From 7f65b1f91047cea920c20bc5ee9e7f1cd2f887dd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Cl=C3=A9ment=20Fournier?= <clement.fournier76@gmail.com>
Date: Mon, 20 Apr 2020 08:07:04 +0200
Subject: [PATCH] Improve pooling stats

---
 .../pmd/lang/ast/impl/javacc/StringPool.java  | 29 +++++++++++--------
 1 file changed, 17 insertions(+), 12 deletions(-)

diff --git a/pmd-core/src/main/java/net/sourceforge/pmd/lang/ast/impl/javacc/StringPool.java b/pmd-core/src/main/java/net/sourceforge/pmd/lang/ast/impl/javacc/StringPool.java
index 7a0d1e464c..a8a238f473 100644
--- a/pmd-core/src/main/java/net/sourceforge/pmd/lang/ast/impl/javacc/StringPool.java
+++ b/pmd-core/src/main/java/net/sourceforge/pmd/lang/ast/impl/javacc/StringPool.java
@@ -5,9 +5,9 @@
 package net.sourceforge.pmd.lang.ast.impl.javacc;
 
 
-import java.util.HashMap;
 import java.util.LongSummaryStatistics;
 import java.util.Map;
+import java.util.WeakHashMap;
 
 import net.sourceforge.pmd.util.document.Chars;
 
@@ -19,12 +19,12 @@ import net.sourceforge.pmd.util.document.Chars;
  */
 public final class StringPool {
 
+    // This will be constant-folded by the JIT, use false in production
     private static final boolean COLLECT_STATS = false;
-    private static final Stats stats = new Stats();
+    private static final Stats STATS = new Stats();
 
     private static final String[] SINGLE_CHARS;
 
-
     static {
         // all ascii characters
         SINGLE_CHARS = new String[128];
@@ -34,12 +34,15 @@ public final class StringPool {
     }
 
 
-    private final Map<Chars, String> pool = new HashMap<>();
+    private final Map<Chars, String> pool = new WeakHashMap<>();
 
     String toString(CharSequence c, boolean doPool) {
         if (c.length() == 1) {
             char fst = c.charAt(0);
             if (fst < 128) {
+                if (COLLECT_STATS) {
+                    STATS.addCacheHit(1);
+                }
                 return SINGLE_CHARS[fst];
             }
         }
@@ -49,10 +52,10 @@ public final class StringPool {
                 return pool.compute((Chars) c,
                                     (chars, s) -> {
                                         if (s != null) {
-                                            stats.addCacheHit(s.length());
+                                            STATS.addCacheHit(s.length());
                                             return s;
                                         } else {
-                                            stats.addCacheMiss(chars.length());
+                                            STATS.addCacheMiss(chars.length());
                                             return chars.toString();
                                         }
                                     });
@@ -61,13 +64,15 @@ public final class StringPool {
             }
         }
         if (COLLECT_STATS) {
-            stats.addPass(c);
+            STATS.addPass(c);
         }
         return c.toString();
     }
 
     public static void printStats() {
-        stats.print();
+        if (COLLECT_STATS) {
+            STATS.print();
+        }
     }
 
     private static class Stats {
@@ -84,15 +89,15 @@ public final class StringPool {
             System.err.println("String pool stats");
             System.err.println("=================");
             System.err.println("Hits: " + hits + " (" + (hits * 100 / (total() + 1)) + "% of " + total() + ")");
-            System.err.println("Hit length (net savings): " + hitLen + " " + toSize(hitLen));
-            System.err.println(
-                "Total pool size: " + poolContents.getCount() + " strings (" + toSize(poolContents.getSum()) + ")");
+            System.err.println("Hit length (net savings): " + hitLen + " (" + toSize(hitLen) + ")");
+            System.err.println("Total pool size: " + poolContents.getCount() + " strings (" + toSize(poolContents.getSum()) + ")");
             System.err.println("Avg pooled string length: " + poolContents.getAverage() + " chars");
             System.err.println("Unpooled string length: " + notPooledAlloc + " chars (" + toSize(notPooledAlloc) + ")");
         }
 
         private String toSize(long charLen) {
-            // assuming all pooled chars are latin-1, so 1B in a compressed string (byte[], not char[])
+            // Assuming all pooled chars are latin-1, so 1B in a compressed string (byte[], not char[])
+            // Before Java 9, it's twice as much
             if (charLen > (1 << 20)) {
                 return (charLen >> 20) + " MB";
             } else if (charLen > (1 << 10)) {