utf16_test.go

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