#36 Add ability to change the number of hash functions

Author: justcoding121

src/Advanced.Algorithms/DataStructures/Set/BloomFilter.cs

@@ -1,5 +1,6 @@
 using System;
 using System.Collections;
+using System.Collections.Generic;
 
 namespace Advanced.Algorithms.DataStructures
 {
@@ -10,12 +11,20 @@ public class BloomFilter<T>
     {
         private readonly BitArray filter;
 
+        private readonly int numberOfHashFunctions;
+
         /// <summary>
         /// Higher the size lower the collision and 
         /// failure probablity.
         /// </summary>
-        public BloomFilter(int size)
+        public BloomFilter(int size, int numberOfHashFunctions = 2)
         {
+            if (size <= numberOfHashFunctions)
+            {
+                throw new ArgumentException("size cannot be less than or equal to numberOfHashFunctions.");
+            }
+
+            this.numberOfHashFunctions = numberOfHashFunctions;
             filter = new BitArray(size);
         }
 
@@ -24,13 +33,9 @@ public BloomFilter(int size)
         /// </summary>
         public void AddKey(T key)
         {
-            var hashCode = key.GetHashCode();
-
-            //set 8 consecutive bits (a byte)
-            for (var i = 0; i < 8; i++)
+            foreach (var hash in getHashes(key))
             {
-                var index = Math.Abs(hashCode + i) % filter.Length;
-                filter[index] = true;
+                filter[hash % filter.Length] = true;
             }
         }
 
@@ -39,21 +44,24 @@ public void AddKey(T key)
         /// </summary>
         public bool KeyExists(T key)
         {
-            var hashCode = key.GetHashCode();
-
-
-            //set 8 consecutive bits (a byte)
-            for (var i = 0; i < 8; i++)
+            foreach (var hash in getHashes(key))
             {
-                var index = Math.Abs(hashCode + i) % filter.Length;
-
-                if (filter[index] == false)
+                if (filter[hash % filter.Length] == false)
                 {
                     return false;
                 }
             }
 
             return true;
         }
+
+        private IEnumerable<int> getHashes(T key)
+        {
+            for (var i = 1; i <= numberOfHashFunctions; i++)
+            {
+                var obj = new { Key = key, InitialValue = i };
+                yield return Math.Abs(obj.GetHashCode());
+            }
+        }
     }
 }

tests/Advanced.Algorithms.Tests/DataStructures/Set/BloomFilter_Tests.cs

@@ -18,5 +18,32 @@ public void BloomFilter_Smoke_Test()
             Assert.IsTrue(filter.KeyExists("cat"));
             Assert.IsFalse(filter.KeyExists("bat"));
         }
+
+        [TestMethod]
+        public void BloomFilter_Accuracy_Test()
+        {
+            var bloomFilter = new BloomFilter<string>(10);
+
+            bloomFilter.AddKey("foo");
+            bloomFilter.AddKey("bar");
+            bloomFilter.AddKey("apple");
+            bloomFilter.AddKey("orange");
+            bloomFilter.AddKey("banana");
+
+            Assert.IsTrue(bloomFilter.KeyExists("bar"));
+            Assert.IsFalse(bloomFilter.KeyExists("ba111r"));
+
+            Assert.IsTrue(bloomFilter.KeyExists("banana"));
+            Assert.IsFalse(bloomFilter.KeyExists("dfs11j"));
+
+            Assert.IsTrue(bloomFilter.KeyExists("foo"));
+            Assert.IsFalse(bloomFilter.KeyExists("1foo"));
+
+            Assert.IsTrue(bloomFilter.KeyExists("apple"));
+            Assert.IsFalse(bloomFilter.KeyExists("applefoo"));
+
+            Assert.IsTrue(bloomFilter.KeyExists("orange"));
+            Assert.IsFalse(bloomFilter.KeyExists("orangew"));
+        }
     }
 }