Add an extra index to the metadata (distinct) tables for better scalability.

Add option for 'case-insensitive' metadata browse tables, as a way to avoid the problems of duplicate entries.

git-svn-id: http://scm.dspace.org/svn/repo/branches/dspace-1_5_x@3216 9c30dcfa-912a-0410-8fc2-9e0234be79fd

dspace-api/src/main/java/org/dspace/browse/BrowseCreateDAOOracle.java

@@ -47,6 +47,7 @@ import java.util.Map;
 
 import org.apache.log4j.Logger;
 import org.dspace.core.Context;
+import org.dspace.core.ConfigurationManager;
 import org.dspace.storage.rdbms.DatabaseManager;
 import org.dspace.storage.rdbms.TableRow;
 import org.dspace.storage.rdbms.TableRowIterator;
@@ -193,10 +194,12 @@ public class BrowseCreateDAOOracle implements BrowseCreateDAO
     {
         try
         {
-            String[] arr = new String[3];
+            String[] arr = new String[5];
-            arr[0] = "CREATE INDEX " + disTable + "_value_idx ON " + disTable + "(sort_value)";
+            arr[0] = "CREATE INDEX " + disTable + "_svalue_idx ON " + disTable + "(sort_value)";
-            arr[1] = "CREATE INDEX " + mapTable + "_item_id_idx ON " + mapTable + "(item_id)";
+            arr[1] = "CREATE INDEX " + disTable + "_value_idx ON " + disTable + "(value)";
-            arr[2] = "CREATE INDEX " + mapTable + "_dist_idx ON " + mapTable + "(distinct_id)";
+            arr[2] = "CREATE INDEX " + disTable + "_uvalue_idx ON " + disTable + "(UPPER(value))";
+            arr[3] = "CREATE INDEX " + mapTable + "_item_id_idx ON " + mapTable + "(item_id)";
+            arr[4] = "CREATE INDEX " + mapTable + "_dist_idx ON " + mapTable + "(distinct_id)";
             
             if (execute)
             {
@@ -520,10 +523,20 @@ public class BrowseCreateDAOOracle implements BrowseCreateDAO
             Object[] params = { value };
             String select = "SELECT id FROM " + table;
             
-            if (isValueColumnClob())
+            if (ConfigurationManager.getBooleanProperty("webui.browse.metadata.case-insensitive", false))
-                select = select + " WHERE TO_CHAR(value)=?";
+            {
+                if (isValueColumnClob())
+                    select = select + " WHERE UPPER(TO_CHAR(value))=UPPER(?)";
+                else
+                    select = select + " WHERE UPPER(value)=UPPER(?)";
+            }
             else
-                select = select + " WHERE value=?";
+            {
+                if (isValueColumnClob())
+                    select = select + " WHERE TO_CHAR(value)=?";
+                else
+                    select = select + " WHERE value=?";
+            }
                
             tri = DatabaseManager.query(context, select, params);
             int distinctID = -1;

dspace-api/src/main/java/org/dspace/browse/BrowseCreateDAOPostgres.java

@@ -47,6 +47,7 @@ import java.util.Map;
 
 import org.apache.log4j.Logger;
 import org.dspace.core.Context;
+import org.dspace.core.ConfigurationManager;
 import org.dspace.storage.rdbms.DatabaseManager;
 import org.dspace.storage.rdbms.TableRow;
 import org.dspace.storage.rdbms.TableRowIterator;
@@ -189,10 +190,12 @@ public class BrowseCreateDAOPostgres implements BrowseCreateDAO
     {
         try
         {
-            String[] arr = new String[3];
+            String[] arr = new String[5];
-            arr[0] = "CREATE INDEX " + disTable + "_value_idx ON " + disTable + "(sort_value);";
+            arr[0] = "CREATE INDEX " + disTable + "_svalue_idx ON " + disTable + "(sort_value)";
-            arr[1] = "CREATE INDEX " + mapTable + "_item_id_idx ON " + mapTable + "(item_id);";
+            arr[1] = "CREATE INDEX " + disTable + "_value_idx ON " + disTable + "(value)";
-            arr[2] = "CREATE INDEX " + mapTable + "_dist_idx ON " + mapTable + "(distinct_id);";
+            arr[2] = "CREATE INDEX " + disTable + "_uvalue_idx ON " + disTable + "(UPPER(value))";
+            arr[3] = "CREATE INDEX " + mapTable + "_item_id_idx ON " + mapTable + "(item_id)";
+            arr[4] = "CREATE INDEX " + mapTable + "_dist_idx ON " + mapTable + "(distinct_id)";
             
             if (execute)
             {
@@ -522,7 +525,17 @@ public class BrowseCreateDAOPostgres implements BrowseCreateDAO
         try
         {
             Object[] params = { value };
-            String select = "SELECT id FROM " + table + " WHERE value = ?";
+            String select = null;
+
+            if (ConfigurationManager.getBooleanProperty("webui.browse.metadata.case-insensitive", false))
+            {
+                select = "SELECT id FROM " + table + " WHERE UPPER(value) = UPPER(?)";
+            }
+            else
+            {
+                select = "SELECT id FROM " + table + " WHERE value = ?";
+            }
+
             tri = DatabaseManager.query(context, select, params);
             int distinctID = -1;
             if (!tri.hasNext())

dspace/config/dspace.cfg

@@ -817,6 +817,24 @@ webui.itemlist.sort-option.1 = title:dc.title:title
 webui.itemlist.sort-option.2 = dateissued:dc.date.issued:date
 webui.itemlist.sort-option.3 = dateaccessioned:dc.date.accessioned:date
 
+# By default, the display of metadata in the browse indexes is case sensitive
+# So, you will get seperate entries for the terms
+#
+#   Olive oil
+#   olive oil
+#
+# However, clicking through from either of these will result in the same set of items
+# (ie. any item that contains either representation in the correct field).
+#
+# Uncommenting the option below will make the metadata items case-insensitive. This will
+# result in a single entry in the example above. However the value displayed may be either 'Olive oil'
+# or 'olive oil' - depending on what representation was present in the first item indexed.
+#
+# If you care about the display of the metadata in the browse index - well, you'll have to go and
+# fix the metadata in your items.
+#
+# webui.browse.metadata.case-insensitive = true
+
 # Set the options for the size (number of characters) of the fields stored in the database.
 #
 # The default is 0, which is unlimited size for fields holding indexed data.  Some