Fixed several bugs with like generation

Author: ArnauPrat

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

@@ -74,8 +74,8 @@ private void generateWall( Person person, ArrayList<Person> block ) {
 		for( Post p : wallPosts ) {
 			export(p);
 			// generate likes to post
-			ArrayList<Like> postLikes = likeGenerator_.generateLikes(randomFarm_.get(RandomGeneratorFarm.Aspect.NUM_LIKE), wall, p, Like.LikeType.POST);
 			if( randomFarm_.get(RandomGeneratorFarm.Aspect.NUM_LIKE).nextDouble() <= 0.1 ) {
+                ArrayList<Like> postLikes = likeGenerator_.generateLikes(randomFarm_.get(RandomGeneratorFarm.Aspect.NUM_LIKE), wall, p, Like.LikeType.POST);
 				for( Like l : postLikes ) {
 					export(l);
 				}
@@ -134,7 +134,7 @@ private void generateGroups( Person person, ArrayList<Person> block ) {
 						export(c);
 						// generate likes to comments
 						if( c.content().length() > 10 && randomFarm_.get(RandomGeneratorFarm.Aspect.NUM_LIKE).nextDouble() <= 0.1 ) {
-							ArrayList<Like> commentLikes = likeGenerator_.generateLikes(randomFarm_.get(RandomGeneratorFarm.Aspect.NUM_LIKE), group, p, Like.LikeType.POST);
+							ArrayList<Like> commentLikes = likeGenerator_.generateLikes(randomFarm_.get(RandomGeneratorFarm.Aspect.NUM_LIKE), group, c, Like.LikeType.COMMENT);
 
 							for( Like l : commentLikes ) {
 								export(l);

tools/validatePairUniqueness.py

@@ -0,0 +1,32 @@
+
+import sys, os
+
+
+if( len(sys.argv) < 4):
+    print("Validates that a pair of columns never appear repeated.")
+    print("Usage: validateIdUniqueness <coulmn1> <column2> <file0> <file1> ... <filen>")
+
+column1=int(sys.argv[1])
+column2=int(sys.argv[2])
+
+ids = {} 
+
+for i in range(3,len(sys.argv)):
+    print("Reading "+sys.argv[i])
+    inputFile = open(sys.argv[i],'r')
+    index = 0
+    for line in inputFile.readlines():
+        if index > 0:
+            firstId = int((line.split('|'))[column1])
+            secondId = int((line.split('|'))[column2])
+            if firstId not in ids:
+              ids[firstId] = set([])
+            s = ids[firstId]
+            if secondId in s:
+              print("ERROR, Id pair not unique")
+              print(str(firstId)+" "+str(secondId))
+              exit(1)
+            s.add(secondId)
+        index+=1
+    inputFile.close()
+print("GREAT! All ids are different.")