ESQL: Compute infrastruture for LEFT JOIN (#118889)

This adds some infrastructure that we can use to run LOOKUP JOIN using
real LEFT JOIN semantics.

Right now if LOOKUP JOIN matches many rows in the `lookup` index we
merge all of the values into a multivalued field. So the number of rows
emitted from LOOKUP JOIN is the same as the number of rows that comes
into LOOKUP JOIN.

This change builds the infrastructure to emit one row per match, mostly
reusing the infrastructure from ENRICH.

Author: nik9000

x-pack/plugin/esql/compute/src/main/generated-src/org/elasticsearch/compute/data/BooleanArrayBlock.java

@@ -212,10 +212,46 @@ default boolean mvSortedAscending() {
     /**
      * Expand multivalued fields into one row per value. Returns the same block if there aren't any multivalued
      * fields to expand. The returned block needs to be closed by the caller to release the block's resources.
-     * TODO: pass BlockFactory
      */
     Block expand();
 
+    /**
+     * Build a {@link Block} with a {@code null} inserted {@code before} each
+     * listed position.
+     * <p>
+     *     Note: {@code before} must be non-decreasing.
+     * </p>
+     */
+    default Block insertNulls(IntVector before) {
+        // TODO remove default and scatter to implementation where it can be a lot more efficient
+        int myCount = getPositionCount();
+        int beforeCount = before.getPositionCount();
+        try (Builder builder = elementType().newBlockBuilder(myCount + beforeCount, blockFactory())) {
+            int beforeP = 0;
+            int nextNull = before.getInt(beforeP);
+            for (int mainP = 0; mainP < myCount; mainP++) {
+                while (mainP == nextNull) {
+                    builder.appendNull();
+                    beforeP++;
+                    if (beforeP >= beforeCount) {
+                        builder.copyFrom(this, mainP, myCount);
+                        return builder.build();
+                    }
+                    nextNull = before.getInt(beforeP);
+                }
+                // This line right below this is the super inefficient one.
+                builder.copyFrom(this, mainP, mainP + 1);
+            }
+            assert nextNull == myCount;
+            while (beforeP < beforeCount) {
+                nextNull = before.getInt(beforeP++);
+                assert nextNull == myCount;
+                builder.appendNull();
+            }
+            return builder.build();
+        }
+    }
+
     /**
      * Builds {@link Block}s. Typically, you use one of it's direct supinterfaces like {@link IntBlock.Builder}.
      * This is {@link Releasable} and should be released after building the block or if building the block fails.

x-pack/plugin/esql/compute/src/main/generated-src/org/elasticsearch/compute/data/BytesRefArrayBlock.java

@@ -149,7 +149,7 @@ $endif$
                 int valueCount = getValueCount(pos);
                 int first = getFirstValueIndex(pos);
                 if (valueCount == 1) {
-                    builder.append$Type$(get$Type$(getFirstValueIndex(pos)$if(BytesRef)$, scratch$endif$));
+                    builder.append$Type$(get$Type$(first$if(BytesRef)$, scratch$endif$));
                 } else {
                     builder.beginPositionEntry();
                     for (int c = 0; c < valueCount; c++) {

x-pack/plugin/esql/compute/src/main/generated-src/org/elasticsearch/compute/data/DoubleArrayBlock.java

@@ -20,28 +20,32 @@
 import java.util.Objects;
 
 /**
- * Combines values at the given blocks with the same positions into a single position for the blocks at the given channels
+ * Combines values at the given blocks with the same positions into a single position
+ * for the blocks at the given channels.
+ * <p>
  * Example, input pages consisting of three blocks:
- * positions    | field-1 | field-2 |
- * -----------------------------------
+ * </p>
+ * <pre>{@code
+ * | positions    | field-1 | field-2 |
+ * ------------------------------------
  * Page 1:
- * 1           |  a,b    |   2020  |
- * 1           |  c      |   2021  |
- * ---------------------------------
+ * | 1            |  a,b    |   2020  |
+ * | 1            |  c      |   2021  |
  * Page 2:
- * 2           |  a,e    |   2021  |
- * ---------------------------------
+ * | 2            |  a,e    |   2021  |
  * Page 3:
- * 4           |  d      |   null  |
- * ---------------------------------
+ * | 4            |  d      |   null  |
+ * }</pre>
  * Output:
+ * <pre>{@code
  * |  field-1   | field-2    |
  * ---------------------------
  * |  null      | null       |
  * |  a,b,c     | 2020,2021  |
  * |  a,e       | 2021       |
  * |  null      | null       |
  * |  d         | 2023       |
+ * }</pre>
  */
 public final class MergePositionsOperator implements Operator {
     private boolean finished = false;