HyperBloomBloom fix

ZeroLevel.UnitTests/BloomFilterTest.cs

@@ -55,7 +55,7 @@ namespace ZeroLevel.UnitTests
                 {
                     if (false == lines.Contains(line))
                     {
-                        collision_count++;                        
+                        collision_count++;
                     }
                 }
             }
@@ -67,7 +67,7 @@ namespace ZeroLevel.UnitTests
         public void HyperBloomBloomFilterTest()
         {
             // Arrange
-            var size = 100000;
+            var size = 1000000;
             var lines = new HashSet<string>(size);
             var lines_another = new HashSet<string>(size);
             for (int i = 0; i < size; i++)
@@ -75,7 +75,7 @@ namespace ZeroLevel.UnitTests
                 lines.Add(RandomString(i % 9 + 5));
                 lines_another.Add(RandomString(i % 9 + 5));
             }
-            var bloom = new HyperBloomBloom(16536 * 1024, true);
+            var bloom = new HyperBloomBloom(64, 16536 * 1024, true);
             // Act
             var sw = new Stopwatch();
             sw.Start();

ZeroLevel/DataStructures/HyperBloomBloom.cs

@@ -1,38 +1,33 @@
-using System;
+using ZeroLevel.Services.HashFunctions;
-using System.Collections.Generic;
 
 namespace ZeroLevel.DataStructures
 {
     public class HyperBloomBloom
     {
-        private BloomFilter _trash;
+        private IHash _shardHash = new XXHashUnsafe();
-        private Dictionary<char, BloomFilter> _shardes = new Dictionary<char, BloomFilter>();
+        private BloomFilter[] _shardes;
 
-        public HyperBloomBloom(int bit_size, bool use_reverse)
+        public HyperBloomBloom(int shardes_size, int bit_size, bool use_reverse)
         {
-            _trash = new BloomFilter(bit_size, use_reverse);
+            _shardes = new BloomFilter[shardes_size];
-            foreach (var ch in "abcdefghijklmnopqrstuvwxyz0123456789-")
+            for (int i = 0; i < shardes_size; i++)
             {
-                _shardes.Add(ch, new BloomFilter(bit_size, use_reverse));
+                _shardes[i] = new BloomFilter(bit_size, use_reverse);
             }
         }
 
         public void Add(string item)
         {
             if (item == null || item.Length == 0) return;
-            var k = Char.ToLowerInvariant(item[0]);
+            int index = (int)(_shardHash.Hash(item) % _shardes.Length);
-            BloomFilter filter;
+            _shardes[index].Add(item);
-            if (_shardes.TryGetValue(k, out filter) == false) filter = _trash;
-            filter.Add(item);
         }
 
         public bool Contains(string item)
         {
             if (item == null || item.Length == 0) return true;
-            var k = Char.ToLowerInvariant(item[0]);
+            int index = (int)(_shardHash.Hash(item) % _shardes.Length);
-            BloomFilter filter;
+            return _shardes[index].Contains(item);
-            if (_shardes.TryGetValue(k, out filter) == false) filter = _trash;
-            return filter.Contains(item);
         }
         /// <summary>
         /// true if added, false if already exists
@@ -40,10 +35,8 @@ namespace ZeroLevel.DataStructures
         public bool TryAdd(string item)
         {
             if (item == null || item.Length == 0) return false;
-            var k = Char.ToLowerInvariant(item[0]);
+            int index = (int)(_shardHash.Hash(item) % _shardes.Length);
-            BloomFilter filter;
+            return _shardes[index].TryAdd(item);
-            if (_shardes.TryGetValue(k, out filter) == false) filter = _trash;
-            return filter.TryAdd(item);
         }
     }
 }