Source file src/unicode/utf8/utf8_test.go

     1  // Copyright 2009 The Go Authors. All rights reserved.
     2  // Use of this source code is governed by a BSD-style
     3  // license that can be found in the LICENSE file.
     4  
     5  package utf8_test
     6  
     7  import (
     8  	"bytes"
     9  	"testing"
    10  	"unicode"
    11  	. "unicode/utf8"
    12  )
    13  
    14  // Validate the constants redefined from unicode.
    15  func init() {
    16  	if MaxRune != unicode.MaxRune {
    17  		panic("utf8.MaxRune is wrong")
    18  	}
    19  	if RuneError != unicode.ReplacementChar {
    20  		panic("utf8.RuneError is wrong")
    21  	}
    22  }
    23  
    24  // Validate the constants redefined from unicode.
    25  func TestConstants(t *testing.T) {
    26  	if MaxRune != unicode.MaxRune {
    27  		t.Errorf("utf8.MaxRune is wrong: %x should be %x", MaxRune, unicode.MaxRune)
    28  	}
    29  	if RuneError != unicode.ReplacementChar {
    30  		t.Errorf("utf8.RuneError is wrong: %x should be %x", RuneError, unicode.ReplacementChar)
    31  	}
    32  }
    33  
    34  type Utf8Map struct {
    35  	r   rune
    36  	str string
    37  }
    38  
    39  var utf8map = []Utf8Map{
    40  	{0x0000, "\x00"},
    41  	{0x0001, "\x01"},
    42  	{0x007e, "\x7e"},
    43  	{0x007f, "\x7f"},
    44  	{0x0080, "\xc2\x80"},
    45  	{0x0081, "\xc2\x81"},
    46  	{0x00bf, "\xc2\xbf"},
    47  	{0x00c0, "\xc3\x80"},
    48  	{0x00c1, "\xc3\x81"},
    49  	{0x00c8, "\xc3\x88"},
    50  	{0x00d0, "\xc3\x90"},
    51  	{0x00e0, "\xc3\xa0"},
    52  	{0x00f0, "\xc3\xb0"},
    53  	{0x00f8, "\xc3\xb8"},
    54  	{0x00ff, "\xc3\xbf"},
    55  	{0x0100, "\xc4\x80"},
    56  	{0x07ff, "\xdf\xbf"},
    57  	{0x0400, "\xd0\x80"},
    58  	{0x0800, "\xe0\xa0\x80"},
    59  	{0x0801, "\xe0\xa0\x81"},
    60  	{0x1000, "\xe1\x80\x80"},
    61  	{0xd000, "\xed\x80\x80"},
    62  	{0xd7ff, "\xed\x9f\xbf"}, // last code point before surrogate half.
    63  	{0xe000, "\xee\x80\x80"}, // first code point after surrogate half.
    64  	{0xfffe, "\xef\xbf\xbe"},
    65  	{0xffff, "\xef\xbf\xbf"},
    66  	{0x10000, "\xf0\x90\x80\x80"},
    67  	{0x10001, "\xf0\x90\x80\x81"},
    68  	{0x40000, "\xf1\x80\x80\x80"},
    69  	{0x10fffe, "\xf4\x8f\xbf\xbe"},
    70  	{0x10ffff, "\xf4\x8f\xbf\xbf"},
    71  	{0xFFFD, "\xef\xbf\xbd"},
    72  }
    73  
    74  var surrogateMap = []Utf8Map{
    75  	{0xd800, "\xed\xa0\x80"}, // surrogate min decodes to (RuneError, 1)
    76  	{0xdfff, "\xed\xbf\xbf"}, // surrogate max decodes to (RuneError, 1)
    77  }
    78  
    79  var testStrings = []string{
    80  	"",
    81  	"abcd",
    82  	"☺☻☹",
    83  	"日a本b語ç日ð本Ê語þ日¥本¼語i日©",
    84  	"日a本b語ç日ð本Ê語þ日¥本¼語i日©日a本b語ç日ð本Ê語þ日¥本¼語i日©日a本b語ç日ð本Ê語þ日¥本¼語i日©",
    85  	"\x80\x80\x80\x80",
    86  }
    87  
    88  func TestFullRune(t *testing.T) {
    89  	for _, m := range utf8map {
    90  		b := []byte(m.str)
    91  		if !FullRune(b) {
    92  			t.Errorf("FullRune(%q) (%U) = false, want true", b, m.r)
    93  		}
    94  		s := m.str
    95  		if !FullRuneInString(s) {
    96  			t.Errorf("FullRuneInString(%q) (%U) = false, want true", s, m.r)
    97  		}
    98  		b1 := b[0 : len(b)-1]
    99  		if FullRune(b1) {
   100  			t.Errorf("FullRune(%q) = true, want false", b1)
   101  		}
   102  		s1 := string(b1)
   103  		if FullRuneInString(s1) {
   104  			t.Errorf("FullRune(%q) = true, want false", s1)
   105  		}
   106  	}
   107  	for _, s := range []string{"\xc0", "\xc1"} {
   108  		b := []byte(s)
   109  		if !FullRune(b) {
   110  			t.Errorf("FullRune(%q) = false, want true", s)
   111  		}
   112  		if !FullRuneInString(s) {
   113  			t.Errorf("FullRuneInString(%q) = false, want true", s)
   114  		}
   115  	}
   116  }
   117  
   118  func TestEncodeRune(t *testing.T) {
   119  	for _, m := range utf8map {
   120  		b := []byte(m.str)
   121  		var buf [10]byte
   122  		n := EncodeRune(buf[0:], m.r)
   123  		b1 := buf[0:n]
   124  		if !bytes.Equal(b, b1) {
   125  			t.Errorf("EncodeRune(%#04x) = %q want %q", m.r, b1, b)
   126  		}
   127  	}
   128  }
   129  
   130  func TestAppendRune(t *testing.T) {
   131  	for _, m := range utf8map {
   132  		if buf := AppendRune(nil, m.r); string(buf) != m.str {
   133  			t.Errorf("AppendRune(nil, %#04x) = %s, want %s", m.r, buf, m.str)
   134  		}
   135  		if buf := AppendRune([]byte("init"), m.r); string(buf) != "init"+m.str {
   136  			t.Errorf("AppendRune(init, %#04x) = %s, want %s", m.r, buf, "init"+m.str)
   137  		}
   138  	}
   139  }
   140  
   141  func TestDecodeRune(t *testing.T) {
   142  	for _, m := range utf8map {
   143  		b := []byte(m.str)
   144  		r, size := DecodeRune(b)
   145  		if r != m.r || size != len(b) {
   146  			t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b, r, size, m.r, len(b))
   147  		}
   148  		s := m.str
   149  		r, size = DecodeRuneInString(s)
   150  		if r != m.r || size != len(b) {
   151  			t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, m.r, len(b))
   152  		}
   153  
   154  		// there's an extra byte that bytes left behind - make sure trailing byte works
   155  		r, size = DecodeRune(b[0:cap(b)])
   156  		if r != m.r || size != len(b) {
   157  			t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b, r, size, m.r, len(b))
   158  		}
   159  		s = m.str + "\x00"
   160  		r, size = DecodeRuneInString(s)
   161  		if r != m.r || size != len(b) {
   162  			t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, m.r, len(b))
   163  		}
   164  
   165  		// make sure missing bytes fail
   166  		wantsize := 1
   167  		if wantsize >= len(b) {
   168  			wantsize = 0
   169  		}
   170  		r, size = DecodeRune(b[0 : len(b)-1])
   171  		if r != RuneError || size != wantsize {
   172  			t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b[0:len(b)-1], r, size, RuneError, wantsize)
   173  		}
   174  		s = m.str[0 : len(m.str)-1]
   175  		r, size = DecodeRuneInString(s)
   176  		if r != RuneError || size != wantsize {
   177  			t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, RuneError, wantsize)
   178  		}
   179  
   180  		// make sure bad sequences fail
   181  		if len(b) == 1 {
   182  			b[0] = 0x80
   183  		} else {
   184  			b[len(b)-1] = 0x7F
   185  		}
   186  		r, size = DecodeRune(b)
   187  		if r != RuneError || size != 1 {
   188  			t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b, r, size, RuneError, 1)
   189  		}
   190  		s = string(b)
   191  		r, size = DecodeRuneInString(s)
   192  		if r != RuneError || size != 1 {
   193  			t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, RuneError, 1)
   194  		}
   195  
   196  	}
   197  }
   198  
   199  func TestDecodeSurrogateRune(t *testing.T) {
   200  	for _, m := range surrogateMap {
   201  		b := []byte(m.str)
   202  		r, size := DecodeRune(b)
   203  		if r != RuneError || size != 1 {
   204  			t.Errorf("DecodeRune(%q) = %x, %d want %x, %d", b, r, size, RuneError, 1)
   205  		}
   206  		s := m.str
   207  		r, size = DecodeRuneInString(s)
   208  		if r != RuneError || size != 1 {
   209  			t.Errorf("DecodeRuneInString(%q) = %x, %d want %x, %d", b, r, size, RuneError, 1)
   210  		}
   211  	}
   212  }
   213  
   214  // Check that DecodeRune and DecodeLastRune correspond to
   215  // the equivalent range loop.
   216  func TestSequencing(t *testing.T) {
   217  	for _, ts := range testStrings {
   218  		for _, m := range utf8map {
   219  			for _, s := range []string{ts + m.str, m.str + ts, ts + m.str + ts} {
   220  				testSequence(t, s)
   221  			}
   222  		}
   223  	}
   224  }
   225  
   226  func runtimeRuneCount(s string) int {
   227  	return len([]rune(s)) // Replaced by gc with call to runtime.countrunes(s).
   228  }
   229  
   230  // Check that a range loop, len([]rune(string)) optimization and
   231  // []rune conversions visit the same runes.
   232  // Not really a test of this package, but the assumption is used here and
   233  // it's good to verify.
   234  func TestRuntimeConversion(t *testing.T) {
   235  	for _, ts := range testStrings {
   236  		count := RuneCountInString(ts)
   237  		if n := runtimeRuneCount(ts); n != count {
   238  			t.Errorf("%q: len([]rune()) counted %d runes; got %d from RuneCountInString", ts, n, count)
   239  			break
   240  		}
   241  
   242  		runes := []rune(ts)
   243  		if n := len(runes); n != count {
   244  			t.Errorf("%q: []rune() has length %d; got %d from RuneCountInString", ts, n, count)
   245  			break
   246  		}
   247  		i := 0
   248  		for _, r := range ts {
   249  			if r != runes[i] {
   250  				t.Errorf("%q[%d]: expected %c (%U); got %c (%U)", ts, i, runes[i], runes[i], r, r)
   251  			}
   252  			i++
   253  		}
   254  	}
   255  }
   256  
   257  var invalidSequenceTests = []string{
   258  	"\xed\xa0\x80\x80", // surrogate min
   259  	"\xed\xbf\xbf\x80", // surrogate max
   260  
   261  	// xx
   262  	"\x91\x80\x80\x80",
   263  
   264  	// s1
   265  	"\xC2\x7F\x80\x80",
   266  	"\xC2\xC0\x80\x80",
   267  	"\xDF\x7F\x80\x80",
   268  	"\xDF\xC0\x80\x80",
   269  
   270  	// s2
   271  	"\xE0\x9F\xBF\x80",
   272  	"\xE0\xA0\x7F\x80",
   273  	"\xE0\xBF\xC0\x80",
   274  	"\xE0\xC0\x80\x80",
   275  
   276  	// s3
   277  	"\xE1\x7F\xBF\x80",
   278  	"\xE1\x80\x7F\x80",
   279  	"\xE1\xBF\xC0\x80",
   280  	"\xE1\xC0\x80\x80",
   281  
   282  	//s4
   283  	"\xED\x7F\xBF\x80",
   284  	"\xED\x80\x7F\x80",
   285  	"\xED\x9F\xC0\x80",
   286  	"\xED\xA0\x80\x80",
   287  
   288  	// s5
   289  	"\xF0\x8F\xBF\xBF",
   290  	"\xF0\x90\x7F\xBF",
   291  	"\xF0\x90\x80\x7F",
   292  	"\xF0\xBF\xBF\xC0",
   293  	"\xF0\xBF\xC0\x80",
   294  	"\xF0\xC0\x80\x80",
   295  
   296  	// s6
   297  	"\xF1\x7F\xBF\xBF",
   298  	"\xF1\x80\x7F\xBF",
   299  	"\xF1\x80\x80\x7F",
   300  	"\xF1\xBF\xBF\xC0",
   301  	"\xF1\xBF\xC0\x80",
   302  	"\xF1\xC0\x80\x80",
   303  
   304  	// s7
   305  	"\xF4\x7F\xBF\xBF",
   306  	"\xF4\x80\x7F\xBF",
   307  	"\xF4\x80\x80\x7F",
   308  	"\xF4\x8F\xBF\xC0",
   309  	"\xF4\x8F\xC0\x80",
   310  	"\xF4\x90\x80\x80",
   311  }
   312  
   313  func runtimeDecodeRune(s string) rune {
   314  	for _, r := range s {
   315  		return r
   316  	}
   317  	return -1
   318  }
   319  
   320  func TestDecodeInvalidSequence(t *testing.T) {
   321  	for _, s := range invalidSequenceTests {
   322  		r1, _ := DecodeRune([]byte(s))
   323  		if want := RuneError; r1 != want {
   324  			t.Errorf("DecodeRune(%#x) = %#04x, want %#04x", s, r1, want)
   325  			return
   326  		}
   327  		r2, _ := DecodeRuneInString(s)
   328  		if want := RuneError; r2 != want {
   329  			t.Errorf("DecodeRuneInString(%q) = %#04x, want %#04x", s, r2, want)
   330  			return
   331  		}
   332  		if r1 != r2 {
   333  			t.Errorf("DecodeRune(%#x) = %#04x mismatch with DecodeRuneInString(%q) = %#04x", s, r1, s, r2)
   334  			return
   335  		}
   336  		r3 := runtimeDecodeRune(s)
   337  		if r2 != r3 {
   338  			t.Errorf("DecodeRuneInString(%q) = %#04x mismatch with runtime.decoderune(%q) = %#04x", s, r2, s, r3)
   339  			return
   340  		}
   341  	}
   342  }
   343  
   344  func testSequence(t *testing.T, s string) {
   345  	type info struct {
   346  		index int
   347  		r     rune
   348  	}
   349  	index := make([]info, len(s))
   350  	b := []byte(s)
   351  	si := 0
   352  	j := 0
   353  	for i, r := range s {
   354  		if si != i {
   355  			t.Errorf("Sequence(%q) mismatched index %d, want %d", s, si, i)
   356  			return
   357  		}
   358  		index[j] = info{i, r}
   359  		j++
   360  		r1, size1 := DecodeRune(b[i:])
   361  		if r != r1 {
   362  			t.Errorf("DecodeRune(%q) = %#04x, want %#04x", s[i:], r1, r)
   363  			return
   364  		}
   365  		r2, size2 := DecodeRuneInString(s[i:])
   366  		if r != r2 {
   367  			t.Errorf("DecodeRuneInString(%q) = %#04x, want %#04x", s[i:], r2, r)
   368  			return
   369  		}
   370  		if size1 != size2 {
   371  			t.Errorf("DecodeRune/DecodeRuneInString(%q) size mismatch %d/%d", s[i:], size1, size2)
   372  			return
   373  		}
   374  		si += size1
   375  	}
   376  	j--
   377  	for si = len(s); si > 0; {
   378  		r1, size1 := DecodeLastRune(b[0:si])
   379  		r2, size2 := DecodeLastRuneInString(s[0:si])
   380  		if size1 != size2 {
   381  			t.Errorf("DecodeLastRune/DecodeLastRuneInString(%q, %d) size mismatch %d/%d", s, si, size1, size2)
   382  			return
   383  		}
   384  		if r1 != index[j].r {
   385  			t.Errorf("DecodeLastRune(%q, %d) = %#04x, want %#04x", s, si, r1, index[j].r)
   386  			return
   387  		}
   388  		if r2 != index[j].r {
   389  			t.Errorf("DecodeLastRuneInString(%q, %d) = %#04x, want %#04x", s, si, r2, index[j].r)
   390  			return
   391  		}
   392  		si -= size1
   393  		if si != index[j].index {
   394  			t.Errorf("DecodeLastRune(%q) index mismatch at %d, want %d", s, si, index[j].index)
   395  			return
   396  		}
   397  		j--
   398  	}
   399  	if si != 0 {
   400  		t.Errorf("DecodeLastRune(%q) finished at %d, not 0", s, si)
   401  	}
   402  }
   403  
   404  // Check that negative runes encode as U+FFFD.
   405  func TestNegativeRune(t *testing.T) {
   406  	errorbuf := make([]byte, UTFMax)
   407  	errorbuf = errorbuf[0:EncodeRune(errorbuf, RuneError)]
   408  	buf := make([]byte, UTFMax)
   409  	buf = buf[0:EncodeRune(buf, -1)]
   410  	if !bytes.Equal(buf, errorbuf) {
   411  		t.Errorf("incorrect encoding [% x] for -1; expected [% x]", buf, errorbuf)
   412  	}
   413  }
   414  
   415  type RuneCountTest struct {
   416  	in  string
   417  	out int
   418  }
   419  
   420  var runecounttests = []RuneCountTest{
   421  	{"abcd", 4},
   422  	{"☺☻☹", 3},
   423  	{"1,2,3,4", 7},
   424  	{"\xe2\x00", 2},
   425  	{"\xe2\x80", 2},
   426  	{"a\xe2\x80", 3},
   427  }
   428  
   429  func TestRuneCount(t *testing.T) {
   430  	for _, tt := range runecounttests {
   431  		if out := RuneCountInString(tt.in); out != tt.out {
   432  			t.Errorf("RuneCountInString(%q) = %d, want %d", tt.in, out, tt.out)
   433  		}
   434  		if out := RuneCount([]byte(tt.in)); out != tt.out {
   435  			t.Errorf("RuneCount(%q) = %d, want %d", tt.in, out, tt.out)
   436  		}
   437  	}
   438  }
   439  
   440  type RuneLenTest struct {
   441  	r    rune
   442  	size int
   443  }
   444  
   445  var runelentests = []RuneLenTest{
   446  	{0, 1},
   447  	{'e', 1},
   448  	{'é', 2},
   449  	{'☺', 3},
   450  	{RuneError, 3},
   451  	{MaxRune, 4},
   452  	{0xD800, -1},
   453  	{0xDFFF, -1},
   454  	{MaxRune + 1, -1},
   455  	{-1, -1},
   456  }
   457  
   458  func TestRuneLen(t *testing.T) {
   459  	for _, tt := range runelentests {
   460  		if size := RuneLen(tt.r); size != tt.size {
   461  			t.Errorf("RuneLen(%#U) = %d, want %d", tt.r, size, tt.size)
   462  		}
   463  	}
   464  }
   465  
   466  type ValidTest struct {
   467  	in  string
   468  	out bool
   469  }
   470  
   471  var validTests = []ValidTest{
   472  	{"", true},
   473  	{"a", true},
   474  	{"abc", true},
   475  	{"Ж", true},
   476  	{"ЖЖ", true},
   477  	{"брэд-ЛГТМ", true},
   478  	{"☺☻☹", true},
   479  	{"aa\xe2", false},
   480  	{string([]byte{66, 250}), false},
   481  	{string([]byte{66, 250, 67}), false},
   482  	{"a\uFFFDb", true},
   483  	{string("\xF4\x8F\xBF\xBF"), true},      // U+10FFFF
   484  	{string("\xF4\x90\x80\x80"), false},     // U+10FFFF+1; out of range
   485  	{string("\xF7\xBF\xBF\xBF"), false},     // 0x1FFFFF; out of range
   486  	{string("\xFB\xBF\xBF\xBF\xBF"), false}, // 0x3FFFFFF; out of range
   487  	{string("\xc0\x80"), false},             // U+0000 encoded in two bytes: incorrect
   488  	{string("\xed\xa0\x80"), false},         // U+D800 high surrogate (sic)
   489  	{string("\xed\xbf\xbf"), false},         // U+DFFF low surrogate (sic)
   490  }
   491  
   492  func TestValid(t *testing.T) {
   493  	for _, tt := range validTests {
   494  		if Valid([]byte(tt.in)) != tt.out {
   495  			t.Errorf("Valid(%q) = %v; want %v", tt.in, !tt.out, tt.out)
   496  		}
   497  		if ValidString(tt.in) != tt.out {
   498  			t.Errorf("ValidString(%q) = %v; want %v", tt.in, !tt.out, tt.out)
   499  		}
   500  	}
   501  }
   502  
   503  type ValidRuneTest struct {
   504  	r  rune
   505  	ok bool
   506  }
   507  
   508  var validrunetests = []ValidRuneTest{
   509  	{0, true},
   510  	{'e', true},
   511  	{'é', true},
   512  	{'☺', true},
   513  	{RuneError, true},
   514  	{MaxRune, true},
   515  	{0xD7FF, true},
   516  	{0xD800, false},
   517  	{0xDFFF, false},
   518  	{0xE000, true},
   519  	{MaxRune + 1, false},
   520  	{-1, false},
   521  }
   522  
   523  func TestValidRune(t *testing.T) {
   524  	for _, tt := range validrunetests {
   525  		if ok := ValidRune(tt.r); ok != tt.ok {
   526  			t.Errorf("ValidRune(%#U) = %t, want %t", tt.r, ok, tt.ok)
   527  		}
   528  	}
   529  }
   530  
   531  func BenchmarkRuneCountTenASCIIChars(b *testing.B) {
   532  	s := []byte("0123456789")
   533  	for i := 0; i < b.N; i++ {
   534  		RuneCount(s)
   535  	}
   536  }
   537  
   538  func BenchmarkRuneCountTenJapaneseChars(b *testing.B) {
   539  	s := []byte("日本語日本語日本語日")
   540  	for i := 0; i < b.N; i++ {
   541  		RuneCount(s)
   542  	}
   543  }
   544  
   545  func BenchmarkRuneCountInStringTenASCIIChars(b *testing.B) {
   546  	for i := 0; i < b.N; i++ {
   547  		RuneCountInString("0123456789")
   548  	}
   549  }
   550  
   551  func BenchmarkRuneCountInStringTenJapaneseChars(b *testing.B) {
   552  	for i := 0; i < b.N; i++ {
   553  		RuneCountInString("日本語日本語日本語日")
   554  	}
   555  }
   556  
   557  func BenchmarkValidTenASCIIChars(b *testing.B) {
   558  	s := []byte("0123456789")
   559  	for i := 0; i < b.N; i++ {
   560  		Valid(s)
   561  	}
   562  }
   563  
   564  func BenchmarkValidTenJapaneseChars(b *testing.B) {
   565  	s := []byte("日本語日本語日本語日")
   566  	for i := 0; i < b.N; i++ {
   567  		Valid(s)
   568  	}
   569  }
   570  
   571  func BenchmarkValidStringTenASCIIChars(b *testing.B) {
   572  	for i := 0; i < b.N; i++ {
   573  		ValidString("0123456789")
   574  	}
   575  }
   576  
   577  func BenchmarkValidStringTenJapaneseChars(b *testing.B) {
   578  	for i := 0; i < b.N; i++ {
   579  		ValidString("日本語日本語日本語日")
   580  	}
   581  }
   582  
   583  func BenchmarkEncodeASCIIRune(b *testing.B) {
   584  	buf := make([]byte, UTFMax)
   585  	for i := 0; i < b.N; i++ {
   586  		EncodeRune(buf, 'a')
   587  	}
   588  }
   589  
   590  func BenchmarkEncodeJapaneseRune(b *testing.B) {
   591  	buf := make([]byte, UTFMax)
   592  	for i := 0; i < b.N; i++ {
   593  		EncodeRune(buf, '本')
   594  	}
   595  }
   596  
   597  func BenchmarkAppendASCIIRune(b *testing.B) {
   598  	buf := make([]byte, UTFMax)
   599  	for i := 0; i < b.N; i++ {
   600  		AppendRune(buf[:0], 'a')
   601  	}
   602  }
   603  
   604  func BenchmarkAppendJapaneseRune(b *testing.B) {
   605  	buf := make([]byte, UTFMax)
   606  	for i := 0; i < b.N; i++ {
   607  		AppendRune(buf[:0], '本')
   608  	}
   609  }
   610  
   611  func BenchmarkDecodeASCIIRune(b *testing.B) {
   612  	a := []byte{'a'}
   613  	for i := 0; i < b.N; i++ {
   614  		DecodeRune(a)
   615  	}
   616  }
   617  
   618  func BenchmarkDecodeJapaneseRune(b *testing.B) {
   619  	nihon := []byte("本")
   620  	for i := 0; i < b.N; i++ {
   621  		DecodeRune(nihon)
   622  	}
   623  }
   624  
   625  // boolSink is used to reference the return value of benchmarked
   626  // functions to avoid dead code elimination.
   627  var boolSink bool
   628  
   629  func BenchmarkFullRune(b *testing.B) {
   630  	benchmarks := []struct {
   631  		name string
   632  		data []byte
   633  	}{
   634  		{"ASCII", []byte("a")},
   635  		{"Incomplete", []byte("\xf0\x90\x80")},
   636  		{"Japanese", []byte("本")},
   637  	}
   638  	for _, bm := range benchmarks {
   639  		b.Run(bm.name, func(b *testing.B) {
   640  			for i := 0; i < b.N; i++ {
   641  				boolSink = FullRune(bm.data)
   642  			}
   643  		})
   644  	}
   645  }
   646  

View as plain text