWordPopCounts[i - 1];
+ }
+ docCumulativeWordPopCounts[numLongs - 1] = BLOCK_SIZE;
+ assert docCumulativeWordPopCounts[numLongs - 2]
+ + Long.bitCount(docBitSet.getBits()[numLongs - 1])
+ == BLOCK_SIZE;
}
- docCumulativeWordPopCounts[numLongs - 1] = BLOCK_SIZE;
- assert docCumulativeWordPopCounts[numLongs - 2]
- + Long.bitCount(docBitSet.getBits()[numLongs - 1])
- == BLOCK_SIZE;
encoding = DeltaEncoding.UNARY;
}
if (indexHasFreq) {
@@ -726,7 +728,7 @@ public final class Lucene101PostingsReader extends PostingsReaderBase {
}
private void doMoveToNextLevel0Block() throws IOException {
- assert docBufferUpto == BLOCK_SIZE;
+ assert doc == level0LastDocID;
if (posIn != null) {
if (level0PosEndFP >= posIn.getFilePointer()) {
posIn.seek(level0PosEndFP);
@@ -912,7 +914,7 @@ public final class Lucene101PostingsReader extends PostingsReaderBase {
@Override
public int nextDoc() throws IOException {
- if (docBufferUpto == BLOCK_SIZE) {
+ if (doc == level0LastDocID) {
moveToNextLevel0Block();
}
@@ -954,13 +956,21 @@ public final class Lucene101PostingsReader extends PostingsReaderBase {
int next = docBitSet.nextSetBit(target - docBitSetBase);
assert next != NO_MORE_DOCS;
this.doc = docBitSetBase + next;
- int wordIndex = next >> 6;
- // Take the cumulative pop count for the given word, and subtract bits on the left of
- // the current doc.
- docBufferUpto =
- 1
- + docCumulativeWordPopCounts[wordIndex]
- - Long.bitCount(docBitSet.getBits()[wordIndex] >>> next);
+ if (needsFreq) {
+ int wordIndex = next >> 6;
+ // Take the cumulative pop count for the given word, and subtract bits on the left of
+ // the current doc.
+ docBufferUpto =
+ 1
+ + docCumulativeWordPopCounts[wordIndex]
+ - Long.bitCount(docBitSet.getBits()[wordIndex] >>> next);
+ } else {
+ // When only docs needed and block is UNARY encoded, we do not need to maintain
+ // docBufferUpTo to record the iteration position in the block.
+ // docBufferUpTo == 0 means the block has not been iterated.
+ // docBufferUpTo != 0 means the block has been iterated.
+ docBufferUpto = 1;
+ }
}
break;
}
@@ -978,7 +988,7 @@ public final class Lucene101PostingsReader extends PostingsReaderBase {
bitSet.set(doc - offset);
for (; ; ) {
- if (docBufferUpto == BLOCK_SIZE) {
+ if (doc == level0LastDocID) {
// refill
moveToNextLevel0Block();
}