ISSUE-1773 CharacterReader: Reuse stringCache

jhy · Jul 1, 2022 · 32ebe91 · 32ebe91
1 parent d85faf6
commit 32ebe91
Showing 1 changed file with 14 additions and 1 deletion.
diff --git a/src/main/java/org/jsoup/parser/CharacterReader.java b/src/main/java/org/jsoup/parser/CharacterReader.java
@@ -19,6 +19,7 @@
  */
 public final class CharacterReader {
     protected static final ThreadLocal<SoftReference<BufferRecycler>> _recyclerRef = new ThreadLocal<>();
+    protected static final ThreadLocal<SoftReference<String[]>> stringCacheRef = new ThreadLocal<>();
 
     public static BufferRecycler getBufferRecycler() {
         SoftReference<BufferRecycler> ref = _recyclerRef.get();
@@ -32,6 +33,18 @@ public static BufferRecycler getBufferRecycler() {
         return br;
     }
 
+    public static String[] getStringCache() {
+        SoftReference<String[]> ref = stringCacheRef.get();
+        String[] stringCache = (ref == null) ? null : ref.get();
+
+        if (stringCache == null) {
+            stringCache = new String[stringCacheSize];
+            ref = new SoftReference<>(stringCache);
+            stringCacheRef.set(ref);
+        }
+        return stringCache;
+    }
+
     static final char EOF = (char) -1;
     private static final int maxStringCacheLen = 12;
     static final int maxBufferLen = 1024 * 32; // visible for testing
@@ -46,7 +59,7 @@ public static BufferRecycler getBufferRecycler() {
     private int readerPos;
     private int bufMark = -1;
     private static final int stringCacheSize = 512;
-    private String[] stringCache = new String[stringCacheSize]; // holds reused strings in this doc, to lessen garbage
+    private String[] stringCache = getStringCache(); // holds reused strings in this doc, to lessen garbage
 
     @Nullable private ArrayList<Integer> newlinePositions = null; // optionally track the pos() position of newlines - scans during bufferUp()
     private int lineNumberOffset = 1; // line numbers start at 1; += newlinePosition[indexof(pos)]