Merge branch 'pr-3385'

docs/pages/release_notes.md

@@ -18,12 +18,15 @@ This is a {{ site.pmd.release_type }} release.
 
 *   apex
     *   [#3329](https://github.com/pmd/pmd/issues/3329): \[apex] ApexCRUDViolation doesn't report SOQL for loops
+*   core
+    *   [#3387](https://github.com/pmd/pmd/issues/3387): \[core] CPD should avoid unnecessary copies when running with --skip-lexical-errors
 
 ### API Changes
 
 ### External Contributions
 
 *   [#3367](https://github.com/pmd/pmd/pull/3367): \[apex] Check SOQL CRUD on for loops - [Jonathan Wiesel](https://github.com/jonathanwiesel)
+*   [#3385](https://github.com/pmd/pmd/pull/3385): \[core] CPD: Optimize --skip-lexical-errors option - [Woongsik Choi](https://github.com/woongsikchoi)
 
 {% endtocmaker %}
 

pmd-core/src/main/java/net/sourceforge/pmd/cpd/CPD.java

@@ -147,13 +147,12 @@ public class CPD {
     }
 
     private void addAndSkipLexicalErrors(SourceCode sourceCode) throws IOException {
-        TokenEntry.State savedTokenEntry = new TokenEntry.State(tokens.getTokens());
+        final TokenEntry.State savedState = tokens.snapshot();
         try {
             addAndThrowLexicalError(sourceCode);
         } catch (TokenMgrError e) {
             System.err.println("Skipping " + sourceCode.getFileName() + ". Reason: " + e.getMessage());
-            tokens.getTokens().clear();
-            tokens.getTokens().addAll(savedTokenEntry.restore());
+            tokens.restore(savedState);
         }
     }
 

pmd-core/src/main/java/net/sourceforge/pmd/cpd/TokenEntry.java

@@ -4,8 +4,8 @@
 
 package net.sourceforge.pmd.cpd;
 
-import java.util.ArrayList;
 import java.util.HashMap;
+import java.util.Iterator;
 import java.util.List;
 import java.util.Map;
 import java.util.concurrent.atomic.AtomicInteger;
@@ -94,21 +94,23 @@ public class TokenEntry implements Comparable<TokenEntry> {
      * entries.
      */
     public static class State {
-        private int tokenCount;
-        private Map<String, Integer> tokens;
-        private List<TokenEntry> entries;
+        private final int tokenCount;
+        private final int tokensMapSize;
 
-        public State(List<TokenEntry> entries) {
+        public State() {
             this.tokenCount = TokenEntry.TOKEN_COUNT.get().intValue();
-            this.tokens = new HashMap<>(TokenEntry.TOKENS.get());
-            this.entries = new ArrayList<>(entries);
+            this.tokensMapSize = TokenEntry.TOKENS.get().size();
         }
 
-        public List<TokenEntry> restore() {
+        public void restore(final List<TokenEntry> entries) {
             TokenEntry.TOKEN_COUNT.get().set(tokenCount);
-            TOKENS.get().clear();
-            TOKENS.get().putAll(tokens);
-            return entries;
+            final Iterator<Map.Entry<String, Integer>> it = TOKENS.get().entrySet().iterator();
+            while (it.hasNext()) {
+                if (it.next().getValue() > tokensMapSize) {
+                    it.remove();
+                }
+            }
+            entries.subList(tokenCount, entries.size()).clear();
         }
     }
 

pmd-core/src/main/java/net/sourceforge/pmd/cpd/Tokens.java

@@ -8,6 +8,8 @@ import java.util.ArrayList;
 import java.util.Iterator;
 import java.util.List;
 
+import net.sourceforge.pmd.cpd.TokenEntry.State;
+
 public class Tokens {
 
     private List<TokenEntry> tokens = new ArrayList<>();
@@ -44,4 +46,12 @@ public class Tokens {
         return tokens;
     }
 
+    public State snapshot() {
+        return new State();
+    }
+
+    public void restore(final State savedState) {
+        savedState.restore(tokens);
+    }
+
 }