[MOD] XQuery, nondeterminism: existence checks, counts

Author: ChristianGruen

basex-core/src/main/java/org/basex/query/func/fn/FnCount.java

@@ -19,12 +19,14 @@
  * @author Christian Gruen
  */
 public final class FnCount extends StandardFunc {
+  /** Nondeterministic input. */
+  private boolean ndt;
+
   @Override
   public Itr item(final QueryContext qc, final InputInfo ii) throws QueryException {
     // if the iterator size is unknown or nondeterministic, iterate through all results
-    final Expr expr = arg(0);
-    final Iter input = expr.iter(qc);
-    long size = expr.has(Flag.NDT) ? -1 : input.size();
+    final Iter input = arg(0).iter(qc);
+    long size = ndt ? -1 : input.size();
     if(size == -1) {
       do ++size; while(qc.next(input) != null);
     }
@@ -38,10 +40,12 @@ protected void simplifyArgs(final CompileContext cc) throws QueryException {
 
   @Override
   protected Expr opt(final CompileContext cc) throws QueryException {
-    // return static result size
     final Expr input = arg(0);
+    ndt = input.has(Flag.NDT);
+
+    // return static result size
     final long size = input.size();
-    if(size >= 0 && !input.has(Flag.NDT)) return Itr.get(size);
+    if(size >= 0 && !ndt) return Itr.get(size);
 
     // count(map:keys(E)) → map:size(E)
     if(_MAP_KEYS.is(input))

basex-core/src/main/java/org/basex/query/func/fn/FnEmpty.java

@@ -54,6 +54,10 @@ protected final Expr opt(final CompileContext cc) throws QueryException {
     if(REPLICATE.is(input) && input.arg(1) instanceof Itr) {
       input = input.arg(0);
     }
+    // exists(head(E)) → exists(E), empty(head(E)) → empty(E)
+    if(HEAD.is(input)) {
+      input = input.arg(0);
+    }
     // rewrite list to union expression:  exists((nodes1, nodes2)) → exists(nodes1 | nodes2)
     if(input instanceof List && input.seqType().type instanceof NodeType) {
       input = new Union(info, input.args()).optimize(cc);
@@ -84,6 +88,13 @@ protected final Expr opt(final CompileContext cc) throws QueryException {
       return new CmpG(info, input, Itr.ZERO, exists ? CmpOp.NE : CmpOp.EQ).optimize(cc);
     }
 
+    // exists(tail(E)) → util:count-within(E, 2), empty(tail(E)) → util:count-within(E, 0, 1)
+    if(TAIL.is(input) && !input.has(Flag.NDT)) {
+      final Expr seq = input.arg(0);
+      return exists ? cc.function(_UTIL_COUNT_WITHIN, info, seq, Itr.get(2)) :
+        cc.function(_UTIL_COUNT_WITHIN, info, seq, Itr.ZERO, Itr.ONE);
+    }
+
     return embed(cc, true);
   }
 

basex-core/src/main/java/org/basex/query/func/util/UtilCountWithin.java

@@ -20,6 +20,9 @@
  * @author Christian Gruen
  */
 public final class UtilCountWithin extends StandardFunc {
+  /** Nondeterministic input. */
+  private boolean ndt;
+
   @Override
   public Bln item(final QueryContext qc, final InputInfo ii) throws QueryException {
     return Bln.get(test(qc, info, 0));
@@ -32,12 +35,10 @@ public boolean test(final QueryContext qc, final InputInfo ii, final long pos)
     final long min = minMax[0], max = minMax[1];
 
     // iterative through the results if an iterator is nondeterministic or its size is unknown
-    final Expr expr = arg(0);
-    final Iter input = expr.iter(qc);
-    final boolean enforce = expr.has(Flag.NDT);
-    long size = enforce ? -1 : input.size();
+    final Iter input = arg(0).iter(qc);
+    long size = ndt ? -1 : input.size();
     if(size == -1) {
-      if(enforce) {
+      if(ndt) {
         do ++size; while(qc.next(input) != null);
       } else if(max == Long.MAX_VALUE) {
         // >= min: skip if minimum is reached
@@ -58,10 +59,11 @@ protected void simplifyArgs(final CompileContext cc) throws QueryException {
   @Override
   protected Expr opt(final CompileContext cc) throws QueryException {
     final Expr input = arg(0);
+    ndt = input.has(Flag.NDT);
 
     // skip rewrites for nondeterministic input, which must be fully evaluated for its side effects
     final long[] minMax = minMaxValues(cc.qc);
-    if(minMax != null && !input.has(Flag.NDT)) {
+    if(minMax != null && !ndt) {
       final long min = minMax[0], max = minMax[1];
       if(min > max) return Bln.FALSE;
       if(min <= 0 && max == Long.MAX_VALUE) return Bln.TRUE;

basex-core/src/test/java/org/basex/query/func/FnModuleTest.java

@@ -1664,6 +1664,25 @@ public final class FnModuleTest extends SandboxTest {
     error(func.args("", ""), HASH_ALGORITHM_X);
   }
 
+  /** Tests the fn:empty and fn:exists rewritings of fn:head and fn:tail arguments. */
+  @Test public void emptyExistsHeadTail() {
+    final String seq = " tokenize(" + wrap("a b c") + ", ' ')";
+    check(EXISTS.args(HEAD.args(seq)), true, root(EXISTS), empty(HEAD));
+    check(EMPTY.args(HEAD.args(seq)), false, root(EMPTY), empty(HEAD));
+    check(EXISTS.args(TAIL.args(seq)), true, root(_UTIL_COUNT_WITHIN), empty(TAIL));
+    check(EMPTY.args(TAIL.args(seq)), false, root(_UTIL_COUNT_WITHIN), empty(TAIL));
+    check(EXISTS.args(SUBSEQUENCE.args(seq, 2)), true, root(_UTIL_COUNT_WITHIN), empty(TAIL));
+
+    final String one = " tokenize(" + wrap("a") + ", ' ')";
+    check(EXISTS.args(TAIL.args(one)), false, root(_UTIL_COUNT_WITHIN));
+    check(EMPTY.args(TAIL.args(one)), true, root(_UTIL_COUNT_WITHIN));
+
+    // head is rewritten even for nondeterministic input; tail is kept
+    final String ndt = " (1 to 3) ! (if(. = 3) then error() else .)";
+    check(EXISTS.args(HEAD.args(ndt)), true, empty(HEAD));
+    check(EXISTS.args(TAIL.args(ndt)), true, exists(TAIL));
+  }
+
   /** Test method. */
   @Test public void head() {
     final Function func = HEAD;