From c653538e5f92ff100e1de4477e120eb79955cc95 Mon Sep 17 00:00:00 2001
From: Julien HENRY <julien.henry@sonarsource.com>
Date: Tue, 17 Feb 2015 11:50:25 +0100
Subject: [PATCH] SONAR-6000 Truncate duplication data to avoid exceeding
 persistit cache max size

---
 .../org/sonar/batch/cpd/JavaCpdEngine.java    | 16 ++++++
 .../batch/mediumtest/cpd/CpdMediumTest.java   | 49 +++++++++++++++++--
 2 files changed, 61 insertions(+), 4 deletions(-)

diff --git a/sonar-batch/src/main/java/org/sonar/batch/cpd/JavaCpdEngine.java b/sonar-batch/src/main/java/org/sonar/batch/cpd/JavaCpdEngine.java
index 2205227cd31..d753fc45e1f 100644
--- a/sonar-batch/src/main/java/org/sonar/batch/cpd/JavaCpdEngine.java
+++ b/sonar-batch/src/main/java/org/sonar/batch/cpd/JavaCpdEngine.java
@@ -83,6 +83,9 @@ public class JavaCpdEngine extends CpdEngine {
    */
   private static final int TIMEOUT = 5 * 60;
 
+  private static final int MAX_CLONE_GROUP_PER_FILE = 100;
+  private static final int MAX_CLONE_PART_PER_GROUP = 100;
+
   private final IndexFactory indexFactory;
   private final FileSystem fs;
   private final Settings settings;
@@ -230,12 +233,25 @@ public class JavaCpdEngine extends CpdEngine {
       .setFromCore()
       .save();
 
+    int cloneGroupCount = 0;
     for (CloneGroup duplication : duplications) {
+      cloneGroupCount++;
+      if (cloneGroupCount > MAX_CLONE_GROUP_PER_FILE) {
+        LOG.warn("Too many duplication groups on file " + inputFile.relativePath() + ". Keep only the first " + MAX_CLONE_GROUP_PER_FILE + " groups.");
+        break;
+      }
       NewDuplication builder = context.newDuplication();
       ClonePart originPart = duplication.getOriginPart();
       builder.originBlock(inputFile, originPart.getStartLine(), originPart.getEndLine());
+      int clonePartCount = 0;
       for (ClonePart part : duplication.getCloneParts()) {
         if (!part.equals(originPart)) {
+          clonePartCount++;
+          if (clonePartCount > MAX_CLONE_PART_PER_GROUP) {
+            LOG.warn("Too many duplication references on file " + inputFile.relativePath() + " for block at line " + originPart.getStartLine() + ". Keep only the first "
+              + MAX_CLONE_PART_PER_GROUP + " references.");
+            break;
+          }
           ((DefaultDuplication) builder).isDuplicatedBy(part.getResourceId(), part.getStartLine(), part.getEndLine());
         }
       }
diff --git a/sonar-batch/src/test/java/org/sonar/batch/mediumtest/cpd/CpdMediumTest.java b/sonar-batch/src/test/java/org/sonar/batch/mediumtest/cpd/CpdMediumTest.java
index 47ebaa81f8a..8f9297078c5 100644
--- a/sonar-batch/src/test/java/org/sonar/batch/mediumtest/cpd/CpdMediumTest.java
+++ b/sonar-batch/src/test/java/org/sonar/batch/mediumtest/cpd/CpdMediumTest.java
@@ -30,6 +30,9 @@ import org.junit.rules.TemporaryFolder;
 import org.sonar.api.batch.fs.InputFile;
 import org.sonar.api.batch.fs.internal.DefaultInputFile;
 import org.sonar.api.batch.sensor.duplication.Duplication;
+import org.sonar.api.batch.sensor.measure.Measure;
+import org.sonar.api.batch.sensor.measure.internal.DefaultMeasure;
+import org.sonar.api.measures.CoreMetrics;
 import org.sonar.batch.mediumtest.BatchMediumTester;
 import org.sonar.batch.mediumtest.TaskResult;
 import org.sonar.xoo.XooPlugin;
@@ -129,6 +132,44 @@ public class CpdMediumTest {
     assertThat(cloneGroupFile2.duplicates().get(0).resourceKey()).isEqualTo(((DefaultInputFile) inputFile1).key());
   }
 
+  // SONAR-6000
+  @Test
+  public void truncateDuplication() throws IOException {
+    File srcDir = new File(baseDir, "src");
+    srcDir.mkdir();
+
+    String duplicatedStuff = "Sample xoo\nfoo\n";
+
+    int blockCount = 10000;
+    File xooFile1 = new File(srcDir, "sample.xoo");
+    for (int i = 0; i < blockCount; i++) {
+      FileUtils.write(xooFile1, duplicatedStuff, true);
+      FileUtils.write(xooFile1, "" + i, true);
+    }
+
+    TaskResult result = tester.newTask()
+      .properties(builder
+        .put("sonar.sources", "src")
+        .put("sonar.cpd.xoo.minimumTokens", "1")
+        .put("sonar.cpd.xoo.minimumLines", "1")
+        .build())
+      .start();
+
+    Measure duplicatedBlocks = null;
+    for (Measure m : result.measures()) {
+      if (m.metric().key().equals("duplicated_blocks")) {
+        duplicatedBlocks = m;
+      }
+    }
+    assertThat(duplicatedBlocks.value()).isEqualTo(blockCount);
+
+    List<Duplication> duplicationGroups = result.duplicationsFor(result.inputFile("src/sample.xoo"));
+    assertThat(duplicationGroups).hasSize(1);
+
+    Duplication cloneGroup = duplicationGroups.get(0);
+    assertThat(cloneGroup.duplicates()).hasSize(100);
+  }
+
   @Test
   public void testIntraFileDuplications() throws IOException {
     File srcDir = new File(baseDir, "src");
@@ -164,10 +205,10 @@ public class CpdMediumTest {
     assertThat(cloneGroup.duplicates().get(0).startLine()).isEqualTo(5);
     assertThat(cloneGroup.duplicates().get(0).length()).isEqualTo(2);
 
-    // assertThat(result.measures()).contains(new DefaultMeasure<String>()
-    // .forMetric(CoreMetrics.DUPLICATION_LINES_DATA)
-    // .onFile(inputFile)
-    // .withValue("1=1;2=1;3=0;4=0;5=1;6=1;7=0"));
+    assertThat(result.measures()).contains(new DefaultMeasure<String>()
+      .forMetric(CoreMetrics.DUPLICATION_LINES_DATA)
+      .onFile(inputFile)
+      .withValue("1=1;2=1;3=0;4=0;5=1;6=1;7=0"));
   }
 
 }
-- 
2.39.5