Use a threadlocal to hold the NodeIterator

Ensures that an Evaluator query can be multi-threaded

Fixes #2088

Author: jhy

CHANGES.md

@@ -27,6 +27,9 @@
 * When generating XML-syntax output from parsed HTML, script nodes containing (pseudo) CData sections would have an
   extraneous CData section added, causing script execution errors. Now, the data content is emitted in a HTML/XML/XHTML
   polyglot format, if the data is not already within a CData section. [2078](https://github.com/jhy/jsoup/issues/2078)
+* The `:has` evaluator held a non-thread-safe Iterator, and so if an Evaluator object was shared across multiple
+  concurrent threads, a NoSuchElement exception may be thrown, and the selected results may be incorrect. Now, the
+  iterator object is a thread-local. [2088](https://github.com/jhy/jsoup/issues/2088)
 
 ---
 Older changes for versions 0.1.1 (2010-Jan-31) through 1.17.1 (2023-Nov-27) may be found in

src/main/java/org/jsoup/select/StructuralEvaluator.java

@@ -58,16 +58,20 @@ public boolean matches(Element root, Element element) {
         }
     }
 
-    static class Has extends StructuralEvaluator {
-        final NodeIterator<Element> it = new NodeIterator<>(new Element("html"), Element.class);
+    final ThreadLocal<NodeIterator<Element>> threadHasIter =
+        ThreadLocal.withInitial(() -> new NodeIterator<>(new Element("html"), Element.class));
         // the element here is just a placeholder so this can be final - gets set in restart()
 
+    static class Has extends StructuralEvaluator {
+
         public Has(Evaluator evaluator) {
             super(evaluator);
         }
 
         @Override public boolean matches(Element root, Element element) {
             // for :has, we only want to match children (or below), not the input element. And we want to minimize GCs
+            NodeIterator<Element> it = threadHasIter.get();
+
             it.restart(element);
             while (it.hasNext()) {
                 Element el = it.next();

src/test/java/org/jsoup/select/SelectorIT.java

@@ -0,0 +1,58 @@
+package org.jsoup.select;
+
+import org.jsoup.Jsoup;
+import org.jsoup.nodes.Document;
+import org.junit.jupiter.api.Test;
+
+import java.util.concurrent.atomic.AtomicInteger;
+
+import static org.junit.jupiter.api.Assertions.assertEquals;
+
+public class SelectorIT {
+
+    @Test
+    public void multiThreadHas() throws InterruptedException {
+        final String html = "<div id=1></div><div id=2><p>One</p><p>Two</p>";
+        final Evaluator eval = QueryParser.parse("div:has(p)");
+
+        int numThreads = 20;
+        int numThreadLoops = 5;
+
+        SelectorIT.ThreadCatcher catcher = new SelectorIT.ThreadCatcher();
+
+        Thread[] threads = new Thread[numThreads];
+        for (int threadNum = 0; threadNum < numThreads; threadNum++) {
+            Thread thread = new Thread(() -> {
+                Document doc = Jsoup.parse(html);
+                for (int loop = 0; loop < numThreadLoops; loop++) {
+                    Elements els = doc.select(eval);
+                    assertEquals(1, els.size());
+                    assertEquals("2", els.get(0).id());
+                }
+            });
+            thread.setName("Runner-" + threadNum);
+            thread.start();
+            thread.setUncaughtExceptionHandler(catcher);
+            threads[threadNum] = thread;
+        }
+
+        // now join them all
+        for (Thread thread : threads) {
+            thread.join();
+        }
+
+        assertEquals(0, catcher.exceptionCount.get());
+    }
+
+    static class ThreadCatcher implements Thread.UncaughtExceptionHandler {
+        AtomicInteger exceptionCount = new AtomicInteger();
+
+        @Override
+        public void uncaughtException(Thread t, Throwable e) {
+
+            e.printStackTrace();
+            exceptionCount.incrementAndGet();
+        }
+    }
+
+}

src/test/java/org/jsoup/select/SelectorTest.java

@@ -2,7 +2,6 @@
 
 import org.jsoup.Jsoup;
 import org.jsoup.MultiLocaleExtension.MultiLocaleTest;
-import org.jsoup.TextUtil;
 import org.jsoup.nodes.Document;
 import org.jsoup.nodes.Element;
 import org.jsoup.parser.Parser;