New feature to allow harvest to set archive status

src/main/java/gov/nasa/pds/harvest/HarvestCli.java

@@ -15,6 +15,7 @@
 import gov.nasa.pds.harvest.cmd.HarvestCmd;
 import gov.nasa.pds.harvest.util.log.Log4jConfigurator;
 import gov.nasa.pds.registry.common.meta.Metadata;
+import gov.nasa.pds.registry.common.util.ArchiveStatus;
 import gov.nasa.pds.registry.common.util.ExceptionUtils;
 import gov.nasa.pds.registry.common.util.ManifestUtils;
 
@@ -108,11 +109,15 @@ public static void printHelp()
         System.out.println("  -V, --version      Print Harvest version");
         System.out.println();
         System.out.println("Optional parameters:");
-        System.out.println("  -o <dir>           Output directory (applies to supplemental products only)."); 
-        System.out.println("                     Default is /tmp/harvest/out");
-        System.out.println("  -l <file>          Log file. Default is /tmp/harvest/harvest.log");
-        System.out.println("  -v <level>         Logger verbosity: DEBUG, INFO (default), WARNING, ERROR");
-        System.out.println("  -O, --overwrite    Overwrite registered products");
+        System.out.println("  -o <dir>              Output directory (applies to supplemental products only)."); 
+        System.out.println("                        Default is /tmp/harvest/out");
+        System.out.println("  -l <file>             Log file. Default is /tmp/harvest/harvest.log");
+        System.out.println("  -v <level>            Logger verbosity: DEBUG, INFO (default), WARNING, ERROR");
+        System.out.println("  -O, --overwrite       Overwrite registered products");
+        System.out.println("  -a, --archive-status  Set the archive status for all products defaulting to staged");
+        for (String name : new ArchiveStatus().statusNames) {
+          System.out.println("     " + name);
+        }
     }
 
 
@@ -199,6 +204,9 @@ private void initOptions()
 
         bld = Option.builder("O").longOpt("overwrite");
         options.addOption(bld.build());
+
+        bld = Option.builder("a").longOpt("archive-status").hasArg().argName("archive_status");
+        options.addOption(bld.build());
     }
 
 }

src/main/java/gov/nasa/pds/harvest/cmd/HarvestCmd.java

@@ -20,6 +20,8 @@
 import gov.nasa.pds.harvest.util.PackageIdGenerator;
 import gov.nasa.pds.harvest.util.log.LogUtils;
 import gov.nasa.pds.harvest.util.out.WriterManager;
+import gov.nasa.pds.registry.common.meta.BasicMetadataExtractor;
+import gov.nasa.pds.registry.common.util.ArchiveStatus;
 
 
 /**
@@ -37,7 +39,7 @@ public class HarvestCmd implements CliCommand
     private BundleProcessor bundleProc;
     private CollectionProcessor colProc;
     private ProductProcessor prodProc;
-
+    private String archive_status = BasicMetadataExtractor.DEFAULT_ARCHIVE_STATUS;
 
     /**
      * Constructor
@@ -57,6 +59,10 @@ public HarvestCmd()
     @Override
     public void run(CommandLine cmdLine) throws Exception
     {
+        if (cmdLine.hasOption("archive-status")) {
+          this.archive_status = cmdLine.getOptionValue("archive-status").toLowerCase();
+          new ArchiveStatus().validateStatusName(archive_status);;
+        }
         configure(cmdLine);
 
         try
@@ -150,17 +156,17 @@ private void configure(CommandLine cmdLine) throws Exception
         // Xpath maps
         XPathCacheLoader xpcLoader = new XPathCacheLoader();
         xpcLoader.load(cfg.getXpathMaps());
-        
+
         // Processors
         if(cfg.getLoad().getDirectories() != null || cfg.getLoad().getFiles() != null)
         {
-            filesProc = new FilesProcessor(cfg);
+            filesProc = new FilesProcessor(cfg, this.archive_status);
         }
         else if(cfg.getLoad().getBundles() != null)
         {
-            bundleProc = new BundleProcessor(cfg);
-            colProc = new CollectionProcessor(cfg);
-            prodProc = new ProductProcessor(cfg);
+            bundleProc = new BundleProcessor(cfg, this.archive_status);
+            colProc = new CollectionProcessor(cfg, this.archive_status);
+            prodProc = new ProductProcessor(cfg, this.archive_status);
         }
     }
 

src/main/java/gov/nasa/pds/harvest/crawler/BaseProcessor.java

@@ -48,14 +48,16 @@ public class BaseProcessor
     private MetadataNormalizer metaNormalizer;
 
     protected String jobId;
+    final protected String archive_status;
 
     /**
      * Constructor.
      * @param config Harvest configuration parameters
      * @throws Exception Generic exception
      */
-    public BaseProcessor(HarvestConfigurationType config) throws Exception
+    public BaseProcessor(HarvestConfigurationType config, String archive_status) throws Exception
     {
+        this.archive_status = archive_status;
         this.config = config;
 
         log = LogManager.getLogger(this.getClass());

src/main/java/gov/nasa/pds/harvest/crawler/BundleProcessor.java

@@ -53,9 +53,9 @@ public class BundleProcessor extends BaseProcessor
      * @param counter document / product counter
      * @throws Exception Generic exception
      */
-    public BundleProcessor(HarvestConfigurationType config) throws Exception
+    public BundleProcessor(HarvestConfigurationType config, String archive_status) throws Exception
     {
-        super(config);
+        super(config, archive_status);
 
         bundleExtractor = new BundleMetadataExtractor();
     }
@@ -128,7 +128,7 @@ private void onBundle(File file) throws Exception
 
     private void processMetadata(File file, Document doc) throws Exception
     {
-        Metadata meta = basicExtractor.extract(file, doc);
+        Metadata meta = basicExtractor.extract(file, doc, this.archive_status);
         meta.setNodeName(ConfigManager.exchangeIndexForNode(RegistryManager.getInstance().getIndexName()));
 
         if(!bundleCfg.getVersions().equalsIgnoreCase("all") && !bundleCfg.getVersions().contains(meta.strVid)) return;

src/main/java/gov/nasa/pds/harvest/crawler/CollectionProcessor.java

@@ -55,9 +55,9 @@ public class CollectionProcessor extends BaseProcessor
      * @param config Harvest configuration parameters
      * @throws Exception Generic exception
      */
-    public CollectionProcessor(HarvestConfigurationType config) throws Exception
+    public CollectionProcessor(HarvestConfigurationType config, String archive_status) throws Exception
     {
-        super(config);
+        super(config, archive_status);
         invWriter = new CollectionInventoryWriter(ConfigManager.exchangeRegistry(config.getRegistry()));
         this.invProc = new CollectionInventoryProcessor(config.getReferences().isPrimaryOnly());
         collectionExtractor = new CollectionMetadataExtractor();
@@ -127,7 +127,7 @@ private void onCollection(File file, BundleType bCfg) throws Exception
 
     private void processMetadata(File file, Document doc, BundleType bCfg) throws Exception
     {
-        Metadata meta = basicExtractor.extract(file, doc);
+        Metadata meta = basicExtractor.extract(file, doc, this.archive_status);
         meta.setNodeName(ConfigManager.exchangeIndexForNode(RegistryManager.getInstance().getIndexName()));
 
         // Collection filter

src/main/java/gov/nasa/pds/harvest/crawler/FilesProcessor.java

@@ -56,9 +56,9 @@ public class FilesProcessor extends BaseProcessor
      * @param config Harvest configuration parameters (from a config file)
      * @throws Exception Generic exception
      */
-    public FilesProcessor(HarvestConfigurationType config) throws Exception
+    public FilesProcessor(HarvestConfigurationType config, String archive_status) throws Exception
     {
-        super(config);
+        super(config, archive_status);
 
         bundleExtractor = new BundleMetadataExtractor();
         collectionExtractor = new CollectionMetadataExtractor();
@@ -191,7 +191,7 @@ private void onFile(File file) throws Exception
     private void processMetadata(File file, Document doc) throws Exception
     {
         // Extract basic metadata
-        Metadata meta = basicExtractor.extract(file, doc);
+        Metadata meta = basicExtractor.extract(file, doc, this.archive_status);
         meta.setNodeName(ConfigManager.exchangeIndexForNode(RegistryManager.getInstance().getIndexName()));
 
         log.info("Processing " + file.getAbsolutePath());

src/main/java/gov/nasa/pds/harvest/crawler/ProductProcessor.java

@@ -47,9 +47,9 @@ public class ProductProcessor extends BaseProcessor
      * @param config Harvest configuration parameters
      * @throws Exception Generic exception
      */
-    public ProductProcessor(HarvestConfigurationType config) throws Exception
+    public ProductProcessor(HarvestConfigurationType config, String archive_status) throws Exception
     {
-        super(config);
+        super(config, archive_status);
     }
 
 
@@ -187,7 +187,7 @@ private void processMetadata(File file, Document doc) throws Exception
         Counter counter = RegistryManager.getInstance().getCounter();
 
         // Extract basic metadata
-        Metadata meta = basicExtractor.extract(file, doc);
+        Metadata meta = basicExtractor.extract(file, doc, this.archive_status);
         meta.setNodeName(ConfigManager.exchangeIndexForNode(RegistryManager.getInstance().getIndexName()));
 
         // Only process primary products from collection inventory