refs #93 - added method for inputting your own metadata when bagging in place, and defaults to include date bagged and payload-oxum

Author: johnscancella

src/main/java/gov/loc/repository/bagit/creator/BagCreator.java

@@ -6,7 +6,10 @@
 import java.nio.file.Path;
 import java.security.MessageDigest;
 import java.security.NoSuchAlgorithmException;
+import java.text.SimpleDateFormat;
 import java.util.Collection;
+import java.util.Date;
+import java.util.Locale;
 import java.util.Map;
 import java.util.ResourceBundle;
 
@@ -16,40 +19,78 @@
 import gov.loc.repository.bagit.annotation.Incubating;
 import gov.loc.repository.bagit.domain.Bag;
 import gov.loc.repository.bagit.domain.Manifest;
+import gov.loc.repository.bagit.domain.Metadata;
 import gov.loc.repository.bagit.domain.Version;
 import gov.loc.repository.bagit.hash.Hasher;
 import gov.loc.repository.bagit.hash.SupportedAlgorithm;
+import gov.loc.repository.bagit.util.PathUtils;
 import gov.loc.repository.bagit.writer.BagitFileWriter;
 import gov.loc.repository.bagit.writer.ManifestWriter;
+import gov.loc.repository.bagit.writer.MetadataWriter;
 
 /**
  * Responsible for creating a bag in place.
  */
+//TODO look at cleaning up this class so we don't have to ignore CPD
 public final class BagCreator {
   private static final Logger logger = LoggerFactory.getLogger(BagCreator.class);
   private static final ResourceBundle messages = ResourceBundle.getBundle("MessageBundle");
+  private static final String DATE_FORMAT = "yyyy-MM-dd";
 
   private BagCreator(){}
 
-  @SuppressWarnings("CPD-START")
   /**
-   * Creates a basic(only required elements) bag in place for version 0.97.
+   * Creates a bag in place for version 0.97.
    * This method moves and creates files, thus if an error is thrown during operation it may leave the filesystem 
    * in an unknown state of transition. Thus this is <b>not thread safe</b>
    * 
    * @param root the directory that will become the base of the bag and where to start searching for content
    * @param algorithms an collection of {@link SupportedAlgorithm} implementations
    * @param includeHidden to include hidden files when generating the bagit files, like the manifests
+   * @param metadata the metadata to include when creating the bag. Payload-Oxum and Bagging-Date will be overwritten 
+   * 
    * @throws NoSuchAlgorithmException if {@link MessageDigest} can't find the algorithm
    * @throws IOException if there is a problem writing or moving file(s)
+   * 
    * @return a {@link Bag} object representing the newly created bagit bag
    */
-  public static Bag bagInPlace(final Path root, final Collection<SupportedAlgorithm> algorithms, final boolean includeHidden) throws NoSuchAlgorithmException, IOException{
+  @SuppressWarnings("CPD-START")
+  public static Bag bagInPlace(final Path root, final Collection<SupportedAlgorithm> algorithms, final boolean includeHidden, final Metadata metadata) throws NoSuchAlgorithmException, IOException{
     final Bag bag = new Bag(new Version(0, 97));
     bag.setRootDir(root);
     logger.info(messages.getString("creating_bag"), bag.getVersion(), root);
 
     final Path dataDir = root.resolve("data");
+    moveFilesToDataDirectory(root, dataDir, includeHidden);
+    
+    BagitFileWriter.writeBagitFile(bag.getVersion(), bag.getFileEncoding(), root);
+    
+    createManifests(root, dataDir, bag, algorithms, includeHidden);
+    
+    createMetadataFile(root, dataDir, bag, metadata);
+    
+    return bag;
+  }
+  
+  /**
+   * Creates a bag in place for version 0.97.
+   * This method moves and creates files, thus if an error is thrown during operation it may leave the filesystem 
+   * in an unknown state of transition. Thus this is <b>not thread safe</b>
+   * 
+   * @param root the directory that will become the base of the bag and where to start searching for content
+   * @param algorithms an collection of {@link SupportedAlgorithm} implementations
+   * @param includeHidden to include hidden files when generating the bagit files, like the manifests
+   * 
+   * @throws NoSuchAlgorithmException if {@link MessageDigest} can't find the algorithm
+   * @throws IOException if there is a problem writing or moving file(s)
+   * 
+   * @return a {@link Bag} object representing the newly created bagit bag
+   */
+  public static Bag bagInPlace(final Path root, final Collection<SupportedAlgorithm> algorithms, final boolean includeHidden) throws NoSuchAlgorithmException, IOException{
+    return bagInPlace(root, algorithms, includeHidden, new Metadata());
+  }
+  
+  private static void moveFilesToDataDirectory(final Path root, final Path dataDir, final boolean includeHidden) throws IOException{
     Files.createDirectory(dataDir);
     try(final DirectoryStream<Path> directoryStream = Files.newDirectoryStream(root)){
       for(final Path path : directoryStream){
@@ -58,14 +99,15 @@ public static Bag bagInPlace(final Path root, final Collection<SupportedAlgorith
         }
       }
     }
-    
+  }
+  
+  private static void createManifests(final Path root, final Path dataDir, final Bag bag, final Collection<SupportedAlgorithm> algorithms, final boolean includeHidden) throws IOException, NoSuchAlgorithmException{
     logger.info(messages.getString("creating_payload_manifests"));
     final Map<Manifest, MessageDigest> payloadFilesMap = Hasher.createManifestToMessageDigestMap(algorithms);
     final CreatePayloadManifestsVistor payloadVisitor = new CreatePayloadManifestsVistor(payloadFilesMap, includeHidden);
     Files.walkFileTree(dataDir, payloadVisitor);
 
     bag.getPayLoadManifests().addAll(payloadFilesMap.keySet());
-    BagitFileWriter.writeBagitFile(bag.getVersion(), bag.getFileEncoding(), root);
     ManifestWriter.writePayloadManifests(bag.getPayLoadManifests(), root, root, bag.getFileEncoding());
 
     logger.info(messages.getString("creating_tag_manifests"));
@@ -75,11 +117,22 @@ public static Bag bagInPlace(final Path root, final Collection<SupportedAlgorith
 
     bag.getTagManifests().addAll(tagFilesMap.keySet());
     ManifestWriter.writeTagManifests(bag.getTagManifests(), root, root, bag.getFileEncoding());
+  }
+  
+  private static void createMetadataFile(final Path root, final Path dataDir, final Bag bag, final Metadata metadata) throws IOException{
+    bag.setMetadata(metadata);
 
-    return bag;
+    logger.debug(messages.getString("calculating_payload_oxum"), dataDir);
+    final String payloadOxum = PathUtils.generatePayloadOxum(PathUtils.getDataDir(bag.getVersion(), root));
+    bag.getMetadata().upsertPayloadOxum(payloadOxum);
+    
+    bag.getMetadata().remove("Bagging-Date");
+    bag.getMetadata().add("Bagging-Date", new SimpleDateFormat(DATE_FORMAT, Locale.ENGLISH).format(new Date()));
+    
+    logger.info(messages.getString("creating_metadata_file"));
+    MetadataWriter.writeBagMetadata(bag.getMetadata(), bag.getVersion(), root, bag.getFileEncoding());
   }
 
-  @SuppressWarnings("CPD-END")
   /**
    * Creates a basic(only required elements) .bagit bag in place.
    * This creates files and directories, thus if an error is thrown during operation it may leave the filesystem 
@@ -88,11 +141,14 @@ public static Bag bagInPlace(final Path root, final Collection<SupportedAlgorith
    * @param root the directory that will become the base of the bag and where to start searching for content
    * @param algorithms an collection of {@link SupportedAlgorithm} implementations
    * @param includeHidden to include hidden files when generating the bagit files, like the manifests
-   * @return a {@link Bag} object representing the newly created bagit bag
+   * 
    * @throws NoSuchAlgorithmException if {@link MessageDigest} can't find the algorithm
    * @throws IOException if there is a problem writing files or .bagit directory
+   * 
+   * @return a {@link Bag} object representing the newly created bagit bag
    */
   @Incubating
+  @SuppressWarnings("CPD-END")
   public static Bag createDotBagit(final Path root, final Collection<SupportedAlgorithm> algorithms, final boolean includeHidden) throws NoSuchAlgorithmException, IOException{
     final Bag bag = new Bag(new Version(2, 0));
     bag.setRootDir(root);
@@ -120,5 +176,4 @@ public static Bag createDotBagit(final Path root, final Collection<SupportedAlgo
 
     return bag;
   }
-  
-}
+}

src/main/java/gov/loc/repository/bagit/util/PathUtils.java

@@ -7,6 +7,7 @@
 
 import gov.loc.repository.bagit.domain.Bag;
 import gov.loc.repository.bagit.domain.Version;
+import gov.loc.repository.bagit.verify.FileCountAndTotalSizeVistor;
 
 /**
  * Convenience class for dealing with various path issues
@@ -123,4 +124,21 @@ public static Path getBagitDir(final Version version, final Path bagRoot){
 
     return bagRoot;
   }
+  
+  /**
+   * Calculate the total file and byte count of the files in the payload directory
+   * 
+   * @param dataDir the directory to calculate the payload-oxum
+   * 
+   * @return the string representation of the payload-oxum value
+   * 
+   * @throws IOException if there is an error reading any of the files
+   */
+  public static String generatePayloadOxum(final Path dataDir) throws IOException{
+    final FileCountAndTotalSizeVistor visitor = new FileCountAndTotalSizeVistor();
+    
+    Files.walkFileTree(dataDir, visitor);
+    
+    return visitor.getTotalSize() + "." + visitor.getCount();
+  }
 }

src/main/java/gov/loc/repository/bagit/writer/BagWriter.java

@@ -6,8 +6,8 @@
 import java.security.MessageDigest;
 import java.security.NoSuchAlgorithmException;
 import java.util.HashSet;
-import java.util.ResourceBundle;
 import java.util.Map.Entry;
+import java.util.ResourceBundle;
 import java.util.Set;
 
 import org.slf4j.Logger;
@@ -17,7 +17,6 @@
 import gov.loc.repository.bagit.domain.Manifest;
 import gov.loc.repository.bagit.hash.Hasher;
 import gov.loc.repository.bagit.util.PathUtils;
-import gov.loc.repository.bagit.verify.FileCountAndTotalSizeVistor;
 
 /**
  * responsible for writing out a {@link Bag}
@@ -47,7 +46,7 @@ public static void write(final Bag bag, final Path outputDir) throws IOException
     final Path bagitDir = PayloadWriter.writeVersionDependentPayloadFiles(bag, outputDir);
 
     logger.debug(messages.getString("upsert_payload_oxum"));
-    final String payloadOxum = generatePayloadOxum(PathUtils.getDataDir(bag.getVersion(), outputDir));
+    final String payloadOxum = PathUtils.generatePayloadOxum(PathUtils.getDataDir(bag.getVersion(), outputDir));
     bag.getMetadata().upsertPayloadOxum(payloadOxum);
 
     logger.debug(messages.getString("writing_bagit_file"));
@@ -73,23 +72,6 @@ public static void write(final Bag bag, final Path outputDir) throws IOException
     }
   }
 
-  /**
-   * Calculate the total file and byte count of the files in the payload directory
-   * 
-   * @param dataDir the directory to calculate the payload-oxum
-   * 
-   * @return the string representation of the payload-oxum value
-   * 
-   * @throws IOException if there is an error reading any of the files
-   */
-  private static String generatePayloadOxum(final Path dataDir) throws IOException{
-    final FileCountAndTotalSizeVistor visitor = new FileCountAndTotalSizeVistor();
-    
-    Files.walkFileTree(dataDir, visitor);
-    
-    return visitor.getTotalSize() + "." + visitor.getCount();
-  }
-  
   /*
    * Update the tag manifest cause the checksum of the other tag files will have changed since we just wrote them out to disk
    */

src/main/resources/MessageBundle.properties

@@ -95,6 +95,8 @@ skipping_ignored_directory=Skipping [{}] since we are ignoring hidden directorie
 creating_bag=Creating a bag with version: [{}] in directory: [{}].
 creating_payload_manifests=Creating payload manifest(s).
 creating_tag_manifests=Creating tag manifest(s).
+calculating_payload_oxum=Calculating the payload oxum of the data directory [{}].
+creating_metadata_file=Creating the bag metadata file (bag-info.txt or package-info.txt).
 
 #for Hasher.java
 adding_checksum=Adding [{}] to manifest with hash [{}].