optimize memory allocations

Author: def

pattern.go

@@ -6,6 +6,7 @@ import (
 	"fmt"
 	"regexp"
 	"strings"
+	"sync"
 )
 
 const (
@@ -14,6 +15,14 @@ const (
 	patternMaxDiff   = 1
 )
 
+var (
+	buffers = sync.Pool{
+		New: func() interface{} {
+			return new(bytes.Buffer)
+		},
+	}
+)
+
 var (
 	squote  = '\''
 	dquote  = '"'
@@ -38,8 +47,17 @@ type Pattern struct {
 
 func (p *Pattern) String() string {
 	if p.str == nil {
-		s := strings.Join(p.words, " ")
+		buf := buffers.Get().(*bytes.Buffer)
+		buf.Reset()
+		for _, w := range p.words {
+			if buf.Len() > 0 {
+				buf.WriteByte(' ')
+			}
+			buf.WriteString(w)
+		}
+		s := buf.String()
 		p.str = &s
+		buffers.Put(buf)
 	}
 	return *p.str
 }
@@ -70,15 +88,17 @@ func (p *Pattern) WeakEqual(other *Pattern) bool {
 
 func NewPattern(input string) *Pattern {
 	pattern := &Pattern{}
-	for _, p := range strings.Fields(removeQuotedAndBrackets(input)) {
+	buf := buffers.Get().(*bytes.Buffer)
+	buf.Reset()
+	for _, p := range strings.Fields(removeQuotedAndBrackets(input, buf)) {
 		p = strings.TrimRight(p, "=:],;")
 		if len(p) < patterMinWordLen {
 			continue
 		}
 		if hexWithPrefix.MatchString(p) || hex.MatchString(p) || uuid.MatchString(p) {
 			continue
 		}
-		p = removeDigits(p)
+		p = removeDigits(p, buf)
 		if !isWord(p) {
 			continue
 		}
@@ -87,6 +107,7 @@ func NewPattern(input string) *Pattern {
 			break
 		}
 	}
+	buffers.Put(buf)
 	return pattern
 }
 
@@ -124,21 +145,19 @@ func isWord(s string) bool {
 	return firstLast == 2
 }
 
-func removeDigits(s string) string {
-	res := bytes.NewBufferString(s)
-	res.Reset()
+func removeDigits(s string, buf *bytes.Buffer) string {
+	buf.Reset()
 	for _, r := range s {
 		if r >= '0' && r <= '9' {
 			continue
 		}
-		res.WriteRune(r)
+		buf.WriteRune(r)
 	}
-	return res.String()
+	return buf.String()
 }
 
-func removeQuotedAndBrackets(s string) string {
-	res := bytes.NewBufferString(s)
-	res.Reset()
+func removeQuotedAndBrackets(s string, buf *bytes.Buffer) string {
+	buf.Reset()
 	var quote, prev rune
 	var seenBrackets []rune
 	var l int
@@ -180,7 +199,7 @@ func removeQuotedAndBrackets(s string) string {
 		if quote != 0 || len(seenBrackets) > 0 {
 			continue
 		}
-		res.WriteRune(r)
+		buf.WriteRune(r)
 	}
-	return res.String()
+	return buf.String()
 }

pattern_test.go

@@ -1,8 +1,10 @@
 package logparser
 
 import (
-	"github.com/stretchr/testify/assert"
+	"bytes"
 	"testing"
+
+	"github.com/stretchr/testify/assert"
 )
 
 func TestPattern(t *testing.T) {
@@ -96,18 +98,19 @@ func BenchmarkPatternWeakEqual(b *testing.B) {
 }
 
 func TestPatternRemoveQuotedAndBrackets(t *testing.T) {
-	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo 'squoted' bar`))
-	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo 'squoted \'baz\'' bar`))
-	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo "dquoted" bar`))
-	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo "dquoted \"baz\"" bar`))
-	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo "dquoted 'squoted' " bar`))
-	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo 'squoted "baz"' bar`))
-
-	assert.Equal(t, " msg", removeQuotedAndBrackets(`[nio-8080-exec-9] msg`))
-	assert.Equal(t, "json: ", removeQuotedAndBrackets(`json: {'arr': ['1', '2'], 'str': 'strval', 'age': 20}`))
+	buf := bytes.NewBuffer(nil)
+	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo 'squoted' bar`, buf))
+	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo 'squoted \'baz\'' bar`, buf))
+	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo "dquoted" bar`, buf))
+	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo "dquoted \"baz\"" bar`, buf))
+	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo "dquoted 'squoted' " bar`, buf))
+	assert.Equal(t, "foo  bar", removeQuotedAndBrackets(`foo 'squoted "baz"' bar`, buf))
+
+	assert.Equal(t, " msg", removeQuotedAndBrackets(`[nio-8080-exec-9] msg`, buf))
+	assert.Equal(t, "json: ", removeQuotedAndBrackets(`json: {'arr': ['1', '2'], 'str': 'strval', 'age': 20}`, buf))
 	assert.Equal(t, " ",
-		removeQuotedAndBrackets(`[Full GC (Allocation Failure) [CMS: 176934K->176934K(176960K), 0.0451364 secs] 253546K->253546K(253632K), [Metaspace: 11797K->11797K(1060864K)], 0.0454767 secs] [Times: user=0.04 sys=0.00, real=0.05 secs]`))
+		removeQuotedAndBrackets(`[Full GC (Allocation Failure) [CMS: 176934K->176934K(176960K), 0.0451364 secs] 253546K->253546K(253632K), [Metaspace: 11797K->11797K(1060864K)], 0.0454767 secs] [Times: user=0.04 sys=0.00, real=0.05 secs]`, buf))
 	assert.Equal(t,
 		"Jun 16 21:41:24 host01 kubelet: W0616 21:41:24.642736     961 reflector.go:341]",
-		removeQuotedAndBrackets(`Jun 16 21:41:24 host01 kubelet[961]: W0616 21:41:24.642736     961 reflector.go:341]`))
+		removeQuotedAndBrackets(`Jun 16 21:41:24 host01 kubelet[961]: W0616 21:41:24.642736     961 reflector.go:341]`, buf))
 }