Fixed a bug in update stream generation which was deleting some updatestream files

Author: ArnauPrat

src/main/java/ldbc/snb/datagen/generator/LDBCDatagen.java

@@ -162,19 +162,19 @@ public int runGenerateJob(Configuration conf) throws Exception {
             List<String> forumStreamsFileNames = new ArrayList<String>();
             for( int i = 0; i < DatagenParams.numThreads; ++i) {
                 int numPartitions = conf.getInt("ldbc.snb.datagen.serializer.numUpdatePartitions", 1);
-                if( i < numBlocks ) {
+                //if( i < numBlocks ) {
                     for (int j = 0; j < numPartitions; ++j) {
                         personStreamsFileNames.add(DatagenParams.hadoopDir + "/temp_updateStream_person_" + i + "_" + j);
                         if( conf.getBoolean("ldbc.snb.datagen.generator.activity", false)) {
                             forumStreamsFileNames.add(DatagenParams.hadoopDir + "/temp_updateStream_forum_" + i + "_" + j);
                         }
                     }
-                } else {
+                /*} else {
                     for (int j = 0; j < numPartitions; ++j) {
                         fs.delete(new Path(DatagenParams.hadoopDir + "/temp_updateStream_person_" + i + "_" + j), true);
                         fs.delete(new Path(DatagenParams.hadoopDir + "/temp_updateStream_forum_" + i + "_" + j), true);
                     }
-                }
+                } */
             }
             HadoopUpdateStreamSorterAndSerializer updateSorterAndSerializer = new HadoopUpdateStreamSorterAndSerializer(conf);
             updateSorterAndSerializer.run(personStreamsFileNames, "person");

src/main/java/ldbc/snb/datagen/objects/Person.java

@@ -13,6 +13,10 @@
  */
 public class Person implements Writable {
 
+	public static interface PersonSimilarity {
+		public double Similarity(Person personA, Person personB);
+	}
+
 	public static class PersonSummary implements Writable {
 		private long accountId_;
 		private long creationDate_;
@@ -462,10 +466,9 @@ public void write(DataOutput arg0) throws IOException {
 		arg0.writeLong(classYear_);
 	}
 
+	public static PersonSimilarity personSimilarity;
+
 	public static float Similarity(Person personA, Person personB) {
-		int zorderA = Dictionaries.places.getZorderID(personA.countryId());
-		int zorderB = Dictionaries.places.getZorderID(personB.countryId());
-		return 1.0f - (Math.abs(zorderA - zorderB) / 256.0f);
 	}
 
 }

src/main/java/ldbc/snb/datagen/objects/similarity/GeoDistanceSimilarity.java

@@ -0,0 +1,15 @@
+package ldbc.snb.datagen.objects.similarity;
+
+import ldbc.snb.datagen.dictionary.Dictionaries;
+import ldbc.snb.datagen.objects.Person;
+
+/**
+ * Created by aprat on 22/01/16.
+ */
+public class GeoDistanceSimilarity implements Person.PersonSimilarity{
+    public double Similarity(Person personA, Person personB) {
+        int zorderA = Dictionaries.places.getZorderID(personA.countryId());
+        int zorderB = Dictionaries.places.getZorderID(personB.countryId());
+        return 1.0f - (Math.abs(zorderA - zorderB) / 256.0f);
+    }
+}

src/main/java/ldbc/snb/datagen/objects/similarity/InterestsSimilarity.java

@@ -0,0 +1,12 @@
+package ldbc.snb.datagen.objects.similarity;
+
+import ldbc.snb.datagen.objects.Person;
+
+/**
+ * Created by aprat on 22/01/16.
+ */
+public class InterestsSimilarity implements Person.PersonSimilarity {
+    public double Similarity(Person personA, Person personB) {
+        return 0;
+    }
+}