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  	"strings"
    10  	"testing"
    11  	"unicode"
    12  	. "unicode/utf8"
    13  )
    14  
    15  // Validate the constants redefined from unicode.
    16  func init() {
    17  	if MaxRune != unicode.MaxRune {
    18  		panic("utf8.MaxRune is wrong")
    19  	}
    20  	if RuneError != unicode.ReplacementChar {
    21  		panic("utf8.RuneError is wrong")
    22  	}
    23  }
    24  
    25  // Validate the constants redefined from unicode.
    26  func TestConstants(t *testing.T) {
    27  	if MaxRune != unicode.MaxRune {
    28  		t.Errorf("utf8.MaxRune is wrong: %x should be %x", MaxRune, unicode.MaxRune)
    29  	}
    30  	if RuneError != unicode.ReplacementChar {
    31  		t.Errorf("utf8.RuneError is wrong: %x should be %x", RuneError, unicode.ReplacementChar)
    32  	}
    33  }
    34  
    35  type Utf8Map struct {
    36  	r   rune
    37  	str string
    38  }
    39  
    40  var utf8map = []Utf8Map{
    41  	{0x0000, "\x00"},
    42  	{0x0001, "\x01"},
    43  	{0x007e, "\x7e"},
    44  	{0x007f, "\x7f"},
    45  	{0x0080, "\xc2\x80"},
    46  	{0x0081, "\xc2\x81"},
    47  	{0x00bf, "\xc2\xbf"},
    48  	{0x00c0, "\xc3\x80"},
    49  	{0x00c1, "\xc3\x81"},
    50  	{0x00c8, "\xc3\x88"},
    51  	{0x00d0, "\xc3\x90"},
    52  	{0x00e0, "\xc3\xa0"},
    53  	{0x00f0, "\xc3\xb0"},
    54  	{0x00f8, "\xc3\xb8"},
    55  	{0x00ff, "\xc3\xbf"},
    56  	{0x0100, "\xc4\x80"},
    57  	{0x07ff, "\xdf\xbf"},
    58  	{0x0400, "\xd0\x80"},
    59  	{0x0800, "\xe0\xa0\x80"},
    60  	{0x0801, "\xe0\xa0\x81"},
    61  	{0x1000, "\xe1\x80\x80"},
    62  	{0xd000, "\xed\x80\x80"},
    63  	{0xd7ff, "\xed\x9f\xbf"}, // last code point before surrogate half.
    64  	{0xe000, "\xee\x80\x80"}, // first code point after surrogate half.
    65  	{0xfffe, "\xef\xbf\xbe"},
    66  	{0xffff, "\xef\xbf\xbf"},
    67  	{0x10000, "\xf0\x90\x80\x80"},
    68  	{0x10001, "\xf0\x90\x80\x81"},
    69  	{0x40000, "\xf1\x80\x80\x80"},
    70  	{0x10fffe, "\xf4\x8f\xbf\xbe"},
    71  	{0x10ffff, "\xf4\x8f\xbf\xbf"},
    72  	{0xFFFD, "\xef\xbf\xbd"},
    73  }
    74  
    75  var surrogateMap = []Utf8Map{
    76  	{0xd800, "\xed\xa0\x80"}, // surrogate min decodes to (RuneError, 1)
    77  	{0xdfff, "\xed\xbf\xbf"}, // surrogate max decodes to (RuneError, 1)
    78  }
    79  
    80  var testStrings = []string{
    81  	"",
    82  	"abcd",
    83  	"☺☻☹",
    84  	"日a本b語ç日ð本Ê語þ日¥本¼語i日©",
    85  	"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©",
    86  	"\x80\x80\x80\x80",
    87  }
    88  
    89  func TestFullRune(t *testing.T) {
    90  	for _, m := range utf8map {
    91  		b := []byte(m.str)
    92  		if !FullRune(b) {
    93  			t.Errorf("FullRune(%q) (%U) = false, want true", b, m.r)
    94  		}
    95  		s := m.str
    96  		if !FullRuneInString(s) {
    97  			t.Errorf("FullRuneInString(%q) (%U) = false, want true", s, m.r)
    98  		}
    99  		b1 := b[0 : len(b)-1]
   100  		if FullRune(b1) {
   101  			t.Errorf("FullRune(%q) = true, want false", b1)
   102  		}
   103  		s1 := string(b1)
   104  		if FullRuneInString(s1) {
   105  			t.Errorf("FullRune(%q) = true, want false", s1)
   106  		}
   107  	}
   108  	for _, s := range []string{"\xc0", "\xc1"} {
   109  		b := []byte(s)
   110  		if !FullRune(b) {
   111  			t.Errorf("FullRune(%q) = false, want true", s)
   112  		}
   113  		if !FullRuneInString(s) {
   114  			t.Errorf("FullRuneInString(%q) = false, want true", s)
   115  		}
   116  	}
   117  }
   118  
   119  func TestEncodeRune(t *testing.T) {
   120  	for _, m := range utf8map {
   121  		b := []byte(m.str)
   122  		var buf [10]byte
   123  		n := EncodeRune(buf[0:], m.r)
   124  		b1 := buf[0:n]
   125  		if !bytes.Equal(b, b1) {
   126  			t.Errorf("EncodeRune(%#04x) = %q want %q", m.r, b1, b)
   127  		}
   128  	}
   129  }
   130  
   131  func TestAppendRune(t *testing.T) {
   132  	for _, m := range utf8map {
   133  		if buf := AppendRune(nil, m.r); string(buf) != m.str {
   134  			t.Errorf("AppendRune(nil, %#04x) = %s, want %s", m.r, buf, m.str)
   135  		}
   136  		if buf := AppendRune([]byte("init"), m.r); string(buf) != "init"+m.str {
   137  			t.Errorf("AppendRune(init, %#04x) = %s, want %s", m.r, buf, "init"+m.str)
   138  		}
   139  	}
   140  }
   141  
   142  func TestDecodeRune(t *testing.T) {
   143  	for _, m := range utf8map {
   144  		b := []byte(m.str)
   145  		r, size := DecodeRune(b)
   146  		if r != m.r || size != len(b) {
   147  			t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b, r, size, m.r, len(b))
   148  		}
   149  		s := m.str
   150  		r, size = DecodeRuneInString(s)
   151  		if r != m.r || size != len(b) {
   152  			t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, m.r, len(b))
   153  		}
   154  
   155  		// there's an extra byte that bytes left behind - make sure trailing byte works
   156  		r, size = DecodeRune(b[0:cap(b)])
   157  		if r != m.r || size != len(b) {
   158  			t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b, r, size, m.r, len(b))
   159  		}
   160  		s = m.str + "\x00"
   161  		r, size = DecodeRuneInString(s)
   162  		if r != m.r || size != len(b) {
   163  			t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, m.r, len(b))
   164  		}
   165  
   166  		// make sure missing bytes fail
   167  		wantsize := 1
   168  		if wantsize >= len(b) {
   169  			wantsize = 0
   170  		}
   171  		r, size = DecodeRune(b[0 : len(b)-1])
   172  		if r != RuneError || size != wantsize {
   173  			t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b[:len(b)-1], r, size, RuneError, wantsize)
   174  		}
   175  		s = m.str[0 : len(m.str)-1]
   176  		r, size = DecodeRuneInString(s)
   177  		if r != RuneError || size != wantsize {
   178  			t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, RuneError, wantsize)
   179  		}
   180  
   181  		// make sure bad sequences fail
   182  		if len(b) == 1 {
   183  			b[0] = 0x80
   184  		} else {
   185  			b[len(b)-1] = 0x7F
   186  		}
   187  		r, size = DecodeRune(b)
   188  		if r != RuneError || size != 1 {
   189  			t.Errorf("DecodeRune(%q) = %#04x, %d want %#04x, %d", b, r, size, RuneError, 1)
   190  		}
   191  		s = string(b)
   192  		r, size = DecodeRuneInString(s)
   193  		if r != RuneError || size != 1 {
   194  			t.Errorf("DecodeRuneInString(%q) = %#04x, %d want %#04x, %d", s, r, size, RuneError, 1)
   195  		}
   196  
   197  	}
   198  }
   199  
   200  func TestDecodeSurrogateRune(t *testing.T) {
   201  	for _, m := range surrogateMap {
   202  		b := []byte(m.str)
   203  		r, size := DecodeRune(b)
   204  		if r != RuneError || size != 1 {
   205  			t.Errorf("DecodeRune(%q) = %x, %d want %x, %d", b, r, size, RuneError, 1)
   206  		}
   207  		s := m.str
   208  		r, size = DecodeRuneInString(s)
   209  		if r != RuneError || size != 1 {
   210  			t.Errorf("DecodeRuneInString(%q) = %x, %d want %x, %d", b, r, size, RuneError, 1)
   211  		}
   212  	}
   213  }
   214  
   215  // Check that DecodeRune and DecodeLastRune correspond to
   216  // the equivalent range loop.
   217  func TestSequencing(t *testing.T) {
   218  	for _, ts := range testStrings {
   219  		for _, m := range utf8map {
   220  			for _, s := range []string{ts + m.str, m.str + ts, ts + m.str + ts} {
   221  				testSequence(t, s)
   222  			}
   223  		}
   224  	}
   225  }
   226  
   227  func runtimeRuneCount(s string) int {
   228  	return len([]rune(s)) // Replaced by gc with call to runtime.countrunes(s).
   229  }
   230  
   231  // Check that a range loop, len([]rune(string)) optimization and
   232  // []rune conversions visit the same runes.
   233  // Not really a test of this package, but the assumption is used here and
   234  // it's good to verify.
   235  func TestRuntimeConversion(t *testing.T) {
   236  	for _, ts := range testStrings {
   237  		count := RuneCountInString(ts)
   238  		if n := runtimeRuneCount(ts); n != count {
   239  			t.Errorf("%q: len([]rune()) counted %d runes; got %d from RuneCountInString", ts, n, count)
   240  			break
   241  		}
   242  
   243  		runes := []rune(ts)
   244  		if n := len(runes); n != count {
   245  			t.Errorf("%q: []rune() has length %d; got %d from RuneCountInString", ts, n, count)
   246  			break
   247  		}
   248  		i := 0
   249  		for _, r := range ts {
   250  			if r != runes[i] {
   251  				t.Errorf("%q[%d]: expected %c (%U); got %c (%U)", ts, i, runes[i], runes[i], r, r)
   252  			}
   253  			i++
   254  		}
   255  	}
   256  }
   257  
   258  var invalidSequenceTests = []string{
   259  	"\xed\xa0\x80\x80", // surrogate min
   260  	"\xed\xbf\xbf\x80", // surrogate max
   261  
   262  	// xx
   263  	"\x91\x80\x80\x80",
   264  
   265  	// s1
   266  	"\xC2\x7F\x80\x80",
   267  	"\xC2\xC0\x80\x80",
   268  	"\xDF\x7F\x80\x80",
   269  	"\xDF\xC0\x80\x80",
   270  
   271  	// s2
   272  	"\xE0\x9F\xBF\x80",
   273  	"\xE0\xA0\x7F\x80",
   274  	"\xE0\xBF\xC0\x80",
   275  	"\xE0\xC0\x80\x80",
   276  
   277  	// s3
   278  	"\xE1\x7F\xBF\x80",
   279  	"\xE1\x80\x7F\x80",
   280  	"\xE1\xBF\xC0\x80",
   281  	"\xE1\xC0\x80\x80",
   282  
   283  	//s4
   284  	"\xED\x7F\xBF\x80",
   285  	"\xED\x80\x7F\x80",
   286  	"\xED\x9F\xC0\x80",
   287  	"\xED\xA0\x80\x80",
   288  
   289  	// s5
   290  	"\xF0\x8F\xBF\xBF",
   291  	"\xF0\x90\x7F\xBF",
   292  	"\xF0\x90\x80\x7F",
   293  	"\xF0\xBF\xBF\xC0",
   294  	"\xF0\xBF\xC0\x80",
   295  	"\xF0\xC0\x80\x80",
   296  
   297  	// s6
   298  	"\xF1\x7F\xBF\xBF",
   299  	"\xF1\x80\x7F\xBF",
   300  	"\xF1\x80\x80\x7F",
   301  	"\xF1\xBF\xBF\xC0",
   302  	"\xF1\xBF\xC0\x80",
   303  	"\xF1\xC0\x80\x80",
   304  
   305  	// s7
   306  	"\xF4\x7F\xBF\xBF",
   307  	"\xF4\x80\x7F\xBF",
   308  	"\xF4\x80\x80\x7F",
   309  	"\xF4\x8F\xBF\xC0",
   310  	"\xF4\x8F\xC0\x80",
   311  	"\xF4\x90\x80\x80",
   312  }
   313  
   314  func runtimeDecodeRune(s string) rune {
   315  	for _, r := range s {
   316  		return r
   317  	}
   318  	return -1
   319  }
   320  
   321  func TestDecodeInvalidSequence(t *testing.T) {
   322  	for _, s := range invalidSequenceTests {
   323  		r1, _ := DecodeRune([]byte(s))
   324  		if want := RuneError; r1 != want {
   325  			t.Errorf("DecodeRune(%#x) = %#04x, want %#04x", s, r1, want)
   326  			return
   327  		}
   328  		r2, _ := DecodeRuneInString(s)
   329  		if want := RuneError; r2 != want {
   330  			t.Errorf("DecodeRuneInString(%q) = %#04x, want %#04x", s, r2, want)
   331  			return
   332  		}
   333  		if r1 != r2 {
   334  			t.Errorf("DecodeRune(%#x) = %#04x mismatch with DecodeRuneInString(%q) = %#04x", s, r1, s, r2)
   335  			return
   336  		}
   337  		r3 := runtimeDecodeRune(s)
   338  		if r2 != r3 {
   339  			t.Errorf("DecodeRuneInString(%q) = %#04x mismatch with runtime.decoderune(%q) = %#04x", s, r2, s, r3)
   340  			return
   341  		}
   342  	}
   343  }
   344  
   345  func testSequence(t *testing.T, s string) {
   346  	type info struct {
   347  		index int
   348  		r     rune
   349  	}
   350  	index := make([]info, len(s))
   351  	b := []byte(s)
   352  	si := 0
   353  	j := 0
   354  	for i, r := range s {
   355  		if si != i {
   356  			t.Errorf("Sequence(%q) mismatched index %d, want %d", s, si, i)
   357  			return
   358  		}
   359  		index[j] = info{i, r}
   360  		j++
   361  		r1, size1 := DecodeRune(b[i:])
   362  		if r != r1 {
   363  			t.Errorf("DecodeRune(%q) = %#04x, want %#04x", s[i:], r1, r)
   364  			return
   365  		}
   366  		r2, size2 := DecodeRuneInString(s[i:])
   367  		if r != r2 {
   368  			t.Errorf("DecodeRuneInString(%q) = %#04x, want %#04x", s[i:], r2, r)
   369  			return
   370  		}
   371  		if size1 != size2 {
   372  			t.Errorf("DecodeRune/DecodeRuneInString(%q) size mismatch %d/%d", s[i:], size1, size2)
   373  			return
   374  		}
   375  		si += size1
   376  	}
   377  	j--
   378  	for si = len(s); si > 0; {
   379  		r1, size1 := DecodeLastRune(b[0:si])
   380  		r2, size2 := DecodeLastRuneInString(s[0:si])
   381  		if size1 != size2 {
   382  			t.Errorf("DecodeLastRune/DecodeLastRuneInString(%q, %d) size mismatch %d/%d", s, si, size1, size2)
   383  			return
   384  		}
   385  		if r1 != index[j].r {
   386  			t.Errorf("DecodeLastRune(%q, %d) = %#04x, want %#04x", s, si, r1, index[j].r)
   387  			return
   388  		}
   389  		if r2 != index[j].r {
   390  			t.Errorf("DecodeLastRuneInString(%q, %d) = %#04x, want %#04x", s, si, r2, index[j].r)
   391  			return
   392  		}
   393  		si -= size1
   394  		if si != index[j].index {
   395  			t.Errorf("DecodeLastRune(%q) index mismatch at %d, want %d", s, si, index[j].index)
   396  			return
   397  		}
   398  		j--
   399  	}
   400  	if si != 0 {
   401  		t.Errorf("DecodeLastRune(%q) finished at %d, not 0", s, si)
   402  	}
   403  }
   404  
   405  // Check that negative runes encode as U+FFFD.
   406  func TestNegativeRune(t *testing.T) {
   407  	errorbuf := make([]byte, UTFMax)
   408  	errorbuf = errorbuf[0:EncodeRune(errorbuf, RuneError)]
   409  	buf := make([]byte, UTFMax)
   410  	buf = buf[0:EncodeRune(buf, -1)]
   411  	if !bytes.Equal(buf, errorbuf) {
   412  		t.Errorf("incorrect encoding [% x] for -1; expected [% x]", buf, errorbuf)
   413  	}
   414  }
   415  
   416  type RuneCountTest struct {
   417  	in  string
   418  	out int
   419  }
   420  
   421  var runecounttests = []RuneCountTest{
   422  	{"abcd", 4},
   423  	{"☺☻☹", 3},
   424  	{"1,2,3,4", 7},
   425  	{"\xe2\x00", 2},
   426  	{"\xe2\x80", 2},
   427  	{"a\xe2\x80", 3},
   428  }
   429  
   430  func TestRuneCount(t *testing.T) {
   431  	for _, tt := range runecounttests {
   432  		if out := RuneCountInString(tt.in); out != tt.out {
   433  			t.Errorf("RuneCountInString(%q) = %d, want %d", tt.in, out, tt.out)
   434  		}
   435  		if out := RuneCount([]byte(tt.in)); out != tt.out {
   436  			t.Errorf("RuneCount(%q) = %d, want %d", tt.in, out, tt.out)
   437  		}
   438  	}
   439  }
   440  
   441  func TestRuneCountNonASCIIAllocation(t *testing.T) {
   442  	if n := testing.AllocsPerRun(10, func() {
   443  		s := []byte("日本語日本語日本語日")
   444  		_ = RuneCount(s)
   445  	}); n > 0 {
   446  		t.Errorf("unexpected RuneCount allocation, got %v, want 0", n)
   447  	}
   448  }
   449  
   450  type RuneLenTest struct {
   451  	r    rune
   452  	size int
   453  }
   454  
   455  var runelentests = []RuneLenTest{
   456  	{0, 1},
   457  	{'e', 1},
   458  	{'é', 2},
   459  	{'☺', 3},
   460  	{RuneError, 3},
   461  	{MaxRune, 4},
   462  	{0xD800, -1},
   463  	{0xDFFF, -1},
   464  	{MaxRune + 1, -1},
   465  	{-1, -1},
   466  }
   467  
   468  func TestRuneLen(t *testing.T) {
   469  	for _, tt := range runelentests {
   470  		if size := RuneLen(tt.r); size != tt.size {
   471  			t.Errorf("RuneLen(%#U) = %d, want %d", tt.r, size, tt.size)
   472  		}
   473  	}
   474  }
   475  
   476  type ValidTest struct {
   477  	in  string
   478  	out bool
   479  }
   480  
   481  var validTests = []ValidTest{
   482  	{"", true},
   483  	{"a", true},
   484  	{"abc", true},
   485  	{"Ж", true},
   486  	{"ЖЖ", true},
   487  	{"брэд-ЛГТМ", true},
   488  	{"☺☻☹", true},
   489  	{"aa\xe2", false},
   490  	{string([]byte{66, 250}), false},
   491  	{string([]byte{66, 250, 67}), false},
   492  	{"a\uFFFDb", true},
   493  	{string("\xF4\x8F\xBF\xBF"), true},      // U+10FFFF
   494  	{string("\xF4\x90\x80\x80"), false},     // U+10FFFF+1; out of range
   495  	{string("\xF7\xBF\xBF\xBF"), false},     // 0x1FFFFF; out of range
   496  	{string("\xFB\xBF\xBF\xBF\xBF"), false}, // 0x3FFFFFF; out of range
   497  	{string("\xc0\x80"), false},             // U+0000 encoded in two bytes: incorrect
   498  	{string("\xed\xa0\x80"), false},         // U+D800 high surrogate (sic)
   499  	{string("\xed\xbf\xbf"), false},         // U+DFFF low surrogate (sic)
   500  }
   501  
   502  func TestValid(t *testing.T) {
   503  	for _, tt := range validTests {
   504  		if Valid([]byte(tt.in)) != tt.out {
   505  			t.Errorf("Valid(%q) = %v; want %v", tt.in, !tt.out, tt.out)
   506  		}
   507  		if ValidString(tt.in) != tt.out {
   508  			t.Errorf("ValidString(%q) = %v; want %v", tt.in, !tt.out, tt.out)
   509  		}
   510  	}
   511  }
   512  
   513  type ValidRuneTest struct {
   514  	r  rune
   515  	ok bool
   516  }
   517  
   518  var validrunetests = []ValidRuneTest{
   519  	{0, true},
   520  	{'e', true},
   521  	{'é', true},
   522  	{'☺', true},
   523  	{RuneError, true},
   524  	{MaxRune, true},
   525  	{0xD7FF, true},
   526  	{0xD800, false},
   527  	{0xDFFF, false},
   528  	{0xE000, true},
   529  	{MaxRune + 1, false},
   530  	{-1, false},
   531  }
   532  
   533  func TestValidRune(t *testing.T) {
   534  	for _, tt := range validrunetests {
   535  		if ok := ValidRune(tt.r); ok != tt.ok {
   536  			t.Errorf("ValidRune(%#U) = %t, want %t", tt.r, ok, tt.ok)
   537  		}
   538  	}
   539  }
   540  
   541  func BenchmarkRuneCountTenASCIIChars(b *testing.B) {
   542  	s := []byte("0123456789")
   543  	for i := 0; i < b.N; i++ {
   544  		RuneCount(s)
   545  	}
   546  }
   547  
   548  func BenchmarkRuneCountTenJapaneseChars(b *testing.B) {
   549  	s := []byte("日本語日本語日本語日")
   550  	for i := 0; i < b.N; i++ {
   551  		RuneCount(s)
   552  	}
   553  }
   554  
   555  func BenchmarkRuneCountInStringTenASCIIChars(b *testing.B) {
   556  	for i := 0; i < b.N; i++ {
   557  		RuneCountInString("0123456789")
   558  	}
   559  }
   560  
   561  func BenchmarkRuneCountInStringTenJapaneseChars(b *testing.B) {
   562  	for i := 0; i < b.N; i++ {
   563  		RuneCountInString("日本語日本語日本語日")
   564  	}
   565  }
   566  
   567  var ascii100000 = strings.Repeat("0123456789", 10000)
   568  
   569  func BenchmarkValidTenASCIIChars(b *testing.B) {
   570  	s := []byte("0123456789")
   571  	for i := 0; i < b.N; i++ {
   572  		Valid(s)
   573  	}
   574  }
   575  
   576  func BenchmarkValid100KASCIIChars(b *testing.B) {
   577  	s := []byte(ascii100000)
   578  	for i := 0; i < b.N; i++ {
   579  		Valid(s)
   580  	}
   581  }
   582  
   583  func BenchmarkValidTenJapaneseChars(b *testing.B) {
   584  	s := []byte("日本語日本語日本語日")
   585  	for i := 0; i < b.N; i++ {
   586  		Valid(s)
   587  	}
   588  }
   589  func BenchmarkValidLongMostlyASCII(b *testing.B) {
   590  	longMostlyASCII := []byte(longStringMostlyASCII)
   591  	for i := 0; i < b.N; i++ {
   592  		Valid(longMostlyASCII)
   593  	}
   594  }
   595  
   596  func BenchmarkValidLongJapanese(b *testing.B) {
   597  	longJapanese := []byte(longStringJapanese)
   598  	for i := 0; i < b.N; i++ {
   599  		Valid(longJapanese)
   600  	}
   601  }
   602  
   603  func BenchmarkValidStringTenASCIIChars(b *testing.B) {
   604  	for i := 0; i < b.N; i++ {
   605  		ValidString("0123456789")
   606  	}
   607  }
   608  
   609  func BenchmarkValidString100KASCIIChars(b *testing.B) {
   610  	for i := 0; i < b.N; i++ {
   611  		ValidString(ascii100000)
   612  	}
   613  }
   614  
   615  func BenchmarkValidStringTenJapaneseChars(b *testing.B) {
   616  	for i := 0; i < b.N; i++ {
   617  		ValidString("日本語日本語日本語日")
   618  	}
   619  }
   620  
   621  func BenchmarkValidStringLongMostlyASCII(b *testing.B) {
   622  	for i := 0; i < b.N; i++ {
   623  		ValidString(longStringMostlyASCII)
   624  	}
   625  }
   626  
   627  func BenchmarkValidStringLongJapanese(b *testing.B) {
   628  	for i := 0; i < b.N; i++ {
   629  		ValidString(longStringJapanese)
   630  	}
   631  }
   632  
   633  var longStringMostlyASCII string // ~100KB, ~97% ASCII
   634  var longStringJapanese string    // ~100KB, non-ASCII
   635  
   636  func init() {
   637  	const japanese = "日本語日本語日本語日"
   638  	var b strings.Builder
   639  	for i := 0; b.Len() < 100_000; i++ {
   640  		if i%100 == 0 {
   641  			b.WriteString(japanese)
   642  		} else {
   643  			b.WriteString("0123456789")
   644  		}
   645  	}
   646  	longStringMostlyASCII = b.String()
   647  	longStringJapanese = strings.Repeat(japanese, 100_000/len(japanese))
   648  }
   649  
   650  func BenchmarkEncodeASCIIRune(b *testing.B) {
   651  	buf := make([]byte, UTFMax)
   652  	for i := 0; i < b.N; i++ {
   653  		EncodeRune(buf, 'a') // 1 byte
   654  	}
   655  }
   656  
   657  func BenchmarkEncodeSpanishRune(b *testing.B) {
   658  	buf := make([]byte, UTFMax)
   659  	for i := 0; i < b.N; i++ {
   660  		EncodeRune(buf, 'Ñ') // 2 bytes
   661  	}
   662  }
   663  
   664  func BenchmarkEncodeJapaneseRune(b *testing.B) {
   665  	buf := make([]byte, UTFMax)
   666  	for i := 0; i < b.N; i++ {
   667  		EncodeRune(buf, '本') // 3 bytes
   668  	}
   669  }
   670  
   671  func BenchmarkEncodeMaxRune(b *testing.B) {
   672  	buf := make([]byte, UTFMax)
   673  	for i := 0; i < b.N; i++ {
   674  		EncodeRune(buf, MaxRune) // 4 bytes
   675  	}
   676  }
   677  
   678  func BenchmarkEncodeInvalidRuneMaxPlusOne(b *testing.B) {
   679  	buf := make([]byte, UTFMax)
   680  	for i := 0; i < b.N; i++ {
   681  		EncodeRune(buf, MaxRune+1) // 3 bytes: RuneError
   682  	}
   683  }
   684  
   685  func BenchmarkEncodeInvalidRuneSurrogate(b *testing.B) {
   686  	buf := make([]byte, UTFMax)
   687  	for i := 0; i < b.N; i++ {
   688  		EncodeRune(buf, 0xD800) // 3 bytes: RuneError
   689  	}
   690  }
   691  
   692  func BenchmarkEncodeInvalidRuneNegative(b *testing.B) {
   693  	buf := make([]byte, UTFMax)
   694  	for i := 0; i < b.N; i++ {
   695  		EncodeRune(buf, -1) // 3 bytes: RuneError
   696  	}
   697  }
   698  
   699  func BenchmarkAppendASCIIRune(b *testing.B) {
   700  	buf := make([]byte, UTFMax)
   701  	for i := 0; i < b.N; i++ {
   702  		AppendRune(buf[:0], 'a') // 1 byte
   703  	}
   704  }
   705  
   706  func BenchmarkAppendSpanishRune(b *testing.B) {
   707  	buf := make([]byte, UTFMax)
   708  	for i := 0; i < b.N; i++ {
   709  		AppendRune(buf[:0], 'Ñ') // 2 bytes
   710  	}
   711  }
   712  
   713  func BenchmarkAppendJapaneseRune(b *testing.B) {
   714  	buf := make([]byte, UTFMax)
   715  	for i := 0; i < b.N; i++ {
   716  		AppendRune(buf[:0], '本') // 3 bytes
   717  	}
   718  }
   719  
   720  func BenchmarkAppendMaxRune(b *testing.B) {
   721  	buf := make([]byte, UTFMax)
   722  	for i := 0; i < b.N; i++ {
   723  		AppendRune(buf[:0], MaxRune) // 4 bytes
   724  	}
   725  }
   726  
   727  func BenchmarkAppendInvalidRuneMaxPlusOne(b *testing.B) {
   728  	buf := make([]byte, UTFMax)
   729  	for i := 0; i < b.N; i++ {
   730  		AppendRune(buf[:0], MaxRune+1) // 3 bytes: RuneError
   731  	}
   732  }
   733  
   734  func BenchmarkAppendInvalidRuneSurrogate(b *testing.B) {
   735  	buf := make([]byte, UTFMax)
   736  	for i := 0; i < b.N; i++ {
   737  		AppendRune(buf[:0], 0xD800) // 3 bytes: RuneError
   738  	}
   739  }
   740  
   741  func BenchmarkAppendInvalidRuneNegative(b *testing.B) {
   742  	buf := make([]byte, UTFMax)
   743  	for i := 0; i < b.N; i++ {
   744  		AppendRune(buf[:0], -1) // 3 bytes: RuneError
   745  	}
   746  }
   747  
   748  func BenchmarkDecodeASCIIRune(b *testing.B) {
   749  	a := []byte{'a'}
   750  	for i := 0; i < b.N; i++ {
   751  		DecodeRune(a)
   752  	}
   753  }
   754  
   755  func BenchmarkDecodeJapaneseRune(b *testing.B) {
   756  	nihon := []byte("本")
   757  	for i := 0; i < b.N; i++ {
   758  		DecodeRune(nihon)
   759  	}
   760  }
   761  
   762  // boolSink is used to reference the return value of benchmarked
   763  // functions to avoid dead code elimination.
   764  var boolSink bool
   765  
   766  func BenchmarkFullRune(b *testing.B) {
   767  	benchmarks := []struct {
   768  		name string
   769  		data []byte
   770  	}{
   771  		{"ASCII", []byte("a")},
   772  		{"Incomplete", []byte("\xf0\x90\x80")},
   773  		{"Japanese", []byte("本")},
   774  	}
   775  	for _, bm := range benchmarks {
   776  		b.Run(bm.name, func(b *testing.B) {
   777  			for i := 0; i < b.N; i++ {
   778  				boolSink = FullRune(bm.data)
   779  			}
   780  		})
   781  	}
   782  }
   783
View as plain text