strconv: add QuoteToGraphic and friends

This version of quoting allows runes in category Zs, such as the
ideographic space characters, to be passed through unquoted.

Still to do (maybe): A way to access this from Printf.

Updates #11511.

Change-Id: I3bae84b1aa0bc1b885318d3f67c5f451099a2a5a
Reviewed-on: https://go-review.googlesource.com/14184Reviewed-by: Marcel van Lohuizen <mpvl@golang.org>

src/strconv/isprint.go

@@ -635,3 +635,23 @@ var isNotPrint32 = []uint16{ // add 0x10000 to each entry
 	0xf57a,
 	0xf5a4,
 }
+
+// isGraphic lists the graphic runes not matched by IsPrint.
+var isGraphic = []uint16{
+	0x00a0,
+	0x1680,
+	0x2000,
+	0x2001,
+	0x2002,
+	0x2003,
+	0x2004,
+	0x2005,
+	0x2006,
+	0x2007,
+	0x2008,
+	0x2009,
+	0x200a,
+	0x202f,
+	0x205f,
+	0x3000,
+}

src/strconv/makeisprint.go

@@ -174,6 +174,23 @@ func main() {
 		}
 		fmt.Fprintf(&buf, "\t%#04x,\n", r-0x10000)
 	}
+	fmt.Fprintf(&buf, "}\n\n")
+
+	// The list of graphic but not "printable" runes is short. Just make one easy table.
+	fmt.Fprintf(&buf, "// isGraphic lists the graphic runes not matched by IsPrint.\n")
+	fmt.Fprintf(&buf, "var isGraphic = []uint16{\n")
+	for r := rune(0); r <= unicode.MaxRune; r++ {
+		if unicode.IsPrint(r) != unicode.IsGraphic(r) {
+			// Sanity check.
+			if !unicode.IsGraphic(r) {
+				log.Fatalf("%U is printable but not graphic\n", r)
+			}
+			if r > 0xFFFF { // We expect only 16-bit values.
+				log.Fatalf("%U too big for isGraphic\n", r)
+			}
+			fmt.Fprintf(&buf, "\t%#04x,\n", r)
+		}
+	}
 	fmt.Fprintf(&buf, "}\n")
 
 	data, err := format.Source(buf.Bytes())

src/strconv/quote.go

@@ -12,7 +12,7 @@ import (
 
 const lowerhex = "0123456789abcdef"
 
-func quoteWith(s string, quote byte, ASCIIonly bool) string {
+func quoteWith(s string, quote byte, ASCIIonly, graphicOnly bool) string {
 	var runeTmp [utf8.UTFMax]byte
 	buf := make([]byte, 0, 3*len(s)/2) // Try to avoid more allocations.
 	buf = append(buf, quote)
@@ -38,7 +38,7 @@ func quoteWith(s string, quote byte, ASCIIonly bool) string {
 				buf = append(buf, byte(r))
 				continue
 			}
-		} else if IsPrint(r) {
+		} else if IsPrint(r) || graphicOnly && isInGraphicList(r) {
 			n := utf8.EncodeRune(runeTmp[:], r)
 			buf = append(buf, runeTmp[:n]...)
 			continue
@@ -90,7 +90,7 @@ func quoteWith(s string, quote byte, ASCIIonly bool) string {
 // control characters and non-printable characters as defined by
 // IsPrint.
 func Quote(s string) string {
-	return quoteWith(s, '"', false)
+	return quoteWith(s, '"', false, false)
 }
 
 // AppendQuote appends a double-quoted Go string literal representing s,
@@ -103,7 +103,7 @@ func AppendQuote(dst []byte, s string) []byte {
 // The returned string uses Go escape sequences (\t, \n, \xFF, \u0100) for
 // non-ASCII characters and non-printable characters as defined by IsPrint.
 func QuoteToASCII(s string) string {
-	return quoteWith(s, '"', true)
+	return quoteWith(s, '"', true, false)
 }
 
 // AppendQuoteToASCII appends a double-quoted Go string literal representing s,
@@ -112,12 +112,25 @@ func AppendQuoteToASCII(dst []byte, s string) []byte {
 	return append(dst, QuoteToASCII(s)...)
 }
 
+// QuoteToGraphic returns a double-quoted Go string literal representing s.
+// The returned string uses Go escape sequences (\t, \n, \xFF, \u0100) for
+// non-ASCII characters and non-printable characters as defined by IsGraphic.
+func QuoteToGraphic(s string) string {
+	return quoteWith(s, '"', false, true)
+}
+
+// AppendQuoteToGraphic appends a double-quoted Go string literal representing s,
+// as generated by QuoteToGraphic, to dst and returns the extended buffer.
+func AppendQuoteToGraphic(dst []byte, s string) []byte {
+	return append(dst, QuoteToGraphic(s)...)
+}
+
 // QuoteRune returns a single-quoted Go character literal representing the
-// rune.  The returned string uses Go escape sequences (\t, \n, \xFF, \u0100)
+// rune. The returned string uses Go escape sequences (\t, \n, \xFF, \u0100)
 // for control characters and non-printable characters as defined by IsPrint.
 func QuoteRune(r rune) string {
 	// TODO: avoid the allocation here.
-	return quoteWith(string(r), '\'', false)
+	return quoteWith(string(r), '\'', false, false)
 }
 
 // AppendQuoteRune appends a single-quoted Go character literal representing the rune,
@@ -127,12 +140,12 @@ func AppendQuoteRune(dst []byte, r rune) []byte {
 }
 
 // QuoteRuneToASCII returns a single-quoted Go character literal representing
-// the rune.  The returned string uses Go escape sequences (\t, \n, \xFF,
+// the rune. The returned string uses Go escape sequences (\t, \n, \xFF,
 // \u0100) for non-ASCII characters and non-printable characters as defined
 // by IsPrint.
 func QuoteRuneToASCII(r rune) string {
 	// TODO: avoid the allocation here.
-	return quoteWith(string(r), '\'', true)
+	return quoteWith(string(r), '\'', true, false)
 }
 
 // AppendQuoteRuneToASCII appends a single-quoted Go character literal representing the rune,
@@ -141,6 +154,21 @@ func AppendQuoteRuneToASCII(dst []byte, r rune) []byte {
 	return append(dst, QuoteRuneToASCII(r)...)
 }
 
+// QuoteRuneToGraphic returns a single-quoted Go character literal representing
+// the rune. The returned string uses Go escape sequences (\t, \n, \xFF,
+// \u0100) for non-ASCII characters and non-printable characters as defined
+// by IsGraphic.
+func QuoteRuneToGraphic(r rune) string {
+	// TODO: avoid the allocation here.
+	return quoteWith(string(r), '\'', false, true)
+}
+
+// AppendQuoteRuneToGraphic appends a single-quoted Go character literal representing the rune,
+// as generated by QuoteRuneToGraphic, to dst and returns the extended buffer.
+func AppendQuoteRuneToGraphic(dst []byte, r rune) []byte {
+	return append(dst, QuoteRuneToGraphic(r)...)
+}
+
 // CanBackquote reports whether the string s can be represented
 // unchanged as a single-line backquoted string without control
 // characters other than tab.
@@ -453,3 +481,26 @@ func IsPrint(r rune) bool {
 	j := bsearch16(isNotPrint, uint16(r))
 	return j >= len(isNotPrint) || isNotPrint[j] != uint16(r)
 }
+
+// IsGraphic reports whether the rune is defined as a Graphic by Unicode. Such
+// characters include letters, marks, numbers, punctuation, symbols, and
+// spaces, from categories L, M, N, P, S, and Zs.
+func IsGraphic(r rune) bool {
+	if IsPrint(r) {
+		return true
+	}
+	return isInGraphicList(r)
+}
+
+// isInGraphicList reports whether the rune is in the isGraphic list. This separation
+// from IsGraphic allows quoteWith to avoid two calls to IsPrint.
+// Should be called only if IsPrint fails.
+func isInGraphicList(r rune) bool {
+	// We know r must fit in 16 bits - see makeisprint.go.
+	if r > 0xFFFF {
+		return false
+	}
+	rr := uint16(r)
+	i := bsearch16(isGraphic, rr)
+	return i < len(isGraphic) && rr == isGraphic[i]
+}

src/strconv/quote_test.go

@@ -10,7 +10,7 @@ import (
 	"unicode"
 )
 
-// Verify that our isPrint agrees with unicode.IsPrint
+// Verify that our IsPrint agrees with unicode.IsPrint.
 func TestIsPrint(t *testing.T) {
 	n := 0
 	for r := rune(0); r <= unicode.MaxRune; r++ {
@@ -24,19 +24,36 @@ func TestIsPrint(t *testing.T) {
 	}
 }
 
+// Verify that our IsGraphic agrees with unicode.IsGraphic.
+func TestIsGraphic(t *testing.T) {
+	n := 0
+	for r := rune(0); r <= unicode.MaxRune; r++ {
+		if IsGraphic(r) != unicode.IsGraphic(r) {
+			t.Errorf("IsGraphic(%U)=%t incorrect", r, IsGraphic(r))
+			n++
+			if n > 10 {
+				return
+			}
+		}
+	}
+}
+
 type quoteTest struct {
-	in    string
-	out   string
-	ascii string
+	in      string
+	out     string
+	ascii   string
+	graphic string
 }
 
 var quotetests = []quoteTest{
-	{"\a\b\f\r\n\t\v", `"\a\b\f\r\n\t\v"`, `"\a\b\f\r\n\t\v"`},
-	{"\\", `"\\"`, `"\\"`},
-	{"abc\xffdef", `"abc\xffdef"`, `"abc\xffdef"`},
-	{"\u263a", `"☺"`, `"\u263a"`},
-	{"\U0010ffff", `"\U0010ffff"`, `"\U0010ffff"`},
-	{"\x04", `"\x04"`, `"\x04"`},
+	{"\a\b\f\r\n\t\v", `"\a\b\f\r\n\t\v"`, `"\a\b\f\r\n\t\v"`, `"\a\b\f\r\n\t\v"`},
+	{"\\", `"\\"`, `"\\"`, `"\\"`},
+	{"abc\xffdef", `"abc\xffdef"`, `"abc\xffdef"`, `"abc\xffdef"`},
+	{"\u263a", `"☺"`, `"\u263a"`, `"☺"`},
+	{"\U0010ffff", `"\U0010ffff"`, `"\U0010ffff"`, `"\U0010ffff"`},
+	{"\x04", `"\x04"`, `"\x04"`, `"\x04"`},
+	// Some non-printable but graphic runes. Final column is double-quoted.
+	{"!\u00a0!\u2000!\u3000!", `"!\u00a0!\u2000!\u3000!"`, `"!\u00a0!\u2000!\u3000!"`, "\"!\u00a0!\u2000!\u3000!\""},
 }
 
 func TestQuote(t *testing.T) {
@@ -61,22 +78,38 @@ func TestQuoteToASCII(t *testing.T) {
 	}
 }
 
+func TestQuoteToGraphic(t *testing.T) {
+	for _, tt := range quotetests {
+		if out := QuoteToGraphic(tt.in); out != tt.graphic {
+			t.Errorf("QuoteToGraphic(%s) = %s, want %s", tt.in, out, tt.graphic)
+		}
+		if out := AppendQuoteToGraphic([]byte("abc"), tt.in); string(out) != "abc"+tt.graphic {
+			t.Errorf("AppendQuoteToGraphic(%q, %s) = %s, want %s", "abc", tt.in, out, "abc"+tt.graphic)
+		}
+	}
+}
+
 type quoteRuneTest struct {
-	in    rune
-	out   string
-	ascii string
+	in      rune
+	out     string
+	ascii   string
+	graphic string
 }
 
 var quoterunetests = []quoteRuneTest{
-	{'a', `'a'`, `'a'`},
-	{'\a', `'\a'`, `'\a'`},
-	{'\\', `'\\'`, `'\\'`},
-	{0xFF, `'ÿ'`, `'\u00ff'`},
-	{0x263a, `'☺'`, `'\u263a'`},
-	{0xfffd, `'�'`, `'\ufffd'`},
-	{0x0010ffff, `'\U0010ffff'`, `'\U0010ffff'`},
-	{0x0010ffff + 1, `'�'`, `'\ufffd'`},
-	{0x04, `'\x04'`, `'\x04'`},
+	{'a', `'a'`, `'a'`, `'a'`},
+	{'\a', `'\a'`, `'\a'`, `'\a'`},
+	{'\\', `'\\'`, `'\\'`, `'\\'`},
+	{0xFF, `'ÿ'`, `'\u00ff'`, `'ÿ'`},
+	{0x263a, `'☺'`, `'\u263a'`, `'☺'`},
+	{0xfffd, `'�'`, `'\ufffd'`, `'�'`},
+	{0x0010ffff, `'\U0010ffff'`, `'\U0010ffff'`, `'\U0010ffff'`},
+	{0x0010ffff + 1, `'�'`, `'\ufffd'`, `'�'`},
+	{0x04, `'\x04'`, `'\x04'`, `'\x04'`},
+	// Some differences between graphic and printable. Note the last column is double-quoted.
+	{'\u00a0', `'\u00a0'`, `'\u00a0'`, "'\u00a0'"},
+	{'\u2000', `'\u2000'`, `'\u2000'`, "'\u2000'"},
+	{'\u3000', `'\u3000'`, `'\u3000'`, "'\u3000'"},
 }
 
 func TestQuoteRune(t *testing.T) {
@@ -101,6 +134,17 @@ func TestQuoteRuneToASCII(t *testing.T) {
 	}
 }
 
+func TestQuoteRuneToGraphic(t *testing.T) {
+	for _, tt := range quoterunetests {
+		if out := QuoteRuneToGraphic(tt.in); out != tt.graphic {
+			t.Errorf("QuoteRuneToGraphic(%U) = %s, want %s", tt.in, out, tt.graphic)
+		}
+		if out := AppendQuoteRuneToGraphic([]byte("abc"), tt.in); string(out) != "abc"+tt.graphic {
+			t.Errorf("AppendQuoteRuneToGraphic(%q, %U) = %s, want %s", "abc", tt.in, out, "abc"+tt.graphic)
+		}
+	}
+}
+
 type canBackquoteTest struct {
 	in  string
 	out bool