Attribute provides means to cache hashcode and sizeInBytes computation (#2986)

apmoriarty · web-flow · commit 4509a1d6f736 · 2025-06-25T10:49:02.000Z
Most implementations of Attribute now cache the computation for hashcode
and computation.

Fix bug in JexlASTHelper's deconstructIdentifier method that caused a
new String to be built, even when not necessary.
diff --git a/warehouse/query-core/src/main/java/datawave/query/attributes/Attribute.java b/warehouse/query-core/src/main/java/datawave/query/attributes/Attribute.java
@@ -37,6 +37,10 @@ public abstract class Attribute<T extends Comparable<T>> implements WritableComp
     protected boolean toKeep = true; // a flag denoting whether this attribute is to be kept in the returned results (transient or not)
     protected boolean fromIndex = true; // Assume attributes are from the index unless specified otherwise.
 
+    // cache computation to avoid repeated calculation
+    protected int hashcode = Integer.MIN_VALUE;
+    protected long sizeInBytes = Long.MIN_VALUE;
+
     public Attribute() {}
 
     public Attribute(Key metadata, boolean toKeep) {
@@ -244,6 +248,7 @@ public boolean equals(Object o) {
 
     @Override
     public int hashCode() {
+        // Note: implementations of Attribute should cache the result of this operation
         HashCodeBuilder hcb = new HashCodeBuilder(145, 11);
         hcb.append(this.isMetadataSet());
         if (isMetadataSet()) {
@@ -285,15 +290,17 @@ public long sizeInBytes() {
         // 8 for the object overhead
         // 4 for the key reference
         // 1 for the keep boolean
-        // all rounded up to the nearest multiple of 8 to make 16 out of 13 bytes
+        // 4 for the hashcode reference
+        // 8 for the sizeInBytes references
+        // all rounded up to the nearest multiple of 8 to make 32 out of 25 bytes
         size += getMetadataSizeInBytes();
         return size;
     }
 
     // for use by subclasses to estimate size
     protected long sizeInBytes(long extra) {
-        return roundUp(extra + 13) + getMetadataSizeInBytes();
-        // 13 is the base size in bytes (see sizeInBytes(), unrounded and without metadata)
+        return roundUp(extra + 25) + getMetadataSizeInBytes();
+        // 25 is the base size in bytes (see sizeInBytes(), unrounded and without metadata)
     }
 
     // a helper method to return the size of a string
diff --git a/warehouse/query-core/src/main/java/datawave/query/attributes/Attributes.java b/warehouse/query-core/src/main/java/datawave/query/attributes/Attributes.java
@@ -90,8 +90,8 @@ public long sizeInBytes() {
     }
 
     public void add(Attribute<? extends Comparable<?>> attr) {
-        if (!this.attributes.contains(attr)) {
-            this.attributes.add(attr);
+        boolean updated = this.attributes.add(attr);
+        if (updated) {
             this._count += attr.size();
             if (trackSizes) {
                 this._bytes += attr.sizeInBytes() + 24 + 24;
diff --git a/warehouse/query-core/src/main/java/datawave/query/attributes/Content.java b/warehouse/query-core/src/main/java/datawave/query/attributes/Content.java
@@ -43,8 +43,11 @@ public Content(String content, Key docKey, boolean toKeep, Attribute<?> source)
 
     @Override
     public long sizeInBytes() {
-        return sizeInBytes(content) + super.sizeInBytes(4);
-        // 4 for string reference
+        if (sizeInBytes == Long.MIN_VALUE) {
+            // 4 for string reference
+            sizeInBytes = sizeInBytes(content) + super.sizeInBytes(4) + 4;
+        }
+        return sizeInBytes;
     }
 
     public String getContent() {
@@ -119,10 +122,15 @@ public boolean equals(Object o) {
 
     @Override
     public int hashCode() {
-        HashCodeBuilder hcb = new HashCodeBuilder(2099, 2129);
-        hcb.append(content).append(super.hashCode());
-
-        return hcb.toHashCode();
+        if (hashcode == Integer.MIN_VALUE) {
+            //  @formatter:off
+            hashcode = new HashCodeBuilder(2099, 2129)
+                    .append(content)
+                    .append(super.hashCode())
+                    .toHashCode();
+            //  @formatter:off
+        }
+        return hashcode;
     }
 
     @Override
diff --git a/warehouse/query-core/src/main/java/datawave/query/attributes/Geometry.java b/warehouse/query-core/src/main/java/datawave/query/attributes/Geometry.java
@@ -47,8 +47,11 @@ public Geometry(org.locationtech.jts.geom.Geometry geometry, Key docKey, boolean
 
     @Override
     public long sizeInBytes() {
-        return ObjectSizeOf.Sizer.getObjectSize(geometry) + super.sizeInBytes(4);
-        // 4 for geometry reference
+        if (sizeInBytes == Long.MIN_VALUE) {
+            // 4 for geometry reference
+            sizeInBytes = ObjectSizeOf.Sizer.getObjectSize(geometry) + super.sizeInBytes(4);
+        }
+        return sizeInBytes;
     }
 
     private byte[] write() {
@@ -128,10 +131,15 @@ public boolean equals(Object o) {
 
     @Override
     public int hashCode() {
-        HashCodeBuilder hcb = new HashCodeBuilder(163, 157);
-        hcb.append(super.hashCode()).append(geometry);
-
-        return hcb.toHashCode();
+        if (hashcode == Integer.MIN_VALUE) {
+            //  @formatter:off
+            hashcode = new HashCodeBuilder(163, 157)
+                    .append(super.hashCode())
+                    .append(geometry)
+                    .toHashCode();
+            //  @formatter:on
+        }
+        return hashcode;
     }
 
     @Override
diff --git a/warehouse/query-core/src/main/java/datawave/query/attributes/IpAddress.java b/warehouse/query-core/src/main/java/datawave/query/attributes/IpAddress.java
@@ -36,8 +36,11 @@ public IpAddress(String ipAddress, Key docKey, boolean toKeep) {
 
     @Override
     public long sizeInBytes() {
-        return sizeInBytes(value.toString()) + sizeInBytes(normalizedValue) + super.sizeInBytes(8);
-        // 8 for string references
+        if (sizeInBytes == Long.MIN_VALUE) {
+            // 8 for string references
+            sizeInBytes = sizeInBytes(value.toString()) + sizeInBytes(normalizedValue) + super.sizeInBytes(8);
+        }
+        return sizeInBytes;
     }
 
     protected void validate() {
@@ -103,10 +106,15 @@ public boolean equals(Object o) {
 
     @Override
     public int hashCode() {
-        HashCodeBuilder hcb = new HashCodeBuilder(163, 157);
-        hcb.append(super.hashCode()).append(this.value);
-
-        return hcb.toHashCode();
+        if (hashcode == Integer.MIN_VALUE) {
+            //  @formatter:off
+            hashcode = new HashCodeBuilder(163, 157)
+                    .append(super.hashCode())
+                    .append(this.value)
+                    .toHashCode();
+            //  @formatter:on
+        }
+        return hashcode;
     }
 
     @Override
diff --git a/warehouse/query-core/src/main/java/datawave/query/attributes/Numeric.java b/warehouse/query-core/src/main/java/datawave/query/attributes/Numeric.java
@@ -49,9 +49,12 @@ public Numeric(Number value, Key docKey, boolean toKeep) {
 
     @Override
     public long sizeInBytes() {
-        return 20 + sizeInBytes(normalizedValue) + super.sizeInBytes(8);
-        // 20 is for a basic int Number
-        // 8 for string references
+        if (sizeInBytes == Long.MIN_VALUE) {
+            // 20 is for a basic int Number
+            // 8 for string references
+            sizeInBytes = 20 + sizeInBytes(normalizedValue) + super.sizeInBytes(8);
+        }
+        return sizeInBytes;
     }
 
     /**
@@ -153,10 +156,15 @@ public boolean equals(Object o) {
 
     @Override
     public int hashCode() {
-        HashCodeBuilder hcb = new HashCodeBuilder(113, 127);
-        hcb.append(super.hashCode()).append(value);
-
-        return hcb.toHashCode();
+        if (hashcode == Integer.MIN_VALUE) {
+            //  @formatter:off
+            hashcode = new HashCodeBuilder(113, 127)
+                    .append(super.hashCode())
+                    .append(value)
+                    .toHashCode();
+            //  @formatter:on
+        }
+        return hashcode;
     }
 
     @Override
diff --git a/warehouse/query-core/src/main/java/datawave/query/attributes/PreNormalizedAttribute.java b/warehouse/query-core/src/main/java/datawave/query/attributes/PreNormalizedAttribute.java
@@ -40,8 +40,11 @@ public PreNormalizedAttribute(String value, Key docKey, boolean toKeep) {
 
     @Override
     public long sizeInBytes() {
-        return sizeInBytes(value) + super.sizeInBytes(4);
-        // 4 for string reference
+        if (sizeInBytes == Long.MIN_VALUE) {
+            // 4 bytes for string reference
+            sizeInBytes = sizeInBytes(value) + super.sizeInBytes(4);
+        }
+        return sizeInBytes;
     }
 
     @Override
@@ -59,10 +62,15 @@ public boolean equals(Object o) {
 
     @Override
     public int hashCode() {
-        HashCodeBuilder hcb = new HashCodeBuilder(2141, 2137);
-        hcb.append(super.hashCode()).append(this.getData());
-
-        return hcb.toHashCode();
+        if (hashcode == Integer.MIN_VALUE) {
+            //  @formatter:off
+            hashcode = new HashCodeBuilder(2141, 2137)
+                    .append(super.hashCode())
+                    .append(this.getData())
+                    .toHashCode();
+            //  @formatter:on
+        }
+        return hashcode;
     }
 
     @Override
diff --git a/warehouse/query-core/src/main/java/datawave/query/attributes/TypeAttribute.java b/warehouse/query-core/src/main/java/datawave/query/attributes/TypeAttribute.java
@@ -51,12 +51,11 @@ public TypeAttribute(Type<T> datawaveType, Key docKey, boolean toKeep) {
 
     @Override
     public long sizeInBytes() {
-        long size = ObjectSizeOf.Sizer.getObjectSize(datawaveType) + super.sizeInBytes(4) + 2L;
-        if (delegateString != null) {
-            size += (2L + delegateString.length());
+        if (sizeInBytes == Long.MAX_VALUE) {
+            // 4 for datawaveType reference
+            sizeInBytes = ObjectSizeOf.Sizer.getObjectSize(datawaveType) + super.sizeInBytes(4);
         }
-        return size;
-        // 4 for datawaveType reference
+        return sizeInBytes;
     }
 
     public Type<T> getType() {
@@ -135,15 +134,15 @@ public boolean equals(Object o) {
 
     @Override
     public int hashCode() {
-        if (this.hashCode == Integer.MIN_VALUE) {
+        if (hashcode == Integer.MIN_VALUE) {
             //  @formatter:off
-            this.hashCode = new HashCodeBuilder(2099, 2129)
+            hashcode = new HashCodeBuilder(2099, 2129)
                     .append(datawaveType.getDelegateAsString())
                     .append(super.hashCode())
                     .toHashCode();
             //  @formatter:on
         }
-        return hashCode;
+        return hashcode;
     }
 
     @Override
diff --git a/warehouse/query-core/src/main/java/datawave/query/common/grouping/GroupingAttribute.java b/warehouse/query-core/src/main/java/datawave/query/common/grouping/GroupingAttribute.java
@@ -58,6 +58,21 @@ private boolean isMetadataRowEqual(Attribute<?> other) {
      */
     @Override
     public int hashCode() {
-        return new HashCodeBuilder(2099, 2129).append(getType().getDelegateAsString()).toHashCode();
+        if (hashcode == Integer.MIN_VALUE) {
+            //  @formatter:off
+            hashcode = new HashCodeBuilder(2099, 2129)
+                    .append(getType().getDelegateAsString())
+                    .toHashCode();
+            //  @formatter:on
+        }
+        return hashcode;
+    }
+
+    @Override
+    public long sizeInBytes() {
+        if (sizeInBytes == Long.MIN_VALUE) {
+            sizeInBytes = super.sizeInBytes();
+        }
+        return sizeInBytes;
     }
 }
diff --git a/warehouse/query-core/src/main/java/datawave/query/jexl/JexlASTHelper.java b/warehouse/query-core/src/main/java/datawave/query/jexl/JexlASTHelper.java
@@ -699,7 +699,7 @@ public static String deconstructIdentifier(String fieldName, boolean includeGrou
                 }
             }
 
-            if (startIndex != 0 || stopLength != -1) {
+            if (startIndex != 0 || stopLength != fieldName.length()) {
                 fieldName = new String(fieldName.getBytes(), startIndex, stopLength);
             }
         }
diff --git a/warehouse/query-core/src/test/java/datawave/query/attributes/AttributeIT.java b/warehouse/query-core/src/test/java/datawave/query/attributes/AttributeIT.java
diff --git a/warehouse/query-core/src/test/java/datawave/query/attributes/DocumentTest.java b/warehouse/query-core/src/test/java/datawave/query/attributes/DocumentTest.java
diff --git a/warehouse/query-core/src/test/java/datawave/query/jexl/JexlASTHelperTest.java b/warehouse/query-core/src/test/java/datawave/query/jexl/JexlASTHelperTest.java

Original file line number	Diff line number	Diff line change
`@@ -699,7 +699,7 @@ public static String deconstructIdentifier(String fieldName, boolean includeGrou`
`699`	`699`	`}`
`700`	`700`	`}`
`701`	`701`
`702`		`- if (startIndex != 0 \|\| stopLength != -1) {`
	`702`	`+ if (startIndex != 0 \|\| stopLength != fieldName.length()) {`
`703`	`703`	`fieldName = new String(fieldName.getBytes(), startIndex, stopLength);`
`704`	`704`	`}`
`705`	`705`	`}`