[MOD] XQuery: create compact sequence if size is known

Author: ChristianGruen

basex-core/src/main/java/org/basex/query/value/ValueBuilder.java

@@ -2,6 +2,7 @@
 
 import org.basex.query.*;
 import org.basex.query.expr.*;
+import org.basex.query.util.list.*;
 import org.basex.query.value.item.*;
 import org.basex.query.value.seq.*;
 import org.basex.query.value.seq.tree.*;
@@ -52,24 +53,23 @@ public ValueBuilder(final QueryContext qc, final long capacity) {
   public ValueBuilder add(final Value value) {
     if(!value.isEmpty()) {
       qc.checkStop();
-      final Value sngl = single;
-      if(sngl != null) {
-        single = null;
-        if(capacity != Integer.MIN_VALUE) {
-          sequence = isStr(sngl) && isStr(value) ? new StrSeqBuilder() :
-                     isAtm(sngl) && isAtm(value) ? new AtmSeqBuilder() :
-                     isInt(sngl) && isInt(value) ? new IntSeqBuilder() :
-                     isDbl(sngl) && isDbl(value) ? new DblSeqBuilder() :
-                     isBln(sngl) && isBln(value) ? new BlnSeqBuilder() : null;
+      if(sequence == null) {
+        final Value sngl = single;
+        if(sngl == null) {
+          single = value;
+          return this;
         }
-        if(sequence == null) sequence = new TreeSeqBuilder();
+        single = null;
+        sequence = capacity == Integer.MIN_VALUE ? new TreeSeqBuilder() :
+                   isStr(sngl) && isStr(value) ? new StrSeqBuilder() :
+                   isAtm(sngl) && isAtm(value) ? new AtmSeqBuilder() :
+                   isInt(sngl) && isInt(value) ? new IntSeqBuilder() :
+                   isDbl(sngl) && isDbl(value) ? new DblSeqBuilder() :
+                   isBln(sngl) && isBln(value) ? new BlnSeqBuilder() :
+                   new ItemSeqBuilder();
         add(sngl);
       }
-      if(sequence != null) {
-        sequence = sequence.add(value, qc);
-      } else {
-        single = value;
-      }
+      sequence = sequence.add(value, qc);
     }
     return this;
   }
@@ -260,4 +260,21 @@ public Value value(final Type type) {
       return BlnSeq.get(values.finish());
     }
   }
+
+  /** Item sequence builder. */
+  final class ItemSeqBuilder implements SeqBuilder {
+    /** Items. */
+    private final ItemList items = new ItemList(capacity);
+
+    @Override
+    public SeqBuilder add(final Item item) {
+      items.add(item);
+      return this;
+    }
+
+    @Override
+    public Value value(final Type type) {
+      return items.value(type);
+    }
+  }
 }

basex-core/src/main/java/org/basex/query/value/array/ArrayBuilder.java

@@ -53,18 +53,18 @@ public ArrayBuilder(final QueryContext qc, final long capacity) {
    * @return reference to this builder for convenience
    */
   public ArrayBuilder add(final Value value) {
-    final Value sngl = single;
-    if(sngl != null) {
+    if(array == null) {
+      final Value sngl = single;
+      if(sngl == null) {
+        single = value;
+        return this;
+      }
       single = null;
-      array = capacity != Integer.MIN_VALUE && sngl.size() == 1 && value.size() == 1 ?
-        new ItemArrayBuilder(qc, capacity) : new TreeArrayBuilder();
+      array = capacity == Integer.MIN_VALUE || sngl.size() != 1 || value.size() != 1 ?
+        new TreeArrayBuilder() : new ItemArrayBuilder(qc, capacity);
       add(sngl);
     }
-    if(array != null) {
-      array = array.add(value);
-    } else {
-      single = value;
-    }
+    array = array.add(value);
     return this;
   }
 

basex-core/src/test/java/org/basex/query/ast/RewritingsTest.java

@@ -1016,7 +1016,7 @@ public final class RewritingsTest extends SandboxTest {
     check("(1, 2) coerce to xs:double+", "1\n2",
         empty(TypeCheck.class), root(DblSeq.class), count(Dbl.class, 2));
     check("(-128, 127) coerce to xs:byte+", "-128\n127",
-        empty(TypeCheck.class), root(SmallSeq.class), count(Itr.class, 2));
+        empty(TypeCheck.class), root(ItemSeq.class), count(Itr.class, 2));
 
     error("<a/> coerce to empty-sequence()", INVTYPE_X);
     error("(1, 128) coerce to xs:byte+", FUNCCAST_X_X_X);
@@ -2121,7 +2121,7 @@ private static String gh1852(final String query) {
 
     check("for $a in (1, 2) group by $a return $a", "1\n2", root(RangeSeq.class));
     check("for $a in (1, 3) group by $a return $a", "1\n3", root(IntSeq.class));
-    check("for $a in (1, 'a', 1) group by $a return $a", "1\na", root(SmallSeq.class));
+    check("for $a in (1, 'a', 1) group by $a return $a", "1\na", root(ItemSeq.class));
 
     check("for $p in (1 to 2)[. >= 0] group by $q := string($p) return $q",
         "1\n2", root(DISTINCT_VALUES), exists(DualMap.class));

basex-core/src/test/java/org/basex/query/expr/MixedTest.java

@@ -13,7 +13,6 @@
 import org.basex.query.value.item.*;
 import org.basex.query.value.map.*;
 import org.basex.query.value.seq.*;
-import org.basex.query.value.seq.tree.*;
 import org.junit.jupiter.api.*;
 import org.junit.jupiter.api.Test;
 
@@ -471,10 +470,10 @@ public final class MixedTest extends SandboxTest {
   /** Tests the data structure before and after shrinking. */
   @Test public void shrink() {
     test("('x', 1 to 3) => remove(1)",
-        new TypeInfo(SmallSeq.class, "xs:anyAtomicType+", 3),
+        new TypeInfo(SubSeq.class, "xs:anyAtomicType+", 3),
         new TypeInfo(RangeSeq.class, "xs:integer+", 3));
     test("('x', 1 to 100) => remove(1)",
-        new TypeInfo(BigSeq.class, "xs:anyAtomicType+", 100),
+        new TypeInfo(SubSeq.class, "xs:anyAtomicType+", 100),
         new TypeInfo(RangeSeq.class, "xs:integer+", 100));
 
     test("map:build(1 to 3) => map:remove(1)",

basex-core/src/test/java/org/basex/query/expr/SimpleMapTest.java

@@ -7,7 +7,6 @@
 import org.basex.query.expr.constr.*;
 import org.basex.query.value.item.*;
 import org.basex.query.value.seq.*;
-import org.basex.query.value.seq.tree.*;
 import org.basex.query.var.*;
 import org.junit.jupiter.api.*;
 
@@ -82,7 +81,7 @@ public final class SimpleMapTest extends SandboxTest {
 
   /** Inline simple expressions into next operand. */
   @Test public void inline() {
-    check("'1' ! (., number())", "1\n1", root(SmallSeq.class));
+    check("'1' ! (., number())", "1\n1", root(ItemSeq.class));
     check("let $a := document { <a/> } return $a ! (., /)", "<a/>\n<a/>", empty(VarRef.class));
     check("let $d := document{} return $d ! /", "", root(CDoc.class));
     check("{ 1: 2 } ! ?*", 2, root(Itr.class));