Source file src/bytes/bytes_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 bytes_test
     6  
     7  import (
     8  	. "bytes"
     9  	"fmt"
    10  	"internal/testenv"
    11  	"math"
    12  	"math/rand"
    13  	"reflect"
    14  	"strings"
    15  	"testing"
    16  	"unicode"
    17  	"unicode/utf8"
    18  )
    19  
    20  func eq(a, b []string) bool {
    21  	if len(a) != len(b) {
    22  		return false
    23  	}
    24  	for i := 0; i < len(a); i++ {
    25  		if a[i] != b[i] {
    26  			return false
    27  		}
    28  	}
    29  	return true
    30  }
    31  
    32  func sliceOfString(s [][]byte) []string {
    33  	result := make([]string, len(s))
    34  	for i, v := range s {
    35  		result[i] = string(v)
    36  	}
    37  	return result
    38  }
    39  
    40  // For ease of reading, the test cases use strings that are converted to byte
    41  // slices before invoking the functions.
    42  
    43  var abcd = "abcd"
    44  var faces = "☺☻☹"
    45  var commas = "1,2,3,4"
    46  var dots = "1....2....3....4"
    47  
    48  type BinOpTest struct {
    49  	a string
    50  	b string
    51  	i int
    52  }
    53  
    54  func TestEqual(t *testing.T) {
    55  	// Run the tests and check for allocation at the same time.
    56  	allocs := testing.AllocsPerRun(10, func() {
    57  		for _, tt := range compareTests {
    58  			eql := Equal(tt.a, tt.b)
    59  			if eql != (tt.i == 0) {
    60  				t.Errorf(`Equal(%q, %q) = %v`, tt.a, tt.b, eql)
    61  			}
    62  		}
    63  	})
    64  	if allocs > 0 {
    65  		t.Errorf("Equal allocated %v times", allocs)
    66  	}
    67  }
    68  
    69  func TestEqualExhaustive(t *testing.T) {
    70  	var size = 128
    71  	if testing.Short() {
    72  		size = 32
    73  	}
    74  	a := make([]byte, size)
    75  	b := make([]byte, size)
    76  	b_init := make([]byte, size)
    77  	// randomish but deterministic data
    78  	for i := 0; i < size; i++ {
    79  		a[i] = byte(17 * i)
    80  		b_init[i] = byte(23*i + 100)
    81  	}
    82  
    83  	for len := 0; len <= size; len++ {
    84  		for x := 0; x <= size-len; x++ {
    85  			for y := 0; y <= size-len; y++ {
    86  				copy(b, b_init)
    87  				copy(b[y:y+len], a[x:x+len])
    88  				if !Equal(a[x:x+len], b[y:y+len]) || !Equal(b[y:y+len], a[x:x+len]) {
    89  					t.Errorf("Equal(%d, %d, %d) = false", len, x, y)
    90  				}
    91  			}
    92  		}
    93  	}
    94  }
    95  
    96  // make sure Equal returns false for minimally different strings. The data
    97  // is all zeros except for a single one in one location.
    98  func TestNotEqual(t *testing.T) {
    99  	var size = 128
   100  	if testing.Short() {
   101  		size = 32
   102  	}
   103  	a := make([]byte, size)
   104  	b := make([]byte, size)
   105  
   106  	for len := 0; len <= size; len++ {
   107  		for x := 0; x <= size-len; x++ {
   108  			for y := 0; y <= size-len; y++ {
   109  				for diffpos := x; diffpos < x+len; diffpos++ {
   110  					a[diffpos] = 1
   111  					if Equal(a[x:x+len], b[y:y+len]) || Equal(b[y:y+len], a[x:x+len]) {
   112  						t.Errorf("NotEqual(%d, %d, %d, %d) = true", len, x, y, diffpos)
   113  					}
   114  					a[diffpos] = 0
   115  				}
   116  			}
   117  		}
   118  	}
   119  }
   120  
   121  var indexTests = []BinOpTest{
   122  	{"", "", 0},
   123  	{"", "a", -1},
   124  	{"", "foo", -1},
   125  	{"fo", "foo", -1},
   126  	{"foo", "baz", -1},
   127  	{"foo", "foo", 0},
   128  	{"oofofoofooo", "f", 2},
   129  	{"oofofoofooo", "foo", 4},
   130  	{"barfoobarfoo", "foo", 3},
   131  	{"foo", "", 0},
   132  	{"foo", "o", 1},
   133  	{"abcABCabc", "A", 3},
   134  	// cases with one byte strings - test IndexByte and special case in Index()
   135  	{"", "a", -1},
   136  	{"x", "a", -1},
   137  	{"x", "x", 0},
   138  	{"abc", "a", 0},
   139  	{"abc", "b", 1},
   140  	{"abc", "c", 2},
   141  	{"abc", "x", -1},
   142  	{"barfoobarfooyyyzzzyyyzzzyyyzzzyyyxxxzzzyyy", "x", 33},
   143  	{"fofofofooofoboo", "oo", 7},
   144  	{"fofofofofofoboo", "ob", 11},
   145  	{"fofofofofofoboo", "boo", 12},
   146  	{"fofofofofofoboo", "oboo", 11},
   147  	{"fofofofofoooboo", "fooo", 8},
   148  	{"fofofofofofoboo", "foboo", 10},
   149  	{"fofofofofofoboo", "fofob", 8},
   150  	{"fofofofofofofoffofoobarfoo", "foffof", 12},
   151  	{"fofofofofoofofoffofoobarfoo", "foffof", 13},
   152  	{"fofofofofofofoffofoobarfoo", "foffofo", 12},
   153  	{"fofofofofoofofoffofoobarfoo", "foffofo", 13},
   154  	{"fofofofofoofofoffofoobarfoo", "foffofoo", 13},
   155  	{"fofofofofofofoffofoobarfoo", "foffofoo", 12},
   156  	{"fofofofofoofofoffofoobarfoo", "foffofoob", 13},
   157  	{"fofofofofofofoffofoobarfoo", "foffofoob", 12},
   158  	{"fofofofofoofofoffofoobarfoo", "foffofooba", 13},
   159  	{"fofofofofofofoffofoobarfoo", "foffofooba", 12},
   160  	{"fofofofofoofofoffofoobarfoo", "foffofoobar", 13},
   161  	{"fofofofofofofoffofoobarfoo", "foffofoobar", 12},
   162  	{"fofofofofoofofoffofoobarfoo", "foffofoobarf", 13},
   163  	{"fofofofofofofoffofoobarfoo", "foffofoobarf", 12},
   164  	{"fofofofofoofofoffofoobarfoo", "foffofoobarfo", 13},
   165  	{"fofofofofofofoffofoobarfoo", "foffofoobarfo", 12},
   166  	{"fofofofofoofofoffofoobarfoo", "foffofoobarfoo", 13},
   167  	{"fofofofofofofoffofoobarfoo", "foffofoobarfoo", 12},
   168  	{"fofofofofoofofoffofoobarfoo", "ofoffofoobarfoo", 12},
   169  	{"fofofofofofofoffofoobarfoo", "ofoffofoobarfoo", 11},
   170  	{"fofofofofoofofoffofoobarfoo", "fofoffofoobarfoo", 11},
   171  	{"fofofofofofofoffofoobarfoo", "fofoffofoobarfoo", 10},
   172  	{"fofofofofoofofoffofoobarfoo", "foobars", -1},
   173  	{"foofyfoobarfoobar", "y", 4},
   174  	{"oooooooooooooooooooooo", "r", -1},
   175  	{"oxoxoxoxoxoxoxoxoxoxoxoy", "oy", 22},
   176  	{"oxoxoxoxoxoxoxoxoxoxoxox", "oy", -1},
   177  	// test fallback to Rabin-Karp.
   178  	{"000000000000000000000000000000000000000000000000000000000000000000000001", "0000000000000000000000000000000000000000000000000000000000000000001", 5},
   179  }
   180  
   181  var lastIndexTests = []BinOpTest{
   182  	{"", "", 0},
   183  	{"", "a", -1},
   184  	{"", "foo", -1},
   185  	{"fo", "foo", -1},
   186  	{"foo", "foo", 0},
   187  	{"foo", "f", 0},
   188  	{"oofofoofooo", "f", 7},
   189  	{"oofofoofooo", "foo", 7},
   190  	{"barfoobarfoo", "foo", 9},
   191  	{"foo", "", 3},
   192  	{"foo", "o", 2},
   193  	{"abcABCabc", "A", 3},
   194  	{"abcABCabc", "a", 6},
   195  }
   196  
   197  var indexAnyTests = []BinOpTest{
   198  	{"", "", -1},
   199  	{"", "a", -1},
   200  	{"", "abc", -1},
   201  	{"a", "", -1},
   202  	{"a", "a", 0},
   203  	{"\x80", "\xffb", 0},
   204  	{"aaa", "a", 0},
   205  	{"abc", "xyz", -1},
   206  	{"abc", "xcz", 2},
   207  	{"ab☺c", "x☺yz", 2},
   208  	{"a☺b☻c☹d", "cx", len("a☺b☻")},
   209  	{"a☺b☻c☹d", "uvw☻xyz", len("a☺b")},
   210  	{"aRegExp*", ".(|)*+?^$[]", 7},
   211  	{dots + dots + dots, " ", -1},
   212  	{"012abcba210", "\xffb", 4},
   213  	{"012\x80bcb\x80210", "\xffb", 3},
   214  	{"0123456\xcf\x80abc", "\xcfb\x80", 10},
   215  }
   216  
   217  var lastIndexAnyTests = []BinOpTest{
   218  	{"", "", -1},
   219  	{"", "a", -1},
   220  	{"", "abc", -1},
   221  	{"a", "", -1},
   222  	{"a", "a", 0},
   223  	{"\x80", "\xffb", 0},
   224  	{"aaa", "a", 2},
   225  	{"abc", "xyz", -1},
   226  	{"abc", "ab", 1},
   227  	{"ab☺c", "x☺yz", 2},
   228  	{"a☺b☻c☹d", "cx", len("a☺b☻")},
   229  	{"a☺b☻c☹d", "uvw☻xyz", len("a☺b")},
   230  	{"a.RegExp*", ".(|)*+?^$[]", 8},
   231  	{dots + dots + dots, " ", -1},
   232  	{"012abcba210", "\xffb", 6},
   233  	{"012\x80bcb\x80210", "\xffb", 7},
   234  	{"0123456\xcf\x80abc", "\xcfb\x80", 10},
   235  }
   236  
   237  // Execute f on each test case.  funcName should be the name of f; it's used
   238  // in failure reports.
   239  func runIndexTests(t *testing.T, f func(s, sep []byte) int, funcName string, testCases []BinOpTest) {
   240  	for _, test := range testCases {
   241  		a := []byte(test.a)
   242  		b := []byte(test.b)
   243  		actual := f(a, b)
   244  		if actual != test.i {
   245  			t.Errorf("%s(%q,%q) = %v; want %v", funcName, a, b, actual, test.i)
   246  		}
   247  	}
   248  	var allocTests = []struct {
   249  		a []byte
   250  		b []byte
   251  		i int
   252  	}{
   253  		// case for function Index.
   254  		{[]byte("000000000000000000000000000000000000000000000000000000000000000000000001"), []byte("0000000000000000000000000000000000000000000000000000000000000000001"), 5},
   255  		// case for function LastIndex.
   256  		{[]byte("000000000000000000000000000000000000000000000000000000000000000010000"), []byte("00000000000000000000000000000000000000000000000000000000000001"), 3},
   257  	}
   258  	allocs := testing.AllocsPerRun(100, func() {
   259  		if i := Index(allocTests[1].a, allocTests[1].b); i != allocTests[1].i {
   260  			t.Errorf("Index([]byte(%q), []byte(%q)) = %v; want %v", allocTests[1].a, allocTests[1].b, i, allocTests[1].i)
   261  		}
   262  		if i := LastIndex(allocTests[0].a, allocTests[0].b); i != allocTests[0].i {
   263  			t.Errorf("LastIndex([]byte(%q), []byte(%q)) = %v; want %v", allocTests[0].a, allocTests[0].b, i, allocTests[0].i)
   264  		}
   265  	})
   266  	if allocs != 0 {
   267  		t.Errorf("expected no allocations, got %f", allocs)
   268  	}
   269  }
   270  
   271  func runIndexAnyTests(t *testing.T, f func(s []byte, chars string) int, funcName string, testCases []BinOpTest) {
   272  	for _, test := range testCases {
   273  		a := []byte(test.a)
   274  		actual := f(a, test.b)
   275  		if actual != test.i {
   276  			t.Errorf("%s(%q,%q) = %v; want %v", funcName, a, test.b, actual, test.i)
   277  		}
   278  	}
   279  }
   280  
   281  func TestIndex(t *testing.T)     { runIndexTests(t, Index, "Index", indexTests) }
   282  func TestLastIndex(t *testing.T) { runIndexTests(t, LastIndex, "LastIndex", lastIndexTests) }
   283  func TestIndexAny(t *testing.T)  { runIndexAnyTests(t, IndexAny, "IndexAny", indexAnyTests) }
   284  func TestLastIndexAny(t *testing.T) {
   285  	runIndexAnyTests(t, LastIndexAny, "LastIndexAny", lastIndexAnyTests)
   286  }
   287  
   288  func TestIndexByte(t *testing.T) {
   289  	for _, tt := range indexTests {
   290  		if len(tt.b) != 1 {
   291  			continue
   292  		}
   293  		a := []byte(tt.a)
   294  		b := tt.b[0]
   295  		pos := IndexByte(a, b)
   296  		if pos != tt.i {
   297  			t.Errorf(`IndexByte(%q, '%c') = %v`, tt.a, b, pos)
   298  		}
   299  		posp := IndexBytePortable(a, b)
   300  		if posp != tt.i {
   301  			t.Errorf(`indexBytePortable(%q, '%c') = %v`, tt.a, b, posp)
   302  		}
   303  	}
   304  }
   305  
   306  func TestLastIndexByte(t *testing.T) {
   307  	testCases := []BinOpTest{
   308  		{"", "q", -1},
   309  		{"abcdef", "q", -1},
   310  		{"abcdefabcdef", "a", len("abcdef")},      // something in the middle
   311  		{"abcdefabcdef", "f", len("abcdefabcde")}, // last byte
   312  		{"zabcdefabcdef", "z", 0},                 // first byte
   313  		{"a☺b☻c☹d", "b", len("a☺")},               // non-ascii
   314  	}
   315  	for _, test := range testCases {
   316  		actual := LastIndexByte([]byte(test.a), test.b[0])
   317  		if actual != test.i {
   318  			t.Errorf("LastIndexByte(%q,%c) = %v; want %v", test.a, test.b[0], actual, test.i)
   319  		}
   320  	}
   321  }
   322  
   323  // test a larger buffer with different sizes and alignments
   324  func TestIndexByteBig(t *testing.T) {
   325  	var n = 1024
   326  	if testing.Short() {
   327  		n = 128
   328  	}
   329  	b := make([]byte, n)
   330  	for i := 0; i < n; i++ {
   331  		// different start alignments
   332  		b1 := b[i:]
   333  		for j := 0; j < len(b1); j++ {
   334  			b1[j] = 'x'
   335  			pos := IndexByte(b1, 'x')
   336  			if pos != j {
   337  				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
   338  			}
   339  			b1[j] = 0
   340  			pos = IndexByte(b1, 'x')
   341  			if pos != -1 {
   342  				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
   343  			}
   344  		}
   345  		// different end alignments
   346  		b1 = b[:i]
   347  		for j := 0; j < len(b1); j++ {
   348  			b1[j] = 'x'
   349  			pos := IndexByte(b1, 'x')
   350  			if pos != j {
   351  				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
   352  			}
   353  			b1[j] = 0
   354  			pos = IndexByte(b1, 'x')
   355  			if pos != -1 {
   356  				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
   357  			}
   358  		}
   359  		// different start and end alignments
   360  		b1 = b[i/2 : n-(i+1)/2]
   361  		for j := 0; j < len(b1); j++ {
   362  			b1[j] = 'x'
   363  			pos := IndexByte(b1, 'x')
   364  			if pos != j {
   365  				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
   366  			}
   367  			b1[j] = 0
   368  			pos = IndexByte(b1, 'x')
   369  			if pos != -1 {
   370  				t.Errorf("IndexByte(%q, 'x') = %v", b1, pos)
   371  			}
   372  		}
   373  	}
   374  }
   375  
   376  // test a small index across all page offsets
   377  func TestIndexByteSmall(t *testing.T) {
   378  	b := make([]byte, 5015) // bigger than a page
   379  	// Make sure we find the correct byte even when straddling a page.
   380  	for i := 0; i <= len(b)-15; i++ {
   381  		for j := 0; j < 15; j++ {
   382  			b[i+j] = byte(100 + j)
   383  		}
   384  		for j := 0; j < 15; j++ {
   385  			p := IndexByte(b[i:i+15], byte(100+j))
   386  			if p != j {
   387  				t.Errorf("IndexByte(%q, %d) = %d", b[i:i+15], 100+j, p)
   388  			}
   389  		}
   390  		for j := 0; j < 15; j++ {
   391  			b[i+j] = 0
   392  		}
   393  	}
   394  	// Make sure matches outside the slice never trigger.
   395  	for i := 0; i <= len(b)-15; i++ {
   396  		for j := 0; j < 15; j++ {
   397  			b[i+j] = 1
   398  		}
   399  		for j := 0; j < 15; j++ {
   400  			p := IndexByte(b[i:i+15], byte(0))
   401  			if p != -1 {
   402  				t.Errorf("IndexByte(%q, %d) = %d", b[i:i+15], 0, p)
   403  			}
   404  		}
   405  		for j := 0; j < 15; j++ {
   406  			b[i+j] = 0
   407  		}
   408  	}
   409  }
   410  
   411  func TestIndexRune(t *testing.T) {
   412  	tests := []struct {
   413  		in   string
   414  		rune rune
   415  		want int
   416  	}{
   417  		{"", 'a', -1},
   418  		{"", '☺', -1},
   419  		{"foo", '☹', -1},
   420  		{"foo", 'o', 1},
   421  		{"foo☺bar", '☺', 3},
   422  		{"foo☺☻☹bar", '☹', 9},
   423  		{"a A x", 'A', 2},
   424  		{"some_text=some_value", '=', 9},
   425  		{"☺a", 'a', 3},
   426  		{"a☻☺b", '☺', 4},
   427  
   428  		// RuneError should match any invalid UTF-8 byte sequence.
   429  		{"�", '�', 0},
   430  		{"\xff", '�', 0},
   431  		{"☻x�", '�', len("☻x")},
   432  		{"☻x\xe2\x98", '�', len("☻x")},
   433  		{"☻x\xe2\x98�", '�', len("☻x")},
   434  		{"☻x\xe2\x98x", '�', len("☻x")},
   435  
   436  		// Invalid rune values should never match.
   437  		{"a☺b☻c☹d\xe2\x98�\xff�\xed\xa0\x80", -1, -1},
   438  		{"a☺b☻c☹d\xe2\x98�\xff�\xed\xa0\x80", 0xD800, -1}, // Surrogate pair
   439  		{"a☺b☻c☹d\xe2\x98�\xff�\xed\xa0\x80", utf8.MaxRune + 1, -1},
   440  	}
   441  	for _, tt := range tests {
   442  		if got := IndexRune([]byte(tt.in), tt.rune); got != tt.want {
   443  			t.Errorf("IndexRune(%q, %d) = %v; want %v", tt.in, tt.rune, got, tt.want)
   444  		}
   445  	}
   446  
   447  	haystack := []byte("test世界")
   448  	allocs := testing.AllocsPerRun(1000, func() {
   449  		if i := IndexRune(haystack, 's'); i != 2 {
   450  			t.Fatalf("'s' at %d; want 2", i)
   451  		}
   452  		if i := IndexRune(haystack, '世'); i != 4 {
   453  			t.Fatalf("'世' at %d; want 4", i)
   454  		}
   455  	})
   456  	if allocs != 0 {
   457  		t.Errorf("expected no allocations, got %f", allocs)
   458  	}
   459  }
   460  
   461  // test count of a single byte across page offsets
   462  func TestCountByte(t *testing.T) {
   463  	b := make([]byte, 5015) // bigger than a page
   464  	windows := []int{1, 2, 3, 4, 15, 16, 17, 31, 32, 33, 63, 64, 65, 128}
   465  	testCountWindow := func(i, window int) {
   466  		for j := 0; j < window; j++ {
   467  			b[i+j] = byte(100)
   468  			p := Count(b[i:i+window], []byte{100})
   469  			if p != j+1 {
   470  				t.Errorf("TestCountByte.Count(%q, 100) = %d", b[i:i+window], p)
   471  			}
   472  		}
   473  	}
   474  
   475  	maxWnd := windows[len(windows)-1]
   476  
   477  	for i := 0; i <= 2*maxWnd; i++ {
   478  		for _, window := range windows {
   479  			if window > len(b[i:]) {
   480  				window = len(b[i:])
   481  			}
   482  			testCountWindow(i, window)
   483  			for j := 0; j < window; j++ {
   484  				b[i+j] = byte(0)
   485  			}
   486  		}
   487  	}
   488  	for i := 4096 - (maxWnd + 1); i < len(b); i++ {
   489  		for _, window := range windows {
   490  			if window > len(b[i:]) {
   491  				window = len(b[i:])
   492  			}
   493  			testCountWindow(i, window)
   494  			for j := 0; j < window; j++ {
   495  				b[i+j] = byte(0)
   496  			}
   497  		}
   498  	}
   499  }
   500  
   501  // Make sure we don't count bytes outside our window
   502  func TestCountByteNoMatch(t *testing.T) {
   503  	b := make([]byte, 5015)
   504  	windows := []int{1, 2, 3, 4, 15, 16, 17, 31, 32, 33, 63, 64, 65, 128}
   505  	for i := 0; i <= len(b); i++ {
   506  		for _, window := range windows {
   507  			if window > len(b[i:]) {
   508  				window = len(b[i:])
   509  			}
   510  			// Fill the window with non-match
   511  			for j := 0; j < window; j++ {
   512  				b[i+j] = byte(100)
   513  			}
   514  			// Try to find something that doesn't exist
   515  			p := Count(b[i:i+window], []byte{0})
   516  			if p != 0 {
   517  				t.Errorf("TestCountByteNoMatch(%q, 0) = %d", b[i:i+window], p)
   518  			}
   519  			for j := 0; j < window; j++ {
   520  				b[i+j] = byte(0)
   521  			}
   522  		}
   523  	}
   524  }
   525  
   526  var bmbuf []byte
   527  
   528  func valName(x int) string {
   529  	if s := x >> 20; s<<20 == x {
   530  		return fmt.Sprintf("%dM", s)
   531  	}
   532  	if s := x >> 10; s<<10 == x {
   533  		return fmt.Sprintf("%dK", s)
   534  	}
   535  	return fmt.Sprint(x)
   536  }
   537  
   538  func benchBytes(b *testing.B, sizes []int, f func(b *testing.B, n int)) {
   539  	for _, n := range sizes {
   540  		if isRaceBuilder && n > 4<<10 {
   541  			continue
   542  		}
   543  		b.Run(valName(n), func(b *testing.B) {
   544  			if len(bmbuf) < n {
   545  				bmbuf = make([]byte, n)
   546  			}
   547  			b.SetBytes(int64(n))
   548  			f(b, n)
   549  		})
   550  	}
   551  }
   552  
   553  var indexSizes = []int{10, 32, 4 << 10, 4 << 20, 64 << 20}
   554  
   555  var isRaceBuilder = strings.HasSuffix(testenv.Builder(), "-race")
   556  
   557  func BenchmarkIndexByte(b *testing.B) {
   558  	benchBytes(b, indexSizes, bmIndexByte(IndexByte))
   559  }
   560  
   561  func BenchmarkIndexBytePortable(b *testing.B) {
   562  	benchBytes(b, indexSizes, bmIndexByte(IndexBytePortable))
   563  }
   564  
   565  func bmIndexByte(index func([]byte, byte) int) func(b *testing.B, n int) {
   566  	return func(b *testing.B, n int) {
   567  		buf := bmbuf[0:n]
   568  		buf[n-1] = 'x'
   569  		for i := 0; i < b.N; i++ {
   570  			j := index(buf, 'x')
   571  			if j != n-1 {
   572  				b.Fatal("bad index", j)
   573  			}
   574  		}
   575  		buf[n-1] = '\x00'
   576  	}
   577  }
   578  
   579  func BenchmarkIndexRune(b *testing.B) {
   580  	benchBytes(b, indexSizes, bmIndexRune(IndexRune))
   581  }
   582  
   583  func BenchmarkIndexRuneASCII(b *testing.B) {
   584  	benchBytes(b, indexSizes, bmIndexRuneASCII(IndexRune))
   585  }
   586  
   587  func bmIndexRuneASCII(index func([]byte, rune) int) func(b *testing.B, n int) {
   588  	return func(b *testing.B, n int) {
   589  		buf := bmbuf[0:n]
   590  		buf[n-1] = 'x'
   591  		for i := 0; i < b.N; i++ {
   592  			j := index(buf, 'x')
   593  			if j != n-1 {
   594  				b.Fatal("bad index", j)
   595  			}
   596  		}
   597  		buf[n-1] = '\x00'
   598  	}
   599  }
   600  
   601  func bmIndexRune(index func([]byte, rune) int) func(b *testing.B, n int) {
   602  	return func(b *testing.B, n int) {
   603  		buf := bmbuf[0:n]
   604  		utf8.EncodeRune(buf[n-3:], '世')
   605  		for i := 0; i < b.N; i++ {
   606  			j := index(buf, '世')
   607  			if j != n-3 {
   608  				b.Fatal("bad index", j)
   609  			}
   610  		}
   611  		buf[n-3] = '\x00'
   612  		buf[n-2] = '\x00'
   613  		buf[n-1] = '\x00'
   614  	}
   615  }
   616  
   617  func BenchmarkEqual(b *testing.B) {
   618  	b.Run("0", func(b *testing.B) {
   619  		var buf [4]byte
   620  		buf1 := buf[0:0]
   621  		buf2 := buf[1:1]
   622  		for i := 0; i < b.N; i++ {
   623  			eq := Equal(buf1, buf2)
   624  			if !eq {
   625  				b.Fatal("bad equal")
   626  			}
   627  		}
   628  	})
   629  
   630  	sizes := []int{1, 6, 9, 15, 16, 20, 32, 4 << 10, 4 << 20, 64 << 20}
   631  	benchBytes(b, sizes, bmEqual(Equal))
   632  }
   633  
   634  func bmEqual(equal func([]byte, []byte) bool) func(b *testing.B, n int) {
   635  	return func(b *testing.B, n int) {
   636  		if len(bmbuf) < 2*n {
   637  			bmbuf = make([]byte, 2*n)
   638  		}
   639  		buf1 := bmbuf[0:n]
   640  		buf2 := bmbuf[n : 2*n]
   641  		buf1[n-1] = 'x'
   642  		buf2[n-1] = 'x'
   643  		for i := 0; i < b.N; i++ {
   644  			eq := equal(buf1, buf2)
   645  			if !eq {
   646  				b.Fatal("bad equal")
   647  			}
   648  		}
   649  		buf1[n-1] = '\x00'
   650  		buf2[n-1] = '\x00'
   651  	}
   652  }
   653  
   654  func BenchmarkIndex(b *testing.B) {
   655  	benchBytes(b, indexSizes, func(b *testing.B, n int) {
   656  		buf := bmbuf[0:n]
   657  		buf[n-1] = 'x'
   658  		for i := 0; i < b.N; i++ {
   659  			j := Index(buf, buf[n-7:])
   660  			if j != n-7 {
   661  				b.Fatal("bad index", j)
   662  			}
   663  		}
   664  		buf[n-1] = '\x00'
   665  	})
   666  }
   667  
   668  func BenchmarkIndexEasy(b *testing.B) {
   669  	benchBytes(b, indexSizes, func(b *testing.B, n int) {
   670  		buf := bmbuf[0:n]
   671  		buf[n-1] = 'x'
   672  		buf[n-7] = 'x'
   673  		for i := 0; i < b.N; i++ {
   674  			j := Index(buf, buf[n-7:])
   675  			if j != n-7 {
   676  				b.Fatal("bad index", j)
   677  			}
   678  		}
   679  		buf[n-1] = '\x00'
   680  		buf[n-7] = '\x00'
   681  	})
   682  }
   683  
   684  func BenchmarkCount(b *testing.B) {
   685  	benchBytes(b, indexSizes, func(b *testing.B, n int) {
   686  		buf := bmbuf[0:n]
   687  		buf[n-1] = 'x'
   688  		for i := 0; i < b.N; i++ {
   689  			j := Count(buf, buf[n-7:])
   690  			if j != 1 {
   691  				b.Fatal("bad count", j)
   692  			}
   693  		}
   694  		buf[n-1] = '\x00'
   695  	})
   696  }
   697  
   698  func BenchmarkCountEasy(b *testing.B) {
   699  	benchBytes(b, indexSizes, func(b *testing.B, n int) {
   700  		buf := bmbuf[0:n]
   701  		buf[n-1] = 'x'
   702  		buf[n-7] = 'x'
   703  		for i := 0; i < b.N; i++ {
   704  			j := Count(buf, buf[n-7:])
   705  			if j != 1 {
   706  				b.Fatal("bad count", j)
   707  			}
   708  		}
   709  		buf[n-1] = '\x00'
   710  		buf[n-7] = '\x00'
   711  	})
   712  }
   713  
   714  func BenchmarkCountSingle(b *testing.B) {
   715  	benchBytes(b, indexSizes, func(b *testing.B, n int) {
   716  		buf := bmbuf[0:n]
   717  		step := 8
   718  		for i := 0; i < len(buf); i += step {
   719  			buf[i] = 1
   720  		}
   721  		expect := (len(buf) + (step - 1)) / step
   722  		for i := 0; i < b.N; i++ {
   723  			j := Count(buf, []byte{1})
   724  			if j != expect {
   725  				b.Fatal("bad count", j, expect)
   726  			}
   727  		}
   728  		for i := 0; i < len(buf); i++ {
   729  			buf[i] = 0
   730  		}
   731  	})
   732  }
   733  
   734  type SplitTest struct {
   735  	s   string
   736  	sep string
   737  	n   int
   738  	a   []string
   739  }
   740  
   741  var splittests = []SplitTest{
   742  	{"", "", -1, []string{}},
   743  	{abcd, "a", 0, nil},
   744  	{abcd, "", 2, []string{"a", "bcd"}},
   745  	{abcd, "a", -1, []string{"", "bcd"}},
   746  	{abcd, "z", -1, []string{"abcd"}},
   747  	{abcd, "", -1, []string{"a", "b", "c", "d"}},
   748  	{commas, ",", -1, []string{"1", "2", "3", "4"}},
   749  	{dots, "...", -1, []string{"1", ".2", ".3", ".4"}},
   750  	{faces, "☹", -1, []string{"☺☻", ""}},
   751  	{faces, "~", -1, []string{faces}},
   752  	{faces, "", -1, []string{"☺", "☻", "☹"}},
   753  	{"1 2 3 4", " ", 3, []string{"1", "2", "3 4"}},
   754  	{"1 2", " ", 3, []string{"1", "2"}},
   755  	{"123", "", 2, []string{"1", "23"}},
   756  	{"123", "", 17, []string{"1", "2", "3"}},
   757  	{"bT", "T", math.MaxInt / 4, []string{"b", ""}},
   758  }
   759  
   760  func TestSplit(t *testing.T) {
   761  	for _, tt := range splittests {
   762  		a := SplitN([]byte(tt.s), []byte(tt.sep), tt.n)
   763  
   764  		// Appending to the results should not change future results.
   765  		var x []byte
   766  		for _, v := range a {
   767  			x = append(v, 'z')
   768  		}
   769  
   770  		result := sliceOfString(a)
   771  		if !eq(result, tt.a) {
   772  			t.Errorf(`Split(%q, %q, %d) = %v; want %v`, tt.s, tt.sep, tt.n, result, tt.a)
   773  			continue
   774  		}
   775  		if tt.n == 0 || len(a) == 0 {
   776  			continue
   777  		}
   778  
   779  		if want := tt.a[len(tt.a)-1] + "z"; string(x) != want {
   780  			t.Errorf("last appended result was %s; want %s", x, want)
   781  		}
   782  
   783  		s := Join(a, []byte(tt.sep))
   784  		if string(s) != tt.s {
   785  			t.Errorf(`Join(Split(%q, %q, %d), %q) = %q`, tt.s, tt.sep, tt.n, tt.sep, s)
   786  		}
   787  		if tt.n < 0 {
   788  			b := Split([]byte(tt.s), []byte(tt.sep))
   789  			if !reflect.DeepEqual(a, b) {
   790  				t.Errorf("Split disagrees withSplitN(%q, %q, %d) = %v; want %v", tt.s, tt.sep, tt.n, b, a)
   791  			}
   792  		}
   793  		if len(a) > 0 {
   794  			in, out := a[0], s
   795  			if cap(in) == cap(out) && &in[:1][0] == &out[:1][0] {
   796  				t.Errorf("Join(%#v, %q) didn't copy", a, tt.sep)
   797  			}
   798  		}
   799  	}
   800  }
   801  
   802  var splitaftertests = []SplitTest{
   803  	{abcd, "a", -1, []string{"a", "bcd"}},
   804  	{abcd, "z", -1, []string{"abcd"}},
   805  	{abcd, "", -1, []string{"a", "b", "c", "d"}},
   806  	{commas, ",", -1, []string{"1,", "2,", "3,", "4"}},
   807  	{dots, "...", -1, []string{"1...", ".2...", ".3...", ".4"}},
   808  	{faces, "☹", -1, []string{"☺☻☹", ""}},
   809  	{faces, "~", -1, []string{faces}},
   810  	{faces, "", -1, []string{"☺", "☻", "☹"}},
   811  	{"1 2 3 4", " ", 3, []string{"1 ", "2 ", "3 4"}},
   812  	{"1 2 3", " ", 3, []string{"1 ", "2 ", "3"}},
   813  	{"1 2", " ", 3, []string{"1 ", "2"}},
   814  	{"123", "", 2, []string{"1", "23"}},
   815  	{"123", "", 17, []string{"1", "2", "3"}},
   816  }
   817  
   818  func TestSplitAfter(t *testing.T) {
   819  	for _, tt := range splitaftertests {
   820  		a := SplitAfterN([]byte(tt.s), []byte(tt.sep), tt.n)
   821  
   822  		// Appending to the results should not change future results.
   823  		var x []byte
   824  		for _, v := range a {
   825  			x = append(v, 'z')
   826  		}
   827  
   828  		result := sliceOfString(a)
   829  		if !eq(result, tt.a) {
   830  			t.Errorf(`Split(%q, %q, %d) = %v; want %v`, tt.s, tt.sep, tt.n, result, tt.a)
   831  			continue
   832  		}
   833  
   834  		if want := tt.a[len(tt.a)-1] + "z"; string(x) != want {
   835  			t.Errorf("last appended result was %s; want %s", x, want)
   836  		}
   837  
   838  		s := Join(a, nil)
   839  		if string(s) != tt.s {
   840  			t.Errorf(`Join(Split(%q, %q, %d), %q) = %q`, tt.s, tt.sep, tt.n, tt.sep, s)
   841  		}
   842  		if tt.n < 0 {
   843  			b := SplitAfter([]byte(tt.s), []byte(tt.sep))
   844  			if !reflect.DeepEqual(a, b) {
   845  				t.Errorf("SplitAfter disagrees withSplitAfterN(%q, %q, %d) = %v; want %v", tt.s, tt.sep, tt.n, b, a)
   846  			}
   847  		}
   848  	}
   849  }
   850  
   851  type FieldsTest struct {
   852  	s string
   853  	a []string
   854  }
   855  
   856  var fieldstests = []FieldsTest{
   857  	{"", []string{}},
   858  	{" ", []string{}},
   859  	{" \t ", []string{}},
   860  	{"  abc  ", []string{"abc"}},
   861  	{"1 2 3 4", []string{"1", "2", "3", "4"}},
   862  	{"1  2  3  4", []string{"1", "2", "3", "4"}},
   863  	{"1\t\t2\t\t3\t4", []string{"1", "2", "3", "4"}},
   864  	{"1\u20002\u20013\u20024", []string{"1", "2", "3", "4"}},
   865  	{"\u2000\u2001\u2002", []string{}},
   866  	{"\n™\t™\n", []string{"™", "™"}},
   867  	{faces, []string{faces}},
   868  }
   869  
   870  func TestFields(t *testing.T) {
   871  	for _, tt := range fieldstests {
   872  		b := []byte(tt.s)
   873  		a := Fields(b)
   874  
   875  		// Appending to the results should not change future results.
   876  		var x []byte
   877  		for _, v := range a {
   878  			x = append(v, 'z')
   879  		}
   880  
   881  		result := sliceOfString(a)
   882  		if !eq(result, tt.a) {
   883  			t.Errorf("Fields(%q) = %v; want %v", tt.s, a, tt.a)
   884  			continue
   885  		}
   886  
   887  		if string(b) != tt.s {
   888  			t.Errorf("slice changed to %s; want %s", string(b), tt.s)
   889  		}
   890  		if len(tt.a) > 0 {
   891  			if want := tt.a[len(tt.a)-1] + "z"; string(x) != want {
   892  				t.Errorf("last appended result was %s; want %s", x, want)
   893  			}
   894  		}
   895  	}
   896  }
   897  
   898  func TestFieldsFunc(t *testing.T) {
   899  	for _, tt := range fieldstests {
   900  		a := FieldsFunc([]byte(tt.s), unicode.IsSpace)
   901  		result := sliceOfString(a)
   902  		if !eq(result, tt.a) {
   903  			t.Errorf("FieldsFunc(%q, unicode.IsSpace) = %v; want %v", tt.s, a, tt.a)
   904  			continue
   905  		}
   906  	}
   907  	pred := func(c rune) bool { return c == 'X' }
   908  	var fieldsFuncTests = []FieldsTest{
   909  		{"", []string{}},
   910  		{"XX", []string{}},
   911  		{"XXhiXXX", []string{"hi"}},
   912  		{"aXXbXXXcX", []string{"a", "b", "c"}},
   913  	}
   914  	for _, tt := range fieldsFuncTests {
   915  		b := []byte(tt.s)
   916  		a := FieldsFunc(b, pred)
   917  
   918  		// Appending to the results should not change future results.
   919  		var x []byte
   920  		for _, v := range a {
   921  			x = append(v, 'z')
   922  		}
   923  
   924  		result := sliceOfString(a)
   925  		if !eq(result, tt.a) {
   926  			t.Errorf("FieldsFunc(%q) = %v, want %v", tt.s, a, tt.a)
   927  		}
   928  
   929  		if string(b) != tt.s {
   930  			t.Errorf("slice changed to %s; want %s", b, tt.s)
   931  		}
   932  		if len(tt.a) > 0 {
   933  			if want := tt.a[len(tt.a)-1] + "z"; string(x) != want {
   934  				t.Errorf("last appended result was %s; want %s", x, want)
   935  			}
   936  		}
   937  	}
   938  }
   939  
   940  // Test case for any function which accepts and returns a byte slice.
   941  // For ease of creation, we write the input byte slice as a string.
   942  type StringTest struct {
   943  	in  string
   944  	out []byte
   945  }
   946  
   947  var upperTests = []StringTest{
   948  	{"", []byte("")},
   949  	{"ONLYUPPER", []byte("ONLYUPPER")},
   950  	{"abc", []byte("ABC")},
   951  	{"AbC123", []byte("ABC123")},
   952  	{"azAZ09_", []byte("AZAZ09_")},
   953  	{"longStrinGwitHmixofsmaLLandcAps", []byte("LONGSTRINGWITHMIXOFSMALLANDCAPS")},
   954  	{"long\u0250string\u0250with\u0250nonascii\u2C6Fchars", []byte("LONG\u2C6FSTRING\u2C6FWITH\u2C6FNONASCII\u2C6FCHARS")},
   955  	{"\u0250\u0250\u0250\u0250\u0250", []byte("\u2C6F\u2C6F\u2C6F\u2C6F\u2C6F")}, // grows one byte per char
   956  	{"a\u0080\U0010FFFF", []byte("A\u0080\U0010FFFF")},                           // test utf8.RuneSelf and utf8.MaxRune
   957  }
   958  
   959  var lowerTests = []StringTest{
   960  	{"", []byte("")},
   961  	{"abc", []byte("abc")},
   962  	{"AbC123", []byte("abc123")},
   963  	{"azAZ09_", []byte("azaz09_")},
   964  	{"longStrinGwitHmixofsmaLLandcAps", []byte("longstringwithmixofsmallandcaps")},
   965  	{"LONG\u2C6FSTRING\u2C6FWITH\u2C6FNONASCII\u2C6FCHARS", []byte("long\u0250string\u0250with\u0250nonascii\u0250chars")},
   966  	{"\u2C6D\u2C6D\u2C6D\u2C6D\u2C6D", []byte("\u0251\u0251\u0251\u0251\u0251")}, // shrinks one byte per char
   967  	{"A\u0080\U0010FFFF", []byte("a\u0080\U0010FFFF")},                           // test utf8.RuneSelf and utf8.MaxRune
   968  }
   969  
   970  const space = "\t\v\r\f\n\u0085\u00a0\u2000\u3000"
   971  
   972  var trimSpaceTests = []StringTest{
   973  	{"", nil},
   974  	{"  a", []byte("a")},
   975  	{"b  ", []byte("b")},
   976  	{"abc", []byte("abc")},
   977  	{space + "abc" + space, []byte("abc")},
   978  	{" ", nil},
   979  	{"\u3000 ", nil},
   980  	{" \u3000", nil},
   981  	{" \t\r\n \t\t\r\r\n\n ", nil},
   982  	{" \t\r\n x\t\t\r\r\n\n ", []byte("x")},
   983  	{" \u2000\t\r\n x\t\t\r\r\ny\n \u3000", []byte("x\t\t\r\r\ny")},
   984  	{"1 \t\r\n2", []byte("1 \t\r\n2")},
   985  	{" x\x80", []byte("x\x80")},
   986  	{" x\xc0", []byte("x\xc0")},
   987  	{"x \xc0\xc0 ", []byte("x \xc0\xc0")},
   988  	{"x \xc0", []byte("x \xc0")},
   989  	{"x \xc0 ", []byte("x \xc0")},
   990  	{"x \xc0\xc0 ", []byte("x \xc0\xc0")},
   991  	{"x ☺\xc0\xc0 ", []byte("x ☺\xc0\xc0")},
   992  	{"x ☺ ", []byte("x ☺")},
   993  }
   994  
   995  // Execute f on each test case.  funcName should be the name of f; it's used
   996  // in failure reports.
   997  func runStringTests(t *testing.T, f func([]byte) []byte, funcName string, testCases []StringTest) {
   998  	for _, tc := range testCases {
   999  		actual := f([]byte(tc.in))
  1000  		if actual == nil && tc.out != nil {
  1001  			t.Errorf("%s(%q) = nil; want %q", funcName, tc.in, tc.out)
  1002  		}
  1003  		if actual != nil && tc.out == nil {
  1004  			t.Errorf("%s(%q) = %q; want nil", funcName, tc.in, actual)
  1005  		}
  1006  		if !Equal(actual, tc.out) {
  1007  			t.Errorf("%s(%q) = %q; want %q", funcName, tc.in, actual, tc.out)
  1008  		}
  1009  	}
  1010  }
  1011  
  1012  func tenRunes(r rune) string {
  1013  	runes := make([]rune, 10)
  1014  	for i := range runes {
  1015  		runes[i] = r
  1016  	}
  1017  	return string(runes)
  1018  }
  1019  
  1020  // User-defined self-inverse mapping function
  1021  func rot13(r rune) rune {
  1022  	const step = 13
  1023  	if r >= 'a' && r <= 'z' {
  1024  		return ((r - 'a' + step) % 26) + 'a'
  1025  	}
  1026  	if r >= 'A' && r <= 'Z' {
  1027  		return ((r - 'A' + step) % 26) + 'A'
  1028  	}
  1029  	return r
  1030  }
  1031  
  1032  func TestMap(t *testing.T) {
  1033  	// Run a couple of awful growth/shrinkage tests
  1034  	a := tenRunes('a')
  1035  
  1036  	// 1.  Grow. This triggers two reallocations in Map.
  1037  	maxRune := func(r rune) rune { return unicode.MaxRune }
  1038  	m := Map(maxRune, []byte(a))
  1039  	expect := tenRunes(unicode.MaxRune)
  1040  	if string(m) != expect {
  1041  		t.Errorf("growing: expected %q got %q", expect, m)
  1042  	}
  1043  
  1044  	// 2. Shrink
  1045  	minRune := func(r rune) rune { return 'a' }
  1046  	m = Map(minRune, []byte(tenRunes(unicode.MaxRune)))
  1047  	expect = a
  1048  	if string(m) != expect {
  1049  		t.Errorf("shrinking: expected %q got %q", expect, m)
  1050  	}
  1051  
  1052  	// 3. Rot13
  1053  	m = Map(rot13, []byte("a to zed"))
  1054  	expect = "n gb mrq"
  1055  	if string(m) != expect {
  1056  		t.Errorf("rot13: expected %q got %q", expect, m)
  1057  	}
  1058  
  1059  	// 4. Rot13^2
  1060  	m = Map(rot13, Map(rot13, []byte("a to zed")))
  1061  	expect = "a to zed"
  1062  	if string(m) != expect {
  1063  		t.Errorf("rot13: expected %q got %q", expect, m)
  1064  	}
  1065  
  1066  	// 5. Drop
  1067  	dropNotLatin := func(r rune) rune {
  1068  		if unicode.Is(unicode.Latin, r) {
  1069  			return r
  1070  		}
  1071  		return -1
  1072  	}
  1073  	m = Map(dropNotLatin, []byte("Hello, 세계"))
  1074  	expect = "Hello"
  1075  	if string(m) != expect {
  1076  		t.Errorf("drop: expected %q got %q", expect, m)
  1077  	}
  1078  
  1079  	// 6. Invalid rune
  1080  	invalidRune := func(r rune) rune {
  1081  		return utf8.MaxRune + 1
  1082  	}
  1083  	m = Map(invalidRune, []byte("x"))
  1084  	expect = "\uFFFD"
  1085  	if string(m) != expect {
  1086  		t.Errorf("invalidRune: expected %q got %q", expect, m)
  1087  	}
  1088  }
  1089  
  1090  func TestToUpper(t *testing.T) { runStringTests(t, ToUpper, "ToUpper", upperTests) }
  1091  
  1092  func TestToLower(t *testing.T) { runStringTests(t, ToLower, "ToLower", lowerTests) }
  1093  
  1094  func BenchmarkToUpper(b *testing.B) {
  1095  	for _, tc := range upperTests {
  1096  		tin := []byte(tc.in)
  1097  		b.Run(tc.in, func(b *testing.B) {
  1098  			for i := 0; i < b.N; i++ {
  1099  				actual := ToUpper(tin)
  1100  				if !Equal(actual, tc.out) {
  1101  					b.Errorf("ToUpper(%q) = %q; want %q", tc.in, actual, tc.out)
  1102  				}
  1103  			}
  1104  		})
  1105  	}
  1106  }
  1107  
  1108  func BenchmarkToLower(b *testing.B) {
  1109  	for _, tc := range lowerTests {
  1110  		tin := []byte(tc.in)
  1111  		b.Run(tc.in, func(b *testing.B) {
  1112  			for i := 0; i < b.N; i++ {
  1113  				actual := ToLower(tin)
  1114  				if !Equal(actual, tc.out) {
  1115  					b.Errorf("ToLower(%q) = %q; want %q", tc.in, actual, tc.out)
  1116  				}
  1117  			}
  1118  		})
  1119  	}
  1120  }
  1121  
  1122  var toValidUTF8Tests = []struct {
  1123  	in   string
  1124  	repl string
  1125  	out  string
  1126  }{
  1127  	{"", "\uFFFD", ""},
  1128  	{"abc", "\uFFFD", "abc"},
  1129  	{"\uFDDD", "\uFFFD", "\uFDDD"},
  1130  	{"a\xffb", "\uFFFD", "a\uFFFDb"},
  1131  	{"a\xffb\uFFFD", "X", "aXb\uFFFD"},
  1132  	{"a☺\xffb☺\xC0\xAFc☺\xff", "", "a☺b☺c☺"},
  1133  	{"a☺\xffb☺\xC0\xAFc☺\xff", "日本語", "a☺日本語b☺日本語c☺日本語"},
  1134  	{"\xC0\xAF", "\uFFFD", "\uFFFD"},
  1135  	{"\xE0\x80\xAF", "\uFFFD", "\uFFFD"},
  1136  	{"\xed\xa0\x80", "abc", "abc"},
  1137  	{"\xed\xbf\xbf", "\uFFFD", "\uFFFD"},
  1138  	{"\xF0\x80\x80\xaf", "☺", "☺"},
  1139  	{"\xF8\x80\x80\x80\xAF", "\uFFFD", "\uFFFD"},
  1140  	{"\xFC\x80\x80\x80\x80\xAF", "\uFFFD", "\uFFFD"},
  1141  }
  1142  
  1143  func TestToValidUTF8(t *testing.T) {
  1144  	for _, tc := range toValidUTF8Tests {
  1145  		got := ToValidUTF8([]byte(tc.in), []byte(tc.repl))
  1146  		if !Equal(got, []byte(tc.out)) {
  1147  			t.Errorf("ToValidUTF8(%q, %q) = %q; want %q", tc.in, tc.repl, got, tc.out)
  1148  		}
  1149  	}
  1150  }
  1151  
  1152  func TestTrimSpace(t *testing.T) { runStringTests(t, TrimSpace, "TrimSpace", trimSpaceTests) }
  1153  
  1154  type RepeatTest struct {
  1155  	in, out string
  1156  	count   int
  1157  }
  1158  
  1159  var RepeatTests = []RepeatTest{
  1160  	{"", "", 0},
  1161  	{"", "", 1},
  1162  	{"", "", 2},
  1163  	{"-", "", 0},
  1164  	{"-", "-", 1},
  1165  	{"-", "----------", 10},
  1166  	{"abc ", "abc abc abc ", 3},
  1167  }
  1168  
  1169  func TestRepeat(t *testing.T) {
  1170  	for _, tt := range RepeatTests {
  1171  		tin := []byte(tt.in)
  1172  		tout := []byte(tt.out)
  1173  		a := Repeat(tin, tt.count)
  1174  		if !Equal(a, tout) {
  1175  			t.Errorf("Repeat(%q, %d) = %q; want %q", tin, tt.count, a, tout)
  1176  			continue
  1177  		}
  1178  	}
  1179  }
  1180  
  1181  func repeat(b []byte, count int) (err error) {
  1182  	defer func() {
  1183  		if r := recover(); r != nil {
  1184  			switch v := r.(type) {
  1185  			case error:
  1186  				err = v
  1187  			default:
  1188  				err = fmt.Errorf("%s", v)
  1189  			}
  1190  		}
  1191  	}()
  1192  
  1193  	Repeat(b, count)
  1194  
  1195  	return
  1196  }
  1197  
  1198  // See Issue golang.org/issue/16237
  1199  func TestRepeatCatchesOverflow(t *testing.T) {
  1200  	tests := [...]struct {
  1201  		s      string
  1202  		count  int
  1203  		errStr string
  1204  	}{
  1205  		0: {"--", -2147483647, "negative"},
  1206  		1: {"", int(^uint(0) >> 1), ""},
  1207  		2: {"-", 10, ""},
  1208  		3: {"gopher", 0, ""},
  1209  		4: {"-", -1, "negative"},
  1210  		5: {"--", -102, "negative"},
  1211  		6: {string(make([]byte, 255)), int((^uint(0))/255 + 1), "overflow"},
  1212  	}
  1213  
  1214  	for i, tt := range tests {
  1215  		err := repeat([]byte(tt.s), tt.count)
  1216  		if tt.errStr == "" {
  1217  			if err != nil {
  1218  				t.Errorf("#%d panicked %v", i, err)
  1219  			}
  1220  			continue
  1221  		}
  1222  
  1223  		if err == nil || !strings.Contains(err.Error(), tt.errStr) {
  1224  			t.Errorf("#%d expected %q got %q", i, tt.errStr, err)
  1225  		}
  1226  	}
  1227  }
  1228  
  1229  func runesEqual(a, b []rune) bool {
  1230  	if len(a) != len(b) {
  1231  		return false
  1232  	}
  1233  	for i, r := range a {
  1234  		if r != b[i] {
  1235  			return false
  1236  		}
  1237  	}
  1238  	return true
  1239  }
  1240  
  1241  type RunesTest struct {
  1242  	in    string
  1243  	out   []rune
  1244  	lossy bool
  1245  }
  1246  
  1247  var RunesTests = []RunesTest{
  1248  	{"", []rune{}, false},
  1249  	{" ", []rune{32}, false},
  1250  	{"ABC", []rune{65, 66, 67}, false},
  1251  	{"abc", []rune{97, 98, 99}, false},
  1252  	{"\u65e5\u672c\u8a9e", []rune{26085, 26412, 35486}, false},
  1253  	{"ab\x80c", []rune{97, 98, 0xFFFD, 99}, true},
  1254  	{"ab\xc0c", []rune{97, 98, 0xFFFD, 99}, true},
  1255  }
  1256  
  1257  func TestRunes(t *testing.T) {
  1258  	for _, tt := range RunesTests {
  1259  		tin := []byte(tt.in)
  1260  		a := Runes(tin)
  1261  		if !runesEqual(a, tt.out) {
  1262  			t.Errorf("Runes(%q) = %v; want %v", tin, a, tt.out)
  1263  			continue
  1264  		}
  1265  		if !tt.lossy {
  1266  			// can only test reassembly if we didn't lose information
  1267  			s := string(a)
  1268  			if s != tt.in {
  1269  				t.Errorf("string(Runes(%q)) = %x; want %x", tin, s, tin)
  1270  			}
  1271  		}
  1272  	}
  1273  }
  1274  
  1275  type TrimTest struct {
  1276  	f            string
  1277  	in, arg, out string
  1278  }
  1279  
  1280  var trimTests = []TrimTest{
  1281  	{"Trim", "abba", "a", "bb"},
  1282  	{"Trim", "abba", "ab", ""},
  1283  	{"TrimLeft", "abba", "ab", ""},
  1284  	{"TrimRight", "abba", "ab", ""},
  1285  	{"TrimLeft", "abba", "a", "bba"},
  1286  	{"TrimLeft", "abba", "b", "abba"},
  1287  	{"TrimRight", "abba", "a", "abb"},
  1288  	{"TrimRight", "abba", "b", "abba"},
  1289  	{"Trim", "<tag>", "<>", "tag"},
  1290  	{"Trim", "* listitem", " *", "listitem"},
  1291  	{"Trim", `"quote"`, `"`, "quote"},
  1292  	{"Trim", "\u2C6F\u2C6F\u0250\u0250\u2C6F\u2C6F", "\u2C6F", "\u0250\u0250"},
  1293  	{"Trim", "\x80test\xff", "\xff", "test"},
  1294  	{"Trim", " Ġ ", " ", "Ġ"},
  1295  	{"Trim", " Ġİ0", "0 ", "Ġİ"},
  1296  	//empty string tests
  1297  	{"Trim", "abba", "", "abba"},
  1298  	{"Trim", "", "123", ""},
  1299  	{"Trim", "", "", ""},
  1300  	{"TrimLeft", "abba", "", "abba"},
  1301  	{"TrimLeft", "", "123", ""},
  1302  	{"TrimLeft", "", "", ""},
  1303  	{"TrimRight", "abba", "", "abba"},
  1304  	{"TrimRight", "", "123", ""},
  1305  	{"TrimRight", "", "", ""},
  1306  	{"TrimRight", "☺\xc0", "☺", "☺\xc0"},
  1307  	{"TrimPrefix", "aabb", "a", "abb"},
  1308  	{"TrimPrefix", "aabb", "b", "aabb"},
  1309  	{"TrimSuffix", "aabb", "a", "aabb"},
  1310  	{"TrimSuffix", "aabb", "b", "aab"},
  1311  }
  1312  
  1313  type TrimNilTest struct {
  1314  	f   string
  1315  	in  []byte
  1316  	arg string
  1317  	out []byte
  1318  }
  1319  
  1320  var trimNilTests = []TrimNilTest{
  1321  	{"Trim", nil, "", nil},
  1322  	{"Trim", []byte{}, "", nil},
  1323  	{"Trim", []byte{'a'}, "a", nil},
  1324  	{"Trim", []byte{'a', 'a'}, "a", nil},
  1325  	{"Trim", []byte{'a'}, "ab", nil},
  1326  	{"Trim", []byte{'a', 'b'}, "ab", nil},
  1327  	{"Trim", []byte("☺"), "☺", nil},
  1328  	{"TrimLeft", nil, "", nil},
  1329  	{"TrimLeft", []byte{}, "", nil},
  1330  	{"TrimLeft", []byte{'a'}, "a", nil},
  1331  	{"TrimLeft", []byte{'a', 'a'}, "a", nil},
  1332  	{"TrimLeft", []byte{'a'}, "ab", nil},
  1333  	{"TrimLeft", []byte{'a', 'b'}, "ab", nil},
  1334  	{"TrimLeft", []byte("☺"), "☺", nil},
  1335  	{"TrimRight", nil, "", nil},
  1336  	{"TrimRight", []byte{}, "", []byte{}},
  1337  	{"TrimRight", []byte{'a'}, "a", []byte{}},
  1338  	{"TrimRight", []byte{'a', 'a'}, "a", []byte{}},
  1339  	{"TrimRight", []byte{'a'}, "ab", []byte{}},
  1340  	{"TrimRight", []byte{'a', 'b'}, "ab", []byte{}},
  1341  	{"TrimRight", []byte("☺"), "☺", []byte{}},
  1342  	{"TrimPrefix", nil, "", nil},
  1343  	{"TrimPrefix", []byte{}, "", []byte{}},
  1344  	{"TrimPrefix", []byte{'a'}, "a", []byte{}},
  1345  	{"TrimPrefix", []byte("☺"), "☺", []byte{}},
  1346  	{"TrimSuffix", nil, "", nil},
  1347  	{"TrimSuffix", []byte{}, "", []byte{}},
  1348  	{"TrimSuffix", []byte{'a'}, "a", []byte{}},
  1349  	{"TrimSuffix", []byte("☺"), "☺", []byte{}},
  1350  }
  1351  
  1352  func TestTrim(t *testing.T) {
  1353  	toFn := func(name string) (func([]byte, string) []byte, func([]byte, []byte) []byte) {
  1354  		switch name {
  1355  		case "Trim":
  1356  			return Trim, nil
  1357  		case "TrimLeft":
  1358  			return TrimLeft, nil
  1359  		case "TrimRight":
  1360  			return TrimRight, nil
  1361  		case "TrimPrefix":
  1362  			return nil, TrimPrefix
  1363  		case "TrimSuffix":
  1364  			return nil, TrimSuffix
  1365  		default:
  1366  			t.Errorf("Undefined trim function %s", name)
  1367  			return nil, nil
  1368  		}
  1369  	}
  1370  
  1371  	for _, tc := range trimTests {
  1372  		name := tc.f
  1373  		f, fb := toFn(name)
  1374  		if f == nil && fb == nil {
  1375  			continue
  1376  		}
  1377  		var actual string
  1378  		if f != nil {
  1379  			actual = string(f([]byte(tc.in), tc.arg))
  1380  		} else {
  1381  			actual = string(fb([]byte(tc.in), []byte(tc.arg)))
  1382  		}
  1383  		if actual != tc.out {
  1384  			t.Errorf("%s(%q, %q) = %q; want %q", name, tc.in, tc.arg, actual, tc.out)
  1385  		}
  1386  	}
  1387  
  1388  	for _, tc := range trimNilTests {
  1389  		name := tc.f
  1390  		f, fb := toFn(name)
  1391  		if f == nil && fb == nil {
  1392  			continue
  1393  		}
  1394  		var actual []byte
  1395  		if f != nil {
  1396  			actual = f(tc.in, tc.arg)
  1397  		} else {
  1398  			actual = fb(tc.in, []byte(tc.arg))
  1399  		}
  1400  		report := func(s []byte) string {
  1401  			if s == nil {
  1402  				return "nil"
  1403  			} else {
  1404  				return fmt.Sprintf("%q", s)
  1405  			}
  1406  		}
  1407  		if len(actual) != 0 {
  1408  			t.Errorf("%s(%s, %q) returned non-empty value", name, report(tc.in), tc.arg)
  1409  		} else {
  1410  			actualNil := actual == nil
  1411  			outNil := tc.out == nil
  1412  			if actualNil != outNil {
  1413  				t.Errorf("%s(%s, %q) got nil %t; want nil %t", name, report(tc.in), tc.arg, actualNil, outNil)
  1414  			}
  1415  		}
  1416  	}
  1417  }
  1418  
  1419  type predicate struct {
  1420  	f    func(r rune) bool
  1421  	name string
  1422  }
  1423  
  1424  var isSpace = predicate{unicode.IsSpace, "IsSpace"}
  1425  var isDigit = predicate{unicode.IsDigit, "IsDigit"}
  1426  var isUpper = predicate{unicode.IsUpper, "IsUpper"}
  1427  var isValidRune = predicate{
  1428  	func(r rune) bool {
  1429  		return r != utf8.RuneError
  1430  	},
  1431  	"IsValidRune",
  1432  }
  1433  
  1434  type TrimFuncTest struct {
  1435  	f        predicate
  1436  	in       string
  1437  	trimOut  []byte
  1438  	leftOut  []byte
  1439  	rightOut []byte
  1440  }
  1441  
  1442  func not(p predicate) predicate {
  1443  	return predicate{
  1444  		func(r rune) bool {
  1445  			return !p.f(r)
  1446  		},
  1447  		"not " + p.name,
  1448  	}
  1449  }
  1450  
  1451  var trimFuncTests = []TrimFuncTest{
  1452  	{isSpace, space + " hello " + space,
  1453  		[]byte("hello"),
  1454  		[]byte("hello " + space),
  1455  		[]byte(space + " hello")},
  1456  	{isDigit, "\u0e50\u0e5212hello34\u0e50\u0e51",
  1457  		[]byte("hello"),
  1458  		[]byte("hello34\u0e50\u0e51"),
  1459  		[]byte("\u0e50\u0e5212hello")},
  1460  	{isUpper, "\u2C6F\u2C6F\u2C6F\u2C6FABCDhelloEF\u2C6F\u2C6FGH\u2C6F\u2C6F",
  1461  		[]byte("hello"),
  1462  		[]byte("helloEF\u2C6F\u2C6FGH\u2C6F\u2C6F"),
  1463  		[]byte("\u2C6F\u2C6F\u2C6F\u2C6FABCDhello")},
  1464  	{not(isSpace), "hello" + space + "hello",
  1465  		[]byte(space),
  1466  		[]byte(space + "hello"),
  1467  		[]byte("hello" + space)},
  1468  	{not(isDigit), "hello\u0e50\u0e521234\u0e50\u0e51helo",
  1469  		[]byte("\u0e50\u0e521234\u0e50\u0e51"),
  1470  		[]byte("\u0e50\u0e521234\u0e50\u0e51helo"),
  1471  		[]byte("hello\u0e50\u0e521234\u0e50\u0e51")},
  1472  	{isValidRune, "ab\xc0a\xc0cd",
  1473  		[]byte("\xc0a\xc0"),
  1474  		[]byte("\xc0a\xc0cd"),
  1475  		[]byte("ab\xc0a\xc0")},
  1476  	{not(isValidRune), "\xc0a\xc0",
  1477  		[]byte("a"),
  1478  		[]byte("a\xc0"),
  1479  		[]byte("\xc0a")},
  1480  	// The nils returned by TrimLeftFunc are odd behavior, but we need
  1481  	// to preserve backwards compatibility.
  1482  	{isSpace, "",
  1483  		nil,
  1484  		nil,
  1485  		[]byte("")},
  1486  	{isSpace, " ",
  1487  		nil,
  1488  		nil,
  1489  		[]byte("")},
  1490  }
  1491  
  1492  func TestTrimFunc(t *testing.T) {
  1493  	for _, tc := range trimFuncTests {
  1494  		trimmers := []struct {
  1495  			name string
  1496  			trim func(s []byte, f func(r rune) bool) []byte
  1497  			out  []byte
  1498  		}{
  1499  			{"TrimFunc", TrimFunc, tc.trimOut},
  1500  			{"TrimLeftFunc", TrimLeftFunc, tc.leftOut},
  1501  			{"TrimRightFunc", TrimRightFunc, tc.rightOut},
  1502  		}
  1503  		for _, trimmer := range trimmers {
  1504  			actual := trimmer.trim([]byte(tc.in), tc.f.f)
  1505  			if actual == nil && trimmer.out != nil {
  1506  				t.Errorf("%s(%q, %q) = nil; want %q", trimmer.name, tc.in, tc.f.name, trimmer.out)
  1507  			}
  1508  			if actual != nil && trimmer.out == nil {
  1509  				t.Errorf("%s(%q, %q) = %q; want nil", trimmer.name, tc.in, tc.f.name, actual)
  1510  			}
  1511  			if !Equal(actual, trimmer.out) {
  1512  				t.Errorf("%s(%q, %q) = %q; want %q", trimmer.name, tc.in, tc.f.name, actual, trimmer.out)
  1513  			}
  1514  		}
  1515  	}
  1516  }
  1517  
  1518  type IndexFuncTest struct {
  1519  	in          string
  1520  	f           predicate
  1521  	first, last int
  1522  }
  1523  
  1524  var indexFuncTests = []IndexFuncTest{
  1525  	{"", isValidRune, -1, -1},
  1526  	{"abc", isDigit, -1, -1},
  1527  	{"0123", isDigit, 0, 3},
  1528  	{"a1b", isDigit, 1, 1},
  1529  	{space, isSpace, 0, len(space) - 3}, // last rune in space is 3 bytes
  1530  	{"\u0e50\u0e5212hello34\u0e50\u0e51", isDigit, 0, 18},
  1531  	{"\u2C6F\u2C6F\u2C6F\u2C6FABCDhelloEF\u2C6F\u2C6FGH\u2C6F\u2C6F", isUpper, 0, 34},
  1532  	{"12\u0e50\u0e52hello34\u0e50\u0e51", not(isDigit), 8, 12},
  1533  
  1534  	// tests of invalid UTF-8
  1535  	{"\x801", isDigit, 1, 1},
  1536  	{"\x80abc", isDigit, -1, -1},
  1537  	{"\xc0a\xc0", isValidRune, 1, 1},
  1538  	{"\xc0a\xc0", not(isValidRune), 0, 2},
  1539  	{"\xc0☺\xc0", not(isValidRune), 0, 4},
  1540  	{"\xc0☺\xc0\xc0", not(isValidRune), 0, 5},
  1541  	{"ab\xc0a\xc0cd", not(isValidRune), 2, 4},
  1542  	{"a\xe0\x80cd", not(isValidRune), 1, 2},
  1543  }
  1544  
  1545  func TestIndexFunc(t *testing.T) {
  1546  	for _, tc := range indexFuncTests {
  1547  		first := IndexFunc([]byte(tc.in), tc.f.f)
  1548  		if first != tc.first {
  1549  			t.Errorf("IndexFunc(%q, %s) = %d; want %d", tc.in, tc.f.name, first, tc.first)
  1550  		}
  1551  		last := LastIndexFunc([]byte(tc.in), tc.f.f)
  1552  		if last != tc.last {
  1553  			t.Errorf("LastIndexFunc(%q, %s) = %d; want %d", tc.in, tc.f.name, last, tc.last)
  1554  		}
  1555  	}
  1556  }
  1557  
  1558  type ReplaceTest struct {
  1559  	in       string
  1560  	old, new string
  1561  	n        int
  1562  	out      string
  1563  }
  1564  
  1565  var ReplaceTests = []ReplaceTest{
  1566  	{"hello", "l", "L", 0, "hello"},
  1567  	{"hello", "l", "L", -1, "heLLo"},
  1568  	{"hello", "x", "X", -1, "hello"},
  1569  	{"", "x", "X", -1, ""},
  1570  	{"radar", "r", "<r>", -1, "<r>ada<r>"},
  1571  	{"", "", "<>", -1, "<>"},
  1572  	{"banana", "a", "<>", -1, "b<>n<>n<>"},
  1573  	{"banana", "a", "<>", 1, "b<>nana"},
  1574  	{"banana", "a", "<>", 1000, "b<>n<>n<>"},
  1575  	{"banana", "an", "<>", -1, "b<><>a"},
  1576  	{"banana", "ana", "<>", -1, "b<>na"},
  1577  	{"banana", "", "<>", -1, "<>b<>a<>n<>a<>n<>a<>"},
  1578  	{"banana", "", "<>", 10, "<>b<>a<>n<>a<>n<>a<>"},
  1579  	{"banana", "", "<>", 6, "<>b<>a<>n<>a<>n<>a"},
  1580  	{"banana", "", "<>", 5, "<>b<>a<>n<>a<>na"},
  1581  	{"banana", "", "<>", 1, "<>banana"},
  1582  	{"banana", "a", "a", -1, "banana"},
  1583  	{"banana", "a", "a", 1, "banana"},
  1584  	{"☺☻☹", "", "<>", -1, "<>☺<>☻<>☹<>"},
  1585  }
  1586  
  1587  func TestReplace(t *testing.T) {
  1588  	for _, tt := range ReplaceTests {
  1589  		in := append([]byte(tt.in), "<spare>"...)
  1590  		in = in[:len(tt.in)]
  1591  		out := Replace(in, []byte(tt.old), []byte(tt.new), tt.n)
  1592  		if s := string(out); s != tt.out {
  1593  			t.Errorf("Replace(%q, %q, %q, %d) = %q, want %q", tt.in, tt.old, tt.new, tt.n, s, tt.out)
  1594  		}
  1595  		if cap(in) == cap(out) && &in[:1][0] == &out[:1][0] {
  1596  			t.Errorf("Replace(%q, %q, %q, %d) didn't copy", tt.in, tt.old, tt.new, tt.n)
  1597  		}
  1598  		if tt.n == -1 {
  1599  			out := ReplaceAll(in, []byte(tt.old), []byte(tt.new))
  1600  			if s := string(out); s != tt.out {
  1601  				t.Errorf("ReplaceAll(%q, %q, %q) = %q, want %q", tt.in, tt.old, tt.new, s, tt.out)
  1602  			}
  1603  		}
  1604  	}
  1605  }
  1606  
  1607  type TitleTest struct {
  1608  	in, out string
  1609  }
  1610  
  1611  var TitleTests = []TitleTest{
  1612  	{"", ""},
  1613  	{"a", "A"},
  1614  	{" aaa aaa aaa ", " Aaa Aaa Aaa "},
  1615  	{" Aaa Aaa Aaa ", " Aaa Aaa Aaa "},
  1616  	{"123a456", "123a456"},
  1617  	{"double-blind", "Double-Blind"},
  1618  	{"ÿøû", "Ÿøû"},
  1619  	{"with_underscore", "With_underscore"},
  1620  	{"unicode \xe2\x80\xa8 line separator", "Unicode \xe2\x80\xa8 Line Separator"},
  1621  }
  1622  
  1623  func TestTitle(t *testing.T) {
  1624  	for _, tt := range TitleTests {
  1625  		if s := string(Title([]byte(tt.in))); s != tt.out {
  1626  			t.Errorf("Title(%q) = %q, want %q", tt.in, s, tt.out)
  1627  		}
  1628  	}
  1629  }
  1630  
  1631  var ToTitleTests = []TitleTest{
  1632  	{"", ""},
  1633  	{"a", "A"},
  1634  	{" aaa aaa aaa ", " AAA AAA AAA "},
  1635  	{" Aaa Aaa Aaa ", " AAA AAA AAA "},
  1636  	{"123a456", "123A456"},
  1637  	{"double-blind", "DOUBLE-BLIND"},
  1638  	{"ÿøû", "ŸØÛ"},
  1639  }
  1640  
  1641  func TestToTitle(t *testing.T) {
  1642  	for _, tt := range ToTitleTests {
  1643  		if s := string(ToTitle([]byte(tt.in))); s != tt.out {
  1644  			t.Errorf("ToTitle(%q) = %q, want %q", tt.in, s, tt.out)
  1645  		}
  1646  	}
  1647  }
  1648  
  1649  var EqualFoldTests = []struct {
  1650  	s, t string
  1651  	out  bool
  1652  }{
  1653  	{"abc", "abc", true},
  1654  	{"ABcd", "ABcd", true},
  1655  	{"123abc", "123ABC", true},
  1656  	{"αβδ", "ΑΒΔ", true},
  1657  	{"abc", "xyz", false},
  1658  	{"abc", "XYZ", false},
  1659  	{"abcdefghijk", "abcdefghijX", false},
  1660  	{"abcdefghijk", "abcdefghij\u212A", true},
  1661  	{"abcdefghijK", "abcdefghij\u212A", true},
  1662  	{"abcdefghijkz", "abcdefghij\u212Ay", false},
  1663  	{"abcdefghijKz", "abcdefghij\u212Ay", false},
  1664  }
  1665  
  1666  func TestEqualFold(t *testing.T) {
  1667  	for _, tt := range EqualFoldTests {
  1668  		if out := EqualFold([]byte(tt.s), []byte(tt.t)); out != tt.out {
  1669  			t.Errorf("EqualFold(%#q, %#q) = %v, want %v", tt.s, tt.t, out, tt.out)
  1670  		}
  1671  		if out := EqualFold([]byte(tt.t), []byte(tt.s)); out != tt.out {
  1672  			t.Errorf("EqualFold(%#q, %#q) = %v, want %v", tt.t, tt.s, out, tt.out)
  1673  		}
  1674  	}
  1675  }
  1676  
  1677  var cutTests = []struct {
  1678  	s, sep        string
  1679  	before, after string
  1680  	found         bool
  1681  }{
  1682  	{"abc", "b", "a", "c", true},
  1683  	{"abc", "a", "", "bc", true},
  1684  	{"abc", "c", "ab", "", true},
  1685  	{"abc", "abc", "", "", true},
  1686  	{"abc", "", "", "abc", true},
  1687  	{"abc", "d", "abc", "", false},
  1688  	{"", "d", "", "", false},
  1689  	{"", "", "", "", true},
  1690  }
  1691  
  1692  func TestCut(t *testing.T) {
  1693  	for _, tt := range cutTests {
  1694  		if before, after, found := Cut([]byte(tt.s), []byte(tt.sep)); string(before) != tt.before || string(after) != tt.after || found != tt.found {
  1695  			t.Errorf("Cut(%q, %q) = %q, %q, %v, want %q, %q, %v", tt.s, tt.sep, before, after, found, tt.before, tt.after, tt.found)
  1696  		}
  1697  	}
  1698  }
  1699  
  1700  func TestBufferGrowNegative(t *testing.T) {
  1701  	defer func() {
  1702  		if err := recover(); err == nil {
  1703  			t.Fatal("Grow(-1) should have panicked")
  1704  		}
  1705  	}()
  1706  	var b Buffer
  1707  	b.Grow(-1)
  1708  }
  1709  
  1710  func TestBufferTruncateNegative(t *testing.T) {
  1711  	defer func() {
  1712  		if err := recover(); err == nil {
  1713  			t.Fatal("Truncate(-1) should have panicked")
  1714  		}
  1715  	}()
  1716  	var b Buffer
  1717  	b.Truncate(-1)
  1718  }
  1719  
  1720  func TestBufferTruncateOutOfRange(t *testing.T) {
  1721  	defer func() {
  1722  		if err := recover(); err == nil {
  1723  			t.Fatal("Truncate(20) should have panicked")
  1724  		}
  1725  	}()
  1726  	var b Buffer
  1727  	b.Write(make([]byte, 10))
  1728  	b.Truncate(20)
  1729  }
  1730  
  1731  var containsTests = []struct {
  1732  	b, subslice []byte
  1733  	want        bool
  1734  }{
  1735  	{[]byte("hello"), []byte("hel"), true},
  1736  	{[]byte("日本語"), []byte("日本"), true},
  1737  	{[]byte("hello"), []byte("Hello, world"), false},
  1738  	{[]byte("東京"), []byte("京東"), false},
  1739  }
  1740  
  1741  func TestContains(t *testing.T) {
  1742  	for _, tt := range containsTests {
  1743  		if got := Contains(tt.b, tt.subslice); got != tt.want {
  1744  			t.Errorf("Contains(%q, %q) = %v, want %v", tt.b, tt.subslice, got, tt.want)
  1745  		}
  1746  	}
  1747  }
  1748  
  1749  var ContainsAnyTests = []struct {
  1750  	b        []byte
  1751  	substr   string
  1752  	expected bool
  1753  }{
  1754  	{[]byte(""), "", false},
  1755  	{[]byte(""), "a", false},
  1756  	{[]byte(""), "abc", false},
  1757  	{[]byte("a"), "", false},
  1758  	{[]byte("a"), "a", true},
  1759  	{[]byte("aaa"), "a", true},
  1760  	{[]byte("abc"), "xyz", false},
  1761  	{[]byte("abc"), "xcz", true},
  1762  	{[]byte("a☺b☻c☹d"), "uvw☻xyz", true},
  1763  	{[]byte("aRegExp*"), ".(|)*+?^$[]", true},
  1764  	{[]byte(dots + dots + dots), " ", false},
  1765  }
  1766  
  1767  func TestContainsAny(t *testing.T) {
  1768  	for _, ct := range ContainsAnyTests {
  1769  		if ContainsAny(ct.b, ct.substr) != ct.expected {
  1770  			t.Errorf("ContainsAny(%s, %s) = %v, want %v",
  1771  				ct.b, ct.substr, !ct.expected, ct.expected)
  1772  		}
  1773  	}
  1774  }
  1775  
  1776  var ContainsRuneTests = []struct {
  1777  	b        []byte
  1778  	r        rune
  1779  	expected bool
  1780  }{
  1781  	{[]byte(""), 'a', false},
  1782  	{[]byte("a"), 'a', true},
  1783  	{[]byte("aaa"), 'a', true},
  1784  	{[]byte("abc"), 'y', false},
  1785  	{[]byte("abc"), 'c', true},
  1786  	{[]byte("a☺b☻c☹d"), 'x', false},
  1787  	{[]byte("a☺b☻c☹d"), '☻', true},
  1788  	{[]byte("aRegExp*"), '*', true},
  1789  }
  1790  
  1791  func TestContainsRune(t *testing.T) {
  1792  	for _, ct := range ContainsRuneTests {
  1793  		if ContainsRune(ct.b, ct.r) != ct.expected {
  1794  			t.Errorf("ContainsRune(%q, %q) = %v, want %v",
  1795  				ct.b, ct.r, !ct.expected, ct.expected)
  1796  		}
  1797  	}
  1798  }
  1799  
  1800  var makeFieldsInput = func() []byte {
  1801  	x := make([]byte, 1<<20)
  1802  	// Input is ~10% space, ~10% 2-byte UTF-8, rest ASCII non-space.
  1803  	for i := range x {
  1804  		switch rand.Intn(10) {
  1805  		case 0:
  1806  			x[i] = ' '
  1807  		case 1:
  1808  			if i > 0 && x[i-1] == 'x' {
  1809  				copy(x[i-1:], "χ")
  1810  				break
  1811  			}
  1812  			fallthrough
  1813  		default:
  1814  			x[i] = 'x'
  1815  		}
  1816  	}
  1817  	return x
  1818  }
  1819  
  1820  var makeFieldsInputASCII = func() []byte {
  1821  	x := make([]byte, 1<<20)
  1822  	// Input is ~10% space, rest ASCII non-space.
  1823  	for i := range x {
  1824  		if rand.Intn(10) == 0 {
  1825  			x[i] = ' '
  1826  		} else {
  1827  			x[i] = 'x'
  1828  		}
  1829  	}
  1830  	return x
  1831  }
  1832  
  1833  var bytesdata = []struct {
  1834  	name string
  1835  	data []byte
  1836  }{
  1837  	{"ASCII", makeFieldsInputASCII()},
  1838  	{"Mixed", makeFieldsInput()},
  1839  }
  1840  
  1841  func BenchmarkFields(b *testing.B) {
  1842  	for _, sd := range bytesdata {
  1843  		b.Run(sd.name, func(b *testing.B) {
  1844  			for j := 1 << 4; j <= 1<<20; j <<= 4 {
  1845  				b.Run(fmt.Sprintf("%d", j), func(b *testing.B) {
  1846  					b.ReportAllocs()
  1847  					b.SetBytes(int64(j))
  1848  					data := sd.data[:j]
  1849  					for i := 0; i < b.N; i++ {
  1850  						Fields(data)
  1851  					}
  1852  				})
  1853  			}
  1854  		})
  1855  	}
  1856  }
  1857  
  1858  func BenchmarkFieldsFunc(b *testing.B) {
  1859  	for _, sd := range bytesdata {
  1860  		b.Run(sd.name, func(b *testing.B) {
  1861  			for j := 1 << 4; j <= 1<<20; j <<= 4 {
  1862  				b.Run(fmt.Sprintf("%d", j), func(b *testing.B) {
  1863  					b.ReportAllocs()
  1864  					b.SetBytes(int64(j))
  1865  					data := sd.data[:j]
  1866  					for i := 0; i < b.N; i++ {
  1867  						FieldsFunc(data, unicode.IsSpace)
  1868  					}
  1869  				})
  1870  			}
  1871  		})
  1872  	}
  1873  }
  1874  
  1875  func BenchmarkTrimSpace(b *testing.B) {
  1876  	tests := []struct {
  1877  		name  string
  1878  		input []byte
  1879  	}{
  1880  		{"NoTrim", []byte("typical")},
  1881  		{"ASCII", []byte("  foo bar  ")},
  1882  		{"SomeNonASCII", []byte("    \u2000\t\r\n x\t\t\r\r\ny\n \u3000    ")},
  1883  		{"JustNonASCII", []byte("\u2000\u2000\u2000☺☺☺☺\u3000\u3000\u3000")},
  1884  	}
  1885  	for _, test := range tests {
  1886  		b.Run(test.name, func(b *testing.B) {
  1887  			for i := 0; i < b.N; i++ {
  1888  				TrimSpace(test.input)
  1889  			}
  1890  		})
  1891  	}
  1892  }
  1893  
  1894  func BenchmarkToValidUTF8(b *testing.B) {
  1895  	tests := []struct {
  1896  		name  string
  1897  		input []byte
  1898  	}{
  1899  		{"Valid", []byte("typical")},
  1900  		{"InvalidASCII", []byte("foo\xffbar")},
  1901  		{"InvalidNonASCII", []byte("日本語\xff日本語")},
  1902  	}
  1903  	replacement := []byte("\uFFFD")
  1904  	b.ResetTimer()
  1905  	for _, test := range tests {
  1906  		b.Run(test.name, func(b *testing.B) {
  1907  			for i := 0; i < b.N; i++ {
  1908  				ToValidUTF8(test.input, replacement)
  1909  			}
  1910  		})
  1911  	}
  1912  }
  1913  
  1914  func makeBenchInputHard() []byte {
  1915  	tokens := [...]string{
  1916  		"<a>", "<p>", "<b>", "<strong>",
  1917  		"</a>", "</p>", "</b>", "</strong>",
  1918  		"hello", "world",
  1919  	}
  1920  	x := make([]byte, 0, 1<<20)
  1921  	for {
  1922  		i := rand.Intn(len(tokens))
  1923  		if len(x)+len(tokens[i]) >= 1<<20 {
  1924  			break
  1925  		}
  1926  		x = append(x, tokens[i]...)
  1927  	}
  1928  	return x
  1929  }
  1930  
  1931  var benchInputHard = makeBenchInputHard()
  1932  
  1933  func benchmarkIndexHard(b *testing.B, sep []byte) {
  1934  	for i := 0; i < b.N; i++ {
  1935  		Index(benchInputHard, sep)
  1936  	}
  1937  }
  1938  
  1939  func benchmarkLastIndexHard(b *testing.B, sep []byte) {
  1940  	for i := 0; i < b.N; i++ {
  1941  		LastIndex(benchInputHard, sep)
  1942  	}
  1943  }
  1944  
  1945  func benchmarkCountHard(b *testing.B, sep []byte) {
  1946  	for i := 0; i < b.N; i++ {
  1947  		Count(benchInputHard, sep)
  1948  	}
  1949  }
  1950  
  1951  func BenchmarkIndexHard1(b *testing.B) { benchmarkIndexHard(b, []byte("<>")) }
  1952  func BenchmarkIndexHard2(b *testing.B) { benchmarkIndexHard(b, []byte("</pre>")) }
  1953  func BenchmarkIndexHard3(b *testing.B) { benchmarkIndexHard(b, []byte("<b>hello world</b>")) }
  1954  func BenchmarkIndexHard4(b *testing.B) {
  1955  	benchmarkIndexHard(b, []byte("<pre><b>hello</b><strong>world</strong></pre>"))
  1956  }
  1957  
  1958  func BenchmarkLastIndexHard1(b *testing.B) { benchmarkLastIndexHard(b, []byte("<>")) }
  1959  func BenchmarkLastIndexHard2(b *testing.B) { benchmarkLastIndexHard(b, []byte("</pre>")) }
  1960  func BenchmarkLastIndexHard3(b *testing.B) { benchmarkLastIndexHard(b, []byte("<b>hello world</b>")) }
  1961  
  1962  func BenchmarkCountHard1(b *testing.B) { benchmarkCountHard(b, []byte("<>")) }
  1963  func BenchmarkCountHard2(b *testing.B) { benchmarkCountHard(b, []byte("</pre>")) }
  1964  func BenchmarkCountHard3(b *testing.B) { benchmarkCountHard(b, []byte("<b>hello world</b>")) }
  1965  
  1966  func BenchmarkSplitEmptySeparator(b *testing.B) {
  1967  	for i := 0; i < b.N; i++ {
  1968  		Split(benchInputHard, nil)
  1969  	}
  1970  }
  1971  
  1972  func BenchmarkSplitSingleByteSeparator(b *testing.B) {
  1973  	sep := []byte("/")
  1974  	for i := 0; i < b.N; i++ {
  1975  		Split(benchInputHard, sep)
  1976  	}
  1977  }
  1978  
  1979  func BenchmarkSplitMultiByteSeparator(b *testing.B) {
  1980  	sep := []byte("hello")
  1981  	for i := 0; i < b.N; i++ {
  1982  		Split(benchInputHard, sep)
  1983  	}
  1984  }
  1985  
  1986  func BenchmarkSplitNSingleByteSeparator(b *testing.B) {
  1987  	sep := []byte("/")
  1988  	for i := 0; i < b.N; i++ {
  1989  		SplitN(benchInputHard, sep, 10)
  1990  	}
  1991  }
  1992  
  1993  func BenchmarkSplitNMultiByteSeparator(b *testing.B) {
  1994  	sep := []byte("hello")
  1995  	for i := 0; i < b.N; i++ {
  1996  		SplitN(benchInputHard, sep, 10)
  1997  	}
  1998  }
  1999  
  2000  func BenchmarkRepeat(b *testing.B) {
  2001  	for i := 0; i < b.N; i++ {
  2002  		Repeat([]byte("-"), 80)
  2003  	}
  2004  }
  2005  
  2006  func BenchmarkBytesCompare(b *testing.B) {
  2007  	for n := 1; n <= 2048; n <<= 1 {
  2008  		b.Run(fmt.Sprint(n), func(b *testing.B) {
  2009  			var x = make([]byte, n)
  2010  			var y = make([]byte, n)
  2011  
  2012  			for i := 0; i < n; i++ {
  2013  				x[i] = 'a'
  2014  			}
  2015  
  2016  			for i := 0; i < n; i++ {
  2017  				y[i] = 'a'
  2018  			}
  2019  
  2020  			b.ResetTimer()
  2021  			for i := 0; i < b.N; i++ {
  2022  				Compare(x, y)
  2023  			}
  2024  		})
  2025  	}
  2026  }
  2027  
  2028  func BenchmarkIndexAnyASCII(b *testing.B) {
  2029  	x := Repeat([]byte{'#'}, 2048) // Never matches set
  2030  	cs := "0123456789abcdefghijklmnopqrstuvwxyz0123456789abcdefghijklmnopqrstuvwxyz"
  2031  	for k := 1; k <= 2048; k <<= 4 {
  2032  		for j := 1; j <= 64; j <<= 1 {
  2033  			b.Run(fmt.Sprintf("%d:%d", k, j), func(b *testing.B) {
  2034  				for i := 0; i < b.N; i++ {
  2035  					IndexAny(x[:k], cs[:j])
  2036  				}
  2037  			})
  2038  		}
  2039  	}
  2040  }
  2041  
  2042  func BenchmarkIndexAnyUTF8(b *testing.B) {
  2043  	x := Repeat([]byte{'#'}, 2048) // Never matches set
  2044  	cs := "你好世界, hello world. 你好世界, hello world. 你好世界, hello world."
  2045  	for k := 1; k <= 2048; k <<= 4 {
  2046  		for j := 1; j <= 64; j <<= 1 {
  2047  			b.Run(fmt.Sprintf("%d:%d", k, j), func(b *testing.B) {
  2048  				for i := 0; i < b.N; i++ {
  2049  					IndexAny(x[:k], cs[:j])
  2050  				}
  2051  			})
  2052  		}
  2053  	}
  2054  }
  2055  
  2056  func BenchmarkLastIndexAnyASCII(b *testing.B) {
  2057  	x := Repeat([]byte{'#'}, 2048) // Never matches set
  2058  	cs := "0123456789abcdefghijklmnopqrstuvwxyz0123456789abcdefghijklmnopqrstuvwxyz"
  2059  	for k := 1; k <= 2048; k <<= 4 {
  2060  		for j := 1; j <= 64; j <<= 1 {
  2061  			b.Run(fmt.Sprintf("%d:%d", k, j), func(b *testing.B) {
  2062  				for i := 0; i < b.N; i++ {
  2063  					LastIndexAny(x[:k], cs[:j])
  2064  				}
  2065  			})
  2066  		}
  2067  	}
  2068  }
  2069  
  2070  func BenchmarkLastIndexAnyUTF8(b *testing.B) {
  2071  	x := Repeat([]byte{'#'}, 2048) // Never matches set
  2072  	cs := "你好世界, hello world. 你好世界, hello world. 你好世界, hello world."
  2073  	for k := 1; k <= 2048; k <<= 4 {
  2074  		for j := 1; j <= 64; j <<= 1 {
  2075  			b.Run(fmt.Sprintf("%d:%d", k, j), func(b *testing.B) {
  2076  				for i := 0; i < b.N; i++ {
  2077  					LastIndexAny(x[:k], cs[:j])
  2078  				}
  2079  			})
  2080  		}
  2081  	}
  2082  }
  2083  
  2084  func BenchmarkTrimASCII(b *testing.B) {
  2085  	cs := "0123456789abcdef"
  2086  	for k := 1; k <= 4096; k <<= 4 {
  2087  		for j := 1; j <= 16; j <<= 1 {
  2088  			b.Run(fmt.Sprintf("%d:%d", k, j), func(b *testing.B) {
  2089  				x := Repeat([]byte(cs[:j]), k) // Always matches set
  2090  				for i := 0; i < b.N; i++ {
  2091  					Trim(x[:k], cs[:j])
  2092  				}
  2093  			})
  2094  		}
  2095  	}
  2096  }
  2097  
  2098  func BenchmarkTrimByte(b *testing.B) {
  2099  	x := []byte("  the quick brown fox   ")
  2100  	for i := 0; i < b.N; i++ {
  2101  		Trim(x, " ")
  2102  	}
  2103  }
  2104  
  2105  func BenchmarkIndexPeriodic(b *testing.B) {
  2106  	key := []byte{1, 1}
  2107  	for _, skip := range [...]int{2, 4, 8, 16, 32, 64} {
  2108  		b.Run(fmt.Sprintf("IndexPeriodic%d", skip), func(b *testing.B) {
  2109  			buf := make([]byte, 1<<16)
  2110  			for i := 0; i < len(buf); i += skip {
  2111  				buf[i] = 1
  2112  			}
  2113  			for i := 0; i < b.N; i++ {
  2114  				Index(buf, key)
  2115  			}
  2116  		})
  2117  	}
  2118  }
  2119  

View as plain text