refs #61 - added more os specific files to warn about

Author: johnscancella

src/main/java/gov/loc/repository/bagit/conformance/BagLinter.java

@@ -37,7 +37,10 @@ public class BagLinter {
   private static final Version LATEST_BAGIT_VERSION = new Version(0, 97);
   private static final String THUMBS_DB_FILE = "[Tt][Hh][Uu][Mm][Bb][Ss]\\.[Dd][Bb]";
   private static final String DS_STORE_FILE = "\\.[Dd][Ss]_[Ss][Tt][Oo][Rr][Ee]";
-  private static final String OS_FILES_REGEX = ".*data/(" + THUMBS_DB_FILE + "|" + DS_STORE_FILE + ")";
+  private static final String SPOTLIGHT_FILE = "\\.[Ss][Pp][Oo][Tt][Ll][Ii][Gg][Hh][Tt]-[Vv]100";
+  private static final String TRASHES_FILE = "\\.(_.)?[Tt][Rr][Aa][Ss][Hh][Ee][Ss]";
+  private static final String FS_EVENTS_FILE = "\\.[Ff][Ss][Ee][Vv][Ee][Nn][Tt][Ss][Dd]";
+  private static final String OS_FILES_REGEX = ".*data/(" + THUMBS_DB_FILE + "|" + DS_STORE_FILE + "|" + SPOTLIGHT_FILE + "|" + TRASHES_FILE + "|" + FS_EVENTS_FILE + ")";
 
   private final BagReader reader;
 
@@ -308,4 +311,12 @@ private void checkForPayloadOxumMetadata(final Path bagitDir, final Charset enco
   public BagReader getReader() {
     return reader;
   }
+
+  /**
+   * Used only for unit testing
+   * @return the regex for os specific files to warn about
+   */
+  protected static String getOsFilesRegex() {
+    return OS_FILES_REGEX;
+  }
 }

src/main/java/gov/loc/repository/bagit/conformance/BagitWarning.java

@@ -19,8 +19,8 @@ public enum BagitWarning {
       "contained in the GNU Coreutils package (md5sum, sha1sum, etc.), collectively referred to here as 'md5sum'. "
       + "This creates slight differences in generated manifests that can cause problems in some implementations."),
   OLD_BAGIT_VERSION("The bagit specification version is not the newest. Consider converting to the latest version."),
-  OS_SPECIFIC_FILES("Files created by the operating system (OS) for its own use. They are non-protable across OS versions "
-      + "and should not be included in any manifest. Examples Thumbs.db on Windows or .DS_Store on OSX"),
+  OS_SPECIFIC_FILES("Files created by the operating system (OS) for its own use. They are non-portable across OS versions "
+      + "and should not be included in any manifest. Examples Thumbs.db on Windows or .DS_Store on OS X"),
   PAYLOAD_OXUM_MISSING("It is recommended to always include the Payload-Oxum in the bag metadata "
       + "since it allows for a 'quick verification' of the bag."),
   TAG_FILES_ENCODING("It is recommended to always use UTF-8"),

src/test/java/gov/loc/repository/bagit/conformance/BagLinterTest.java

@@ -31,6 +31,17 @@ public void setup(){
     sut = new BagLinter(mapping);
   }
 
+  @Test
+  public void testOSSpecificFilesRegex(){
+    String regex = BagLinter.getOsFilesRegex();
+    String[] osFilesToTest = new String[]{"data/Thumbs.db", "data/.DS_Store", "data/.Spotlight-V100", "data/.Trashes", 
+        "data/._.Trashes", "data/.fseventsd"};
+    
+    for(String osFileToTest : osFilesToTest){
+      assertTrue(osFileToTest + " should match regex but it doesn't", osFileToTest.matches(regex));
+    }
+  }
+  
   @Test
   public void testLintBag() throws Exception{
     Set<BagitWarning> expectedWarnings = new HashSet<>();