fmt: avoid allocation when formatting byte slice arguments with verb s

fmtBytes is in the top 10 callers of runtime.slicebytetostring according
to Google wide profiling data.

Avoid the string conversion of the input byte slice in fmtBytes by calling
a newly added specialized fmtS function for byte slices.

Expand tests for verb s with widths to test strings and byte slice arguments.

SprintfTruncateString     157ns ± 4%     156ns ± 3%     ~     (p=0.122 n=20+20)
SprintfTruncateBytes      188ns ± 2%     155ns ± 3%  -18.00%  (p=0.000 n=20+19)

name                   old alloc/op   new alloc/op   delta
SprintfTruncateString     16.0B ± 0%     16.0B ± 0%     ~     (all equal)
SprintfTruncateBytes      64.0B ± 0%     16.0B ± 0%  -75.00%  (p=0.000 n=20+20)

name                   old allocs/op  new allocs/op  delta
SprintfTruncateString      1.00 ± 0%      1.00 ± 0%     ~     (all equal)
SprintfTruncateBytes       2.00 ± 0%      1.00 ± 0%  -50.00%  (p=0.000 n=20+20)

Change-Id: I461bf514d4232b39bd9c812f7faa4e5ef693a03b
Reviewed-on: https://go-review.googlesource.com/c/145284
Run-TryBot: Martin Möhrmann <martisch@uos.de>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Rob Pike <r@golang.org>

3 files changed, 55 insertions, 10 deletions

diff --git a/src/fmt/fmt_test.go b/src/fmt/fmt_test.go
index e97372225c..1907268c74 100644
--- a/src/fmt/fmt_test.go
+++ b/src/fmt/fmt_test.go
@@ -298,20 +298,30 @@ var fmtTests = []struct {
 
 	// width
 	{"%5s", "abc", "  abc"},
+	{"%5s", []byte("abc"), "  abc"},
 	{"%2s", "\u263a", " ☺"},
+	{"%2s", []byte("\u263a"), " ☺"},
 	{"%-5s", "abc", "abc  "},
-	{"%-8q", "abc", `"abc"   `},
+	{"%-5s", []byte("abc"), "abc  "},
 	{"%05s", "abc", "00abc"},
-	{"%08q", "abc", `000"abc"`},
+	{"%05s", []byte("abc"), "00abc"},
 	{"%5s", "abcdefghijklmnopqrstuvwxyz", "abcdefghijklmnopqrstuvwxyz"},
+	{"%5s", []byte("abcdefghijklmnopqrstuvwxyz"), "abcdefghijklmnopqrstuvwxyz"},
 	{"%.5s", "abcdefghijklmnopqrstuvwxyz", "abcde"},
+	{"%.5s", []byte("abcdefghijklmnopqrstuvwxyz"), "abcde"},
 	{"%.0s", "日本語日本語", ""},
+	{"%.0s", []byte("日本語日本語"), ""},
 	{"%.5s", "日本語日本語", "日本語日本"},
-	{"%.10s", "日本語日本語", "日本語日本語"},
 	{"%.5s", []byte("日本語日本語"), "日本語日本"},
+	{"%.10s", "日本語日本語", "日本語日本語"},
+	{"%.10s", []byte("日本語日本語"), "日本語日本語"},
+	{"%08q", "abc", `000"abc"`},
+	{"%08q", []byte("abc"), `000"abc"`},
+	{"%-8q", "abc", `"abc"   `},
+	{"%-8q", []byte("abc"), `"abc"   `},
 	{"%.5q", "abcdefghijklmnopqrstuvwxyz", `"abcde"`},
-	{"%.5x", "abcdefghijklmnopqrstuvwxyz", "6162636465"},
 	{"%.5q", []byte("abcdefghijklmnopqrstuvwxyz"), `"abcde"`},
+	{"%.5x", "abcdefghijklmnopqrstuvwxyz", "6162636465"},
 	{"%.5x", []byte("abcdefghijklmnopqrstuvwxyz"), "6162636465"},
 	{"%.3q", "日本語日本語", `"日本語"`},
 	{"%.3q", []byte("日本語日本語"), `"日本語"`},
@@ -320,6 +330,7 @@ var fmtTests = []struct {
 	{"%.1x", "日本語", "e6"},
 	{"%.1X", []byte("日本語"), "E6"},
 	{"%10.1q", "日本語日本語", `       "日"`},
+	{"%10.1q", []byte("日本語日本語"), `       "日"`},
 	{"%10v", nil, "     <nil>"},
 	{"%-10v", nil, "<nil>     "},
 
@@ -1211,7 +1222,16 @@ func BenchmarkSprintfString(b *testing.B) {
 func BenchmarkSprintfTruncateString(b *testing.B) {
 	b.RunParallel(func(pb *testing.PB) {
 		for pb.Next() {
-			Sprintf("%.3s", "日本語日本語日本語")
+			Sprintf("%.3s", "日本語日本語日本語日本語")
+		}
+	})
+}
+
+func BenchmarkSprintfTruncateBytes(b *testing.B) {
+	var bytes interface{} = []byte("日本語日本語日本語日本語")
+	b.RunParallel(func(pb *testing.PB) {
+		for pb.Next() {
+			Sprintf("%.3s", bytes)
 		}
 	})
 }
diff --git a/src/fmt/format.go b/src/fmt/format.go
index 91103f2c07..d6da8aed1e 100644
--- a/src/fmt/format.go
+++ b/src/fmt/format.go
@@ -308,8 +308,8 @@ func (f *fmt) fmtInteger(u uint64, base int, isSigned bool, digits string) {
 	f.zero = oldZero
 }
 
-// truncate truncates the string to the specified precision, if present.
-func (f *fmt) truncate(s string) string {
+// truncate truncates the string s to the specified precision, if present.
+func (f *fmt) truncateString(s string) string {
 	if f.precPresent {
 		n := f.prec
 		for i := range s {
@@ -322,12 +322,37 @@ func (f *fmt) truncate(s string) string {
 	return s
 }
 
+// truncate truncates the byte slice b as a string of the specified precision, if present.
+func (f *fmt) truncate(b []byte) []byte {
+	if f.precPresent {
+		n := f.prec
+		for i := 0; i < len(b); {
+			n--
+			if n < 0 {
+				return b[:i]
+			}
+			wid := 1
+			if b[i] >= utf8.RuneSelf {
+				_, wid = utf8.DecodeRune(b[i:])
+			}
+			i += wid
+		}
+	}
+	return b
+}
+
 // fmtS formats a string.
 func (f *fmt) fmtS(s string) {
-	s = f.truncate(s)
+	s = f.truncateString(s)
 	f.padString(s)
 }
 
+// fmtBs formats the byte slice b as if it was formatted as string with fmtS.
+func (f *fmt) fmtBs(b []byte) {
+	b = f.truncate(b)
+	f.pad(b)
+}
+
 // fmtSbx formats a string or byte slice as a hexadecimal encoding of its bytes.
 func (f *fmt) fmtSbx(s string, b []byte, digits string) {
 	length := len(b)
@@ -408,7 +433,7 @@ func (f *fmt) fmtBx(b []byte, digits string) {
 // If f.sharp is set a raw (backquoted) string may be returned instead
 // if the string does not contain any control characters other than tab.
 func (f *fmt) fmtQ(s string) {
-	s = f.truncate(s)
+	s = f.truncateString(s)
 	if f.sharp && strconv.CanBackquote(s) {
 		f.padString("`" + s + "`")
 		return
diff --git a/src/fmt/print.go b/src/fmt/print.go
index 22dc52ccdc..5df34a25e5 100644
--- a/src/fmt/print.go
+++ b/src/fmt/print.go
@@ -488,7 +488,7 @@ func (p *pp) fmtBytes(v []byte, verb rune, typeString string) {
 			p.buf.WriteByte(']')
 		}
 	case 's':
-		p.fmt.fmtS(string(v))
+		p.fmt.fmtBs(v)
 	case 'x':
 		p.fmt.fmtBx(v, ldigits)
 	case 'X':