Skip to content

Apache Lucene open-source search software

Avoid unnecessary sorting and instantiations in readMapOfStrings (#15457)

Benjamin Lererfe761a06f026
diff --git a/lucene/CHANGES.txt b/lucene/CHANGES.txt
index 98d26031709..131abf0a09c 100644
--- a/lucene/CHANGES.txt
+++ b/lucene/CHANGES.txt
@@ -69,6 +69,8 @@ Improvements
* GITHUB#15124: Use RamUsageEstimator to calculate size for non-accountable queries. (Sagar Upadhyaya)
+* GITHUB#15453: Avoid unnecessary sorting and instantiations in readMapOfStrings. (Benjamin Lerer)
+
Optimizations
---------------------
* GITHUB#14011: Reduce allocation rate in HNSW concurrent merge. (Viliam Durina)
diff --git a/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldDocValuesFormat.java b/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldDocValuesFormat.java
index 58407de98c3..9b00de95044 100644
--- a/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldDocValuesFormat.java
+++ b/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldDocValuesFormat.java
@@ -185,7 +185,6 @@ public abstract class PerFieldDocValuesFormat extends DocValuesFormat {
}
final String formatName = format.getName();
- field.putAttribute(PER_FIELD_FORMAT_KEY, formatName);
Integer suffix = null;
ConsumerAndSuffix consumer = formats.get(format);
@@ -229,8 +228,8 @@ public abstract class PerFieldDocValuesFormat extends DocValuesFormat {
assert suffixes.containsKey(formatName);
suffix = consumer.suffix;
}
-
- field.putAttribute(PER_FIELD_SUFFIX_KEY, Integer.toString(suffix));
+ field.putAttributes(
+ Map.of(PER_FIELD_FORMAT_KEY, formatName, PER_FIELD_SUFFIX_KEY, Integer.toString(suffix)));
// TODO: we should only provide the "slice" of FIS
// that this DVF actually sees ...
return consumer.consumer;
diff --git a/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldKnnVectorsFormat.java b/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldKnnVectorsFormat.java
index c36ab3b7df1..16ff34042fa 100644
--- a/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldKnnVectorsFormat.java
+++ b/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldKnnVectorsFormat.java
@@ -148,7 +148,6 @@ public abstract class PerFieldKnnVectorsFormat extends KnnVectorsFormat {
}
final String formatName = format.getName();
- field.putAttribute(PER_FIELD_FORMAT_KEY, formatName);
Integer suffix;
WriterAndSuffix writerAndSuffix = formats.get(format);
@@ -176,7 +175,8 @@ public abstract class PerFieldKnnVectorsFormat extends KnnVectorsFormat {
assert suffixes.containsKey(formatName);
suffix = writerAndSuffix.suffix;
}
- field.putAttribute(PER_FIELD_SUFFIX_KEY, Integer.toString(suffix));
+ field.putAttributes(
+ Map.of(PER_FIELD_FORMAT_KEY, formatName, PER_FIELD_SUFFIX_KEY, Integer.toString(suffix)));
return writerAndSuffix.writer;
}
diff --git a/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldPostingsFormat.java b/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldPostingsFormat.java
index e5efab8a518..5e247c50b5c 100644
--- a/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldPostingsFormat.java
+++ b/lucene/core/src/java/org/apache/lucene/codecs/perfield/PerFieldPostingsFormat.java
@@ -245,8 +245,12 @@ public abstract class PerFieldPostingsFormat extends PostingsFormat {
groupBuilder.addField(field);
- fieldInfo.putAttribute(PER_FIELD_FORMAT_KEY, formatName);
- fieldInfo.putAttribute(PER_FIELD_SUFFIX_KEY, Integer.toString(groupBuilder.suffix));
+ fieldInfo.putAttributes(
+ Map.of(
+ PER_FIELD_FORMAT_KEY,
+ formatName,
+ PER_FIELD_SUFFIX_KEY,
+ Integer.toString(groupBuilder.suffix)));
}
Map<PostingsFormat, FieldsGroup> formatToGroups =
diff --git a/lucene/core/src/java/org/apache/lucene/index/FieldInfo.java b/lucene/core/src/java/org/apache/lucene/index/FieldInfo.java
index e6bdddc9239..e87bd2e8c16 100644
--- a/lucene/core/src/java/org/apache/lucene/index/FieldInfo.java
+++ b/lucene/core/src/java/org/apache/lucene/index/FieldInfo.java
@@ -113,7 +113,7 @@ public final class FieldInfo {
this.omitNorms = false;
}
this.dvGen = dvGen;
- this.attributes = Objects.requireNonNull(attributes);
+ this.attributes = Map.copyOf(Objects.requireNonNull(attributes));
this.pointDimensionCount = pointDimensionCount;
this.pointIndexDimensionCount = pointIndexDimensionCount;
this.pointNumBytes = pointNumBytes;
@@ -679,6 +679,20 @@ public final class FieldInfo {
return oldValue;
}
+ /**
+ * Puts some codec attribute values.
+ *
+ * <p>If multiples attributes need to be added {@code putAttributes} is more efficient than
+ * calling {@code putAttribute} multiple times as it avoid unnecessary copies and synchronisation.
+ */
+ public synchronized void putAttributes(Map<String, String> map) {
+ HashMap<String, String> newMap = new HashMap<>(attributes);
+ newMap.putAll(map);
+ // This needs to be thread-safe as multiple threads may be updating (different) attributes
+ // concurrently due to concurrent merging.
+ attributes = Collections.unmodifiableMap(newMap);
+ }
+
/** Returns internal codec attributes map. */
public synchronized Map<String, String> attributes() {
return attributes;
diff --git a/lucene/core/src/java/org/apache/lucene/index/FieldInfos.java b/lucene/core/src/java/org/apache/lucene/index/FieldInfos.java
index ee37b837e72..269f24bf1fd 100644
--- a/lucene/core/src/java/org/apache/lucene/index/FieldInfos.java
+++ b/lucene/core/src/java/org/apache/lucene/index/FieldInfos.java
@@ -775,7 +775,7 @@ public class FieldInfos implements Iterable<FieldInfo> {
if (curFi != null) {
curFi.verifySameSchema(fi);
if (fi.attributes() != null) {
- fi.attributes().forEach((k, v) -> curFi.putAttribute(k, v));
+ curFi.putAttributes(fi.attributes());
}
if (fi.hasPayloads()) {
curFi.setStorePayloads();
diff --git a/lucene/core/src/java/org/apache/lucene/index/SegmentInfo.java b/lucene/core/src/java/org/apache/lucene/index/SegmentInfo.java
index 9e502b640b7..b85a6cb9e63 100644
--- a/lucene/core/src/java/org/apache/lucene/index/SegmentInfo.java
+++ b/lucene/core/src/java/org/apache/lucene/index/SegmentInfo.java
@@ -362,6 +362,7 @@ public final class SegmentInfo {
// This needs to be thread-safe because multiple threads may be updating (different) attributes
// at the same time due to concurrent merging, plus some threads may be calling toString() on
// segment info while other threads are updating attributes.
+ // We use unmodifiableMap instead of Map.copyOf to avoid an unnecessary copy.
attributes = Collections.unmodifiableMap(newMap);
return oldValue;
}
diff --git a/lucene/core/src/java/org/apache/lucene/store/DataInput.java b/lucene/core/src/java/org/apache/lucene/store/DataInput.java
index 55bd781aee6..62ea389a835 100644
--- a/lucene/core/src/java/org/apache/lucene/store/DataInput.java
+++ b/lucene/core/src/java/org/apache/lucene/store/DataInput.java
@@ -18,14 +18,9 @@ package org.apache.lucene.store;
import java.io.IOException;
import java.nio.charset.StandardCharsets;
-import java.util.Collections;
-import java.util.HashMap;
-import java.util.HashSet;
import java.util.Map;
import java.util.Objects;
import java.util.Set;
-import java.util.TreeMap;
-import java.util.TreeSet;
import org.apache.lucene.util.BitUtil;
/**
@@ -248,17 +243,17 @@ public abstract class DataInput implements Cloneable {
public Map<String, String> readMapOfStrings() throws IOException {
int count = readVInt();
if (count == 0) {
- return Collections.emptyMap();
+ return Map.of();
} else if (count == 1) {
- return Collections.singletonMap(readString(), readString());
+ return Map.of(readString(), readString());
} else {
- Map<String, String> map = count > 10 ? new HashMap<>() : new TreeMap<>();
+ @SuppressWarnings("unchecked")
+ Map.Entry<String, String>[] entries =
+ (Map.Entry<String, String>[]) new Map.Entry<?, ?>[count];
for (int i = 0; i < count; i++) {
- final String key = readString();
- final String val = readString();
- map.put(key, val);
+ entries[i] = Map.entry(readString(), readString());
}
- return Collections.unmodifiableMap(map);
+ return Map.ofEntries(entries);
}
}
@@ -270,15 +265,15 @@ public abstract class DataInput implements Cloneable {
public Set<String> readSetOfStrings() throws IOException {
int count = readVInt();
if (count == 0) {
- return Collections.emptySet();
+ return Set.of();
} else if (count == 1) {
- return Collections.singleton(readString());
+ return Set.of(readString());
} else {
- Set<String> set = count > 10 ? new HashSet<>() : new TreeSet<>();
+ String[] set = new String[count];
for (int i = 0; i < count; i++) {
- set.add(readString());
+ set[i] = readString();
}
- return Collections.unmodifiableSet(set);
+ return Set.of(set);
}
}