extract newCapacity

Author: dsa0x

bloom.go

@@ -49,7 +49,7 @@ func NewBloom(err_rate float64, capacity int, database Store) *BloomFilter {
 	if err_rate <= 0 || err_rate >= 1 {
 		panic("Error rate must be between 0 and 1")
 	}
-	if capacity < 0 {
+	if capacity <= 0 {
 		panic("Capacity must be greater than 0")
 	}
 
@@ -62,13 +62,12 @@ func NewBloom(err_rate float64, capacity int, database Store) *BloomFilter {
 	ln22 := math.Pow(math.Ln2, 2)
 
 	// M
-	bit_width := int(math.Ceil((float64(capacity) * math.Abs(math.Log(err_rate))) /
-		ln22))
+	bit_width := int((float64(capacity) * math.Abs(math.Log(err_rate)) / ln22))
+
 	//m
 	bits_per_slice := bit_width / numHashFn
 
 	seeds := make([]int64, numHashFn)
-
 	for i := 0; i < len(seeds); i++ {
 		seeds[i] = int64((i + 1) << 16)
 	}

cmd/main.go

@@ -2,6 +2,7 @@ package main
 
 import (
 	"encoding/binary"
+	"fmt"
 
 	"github.com/dsa0x/gobloomgo"
 )
@@ -17,17 +18,18 @@ func main() {
 	// opts := badger.DefaultOptions("/tmp/bloom.db")
 	// db := gobloomgo.NewBadger(opts)
 
+	num := 18232
 	// bf := gobloomgo.NewBloom(0.1, 50000, nil)
-	bf := gobloomgo.NewScalableBloom(0.000001, 10, nil)
+	bf := gobloomgo.NewScalableBloom(0.001, num, nil)
 
-	for i := 0; i < 100000; i++ {
+	for i := 0; i < num; i++ {
 		var by [4]byte
 		binary.LittleEndian.PutUint32(by[:], uint32(i))
 		bf.Add(by[:], []byte("bar"))
 	}
 	bf.Add([]byte("foo"), []byte("var"))
 
-	// fmt.Printf("Count: %d, Capacity: %d, ExpCap: %.f\n", bf.Count(), bf.Capacity(), bf.ExpCapacity())
+	fmt.Printf("Count: %d, Capacity: %d,\n", bf.Count(), bf.Capacity())
 	// fmt.Println(bf.Capacity(), bf.ExpCapacity(), bf.Count(), bf.Prob())
 
 }

scalable_bloom.go

@@ -97,10 +97,15 @@ func (sbf *ScalableBloomFilter) Top() *BloomFilter {
 // grow increases the capacity of the bloom filter by adding a new filter
 func (sbf *ScalableBloomFilter) grow() {
 	err_rate := sbf.err_rate * math.Pow(sbf.ratio, float64(len(sbf.filters)))
+	newCapacity := sbf.getNewCap()
+	newFilter := NewBloom(err_rate, newCapacity, sbf.db)
+	sbf.filters = append(sbf.filters, newFilter)
+}
+
+func (sbf *ScalableBloomFilter) getNewCap() int {
 	i := float64(len(sbf.filters)) - 1.0
 	newCapacity := float64(sbf.m0) * float64(math.Pow(float64(sbf.growth_rate), i)) * math.Ln2
-	newFilter := NewBloom(err_rate, int(newCapacity), sbf.db)
-	sbf.filters = append(sbf.filters, newFilter)
+	return int(newCapacity)
 }
 
 // Size returns the total capacity of the scalable bloom filter