Merge pull request #16 from ubbdst/feature/replace-uri

Feature/replace uri

Author: hemed

pom.xml

@@ -3,7 +3,7 @@
     <modelVersion>4.0.0</modelVersion>
     <groupId>org.elasticsearch.plugin.river.ubb</groupId>
     <artifactId>ubb-rdf-river-plugin</artifactId>
-    <version>1.7.7</version>
+    <version>1.7.9</version>
 
     <packaging>jar</packaging>
 

src/main/java/org/elasticsearch/river/ubb/RDFRiver.java

@@ -109,6 +109,8 @@ private void buildHarvester(RiverSettings settings) {
                         rdfSettings.get("tdbLocation"), ""))
                 .rdfQueryPath(XContentMapValues.nodeStringValue(
                         rdfSettings.get("queryPath"), ""))
+                .replaceResourceURI(XContentMapValues.nodeStringValue(
+                        rdfSettings.get("replaceResourceURI"), ""))
                 .rdfNumberOfBulkActions(XContentMapValues.nodeLongValue(
                         rdfSettings.get("bulkActions"),
                         Defaults.DEFAULT_NUMBER_OF_BULK_ACTIONS))
@@ -193,6 +195,13 @@ private void buildHarvester(RiverSettings settings) {
         if (rdfSettings.containsKey("whiteMap")) {
             harvester.rdfWhiteMap(getStrObjMapFromSettings(rdfSettings, "whiteMap"));
         }
+        if (rdfSettings.containsKey("indexTextFrom")) {
+            harvester.textField(XContentMapValues.nodeStringValue(rdfSettings.get("indexTextFrom"), ""));
+        }
+        if (rdfSettings.containsKey("embedResourceUsingProperty")) {
+            harvester.embedResource(XContentMapValues.nodeStringValue(
+                    rdfSettings.get("embedResourceUsingProperty"), ""));
+        }
         if (settings.settings().containsKey("index")) {
             Map<String, Object> indexSettings = extractSettings(settings, "index");
             harvester.index(XContentMapValues.nodeStringValue(indexSettings.get("index"),

src/main/java/org/elasticsearch/river/ubb/settings/Defaults.java

@@ -39,7 +39,7 @@ public final class Defaults {
             "\"http://www.w3.org/1999/02/22-rdf-syntax-ns#about\"" + "]";
     public final static String DEFAULT_RESOURCE_URI = "http://www.w3.org/1999/02/22-rdf-syntax-ns#about";
     public final static String DEFAULT_LIST_TYPE = "white";
-    public final static Boolean DEFAULT_ADD_LANGUAGE = true;
+    public final static Boolean DEFAULT_ADD_LANGUAGE = false;
     public final static String DEFAULT_LANGUAGE = "no";
     public final static Boolean DEFAULT_ADD_URI = true;
     public final static Boolean DEFAULT_UPDATE_DOCUMENTS = false;

src/main/java/org/elasticsearch/river/ubb/settings/RiverUtils.java

@@ -1,5 +1,8 @@
 package org.elasticsearch.river.ubb.settings;
 
+import org.elasticsearch.common.Strings;
+import org.elasticsearch.common.lang3.StringUtils;
+
 import java.text.DecimalFormat;
 import java.util.regex.Pattern;
 
@@ -86,16 +89,16 @@ public static boolean isNullOrEmpty(CharSequence s) {
      */
     public static String getTimeString(long timeInMilliSeconds) {
         //Time in seconds
-        double timeInSeconds = timeInMilliSeconds/1000.0;
+        double timeInSeconds = timeInMilliSeconds / 1000.0;
         //Format to 2 decimal places
         DecimalFormat df = new DecimalFormat(".##");
         //In minutes
         if (timeInSeconds >= 60 && timeInSeconds < 60 * 60) {
-            return df.format(timeInSeconds/60 )+ " minutes";
+            return df.format(timeInSeconds / 60) + " minutes";
         }
         //In hours
         if (timeInSeconds >= 60 * 60 && timeInSeconds < 24 * 3600) {
-            return df.format(timeInSeconds/3600) + " hours";
+            return df.format(timeInSeconds / 3600) + " hours";
         }
         //default unit
         return df.format(timeInSeconds) + " seconds";
@@ -126,4 +129,24 @@ public static String removeIllegalXMLChar(String text) {
         invalidXMLChars.matcher(text).replaceAll("");
         return text;
     }
+
+    /**
+     * Replaces one resource URI based on the list of comma separated fragments
+     *
+     * @param resourceUri             a resource URI to be replaced
+     * @param fragmentsCommaSeparated a list of two comma separated fragments. If more than 2 fragments are found,
+     *                                only the first 2 will be considered.
+     */
+    public static String replaceResourceURI(String resourceUri, String fragmentsCommaSeparated) {
+        if (Strings.hasText(resourceUri)) {
+            String[] frags = Strings.splitStringByCommaToArray(fragmentsCommaSeparated);
+            if (frags != null && frags.length >= 2) { // only if we have something to replace
+                return Strings.replace(resourceUri,
+                        StringUtils.deleteWhitespace(frags[0]),
+                        StringUtils.deleteWhitespace(frags[1]));
+            }
+        }
+        return resourceUri;
+    }
+
 }

src/main/java/org/elasticsearch/river/ubb/support/FlatContextTransformer.java

@@ -94,7 +94,7 @@ public Map<String, String> transform(JsonElement context) {
      * Main method for easy debugging
      */
     public static void main(String[] args) {
-        String s = FileManager.read("http://data.ub.uib.no/momayo/context.json", 20);
+        String s = FileManager.readAsUTF8("http://data.ub.uib.no/momayo/context.json", 20);
         Map<String, String> props = ContextFactory.flatContext().transform(s);
         System.out.println(new GsonBuilder()
                 .setPrettyPrinting()