utf16_test.go

Documentation: unicode/utf16

     1  // Copyright 2010 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 utf16_test
     6  
     7  import (
     8  	"internal/testenv"
     9  	"reflect"
    10  	"testing"
    11  	"unicode"
    12  	. "unicode/utf16"
    13  )
    14  
    15  // Validate the constants redefined from unicode.
    16  func TestConstants(t *testing.T) {
    17  	if MaxRune != unicode.MaxRune {
    18  		t.Errorf("utf16.maxRune is wrong: %x should be %x", MaxRune, unicode.MaxRune)
    19  	}
    20  	if ReplacementChar != unicode.ReplacementChar {
    21  		t.Errorf("utf16.replacementChar is wrong: %x should be %x", ReplacementChar, unicode.ReplacementChar)
    22  	}
    23  }
    24  
    25  type encodeTest struct {
    26  	in  []rune
    27  	out []uint16
    28  }
    29  
    30  var encodeTests = []encodeTest{
    31  	{[]rune{1, 2, 3, 4}, []uint16{1, 2, 3, 4}},
    32  	{[]rune{0xffff, 0x10000, 0x10001, 0x12345, 0x10ffff},
    33  		[]uint16{0xffff, 0xd800, 0xdc00, 0xd800, 0xdc01, 0xd808, 0xdf45, 0xdbff, 0xdfff}},
    34  	{[]rune{'a', 'b', 0xd7ff, 0xd800, 0xdfff, 0xe000, 0x110000, -1},
    35  		[]uint16{'a', 'b', 0xd7ff, 0xfffd, 0xfffd, 0xe000, 0xfffd, 0xfffd}},
    36  }
    37  
    38  func TestEncode(t *testing.T) {
    39  	for _, tt := range encodeTests {
    40  		out := Encode(tt.in)
    41  		if !reflect.DeepEqual(out, tt.out) {
    42  			t.Errorf("Encode(%x) = %x; want %x", tt.in, out, tt.out)
    43  		}
    44  	}
    45  }
    46  
    47  func TestAppendRune(t *testing.T) {
    48  	for _, tt := range encodeTests {
    49  		var out []uint16
    50  		for _, u := range tt.in {
    51  			out = AppendRune(out, u)
    52  		}
    53  		if !reflect.DeepEqual(out, tt.out) {
    54  			t.Errorf("AppendRune(%x) = %x; want %x", tt.in, out, tt.out)
    55  		}
    56  	}
    57  }
    58  
    59  func TestEncodeRune(t *testing.T) {
    60  	for i, tt := range encodeTests {
    61  		j := 0
    62  		for _, r := range tt.in {
    63  			r1, r2 := EncodeRune(r)
    64  			if r < 0x10000 || r > unicode.MaxRune {
    65  				if j >= len(tt.out) {
    66  					t.Errorf("#%d: ran out of tt.out", i)
    67  					break
    68  				}
    69  				if r1 != unicode.ReplacementChar || r2 != unicode.ReplacementChar {
    70  					t.Errorf("EncodeRune(%#x) = %#x, %#x; want 0xfffd, 0xfffd", r, r1, r2)
    71  				}
    72  				j++
    73  			} else {
    74  				if j+1 >= len(tt.out) {
    75  					t.Errorf("#%d: ran out of tt.out", i)
    76  					break
    77  				}
    78  				if r1 != rune(tt.out[j]) || r2 != rune(tt.out[j+1]) {
    79  					t.Errorf("EncodeRune(%#x) = %#x, %#x; want %#x, %#x", r, r1, r2, tt.out[j], tt.out[j+1])
    80  				}
    81  				j += 2
    82  				dec := DecodeRune(r1, r2)
    83  				if dec != r {
    84  					t.Errorf("DecodeRune(%#x, %#x) = %#x; want %#x", r1, r2, dec, r)
    85  				}
    86  			}
    87  		}
    88  		if j != len(tt.out) {
    89  			t.Errorf("#%d: EncodeRune didn't generate enough output", i)
    90  		}
    91  	}
    92  }
    93  
    94  type decodeTest struct {
    95  	in  []uint16
    96  	out []rune
    97  }
    98  
    99  var decodeTests = []decodeTest{
   100  	{[]uint16{1, 2, 3, 4}, []rune{1, 2, 3, 4}},
   101  	{[]uint16{0xffff, 0xd800, 0xdc00, 0xd800, 0xdc01, 0xd808, 0xdf45, 0xdbff, 0xdfff},
   102  		[]rune{0xffff, 0x10000, 0x10001, 0x12345, 0x10ffff}},
   103  	{[]uint16{0xd800, 'a'}, []rune{0xfffd, 'a'}},
   104  	{[]uint16{0xdfff}, []rune{0xfffd}},
   105  }
   106  
   107  func TestAllocationsDecode(t *testing.T) {
   108  	testenv.SkipIfOptimizationOff(t)
   109  
   110  	for _, tt := range decodeTests {
   111  		allocs := testing.AllocsPerRun(10, func() {
   112  			out := Decode(tt.in)
   113  			if out == nil {
   114  				t.Errorf("Decode(%x) = nil", tt.in)
   115  			}
   116  		})
   117  		if allocs > 0 {
   118  			t.Errorf("Decode allocated %v times", allocs)
   119  		}
   120  	}
   121  }
   122  
   123  func TestDecode(t *testing.T) {
   124  	for _, tt := range decodeTests {
   125  		out := Decode(tt.in)
   126  		if !reflect.DeepEqual(out, tt.out) {
   127  			t.Errorf("Decode(%x) = %x; want %x", tt.in, out, tt.out)
   128  		}
   129  	}
   130  }
   131  
   132  var decodeRuneTests = []struct {
   133  	r1, r2 rune
   134  	want   rune
   135  }{
   136  	{0xd800, 0xdc00, 0x10000},
   137  	{0xd800, 0xdc01, 0x10001},
   138  	{0xd808, 0xdf45, 0x12345},
   139  	{0xdbff, 0xdfff, 0x10ffff},
   140  	{0xd800, 'a', 0xfffd}, // illegal, replacement rune substituted
   141  }
   142  
   143  func TestDecodeRune(t *testing.T) {
   144  	for i, tt := range decodeRuneTests {
   145  		got := DecodeRune(tt.r1, tt.r2)
   146  		if got != tt.want {
   147  			t.Errorf("%d: DecodeRune(%q, %q) = %v; want %v", i, tt.r1, tt.r2, got, tt.want)
   148  		}
   149  	}
   150  }
   151  
   152  var surrogateTests = []struct {
   153  	r    rune
   154  	want bool
   155  }{
   156  	// from https://en.wikipedia.org/wiki/UTF-16
   157  	{'\u007A', false},     // LATIN SMALL LETTER Z
   158  	{'\u6C34', false},     // CJK UNIFIED IDEOGRAPH-6C34 (water)
   159  	{'\uFEFF', false},     // Byte Order Mark
   160  	{'\U00010000', false}, // LINEAR B SYLLABLE B008 A (first non-BMP code point)
   161  	{'\U0001D11E', false}, // MUSICAL SYMBOL G CLEF
   162  	{'\U0010FFFD', false}, // PRIVATE USE CHARACTER-10FFFD (last Unicode code point)
   163  
   164  	{rune(0xd7ff), false}, // surr1-1
   165  	{rune(0xd800), true},  // surr1
   166  	{rune(0xdc00), true},  // surr2
   167  	{rune(0xe000), false}, // surr3
   168  	{rune(0xdfff), true},  // surr3-1
   169  }
   170  
   171  func TestIsSurrogate(t *testing.T) {
   172  	for i, tt := range surrogateTests {
   173  		got := IsSurrogate(tt.r)
   174  		if got != tt.want {
   175  			t.Errorf("%d: IsSurrogate(%q) = %v; want %v", i, tt.r, got, tt.want)
   176  		}
   177  	}
   178  }
   179  
   180  func BenchmarkDecodeValidASCII(b *testing.B) {
   181  	// "hello world"
   182  	data := []uint16{104, 101, 108, 108, 111, 32, 119, 111, 114, 108, 100}
   183  	for i := 0; i < b.N; i++ {
   184  		Decode(data)
   185  	}
   186  }
   187  
   188  func BenchmarkDecodeValidJapaneseChars(b *testing.B) {
   189  	// "日本語日本語日本語"
   190  	data := []uint16{26085, 26412, 35486, 26085, 26412, 35486, 26085, 26412, 35486}
   191  	for i := 0; i < b.N; i++ {
   192  		Decode(data)
   193  	}
   194  }
   195  
   196  func BenchmarkDecodeRune(b *testing.B) {
   197  	rs := make([]rune, 10)
   198  	// U+1D4D0 to U+1D4D4: MATHEMATICAL BOLD SCRIPT CAPITAL LETTERS
   199  	for i, u := range []rune{'𝓐', '𝓑', '𝓒', '𝓓', '𝓔'} {
   200  		rs[2*i], rs[2*i+1] = EncodeRune(u)
   201  	}
   202  
   203  	b.ResetTimer()
   204  	for i := 0; i < b.N; i++ {
   205  		for j := 0; j < 5; j++ {
   206  			DecodeRune(rs[2*j], rs[2*j+1])
   207  		}
   208  	}
   209  }
   210  
   211  func BenchmarkEncodeValidASCII(b *testing.B) {
   212  	data := []rune{'h', 'e', 'l', 'l', 'o'}
   213  	for i := 0; i < b.N; i++ {
   214  		Encode(data)
   215  	}
   216  }
   217  
   218  func BenchmarkEncodeValidJapaneseChars(b *testing.B) {
   219  	data := []rune{'日', '本', '語'}
   220  	for i := 0; i < b.N; i++ {
   221  		Encode(data)
   222  	}
   223  }
   224  
   225  func BenchmarkAppendRuneValidASCII(b *testing.B) {
   226  	data := []rune{'h', 'e', 'l', 'l', 'o'}
   227  	a := make([]uint16, 0, len(data)*2)
   228  	for i := 0; i < b.N; i++ {
   229  		for _, u := range data {
   230  			a = AppendRune(a, u)
   231  		}
   232  		a = a[:0]
   233  	}
   234  }
   235  
   236  func BenchmarkAppendRuneValidJapaneseChars(b *testing.B) {
   237  	data := []rune{'日', '本', '語'}
   238  	a := make([]uint16, 0, len(data)*2)
   239  	for i := 0; i < b.N; i++ {
   240  		for _, u := range data {
   241  			a = AppendRune(a, u)
   242  		}
   243  		a = a[:0]
   244  	}
   245  }
   246  
   247  func BenchmarkEncodeRune(b *testing.B) {
   248  	for i := 0; i < b.N; i++ {
   249  		for _, u := range []rune{'𝓐', '𝓑', '𝓒', '𝓓', '𝓔'} {
   250  			EncodeRune(u)
   251  		}
   252  	}
   253  }
   254
View as plain text