Optimized the ~ (any preceding sibling) selector

Changed the order of evaluation so previous matches return earlier - for positive hits this significantly reduces the number of sub-queries. Also memoize the results -- speeds negative matches. Fixes #1956
jhy · May 29, 2023 · 10ef981 · 10ef981
1 parent e4f4b86
commit 10ef981
Show file tree

Hide file tree

Showing 4 changed files with 65 additions and 46 deletions.
diff --git a/CHANGES b/CHANGES
@@ -6,6 +6,9 @@ Release 1.16.2 [PENDING]
     `W3CDom#namespaceAware(false)`.
     <https://github.com/jhy/jsoup/pull/1848>
 
+  * Improvement: speed optimized the `~` (any preceeding sibling) and `:nth-of-type` selectors.
+    <https://github.com/jhy/jsoup/issues/1956>
+
   * Bugfix: `form` elements and empty elements (such as `img`) did not have their attributes de-duplicated.
     <https://github.com/jhy/jsoup/pull/1950>
 

diff --git a/src/main/java/org/jsoup/nodes/Comment.java b/src/main/java/org/jsoup/nodes/Comment.java
@@ -88,7 +88,7 @@ private static boolean isXmlDeclarationData(String data) {
         String fragment = "<" + declContent + ">";
         // use the HTML parser not XML, so we don't get into a recursive XML Declaration on contrived data
         Document doc = Parser.htmlParser().settings(ParseSettings.preserveCase).parseInput(fragment, baseUri());
-        if (doc.body().children().size() > 0) {
+        if (doc.body().childrenSize() > 0) {
             Element el = doc.body().child(0);
             decl = new XmlDeclaration(NodeUtils.parser(doc).settings().normalizeTag(el.tagName()), data.startsWith("!"));
             decl.attributes().addAll(el.attributes());

diff --git a/src/main/java/org/jsoup/select/Evaluator.java b/src/main/java/org/jsoup/select/Evaluator.java
@@ -407,7 +407,7 @@ public static final class IsLastChild extends Evaluator {
 		@Override
 		public boolean matches(Element root, Element element) {
 			final Element p = element.parent();
-			return p != null && !(p instanceof Document) && element.elementSiblingIndex() == p.children().size()-1;
+			return p != null && !(p instanceof Document) && element.elementSiblingIndex() == p.childrenSize()-1;
 		}
 
 		@Override
@@ -508,7 +508,7 @@ public IsNthLastChild(int a, int b) {
         protected int calculatePosition(Element root, Element element) {
     	    if (element.parent() == null)
     	        return 0;
-        	return element.parent().children().size() - element.elementSiblingIndex();
+        	return element.parent().childrenSize()- element.elementSiblingIndex();
         }
 
 		@Override
@@ -522,50 +522,55 @@ protected String getPseudoClass() {
      *
      */
     public static class IsNthOfType extends CssNthEvaluator {
-    	public IsNthOfType(int a, int b) {
-    		super(a,b);
-    	}
+        public IsNthOfType(int a, int b) {
+            super(a, b);
+        }
 
-		protected int calculatePosition(Element root, Element element) {
-			int pos = 0;
-            if (element.parent() == null)
+        protected int calculatePosition(Element root, Element element) {
+            Element parent = element.parent();
+            if (parent == null)
                 return 0;
-        	Elements family = element.parent().children();
-            for (Element el : family) {
-                if (el.tag().equals(element.tag())) pos++;
-                if (el == element) break;
+
+            int pos = 0;
+            final int size = parent.childNodeSize();
+            for (int i = 0; i < size; i++) {
+                Node node = parent.childNode(i);
+                if (node.normalName().equals(element.normalName())) pos++;
+                if (node == element) break;
             }
-			return pos;
-		}
+            return pos;
+        }
 
-		@Override
-		protected String getPseudoClass() {
-			return "nth-of-type";
-		}
+        @Override
+        protected String getPseudoClass() {
+            return "nth-of-type";
+        }
     }
 
     public static class IsNthLastOfType extends CssNthEvaluator {
 
-		public IsNthLastOfType(int a, int b) {
-			super(a, b);
-		}
+        public IsNthLastOfType(int a, int b) {
+            super(a, b);
+        }
 
-		@Override
-		protected int calculatePosition(Element root, Element element) {
-			int pos = 0;
-            if (element.parent() == null)
+        @Override
+        protected int calculatePosition(Element root, Element element) {
+            Element parent = element.parent();
+            if (parent == null)
                 return 0;
-        	Elements family = element.parent().children();
-        	for (int i = element.elementSiblingIndex(); i < family.size(); i++) {
-        		if (family.get(i).tag().equals(element.tag())) pos++;
-        	}
-			return pos;
-		}
 
-		@Override
-		protected String getPseudoClass() {
-			return "nth-last-of-type";
-		}
+            int pos = 0;
+            int size = parent.childrenSize();
+            for (int i = element.elementSiblingIndex(); i < size; i++) {
+                if (parent.child(i).normalName().equals(element.normalName())) pos++;
+            }
+            return pos;
+        }
+
+        @Override
+        protected String getPseudoClass() {
+            return "nth-last-of-type";
+        }
     }
 
     /**
@@ -620,9 +625,12 @@ public boolean matches(Element root, Element element) {
 			if (p==null || p instanceof Document) return false;
 
 			int pos = 0;
-        	Elements family = p.children();
-            for (Element el : family) {
-                if (el.tag().equals(element.tag())) pos++;
+            int size = p.childrenSize();
+            for (int i = 0; i < size; i++) {
+                if (p.child(i).normalName().equals(element.normalName()))
+                    pos++;
+                if (pos > 1)
+                    break;
             }
         	return pos == 1;
 		}

diff --git a/src/main/java/org/jsoup/select/StructuralEvaluator.java b/src/main/java/org/jsoup/select/StructuralEvaluator.java
@@ -3,6 +3,8 @@
 import org.jsoup.nodes.Element;
 import org.jsoup.nodes.Node;
 
+import java.util.IdentityHashMap;
+
 /**
  * Base structural evaluator.
  */
@@ -108,22 +110,28 @@ public String toString() {
     }
 
     static class PreviousSibling extends StructuralEvaluator {
+        private final IdentityHashMap<Element, Boolean> memo = new IdentityHashMap<>(); // memoize results
+
         public PreviousSibling(Evaluator evaluator) {
             this.evaluator = evaluator;
         }
 
         @Override
         public boolean matches(Element root, Element element) {
-            if (root == element)
+            final Element parent = element.parent();
+            if (root == element || parent == null)
                 return false;
 
-            Element prev = element.previousElementSibling();
-
-            while (prev != null) {
-                if (evaluator.matches(root, prev))
+            final int size = element.elementSiblingIndex();
+            for (int i = 0; i < size; i++) {
+                final Element el = parent.child(i);
+                Boolean matches = memo.get(el);
+                if (matches == null) {
+                    matches = evaluator.matches(root, el);
+                    memo.put(el, matches);
+                }
+                if (matches)
                     return true;
-
-                prev = prev.previousElementSibling();
             }
             return false;
         }