pmd: fixed #1068 CPD fails on broken symbolic links

pmd/etc/changelog.txt

@@ -1,3 +1,9 @@
+????? ??, 2013 - 5.0.3:
+
+Fixed bug 1068: CPD fails on broken symbolic links
+
+
+
 February 3, 2013 - 5.0.2:
 
 Fixed bug  878: False positive: UnusedFormalParameter for abstract methods

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

@@ -93,11 +93,16 @@ public class CPD {
             current.add(signature);
         }
 
-        if (!file.getCanonicalPath().equals(new File(file.getAbsolutePath()).getCanonicalPath())) {
+        if (!file.getCanonicalPath().equals(file.getAbsolutePath())) {
             System.err.println("Skipping " + file + " since it appears to be a symlink");
             return;
         }
 
+        if (!file.exists()) {
+            System.err.println("Skipping " + file + " since it doesn't exist (broken symlink?)");
+            return;
+        }
+
         listener.addedFile(fileCount, file);
         SourceCode sourceCode = configuration.sourceCodeFor(file);
         configuration.tokenizer().tokenize(sourceCode, tokens);

pmd/src/test/java/net/sourceforge/pmd/cpd/CPDTest.java

@@ -0,0 +1,71 @@
+/**
+ * BSD-style license; for more info see http://pmd.sourceforge.net/license.html
+ */
+package net.sourceforge.pmd.cpd;
+
+import java.io.File;
+
+import org.junit.Assert;
+import org.junit.Before;
+import org.junit.Test;
+
+/**
+ * Unit test for {@link CPD}
+ */
+public class CPDTest {
+
+    private static final String BASE_TEST_RESOURCE_PATH = "src/test/resources/net/sourceforge/pmd/cpd/files/";
+    
+    private CPD cpd;
+
+    @Before
+    public void setup() {
+        CPDConfiguration theConfiguration = new CPDConfiguration(new String[] {"--language", "java",
+                "--minimum-tokens", "10"});
+        cpd = new CPD(theConfiguration);
+    }
+
+    /**
+     * A broken symlink (which is basically a not existing file), should be skipped.
+     * @throws Exception any error
+     */
+    @Test
+    public void testFileSectionWithBrokenSymlinks() throws Exception {
+        cpd.setCpdListener(new NoFileAssertListener(0));
+
+        cpd.add(new File(BASE_TEST_RESOURCE_PATH, "this-is-a-broken-sym-link-for-test"));
+    }
+
+    /**
+     * A file should be added only once - even if it was found twice, because of a sym link.
+     * @throws Exception any error
+     */
+    @Test
+    public void testFileAddedAsSymlinkAndReal() throws Exception {
+        cpd.setCpdListener(new NoFileAssertListener(1));
+
+        cpd.add(new File(BASE_TEST_RESOURCE_PATH, "real-file.txt"));
+        cpd.add(new File(BASE_TEST_RESOURCE_PATH, "symlink-for-real-file.txt"));
+    }
+
+    /**
+     * Simple listener that fails, if to many files were added and not skipped.
+     */
+    private static class NoFileAssertListener implements CPDListener {
+        private int maximumFilesAllowed;
+        private int files;
+        public NoFileAssertListener(int maximumFilesAllowed) {
+            this.maximumFilesAllowed = maximumFilesAllowed;
+            this.files = 0;
+        }
+        public void addedFile(int fileCount, File file) {
+            files++;
+            if (files > maximumFilesAllowed) {
+                Assert.fail("File was added! - " + file);
+            }
+        }
+        public void phaseUpdate(int phase) {
+            // not needed for this test
+        }
+    }
+}

pmd/src/test/resources/net/sourceforge/pmd/cpd/files/symlink-for-real-file.txt

@@ -0,0 +1 @@
+real-file.txt

pmd/src/test/resources/net/sourceforge/pmd/cpd/files/this-is-a-broken-sym-link-for-test

@@ -0,0 +1 @@
+broken-sym-link