From d91f2b03e9acd1c69bec415f2840619890a53a3e Mon Sep 17 00:00:00 2001
From: Akshay Vadher <akshay@incubyte.co>
Date: Fri, 19 Jul 2024 19:44:52 +0530
Subject: [PATCH 1/2] Added histogram test

---
 cuid2_collision_test.go | 14 +++++--
 cuid2_histogram_test.go | 83 +++++++++++++++++++++++++++++++++++++++++
 2 files changed, 94 insertions(+), 3 deletions(-)
 create mode 100644 cuid2_histogram_test.go

diff --git a/cuid2_collision_test.go b/cuid2_collision_test.go
index 3d3ffb9..f90ab29 100644
--- a/cuid2_collision_test.go
+++ b/cuid2_collision_test.go
@@ -32,6 +32,11 @@ func TestCollision(t *testing.T) {
 		checkHistogram(t, n/numPools, v.Histogram)
 		m.Unlock()
 	}
+	CheckCollision(t, ids)
+	fmt.Printf("Sample ids %v\n", ids[:10])
+}
+
+func CheckCollision(t *testing.T, ids []string) {
 	set := make(map[string]struct{}, len(ids))
 	for _, id := range ids {
 		set[id] = struct{}{}
@@ -39,7 +44,6 @@ func TestCollision(t *testing.T) {
 	if len(set) < len(ids) {
 		t.Errorf("Collision detected. len(set) %d, len(ids) %d", len(set), len(ids))
 	}
-	fmt.Printf("Sample ids %v\n", ids[:10])
 }
 
 func checkHistogram(t *testing.T, numberOfIds int, histogram []int64) {
@@ -63,7 +67,11 @@ type IdPoolResponse struct {
 }
 
 func createIdPool(t *testing.T, max int, poolId int, idPoolResponseChan chan *IdPoolResponse, wg *sync.WaitGroup) {
-	defer wg.Done()
+	idPoolResponseChan <- CreateIdPool(t, max, poolId)
+	wg.Done()
+}
+
+func CreateIdPool(t *testing.T, max int, poolId int) *IdPoolResponse {
 	set := make(map[string]struct{}, max)
 	for i := 0; i < max; i++ {
 		id := CreateId()
@@ -94,7 +102,7 @@ func createIdPool(t *testing.T, max int, poolId int, idPoolResponseChan chan *Id
 	bucketCount := 20
 	histogram := buildHistogram(numbers, bucketCount)
 	fmt.Printf("Histogram created for pool %d\n", poolId)
-	idPoolResponseChan <- &IdPoolResponse{
+	return &IdPoolResponse{
 		Ids:       ids,
 		Numbers:   numbers,
 		Histogram: histogram,
diff --git a/cuid2_histogram_test.go b/cuid2_histogram_test.go
new file mode 100644
index 0000000..9936304
--- /dev/null
+++ b/cuid2_histogram_test.go
@@ -0,0 +1,83 @@
+package cuid2
+
+import (
+	"fmt"
+	"math"
+	"math/rand/v2"
+	"strings"
+	"testing"
+)
+
+func TestHistogram(t *testing.T) {
+	n := 100000
+	fmt.Printf("Testing %d unique ids\n", n)
+	poolId := rand.IntN(100)
+	poolResponse := CreateIdPool(t, n, poolId)
+	ids := poolResponse.Ids
+	sampleIds := ids[:10]
+	fmt.Printf("Sample ids %v\n", sampleIds)
+	t.Run("Test collision", func(t *testing.T) {
+		CheckCollision(t, ids)
+	})
+	t.Run("Test char frequency", func(t *testing.T) {
+		testCharFrequency(t, n, ids)
+	})
+	t.Run("Test histogram", func(t *testing.T) {
+		testHistogram(t, poolResponse, n)
+	})
+
+}
+
+func testCharFrequency(t *testing.T, n int, ids []string) {
+	tolerance := 0.1
+	idLength := 23
+	totalLetters := idLength * n
+	base := 36
+	expectedBinSize := math.Ceil(float64(totalLetters) / float64(base))
+	minBinSize := math.Round(expectedBinSize * (1 - tolerance))
+	maxBinSize := math.Round(expectedBinSize * (1 + tolerance))
+
+	// Drop the first character because it will always be a letter, making
+	// the letter frequency skewed.
+	testIds := make([]string, len(ids))
+	for i, id := range ids {
+		testIds[i] = id[2:]
+	}
+	charFrequencies := make(map[string]int)
+	for _, id := range testIds {
+		chars := strings.Split(id, "")
+		for _, char := range chars {
+			charFrequencies[char] += 1
+		}
+	}
+	fmt.Println("Testing character frequency...")
+	fmt.Printf("expectedBinSize %v\n", expectedBinSize)
+	fmt.Printf("minBinSize %v\n", minBinSize)
+	fmt.Printf("maxBinSize %v\n", maxBinSize)
+	fmt.Printf("charFrequencies %v\n", charFrequencies)
+	for k, v := range charFrequencies {
+		if float64(v) < minBinSize || float64(v) > maxBinSize {
+			t.Errorf("The char %v is out of the expected bin size with value %v\n", k, v)
+		}
+	}
+	if len(charFrequencies) != base {
+		t.Errorf("Not all of the chars are presention in ids. Got only %v\n", len(charFrequencies))
+	}
+}
+
+func testHistogram(t *testing.T, poolResponse *IdPoolResponse, n int) {
+	histogram := poolResponse.Histogram
+	expectedBinSize := math.Ceil(float64(n) / float64(len(histogram)))
+	tolerance := 0.1
+	minBinSize := math.Round(expectedBinSize * (1 - tolerance))
+	maxBinSize := math.Round(expectedBinSize * (1 + tolerance))
+	fmt.Printf("Histogram %v\n", histogram)
+	fmt.Printf("expectedBinSize %v\n", expectedBinSize)
+	fmt.Printf("minBinSize %v\n", minBinSize)
+	fmt.Printf("maxBinSize %v\n", maxBinSize)
+	for _, i := range histogram {
+		if float64(i) < minBinSize || float64(i) > maxBinSize {
+			t.Errorf("Histogram is out of distribution tolerance")
+		}
+	}
+}

From 9b8fd246b00a092a4c89e40938ef029324d795f2 Mon Sep 17 00:00:00 2001
From: Akshay Vadher <akshay@incubyte.co>
Date: Fri, 19 Jul 2024 20:13:27 +0530
Subject: [PATCH 2/2] Review comment

---
 cuid2_histogram_test.go | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/cuid2_histogram_test.go b/cuid2_histogram_test.go
index 9936304..688559f 100644
--- a/cuid2_histogram_test.go
+++ b/cuid2_histogram_test.go
@@ -41,7 +41,7 @@ func testCharFrequency(t *testing.T, n int, ids []string) {
 	// the letter frequency skewed.
 	testIds := make([]string, len(ids))
 	for i, id := range ids {
-		testIds[i] = id[2:]
+		testIds[i] = id[1:]
 	}
 	charFrequencies := make(map[string]int)
 	for _, id := range testIds {