Merge pull request #4 from devinmcgloin/improved-tests

Improved testing

pkg/count/sketch_test.go

@@ -12,10 +12,15 @@ func TestCardinality(t *testing.T) {
 
 	url := []byte("https://devinmcgloin.com")
 	b.Increment(url)
-	if b.Count(url) == 0 {
+	if b.Count(url) != 1 {
 		t.Errorf("expected estimate count to be %d got %d\n", 1, b.Count(url))
 	}
 
+	b.Increment(url)
+	if b.Count(url) != 2 {
+		t.Errorf("expected estimate count to be %d got %d\n", 2, b.Count(url))
+	}
+
 	items := generator.RandomStrings(500)
 	for _, v := range items {
 		b.Increment([]byte(v))

pkg/generator/generator.go

@@ -68,3 +68,12 @@ func RandomStrings(n int) []string {
 	}
 	return words
 }
+
+// RandomSimilarStrings returns strings with some probabilistic overlap.
+func RandomSimilarStrings(n int, length int) []string {
+	words := make([]string, n)
+	for i := 0; i < n; i++ {
+		words[i] = RandString(length)
+	}
+	return words
+}

pkg/min-hash/hash_test.go

@@ -10,12 +10,15 @@ import (
 func TestJaccard(t *testing.T) {
 
 	mh := New(0.05)
-	a := generator.RandomStrings(5000)
-	b := generator.RandomStrings(5000)
+	a := generator.RandomSimilarStrings(5000, 3)
+	b := generator.RandomSimilarStrings(5000, 3)
 
 	naive := NaiveJaccard(a, b)
 
 	estimate := mh.Estimate(toInterface(a), toInterface(b))
+	if naive == 0 {
+		t.Error("naive calculated zero overlap")
+	}
 	if math.Abs(naive-estimate) > mh.ErrorRate() {
 		t.Error()
 	}