Merge pull request DSpace#9653 from AndrewAlesik/media-filter-new-fun…

…ctionality New parameter fromdate for media-filter script
4Science · Sep 4, 2024 · be1ee8b · be1ee8b
2 parents d4f6cd6 + 9c7b20f
commit be1ee8b
Show file tree

Hide file tree

Showing 5 changed files with 37 additions and 3 deletions.
diff --git a/dspace-api/src/main/java/org/dspace/app/mediafilter/MediaFilterScript.java b/dspace-api/src/main/java/org/dspace/app/mediafilter/MediaFilterScript.java
@@ -7,6 +7,7 @@
  */
 package org.dspace.app.mediafilter;
 
+import java.time.LocalDate;
 import java.util.ArrayList;
 import java.util.Arrays;
 import java.util.HashMap;
@@ -37,8 +38,9 @@
  * MFM: -v verbose outputs all extracted text to STDOUT; -f force forces all
  * bitstreams to be processed, even if they have been before; -n noindex does not
  * recreate index after processing bitstreams; -i [identifier] limits processing
- * scope to a community, collection or item; and -m [max] limits processing to a
- * maximum number of items.
+ * scope to a community, collection or item; -m [max] limits processing to a
+ * maximum number of items; -fd [fromdate] takes only items starting from this date,
+ * filtering by last_modified in the item table.
  */
 public class MediaFilterScript extends DSpaceRunnable<MediaFilterScriptConfiguration> {
 
@@ -60,6 +62,7 @@ public class MediaFilterScript extends DSpaceRunnable<MediaFilterScriptConfigura
     private String[] filterNames;
     private String[] skipIds = null;
     private Map<String, List<String>> filterFormats = new HashMap<>();
+    private LocalDate fromDate = null;
 
     public MediaFilterScriptConfiguration getScriptConfiguration() {
         return new DSpace().getServiceManager()
@@ -112,6 +115,10 @@ public void setup() throws ParseException {
             skipIds = commandLine.getOptionValues('s');
         }
 
+        if (commandLine.hasOption('d')) {
+            fromDate = LocalDate.parse(commandLine.getOptionValue('d'));
+        }
+
 
     }
 
@@ -215,6 +222,10 @@ public void internalRun() throws Exception {
             mediaFilterService.setSkipList(Arrays.asList(skipIds));
         }
 
+        if (fromDate != null) {
+            mediaFilterService.setFromDate(fromDate);
+        }
+
         Context c = null;
 
         try {

diff --git a/dspace-api/src/main/java/org/dspace/app/mediafilter/MediaFilterScriptConfiguration.java b/dspace-api/src/main/java/org/dspace/app/mediafilter/MediaFilterScriptConfiguration.java
@@ -52,6 +52,8 @@ public Options getOptions() {
                                     .build();
         options.addOption(pluginOption);
 
+        options.addOption("d", "fromdate", true, "Process only item from specified last modified date");
+
         Option skipOption = Option.builder("s")
                                   .longOpt("skip")
                                   .hasArg()

diff --git a/dspace-api/src/main/java/org/dspace/app/mediafilter/MediaFilterServiceImpl.java b/dspace-api/src/main/java/org/dspace/app/mediafilter/MediaFilterServiceImpl.java
@@ -9,8 +9,11 @@
 
 import java.io.InputStream;
 import java.sql.SQLException;
+import java.time.LocalDate;
+import java.time.ZoneId;
 import java.util.ArrayList;
 import java.util.Collections;
+import java.util.Date;
 import java.util.HashMap;
 import java.util.Iterator;
 import java.util.List;
@@ -93,6 +96,7 @@ public class MediaFilterServiceImpl implements MediaFilterService, InitializingB
     protected boolean isVerbose = false;
     protected boolean isQuiet = false;
     protected boolean isForce = false; // default to not forced
+    protected LocalDate fromDate = null;
 
     protected MediaFilterServiceImpl() {
 
@@ -120,6 +124,15 @@ public void applyFiltersAllItems(Context context) throws Exception {
             for (Community topLevelCommunity : topLevelCommunities) {
                 applyFiltersCommunity(context, topLevelCommunity);
             }
+        } else if (fromDate != null) {
+            Iterator<Item> itemIterator =
+                    itemService.findByLastModifiedSince(
+                            context,
+                            Date.from(fromDate.atStartOfDay(ZoneId.systemDefault()).toInstant())
+                    );
+            while (itemIterator.hasNext() && processed < max2Process) {
+                applyFiltersItem(context, itemIterator.next());
+            }
         } else {
             //otherwise, just find every item and process
             Iterator<Item> itemIterator = itemService.findAll(context);
@@ -588,4 +601,9 @@ public void setFilterFormats(Map<String, List<String>> filterFormats) {
     public void setLogHandler(DSpaceRunnableHandler handler) {
         this.handler = handler;
     }
+
+    @Override
+    public void setFromDate(LocalDate fromDate) {
+        this.fromDate = fromDate;
+    }
 }
diff --git a/dspace-api/src/main/java/org/dspace/app/mediafilter/service/MediaFilterService.java b/dspace-api/src/main/java/org/dspace/app/mediafilter/service/MediaFilterService.java
@@ -8,6 +8,7 @@
 package org.dspace.app.mediafilter.service;
 
 import java.sql.SQLException;
+import java.time.LocalDate;
 import java.util.List;
 import java.util.Map;
 
@@ -149,4 +150,6 @@ public void updatePoliciesOfDerivativeBitstreams(Context context, Item item, Bit
      * @param handler
      */
     public void setLogHandler(DSpaceRunnableHandler handler);
+
+    public void setFromDate(LocalDate fromDate);
 }
diff --git a/dspace-api/src/main/java/org/dspace/content/dao/impl/ItemDAOImpl.java b/dspace-api/src/main/java/org/dspace/content/dao/impl/ItemDAOImpl.java
@@ -441,7 +441,7 @@ public int countItems(Context context, List<Collection> collections, boolean inc
     public Iterator<Item> findByLastModifiedSince(Context context, Date since)
         throws SQLException {
         Query query = createQuery(context,
-                "SELECT i.id FROM Item i WHERE last_modified > :last_modified ORDER BY id");
+                "SELECT i.id FROM Item i WHERE lastModified > :last_modified ORDER BY id");
         query.setParameter("last_modified", since, TemporalType.TIMESTAMP);
         @SuppressWarnings("unchecked")
         List<UUID> uuids = query.getResultList();