Expose Merge expression in vortex-jni

Signed-off-by: Robert Kruszewski <github@robertk.io>

Author: robert3005

java/vortex-jni/src/main/java/dev/vortex/api/Expression.java

@@ -84,6 +84,25 @@ public static Expression pack(String[] fieldNames, Expression[] expressions, boo
         return new Expression(NativeExpression.pack(fieldNames, nativePointers(expressions), nullable));
     }
 
+    /**
+     * Merge struct expressions into a single struct, combining their fields in order.
+     *
+     * <p>Every input must evaluate to a non-nullable struct. When the same field name appears in more than one input,
+     * {@code duplicateHandling} decides the result. Fields are <em>not</em> merged recursively — a later field replaces
+     * an earlier one with the same name. Merging zero expressions yields an empty struct.
+     */
+    public static Expression merge(DuplicateHandling duplicateHandling, Expression... expressions) {
+        return new Expression(NativeExpression.merge(nativePointers(expressions), duplicateHandling.tag()));
+    }
+
+    /**
+     * Merge struct expressions, failing if any field name is duplicated. Equivalent to {@link #merge(DuplicateHandling,
+     * Expression...)} with {@link DuplicateHandling#ERROR}.
+     */
+    public static Expression merge(Expression... expressions) {
+        return merge(DuplicateHandling.ERROR, expressions);
+    }
+
     /** Logical AND. Requires at least one operand. */
     public static Expression and(Expression... operands) {
         Preconditions.checkArgument(operands.length > 0, "and requires at least one operand");
@@ -292,6 +311,27 @@ public byte code() {
         }
     }
 
+    /**
+     * Strategy for resolving duplicate field names in {@link #merge(DuplicateHandling, Expression...)}. Tag values must
+     * match the Rust {@code parse_duplicate_handling} table.
+     */
+    public enum DuplicateHandling {
+        /** When two structs share a field name, keep the value from the right-most (later) struct. */
+        RIGHT_MOST((byte) 0),
+        /** When two structs share a field name, fail with an error. */
+        ERROR((byte) 1);
+
+        private final byte tag;
+
+        DuplicateHandling(byte tag) {
+            this.tag = tag;
+        }
+
+        public byte tag() {
+            return tag;
+        }
+    }
+
     /** Time units for Date/Timestamp literals. Tag values must match the Rust {@code parse_time_unit} table. */
     public enum TimeUnit {
         NANOSECONDS((byte) 0),

java/vortex-jni/src/main/java/dev/vortex/jni/NativeExpression.java

@@ -21,6 +21,8 @@ private NativeExpression() {}
 
     public static native long pack(String[] fieldNames, long[] expressions, boolean nullable);
 
+    public static native long merge(long[] expressions, byte duplicateHandling);
+
     public static native long and(long[] operandPointers);
 
     public static native long or(long[] operandPointers);

java/vortex-jni/src/test/java/dev/vortex/api/ExpressionTest.java

@@ -44,4 +44,15 @@ public void packComposes() {
                 new Expression[] {Expression.column("a"), Expression.literal(5L), Expression.rowIdx()},
                 true));
     }
+
+    @Test
+    public void mergeComposes() {
+        // Default duplicate handling (ERROR).
+        assertNotNull(Expression.merge(Expression.column("a"), Expression.column("b")));
+        // Explicit duplicate handling.
+        assertNotNull(Expression.merge(
+                Expression.DuplicateHandling.RIGHT_MOST, Expression.column("a"), Expression.column("b")));
+        // Merging zero expressions is valid and yields an empty struct.
+        assertNotNull(Expression.merge());
+    }
 }

vortex-jni/src/expression.rs

@@ -39,6 +39,7 @@ use vortex::expr::get_item;
 use vortex::expr::is_not_null;
 use vortex::expr::is_null;
 use vortex::expr::lit;
+use vortex::expr::merge_opts;
 use vortex::expr::not;
 use vortex::expr::or_collect;
 use vortex::expr::pack;
@@ -59,6 +60,7 @@ use vortex::scalar_fn::fns::between::StrictComparison;
 use vortex::scalar_fn::fns::binary::Binary;
 use vortex::scalar_fn::fns::like::Like;
 use vortex::scalar_fn::fns::like::LikeOptions;
+use vortex::scalar_fn::fns::merge::DuplicateHandling;
 use vortex::scalar_fn::fns::operators::Operator;
 
 use crate::errors::JNIError;
@@ -97,6 +99,17 @@ fn parse_time_unit(tag: jbyte) -> Result<TimeUnit, JNIError> {
     TimeUnit::try_from(tag as u8).map_err(JNIError::from)
 }
 
+/// Parse a merge [`DuplicateHandling`] strategy from its wire-encoded byte tag.
+///
+/// See `dev.vortex.api.Expression.DuplicateHandling` on the Java side for the source of truth.
+fn parse_duplicate_handling(tag: jbyte) -> Result<DuplicateHandling, JNIError> {
+    Ok(match tag {
+        0 => DuplicateHandling::RightMost,
+        1 => DuplicateHandling::Error,
+        other => throw_runtime!("unknown duplicate handling code: {other}"),
+    })
+}
+
 #[unsafe(no_mangle)]
 pub extern "system" fn Java_dev_vortex_jni_NativeExpression_free(
     _env: EnvUnowned,
@@ -192,6 +205,24 @@ pub extern "system" fn Java_dev_vortex_jni_NativeExpression_pack(
     })
 }
 
+/// Merge zero or more struct-returning expressions into a single struct.
+///
+/// `duplicate_handling` selects how shared field names are resolved (see
+/// [`parse_duplicate_handling`]). An empty `expressions` array yields an empty struct.
+#[unsafe(no_mangle)]
+pub extern "system" fn Java_dev_vortex_jni_NativeExpression_merge(
+    mut env: EnvUnowned,
+    _class: JClass,
+    expressions: JLongArray,
+    duplicate_handling: jbyte,
+) -> jlong {
+    try_or_throw(&mut env, |env| {
+        let exprs = collect_operands(env, &expressions)?;
+        let handling = parse_duplicate_handling(duplicate_handling)?;
+        Ok(into_raw(merge_opts(exprs, handling)))
+    })
+}
+
 #[unsafe(no_mangle)]
 pub extern "system" fn Java_dev_vortex_jni_NativeExpression_and(
     mut env: EnvUnowned,