Changeset 6321

Timestamp:

06/23/15 14:36:16 (9 years ago)

Author:

davor.ostojic@oeaw.ac.at

Message:

Issue #768
https://trac.clarin.eu/ticket/768

"Maybe also add a command line option to specify which data roots to (try to) import (import all by default)"

the option is added: -l <list_of_dataroot_paths> (space separated)
program compares canonical paths

File:

• vlo/branches/vlo-3.3-oeaw/vlo-importer/src/main/java/eu/clarin/cmdi/vlo/importer/MetadataImporter.java (modified) (7 diffs)

vlo/branches/vlo-3.3-oeaw/vlo-importer/src/main/java/eu/clarin/cmdi/vlo/importer/MetadataImporter.java

@@ -1 +1 @@
 package eu.clarin.cmdi.vlo.importer;
-
-import eu.clarin.cmdi.vlo.LanguageCodeUtils;
-import eu.clarin.cmdi.vlo.CommonUtils;
-import eu.clarin.cmdi.vlo.FacetConstants;
-import eu.clarin.cmdi.vlo.config.DataRoot;
-import eu.clarin.cmdi.vlo.config.VloConfig;
-import eu.clarin.cmdi.vlo.config.XmlVloConfigFactory;
 
 import java.io.File;
@@ -38 +31 @@
 import org.slf4j.LoggerFactory;
 
+import eu.clarin.cmdi.vlo.CommonUtils;
+import eu.clarin.cmdi.vlo.FacetConstants;
+import eu.clarin.cmdi.vlo.LanguageCodeUtils;
+import eu.clarin.cmdi.vlo.config.DataRoot;
+import eu.clarin.cmdi.vlo.config.VloConfig;
+import eu.clarin.cmdi.vlo.config.XmlVloConfigFactory;
+
 /**
  * The main metadataImporter class. Also contains the main function.
@@ -121 +121 @@
         initSolrServer();
         List<DataRoot> dataRoots = checkDataRoots();
+        
+        dataRoots = filterDataRootsWithCLArgs(dataRoots);
+        
         long start = System.currentTimeMillis();
         try {
@@ -204 +207 @@
         for (DataRoot dataRoot : dataRoots) {
             if (!dataRoot.getRootFile().exists()) {
-                LOG.error("Root file " + dataRoot.getRootFile() + " does not exist. Could be configuration error (see VLOConfig.xml)! Proceeding with next ...");
-//              LOG.error("Root file " + dataRoot.getRootFile() + " does not exist. Probable configuration error so stopping import.");
-//                System.exit(1);
+                LOG.error("Root file " + dataRoot.getRootFile() + " does not exist. It could be configuration error! Proceeding with next ...");
             } else{
                 existingDataRoots.add(dataRoot);
             }
+            
         }
         return existingDataRoots;
+    }
+    
+    /**
+     * if user specified which data roots should be imported,
+     * list of existing data roots will be filtered with the list from user
+     * 
+     * @return
+     */
+    protected List<DataRoot> filterDataRootsWithCLArgs(List<DataRoot> dataRoots){
+        if(CL_DATAROOTS_LIST == null)
+                return dataRoots;
+        
+        //filter data
+        
+        LOG.info("Filtering configured data root files with command line arguments: " + CL_DATAROOTS_LIST) ;
+        
+        LinkedList<File> cl_dataroots = new LinkedList<File>();
+        
+        List<String> paths = Arrays.asList((CL_DATAROOTS_LIST.split("\\s+")));
+        
+        for(String path: paths)
+                cl_dataroots.add(new File(path));
+        
+        List<DataRoot> filteredDataRoots = new LinkedList<DataRoot>();
+        try{
+        dr: for(DataRoot dataRoot: dataRoots){
+                        for(File cldr: cl_dataroots){
+                        if(cldr.getCanonicalPath().equals(dataRoot.getRootFile().getCanonicalPath())){
+                                filteredDataRoots.add(dataRoot);
+                                cl_dataroots.remove(cldr);
+                                continue dr;
+                        }
+                        }
+                        LOG.info("Root file " + dataRoot.getRootFile() + " will be omitted from processing");
+                }
+        }catch (IOException e){
+                filteredDataRoots = dataRoots;
+        }
+
+        
+                return filteredDataRoots;
     }
 
@@ -522 +565 @@
     public static LanguageCodeUtils languageCodeUtils;
     
+    //data roots passed from command line    
+    public static String CL_DATAROOTS_LIST = null;
+    
     /**
      * @param args
@@ -540 +586 @@
          */
         options.addOption("c", true, "-c <file> : use parameters specified in <file>");
+        options.addOption("l", true, "-l <dataroot> [ ' ' <dataroot> ]* :  space separated list of dataroots to be processed.\n"
+                        + "If dataroot is not specified in config file it will be ignored.");
+        options.getOption("l").setOptionalArg(true);
         
         CommandLineParser parser = new PosixParser();
@@ -550 +599 @@
                 // the "c" option was specified, now get its value
                 configFile = cmd.getOptionValue("c");
+            }
+            
+            if(cmd.hasOption("l")){
+                CL_DATAROOTS_LIST = cmd.getOptionValue("l");
             }