From f40fe07254d297173d8f3ae9666ea8226b22f94b Mon Sep 17 00:00:00 2001
From: mathetake <cz.rk.t0415y.g@gmail.com>
Date: Fri, 19 Apr 2019 13:26:45 +0900
Subject: [PATCH] fix readme

---
 README.md         | 45 ++++++++++++++++++++++++++++++---------------
 metrics/cosine.go |  2 +-
 2 files changed, 31 insertions(+), 16 deletions(-)

diff --git a/README.md b/README.md
index 0b37e3c..ab42d36 100644
--- a/README.md
+++ b/README.md
@@ -15,38 +15,53 @@ The implemented algorithm is truly inspired by Annoy (https://github.com/spotify
 ## usage
 
 ```golang
+package main
+
 import (
 	"fmt"
 	"math/rand"
+	"testing"
 	"time"
-	
-	"github.com/mathetake/gann/index"
+
+	"github.com/mathetake/gann"
+	"github.com/mathetake/gann/metrics"
 )
 
-func main() {
-	var dim = 3
-	var nTrees = 2
-	var k = 10
-	var nItem = 1000
+var (
+	dim    = 3
+	nTrees = 2
+	k      = 10
+	nItem  = 1000
+)
 
-	rawItems := make([][]float32, 0, nItem)
+func main(t *testing.T) {
+	rawItems := make([][]float64, 0, nItem)
 	rand.Seed(time.Now().UnixNano())
 
 	for i := 0; i < nItem; i++ {
-		item := make([]float32, 0, dim)
+		item := make([]float64, 0, dim)
 		for j := 0; j < dim; j++ {
-			item = append(item, rand.Float32())
+			item = append(item, rand.Float64())
 		}
 		rawItems = append(rawItems, item)
 	}
 
 	// create index
-	gIDx := index.GetIndex(rawItems, dim, nTrees, k, true)
-	gIDx.Build()
+	idx, err := gann.CreateNewIndex(rawItems, dim, nTrees, k, metrics.TypeCosineDistance)
+	if err != nil {
+		// error handling
+		return
+	}
+
+	// search
+	q := []float64{0.1, 0.02, 0.001}
+	res, err := idx.GetANNbyVector(q, 5, 10)
+	if err != nil {
+		// error handling
+		return
+	}
 
-	// do search
-	q := []float32{0.1, 0.02, 0.001}
-	ann, _ := gIDx.GetANNbyVector(q, 5, 10)
+	fmt.Printf("res: %v\n", res)
 }
 ```
 
diff --git a/metrics/cosine.go b/metrics/cosine.go
index c329606..71dc153 100644
--- a/metrics/cosine.go
+++ b/metrics/cosine.go
@@ -90,7 +90,7 @@ func (c *cosineDistance) GetSplittingVector(vs [][]float64) []float64 {
 
 		c1 = make([]float64, c.dim)
 		it1 := int(float64(lvs)*cosineMetricsCentroidCalcRatio + 1)
-		for i := 0; i < int(float32(lc1)*cosineMetricsCentroidCalcRatio+1); i++ {
+		for i := 0; i < int(float64(lc1)*cosineMetricsCentroidCalcRatio+1); i++ {
 			for d := 0; d < c.dim; d++ {
 				c1[d] += clusterToVecs[1][rand.Intn(lc1)][d] / float64(it1)
 			}