[#13] Allow to use english text in the payload

Signed-off-by: Evgenii Stratonikov <e.stratonikov@yadro.com>

go.mod

@@ -11,6 +11,7 @@ require (
 	github.com/aws/aws-sdk-go-v2/config v1.15.5
 	github.com/aws/aws-sdk-go-v2/service/s3 v1.26.9
 	github.com/dop251/goja v0.0.0-20230427124612-428fc442ff5f
+	github.com/go-loremipsum/loremipsum v1.1.3
 	github.com/google/uuid v1.3.0
 	github.com/joho/godotenv v1.5.1
 	github.com/nspcc-dev/neo-go v0.101.1

internal/datagen/datagen.go

@@ -1,6 +1,8 @@
 package datagen
 
 import (
+	"strings"
+
 	"go.k6.io/k6/js/modules"
 )
 
@@ -36,7 +38,7 @@ func (d *Datagen) Exports() modules.Exports {
 	return modules.Exports{Default: d}
 }
 
-func (d *Datagen) Generator(size int) *Generator {
-	g := NewGenerator(d.vu, size)
+func (d *Datagen) Generator(size int, typ string) *Generator {
+	g := NewGenerator(d.vu, size, strings.ToLower(typ))
 	return &g
 }

internal/datagen/generator.go

@@ -1,12 +1,13 @@
 package datagen
 
 import (
+	"bytes"
 	"crypto/sha256"
 	"encoding/hex"
 	"math/rand"
-	"time"
 
 	"github.com/dop251/goja"
+	"github.com/go-loremipsum/loremipsum"
 	"go.k6.io/k6/js/modules"
 )
 
@@ -24,6 +25,7 @@ type (
 		size   int
 		rand   *rand.Rand
 		buf    []byte
+		typ    string
 		offset int
 	}
 
@@ -36,20 +38,27 @@ type (
 // TailSize specifies number of extra random bytes in the buffer tail.
 const TailSize = 1024
 
-func NewGenerator(vu modules.VU, size int) Generator {
+var payloadTypes = []string{
+	"text",
+	"random",
+	"",
+}
+
+func NewGenerator(vu modules.VU, size int, typ string) Generator {
 	if size <= 0 {
 		panic("size should be positive")
 	}
 
-	r := rand.New(rand.NewSource(time.Now().UnixNano()))
-	buf := make([]byte, size+TailSize)
-	r.Read(buf)
-	return Generator{
-		vu:   vu,
-		size: size,
-		rand: r,
-		buf:  buf,
+	var found bool
+	for i := range payloadTypes {
+		if payloadTypes[i] == typ {
+			found = true
+		}
 	}
+	if !found {
+		vu.InitEnv().Logger.Info("Unknown payload type '%s', random will be used.", typ)
+	}
+	return Generator{vu: vu, size: size, buf: nil, typ: typ, offset: 0}
 }
 
 func (g *Generator) GenPayload(calcHash bool) GenPayloadResponse {
@@ -66,9 +75,24 @@ func (g *Generator) GenPayload(calcHash bool) GenPayloadResponse {
 }
 
 func (g *Generator) nextSlice() []byte {
-	if g.offset >= TailSize {
-		g.offset = 0
-		g.rand.Read(g.buf) // Per docs, err is always nil here
+	if g.buf == nil {
+		// Allocate buffer with extra tail for sliding and populate it with random bytes
+		g.buf = make([]byte, g.size+TailSize)
+		rand.Read(g.buf) // Per docs, err is always nil here
+		switch g.typ {
+		case "text":
+			li := loremipsum.New()
+			b := bytes.NewBuffer(nil)
+			for b.Len() < g.size+TailSize {
+				b.WriteString(li.Paragraph())
+				b.WriteRune('\n')
+			}
+			g.buf = b.Bytes()
+		default:
+			// Allocate buffer with extra tail for sliding and populate it with random bytes
+			g.buf = make([]byte, g.size+TailSize)
+			rand.Read(g.buf) // Per docs, err is always nil here
+		}
 	}
 
 	result := g.buf[g.offset : g.offset+g.size]
@@ -76,5 +100,10 @@ func (g *Generator) nextSlice() []byte {
 	// Shift the offset for the next call. If we've used our entire tail, then erase
 	// the buffer so that on the next call it is regenerated anew
 	g.offset += 1
+	if g.offset+g.size >= len(g.buf) {
+		g.buf = nil
+		g.offset = 0
+	}
+
 	return result
 }

internal/datagen/generator_test.go

@@ -16,25 +16,25 @@ func TestGenerator(t *testing.T) {
 
 	t.Run("fails on negative size", func(t *testing.T) {
 		require.Panics(t, func() {
-			_ = NewGenerator(vu, -1)
+			_ = NewGenerator(vu, -1, "")
 		})
 	})
 
 	t.Run("fails on zero size", func(t *testing.T) {
 		require.Panics(t, func() {
-			_ = NewGenerator(vu, 0)
+			_ = NewGenerator(vu, 0, "")
 		})
 	})
 
 	t.Run("creates slice of specified size", func(t *testing.T) {
 		size := 10
-		g := NewGenerator(vu, size)
+		g := NewGenerator(vu, size, "")
 		slice := g.nextSlice()
 		require.Len(t, slice, size)
 	})
 
 	t.Run("creates a different slice on each call", func(t *testing.T) {
-		g := NewGenerator(vu, 1000)
+		g := NewGenerator(vu, 1000, "")
 		slice1 := g.nextSlice()
 		slice2 := g.nextSlice()
 		// Each slice should be unique (assuming that 1000 random bytes will never coincide
@@ -43,7 +43,7 @@ func TestGenerator(t *testing.T) {
 	})
 
 	t.Run("keeps generating slices after consuming entire tail", func(t *testing.T) {
-		g := NewGenerator(vu, 1000)
+		g := NewGenerator(vu, 1000, "")
 		initialSlice := g.nextSlice()
 		for i := 0; i < TailSize; i++ {
 			g.nextSlice()

scenarios/grpc.js

@@ -47,7 +47,7 @@ if (registry_enabled && delete_age) {
 }
 
 
-const generator = datagen.generator(1024 * parseInt(__ENV.WRITE_OBJ_SIZE));
+const generator = datagen.generator(1024 * parseInt(__ENV.WRITE_OBJ_SIZE), __ENV.PAYLOAD_TYPE || "");
 
 const scenarios = {};
 

scenarios/http.js

@@ -31,7 +31,7 @@ const obj_registry = registry_enabled ? registry.open(__ENV.REGISTRY_FILE) : und
 
 const duration = __ENV.DURATION;
 
-const generator = datagen.generator(1024 * parseInt(__ENV.WRITE_OBJ_SIZE));
+const generator = datagen.generator(1024 * parseInt(__ENV.WRITE_OBJ_SIZE), __ENV.PAYLOAD_TYPE || "");
 
 const scenarios = {};
 

scenarios/run_scenarios.md

@@ -18,6 +18,7 @@ Scenarios `grpc.js`, `local.js`, `http.js` and `s3.js` support the following opt
   * `SLEEP_WRITE` - time interval (in seconds) between writing VU iterations.
   * `SLEEP_READ` - time interval (in seconds) between reading VU iterations.
   * `SELECTION_SIZE` - size of batch to select for deletion (default: 1000).
+  * `PAYLOAD_TYPE` - type of an object payload ("random" or "text", default: "random").
 
 Additionally, the profiling extension can be enabled to generate CPU and memory profiles which can be inspected with `go tool pprof file.prof`:
 ```shell

scenarios/s3.js

@@ -46,7 +46,7 @@ if (registry_enabled && delete_age) {
     );
 }
 
-const generator = datagen.generator(1024 * parseInt(__ENV.WRITE_OBJ_SIZE));
+const generator = datagen.generator(1024 * parseInt(__ENV.WRITE_OBJ_SIZE), __ENV.PAYLOAD_TYPE || "");
 
 const scenarios = {};