Merge pull request #29 from rivo/graphemestringwidth

Fixed StringWidth() implementation by using proper Unicode grapheme cluster segmentation. Fixes #28

go.mod

@@ -1,3 +1,5 @@
 module github.com/mattn/go-runewidth
 
 go 1.9
+
+require github.com/rivo/uniseg v0.1.0

go.sum

@@ -0,0 +1,2 @@
+github.com/rivo/uniseg v0.1.0 h1:+2KBaVoUmb9XzDsrx/Ct0W/EYOSFf/nWTauy++DprtY=
+github.com/rivo/uniseg v0.1.0/go.mod h1:J6wj4VEh+S6ZtnVlnTBMWIodfgj8LQOQFoIToxlJtxc=

runewidth.go

@@ -2,6 +2,8 @@ package runewidth
 
 import (
 	"os"
+
+	"github.com/rivo/uniseg"
 )
 
 //go:generate go run script/generate.go
@@ -10,9 +12,6 @@ var (
 	// EastAsianWidth will be set true if the current locale is CJK
 	EastAsianWidth bool
 
-	// ZeroWidthJoiner is flag to set to use UTR#51 ZWJ
-	ZeroWidthJoiner bool
-
 	// DefaultCondition is a condition in current locale
 	DefaultCondition = &Condition{}
 )
@@ -30,7 +29,6 @@ func handleEnv() {
 	}
 	// update DefaultCondition
 	DefaultCondition.EastAsianWidth = EastAsianWidth
-	DefaultCondition.ZeroWidthJoiner = ZeroWidthJoiner
 }
 
 type interval struct {
@@ -85,15 +83,13 @@ var nonprint = table{
 
 // Condition have flag EastAsianWidth whether the current locale is CJK or not.
 type Condition struct {
-	EastAsianWidth  bool
-	ZeroWidthJoiner bool
+	EastAsianWidth bool
 }
 
 // NewCondition return new instance of Condition which is current locale.
 func NewCondition() *Condition {
 	return &Condition{
-		EastAsianWidth:  EastAsianWidth,
-		ZeroWidthJoiner: ZeroWidthJoiner,
+		EastAsianWidth: EastAsianWidth,
 	}
 }
 
@@ -110,66 +106,54 @@ func (c *Condition) RuneWidth(r rune) int {
 	}
 }
 
-func (c *Condition) stringWidth(s string) (width int) {
-	for _, r := range []rune(s) {
-		width += c.RuneWidth(r)
-	}
-	return width
-}
-
-func (c *Condition) stringWidthZeroJoiner(s string) (width int) {
-	r1, r2 := rune(0), rune(0)
-	for _, r := range []rune(s) {
-		if r == 0xFE0E || r == 0xFE0F {
-			continue
-		}
-		w := c.RuneWidth(r)
-		if r2 == 0x200D && inTables(r, emoji) && inTables(r1, emoji) {
-			if width < w {
-				width = w
-			}
-		} else {
-			width += w
-		}
-		r1, r2 = r2, r
-	}
-	return width
-}
-
 // StringWidth return width as you can see
 func (c *Condition) StringWidth(s string) (width int) {
-	if c.ZeroWidthJoiner {
-		return c.stringWidthZeroJoiner(s)
+	g := uniseg.NewGraphemes(s)
+	for g.Next() {
+		var chWidth int
+		for _, r := range g.Runes() {
+			chWidth = c.RuneWidth(r)
+			if chWidth > 0 {
+				break // Our best guess at this point is to use the width of the first non-zero-width rune.
+			}
+		}
+		width += chWidth
 	}
-	return c.stringWidth(s)
+	return
 }
 
 // Truncate return string truncated with w cells
 func (c *Condition) Truncate(s string, w int, tail string) string {
 	if c.StringWidth(s) <= w {
 		return s
 	}
-	r := []rune(s)
-	tw := c.StringWidth(tail)
-	w -= tw
-	width := 0
-	i := 0
-	for ; i < len(r); i++ {
-		cw := c.RuneWidth(r[i])
-		if width+cw > w {
+	w -= c.StringWidth(tail)
+	var width int
+	pos := len(s)
+	g := uniseg.NewGraphemes(s)
+	for g.Next() {
+		var chWidth int
+		for _, r := range g.Runes() {
+			chWidth = c.RuneWidth(r)
+			if chWidth > 0 {
+				break // See StringWidth() for details.
+			}
+		}
+		if width+chWidth > w {
+			pos, _ = g.Positions()
 			break
 		}
-		width += cw
+		width += chWidth
 	}
-	return string(r[0:i]) + tail
+	return s[:pos] + tail
 }
 
 // Wrap return string wrapped with w cells
 func (c *Condition) Wrap(s string, w int) string {
 	width := 0
 	out := ""
 	for _, r := range []rune(s) {
-		cw := RuneWidth(r)
+		cw := c.RuneWidth(r)
 		if r == '\n' {
 			out += string(r)
 			width = 0

runewidth_test.go

@@ -243,7 +243,7 @@ func TestStringWidth(t *testing.T) {
 	c.EastAsianWidth = true
 	for _, tt := range stringwidthtests {
 		if out := c.StringWidth(tt.in); out != tt.eaout {
-			t.Errorf("StringWidth(%q) = %d, want %d", tt.in, out, tt.eaout)
+			t.Errorf("StringWidth(%q) = %d, want %d (EA)", tt.in, out, tt.eaout)
 		}
 	}
 }
@@ -400,9 +400,8 @@ func TestEnv(t *testing.T) {
 	}
 }
 
-func TestZeroWidthJointer(t *testing.T) {
+func TestZeroWidthJoiner(t *testing.T) {
 	c := NewCondition()
-	c.ZeroWidthJoiner = true
 
 	var tests = []struct {
 		in   string
@@ -414,7 +413,7 @@ func TestZeroWidthJointer(t *testing.T) {
 		{"‍🍳", 2},
 		{"👨‍👨", 2},
 		{"👨‍👨‍👧", 2},
-		{"🏳️‍🌈", 2},
+		{"🏳️‍🌈", 1},
 		{"あ👩‍🍳い", 6},
 		{"あ‍🍳い", 6},
 		{"あ‍い", 4},