Source file src/strings/strings.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 strings implements simple functions to manipulate UTF-8 encoded strings.
     6  //
     7  // For information about UTF-8 strings in Go, see https://blog.golang.org/strings.
     8  package strings
     9  
    10  import (
    11  	"internal/bytealg"
    12  	"unicode"
    13  	"unicode/utf8"
    14  )
    15  
    16  const maxInt = int(^uint(0) >> 1)
    17  
    18  // explode splits s into a slice of UTF-8 strings,
    19  // one string per Unicode character up to a maximum of n (n < 0 means no limit).
    20  // Invalid UTF-8 bytes are sliced individually.
    21  func explode(s string, n int) []string {
    22  	l := utf8.RuneCountInString(s)
    23  	if n < 0 || n > l {
    24  		n = l
    25  	}
    26  	a := make([]string, n)
    27  	for i := 0; i < n-1; i++ {
    28  		_, size := utf8.DecodeRuneInString(s)
    29  		a[i] = s[:size]
    30  		s = s[size:]
    31  	}
    32  	if n > 0 {
    33  		a[n-1] = s
    34  	}
    35  	return a
    36  }
    37  
    38  // Count counts the number of non-overlapping instances of substr in s.
    39  // If substr is an empty string, Count returns 1 + the number of Unicode code points in s.
    40  func Count(s, substr string) int {
    41  	// special case
    42  	if len(substr) == 0 {
    43  		return utf8.RuneCountInString(s) + 1
    44  	}
    45  	if len(substr) == 1 {
    46  		return bytealg.CountString(s, substr[0])
    47  	}
    48  	n := 0
    49  	for {
    50  		i := Index(s, substr)
    51  		if i == -1 {
    52  			return n
    53  		}
    54  		n++
    55  		s = s[i+len(substr):]
    56  	}
    57  }
    58  
    59  // Contains reports whether substr is within s.
    60  func Contains(s, substr string) bool {
    61  	return Index(s, substr) >= 0
    62  }
    63  
    64  // ContainsAny reports whether any Unicode code points in chars are within s.
    65  func ContainsAny(s, chars string) bool {
    66  	return IndexAny(s, chars) >= 0
    67  }
    68  
    69  // ContainsRune reports whether the Unicode code point r is within s.
    70  func ContainsRune(s string, r rune) bool {
    71  	return IndexRune(s, r) >= 0
    72  }
    73  
    74  // ContainsFunc reports whether any Unicode code points r within s satisfy f(r).
    75  func ContainsFunc(s string, f func(rune) bool) bool {
    76  	return IndexFunc(s, f) >= 0
    77  }
    78  
    79  // LastIndex returns the index of the last instance of substr in s, or -1 if substr is not present in s.
    80  func LastIndex(s, substr string) int {
    81  	n := len(substr)
    82  	switch {
    83  	case n == 0:
    84  		return len(s)
    85  	case n == 1:
    86  		return LastIndexByte(s, substr[0])
    87  	case n == len(s):
    88  		if substr == s {
    89  			return 0
    90  		}
    91  		return -1
    92  	case n > len(s):
    93  		return -1
    94  	}
    95  	// Rabin-Karp search from the end of the string
    96  	hashss, pow := bytealg.HashStrRev(substr)
    97  	last := len(s) - n
    98  	var h uint32
    99  	for i := len(s) - 1; i >= last; i-- {
   100  		h = h*bytealg.PrimeRK + uint32(s[i])
   101  	}
   102  	if h == hashss && s[last:] == substr {
   103  		return last
   104  	}
   105  	for i := last - 1; i >= 0; i-- {
   106  		h *= bytealg.PrimeRK
   107  		h += uint32(s[i])
   108  		h -= pow * uint32(s[i+n])
   109  		if h == hashss && s[i:i+n] == substr {
   110  			return i
   111  		}
   112  	}
   113  	return -1
   114  }
   115  
   116  // IndexByte returns the index of the first instance of c in s, or -1 if c is not present in s.
   117  func IndexByte(s string, c byte) int {
   118  	return bytealg.IndexByteString(s, c)
   119  }
   120  
   121  // IndexRune returns the index of the first instance of the Unicode code point
   122  // r, or -1 if rune is not present in s.
   123  // If r is utf8.RuneError, it returns the first instance of any
   124  // invalid UTF-8 byte sequence.
   125  func IndexRune(s string, r rune) int {
   126  	switch {
   127  	case 0 <= r && r < utf8.RuneSelf:
   128  		return IndexByte(s, byte(r))
   129  	case r == utf8.RuneError:
   130  		for i, r := range s {
   131  			if r == utf8.RuneError {
   132  				return i
   133  			}
   134  		}
   135  		return -1
   136  	case !utf8.ValidRune(r):
   137  		return -1
   138  	default:
   139  		return Index(s, string(r))
   140  	}
   141  }
   142  
   143  // IndexAny returns the index of the first instance of any Unicode code point
   144  // from chars in s, or -1 if no Unicode code point from chars is present in s.
   145  func IndexAny(s, chars string) int {
   146  	if chars == "" {
   147  		// Avoid scanning all of s.
   148  		return -1
   149  	}
   150  	if len(chars) == 1 {
   151  		// Avoid scanning all of s.
   152  		r := rune(chars[0])
   153  		if r >= utf8.RuneSelf {
   154  			r = utf8.RuneError
   155  		}
   156  		return IndexRune(s, r)
   157  	}
   158  	if len(s) > 8 {
   159  		if as, isASCII := makeASCIISet(chars); isASCII {
   160  			for i := 0; i < len(s); i++ {
   161  				if as.contains(s[i]) {
   162  					return i
   163  				}
   164  			}
   165  			return -1
   166  		}
   167  	}
   168  	for i, c := range s {
   169  		if IndexRune(chars, c) >= 0 {
   170  			return i
   171  		}
   172  	}
   173  	return -1
   174  }
   175  
   176  // LastIndexAny returns the index of the last instance of any Unicode code
   177  // point from chars in s, or -1 if no Unicode code point from chars is
   178  // present in s.
   179  func LastIndexAny(s, chars string) int {
   180  	if chars == "" {
   181  		// Avoid scanning all of s.
   182  		return -1
   183  	}
   184  	if len(s) == 1 {
   185  		rc := rune(s[0])
   186  		if rc >= utf8.RuneSelf {
   187  			rc = utf8.RuneError
   188  		}
   189  		if IndexRune(chars, rc) >= 0 {
   190  			return 0
   191  		}
   192  		return -1
   193  	}
   194  	if len(s) > 8 {
   195  		if as, isASCII := makeASCIISet(chars); isASCII {
   196  			for i := len(s) - 1; i >= 0; i-- {
   197  				if as.contains(s[i]) {
   198  					return i
   199  				}
   200  			}
   201  			return -1
   202  		}
   203  	}
   204  	if len(chars) == 1 {
   205  		rc := rune(chars[0])
   206  		if rc >= utf8.RuneSelf {
   207  			rc = utf8.RuneError
   208  		}
   209  		for i := len(s); i > 0; {
   210  			r, size := utf8.DecodeLastRuneInString(s[:i])
   211  			i -= size
   212  			if rc == r {
   213  				return i
   214  			}
   215  		}
   216  		return -1
   217  	}
   218  	for i := len(s); i > 0; {
   219  		r, size := utf8.DecodeLastRuneInString(s[:i])
   220  		i -= size
   221  		if IndexRune(chars, r) >= 0 {
   222  			return i
   223  		}
   224  	}
   225  	return -1
   226  }
   227  
   228  // LastIndexByte returns the index of the last instance of c in s, or -1 if c is not present in s.
   229  func LastIndexByte(s string, c byte) int {
   230  	for i := len(s) - 1; i >= 0; i-- {
   231  		if s[i] == c {
   232  			return i
   233  		}
   234  	}
   235  	return -1
   236  }
   237  
   238  // Generic split: splits after each instance of sep,
   239  // including sepSave bytes of sep in the subarrays.
   240  func genSplit(s, sep string, sepSave, n int) []string {
   241  	if n == 0 {
   242  		return nil
   243  	}
   244  	if sep == "" {
   245  		return explode(s, n)
   246  	}
   247  	if n < 0 {
   248  		n = Count(s, sep) + 1
   249  	}
   250  
   251  	if n > len(s)+1 {
   252  		n = len(s) + 1
   253  	}
   254  	a := make([]string, n)
   255  	n--
   256  	i := 0
   257  	for i < n {
   258  		m := Index(s, sep)
   259  		if m < 0 {
   260  			break
   261  		}
   262  		a[i] = s[:m+sepSave]
   263  		s = s[m+len(sep):]
   264  		i++
   265  	}
   266  	a[i] = s
   267  	return a[:i+1]
   268  }
   269  
   270  // SplitN slices s into substrings separated by sep and returns a slice of
   271  // the substrings between those separators.
   272  //
   273  // The count determines the number of substrings to return:
   274  //
   275  //	n > 0: at most n substrings; the last substring will be the unsplit remainder.
   276  //	n == 0: the result is nil (zero substrings)
   277  //	n < 0: all substrings
   278  //
   279  // Edge cases for s and sep (for example, empty strings) are handled
   280  // as described in the documentation for Split.
   281  //
   282  // To split around the first instance of a separator, see Cut.
   283  func SplitN(s, sep string, n int) []string { return genSplit(s, sep, 0, n) }
   284  
   285  // SplitAfterN slices s into substrings after each instance of sep and
   286  // returns a slice of those substrings.
   287  //
   288  // The count determines the number of substrings to return:
   289  //
   290  //	n > 0: at most n substrings; the last substring will be the unsplit remainder.
   291  //	n == 0: the result is nil (zero substrings)
   292  //	n < 0: all substrings
   293  //
   294  // Edge cases for s and sep (for example, empty strings) are handled
   295  // as described in the documentation for SplitAfter.
   296  func SplitAfterN(s, sep string, n int) []string {
   297  	return genSplit(s, sep, len(sep), n)
   298  }
   299  
   300  // Split slices s into all substrings separated by sep and returns a slice of
   301  // the substrings between those separators.
   302  //
   303  // If s does not contain sep and sep is not empty, Split returns a
   304  // slice of length 1 whose only element is s.
   305  //
   306  // If sep is empty, Split splits after each UTF-8 sequence. If both s
   307  // and sep are empty, Split returns an empty slice.
   308  //
   309  // It is equivalent to SplitN with a count of -1.
   310  //
   311  // To split around the first instance of a separator, see Cut.
   312  func Split(s, sep string) []string { return genSplit(s, sep, 0, -1) }
   313  
   314  // SplitAfter slices s into all substrings after each instance of sep and
   315  // returns a slice of those substrings.
   316  //
   317  // If s does not contain sep and sep is not empty, SplitAfter returns
   318  // a slice of length 1 whose only element is s.
   319  //
   320  // If sep is empty, SplitAfter splits after each UTF-8 sequence. If
   321  // both s and sep are empty, SplitAfter returns an empty slice.
   322  //
   323  // It is equivalent to SplitAfterN with a count of -1.
   324  func SplitAfter(s, sep string) []string {
   325  	return genSplit(s, sep, len(sep), -1)
   326  }
   327  
   328  var asciiSpace = [256]uint8{'\t': 1, '\n': 1, '\v': 1, '\f': 1, '\r': 1, ' ': 1}
   329  
   330  // Fields splits the string s around each instance of one or more consecutive white space
   331  // characters, as defined by unicode.IsSpace, returning a slice of substrings of s or an
   332  // empty slice if s contains only white space.
   333  func Fields(s string) []string {
   334  	// First count the fields.
   335  	// This is an exact count if s is ASCII, otherwise it is an approximation.
   336  	n := 0
   337  	wasSpace := 1
   338  	// setBits is used to track which bits are set in the bytes of s.
   339  	setBits := uint8(0)
   340  	for i := 0; i < len(s); i++ {
   341  		r := s[i]
   342  		setBits |= r
   343  		isSpace := int(asciiSpace[r])
   344  		n += wasSpace & ^isSpace
   345  		wasSpace = isSpace
   346  	}
   347  
   348  	if setBits >= utf8.RuneSelf {
   349  		// Some runes in the input string are not ASCII.
   350  		return FieldsFunc(s, unicode.IsSpace)
   351  	}
   352  	// ASCII fast path
   353  	a := make([]string, n)
   354  	na := 0
   355  	fieldStart := 0
   356  	i := 0
   357  	// Skip spaces in the front of the input.
   358  	for i < len(s) && asciiSpace[s[i]] != 0 {
   359  		i++
   360  	}
   361  	fieldStart = i
   362  	for i < len(s) {
   363  		if asciiSpace[s[i]] == 0 {
   364  			i++
   365  			continue
   366  		}
   367  		a[na] = s[fieldStart:i]
   368  		na++
   369  		i++
   370  		// Skip spaces in between fields.
   371  		for i < len(s) && asciiSpace[s[i]] != 0 {
   372  			i++
   373  		}
   374  		fieldStart = i
   375  	}
   376  	if fieldStart < len(s) { // Last field might end at EOF.
   377  		a[na] = s[fieldStart:]
   378  	}
   379  	return a
   380  }
   381  
   382  // FieldsFunc splits the string s at each run of Unicode code points c satisfying f(c)
   383  // and returns an array of slices of s. If all code points in s satisfy f(c) or the
   384  // string is empty, an empty slice is returned.
   385  //
   386  // FieldsFunc makes no guarantees about the order in which it calls f(c)
   387  // and assumes that f always returns the same value for a given c.
   388  func FieldsFunc(s string, f func(rune) bool) []string {
   389  	// A span is used to record a slice of s of the form s[start:end].
   390  	// The start index is inclusive and the end index is exclusive.
   391  	type span struct {
   392  		start int
   393  		end   int
   394  	}
   395  	spans := make([]span, 0, 32)
   396  
   397  	// Find the field start and end indices.
   398  	// Doing this in a separate pass (rather than slicing the string s
   399  	// and collecting the result substrings right away) is significantly
   400  	// more efficient, possibly due to cache effects.
   401  	start := -1 // valid span start if >= 0
   402  	for end, rune := range s {
   403  		if f(rune) {
   404  			if start >= 0 {
   405  				spans = append(spans, span{start, end})
   406  				// Set start to a negative value.
   407  				// Note: using -1 here consistently and reproducibly
   408  				// slows down this code by a several percent on amd64.
   409  				start = ^start
   410  			}
   411  		} else {
   412  			if start < 0 {
   413  				start = end
   414  			}
   415  		}
   416  	}
   417  
   418  	// Last field might end at EOF.
   419  	if start >= 0 {
   420  		spans = append(spans, span{start, len(s)})
   421  	}
   422  
   423  	// Create strings from recorded field indices.
   424  	a := make([]string, len(spans))
   425  	for i, span := range spans {
   426  		a[i] = s[span.start:span.end]
   427  	}
   428  
   429  	return a
   430  }
   431  
   432  // Join concatenates the elements of its first argument to create a single string. The separator
   433  // string sep is placed between elements in the resulting string.
   434  func Join(elems []string, sep string) string {
   435  	switch len(elems) {
   436  	case 0:
   437  		return ""
   438  	case 1:
   439  		return elems[0]
   440  	}
   441  
   442  	var n int
   443  	if len(sep) > 0 {
   444  		if len(sep) >= maxInt/(len(elems)-1) {
   445  			panic("strings: Join output length overflow")
   446  		}
   447  		n += len(sep) * (len(elems) - 1)
   448  	}
   449  	for _, elem := range elems {
   450  		if len(elem) > maxInt-n {
   451  			panic("strings: Join output length overflow")
   452  		}
   453  		n += len(elem)
   454  	}
   455  
   456  	var b Builder
   457  	b.Grow(n)
   458  	b.WriteString(elems[0])
   459  	for _, s := range elems[1:] {
   460  		b.WriteString(sep)
   461  		b.WriteString(s)
   462  	}
   463  	return b.String()
   464  }
   465  
   466  // HasPrefix tests whether the string s begins with prefix.
   467  func HasPrefix(s, prefix string) bool {
   468  	return len(s) >= len(prefix) && s[0:len(prefix)] == prefix
   469  }
   470  
   471  // HasSuffix tests whether the string s ends with suffix.
   472  func HasSuffix(s, suffix string) bool {
   473  	return len(s) >= len(suffix) && s[len(s)-len(suffix):] == suffix
   474  }
   475  
   476  // Map returns a copy of the string s with all its characters modified
   477  // according to the mapping function. If mapping returns a negative value, the character is
   478  // dropped from the string with no replacement.
   479  func Map(mapping func(rune) rune, s string) string {
   480  	// In the worst case, the string can grow when mapped, making
   481  	// things unpleasant. But it's so rare we barge in assuming it's
   482  	// fine. It could also shrink but that falls out naturally.
   483  
   484  	// The output buffer b is initialized on demand, the first
   485  	// time a character differs.
   486  	var b Builder
   487  
   488  	for i, c := range s {
   489  		r := mapping(c)
   490  		if r == c && c != utf8.RuneError {
   491  			continue
   492  		}
   493  
   494  		var width int
   495  		if c == utf8.RuneError {
   496  			c, width = utf8.DecodeRuneInString(s[i:])
   497  			if width != 1 && r == c {
   498  				continue
   499  			}
   500  		} else {
   501  			width = utf8.RuneLen(c)
   502  		}
   503  
   504  		b.Grow(len(s) + utf8.UTFMax)
   505  		b.WriteString(s[:i])
   506  		if r >= 0 {
   507  			b.WriteRune(r)
   508  		}
   509  
   510  		s = s[i+width:]
   511  		break
   512  	}
   513  
   514  	// Fast path for unchanged input
   515  	if b.Cap() == 0 { // didn't call b.Grow above
   516  		return s
   517  	}
   518  
   519  	for _, c := range s {
   520  		r := mapping(c)
   521  
   522  		if r >= 0 {
   523  			// common case
   524  			// Due to inlining, it is more performant to determine if WriteByte should be
   525  			// invoked rather than always call WriteRune
   526  			if r < utf8.RuneSelf {
   527  				b.WriteByte(byte(r))
   528  			} else {
   529  				// r is not a ASCII rune.
   530  				b.WriteRune(r)
   531  			}
   532  		}
   533  	}
   534  
   535  	return b.String()
   536  }
   537  
   538  // Repeat returns a new string consisting of count copies of the string s.
   539  //
   540  // It panics if count is negative or if the result of (len(s) * count)
   541  // overflows.
   542  func Repeat(s string, count int) string {
   543  	switch count {
   544  	case 0:
   545  		return ""
   546  	case 1:
   547  		return s
   548  	}
   549  
   550  	// Since we cannot return an error on overflow,
   551  	// we should panic if the repeat will generate an overflow.
   552  	// See golang.org/issue/16237.
   553  	if count < 0 {
   554  		panic("strings: negative Repeat count")
   555  	}
   556  	if len(s) >= maxInt/count {
   557  		panic("strings: Repeat output length overflow")
   558  	}
   559  	n := len(s) * count
   560  
   561  	if len(s) == 0 {
   562  		return ""
   563  	}
   564  
   565  	// Past a certain chunk size it is counterproductive to use
   566  	// larger chunks as the source of the write, as when the source
   567  	// is too large we are basically just thrashing the CPU D-cache.
   568  	// So if the result length is larger than an empirically-found
   569  	// limit (8KB), we stop growing the source string once the limit
   570  	// is reached and keep reusing the same source string - that
   571  	// should therefore be always resident in the L1 cache - until we
   572  	// have completed the construction of the result.
   573  	// This yields significant speedups (up to +100%) in cases where
   574  	// the result length is large (roughly, over L2 cache size).
   575  	const chunkLimit = 8 * 1024
   576  	chunkMax := n
   577  	if n > chunkLimit {
   578  		chunkMax = chunkLimit / len(s) * len(s)
   579  		if chunkMax == 0 {
   580  			chunkMax = len(s)
   581  		}
   582  	}
   583  
   584  	var b Builder
   585  	b.Grow(n)
   586  	b.WriteString(s)
   587  	for b.Len() < n {
   588  		chunk := n - b.Len()
   589  		if chunk > b.Len() {
   590  			chunk = b.Len()
   591  		}
   592  		if chunk > chunkMax {
   593  			chunk = chunkMax
   594  		}
   595  		b.WriteString(b.String()[:chunk])
   596  	}
   597  	return b.String()
   598  }
   599  
   600  // ToUpper returns s with all Unicode letters mapped to their upper case.
   601  func ToUpper(s string) string {
   602  	isASCII, hasLower := true, false
   603  	for i := 0; i < len(s); i++ {
   604  		c := s[i]
   605  		if c >= utf8.RuneSelf {
   606  			isASCII = false
   607  			break
   608  		}
   609  		hasLower = hasLower || ('a' <= c && c <= 'z')
   610  	}
   611  
   612  	if isASCII { // optimize for ASCII-only strings.
   613  		if !hasLower {
   614  			return s
   615  		}
   616  		var (
   617  			b   Builder
   618  			pos int
   619  		)
   620  		b.Grow(len(s))
   621  		for i := 0; i < len(s); i++ {
   622  			c := s[i]
   623  			if 'a' <= c && c <= 'z' {
   624  				c -= 'a' - 'A'
   625  				if pos < i {
   626  					b.WriteString(s[pos:i])
   627  				}
   628  				b.WriteByte(c)
   629  				pos = i + 1
   630  			}
   631  		}
   632  		if pos < len(s) {
   633  			b.WriteString(s[pos:])
   634  		}
   635  		return b.String()
   636  	}
   637  	return Map(unicode.ToUpper, s)
   638  }
   639  
   640  // ToLower returns s with all Unicode letters mapped to their lower case.
   641  func ToLower(s string) string {
   642  	isASCII, hasUpper := true, false
   643  	for i := 0; i < len(s); i++ {
   644  		c := s[i]
   645  		if c >= utf8.RuneSelf {
   646  			isASCII = false
   647  			break
   648  		}
   649  		hasUpper = hasUpper || ('A' <= c && c <= 'Z')
   650  	}
   651  
   652  	if isASCII { // optimize for ASCII-only strings.
   653  		if !hasUpper {
   654  			return s
   655  		}
   656  		var (
   657  			b   Builder
   658  			pos int
   659  		)
   660  		b.Grow(len(s))
   661  		for i := 0; i < len(s); i++ {
   662  			c := s[i]
   663  			if 'A' <= c && c <= 'Z' {
   664  				c += 'a' - 'A'
   665  				if pos < i {
   666  					b.WriteString(s[pos:i])
   667  				}
   668  				b.WriteByte(c)
   669  				pos = i + 1
   670  			}
   671  		}
   672  		if pos < len(s) {
   673  			b.WriteString(s[pos:])
   674  		}
   675  		return b.String()
   676  	}
   677  	return Map(unicode.ToLower, s)
   678  }
   679  
   680  // ToTitle returns a copy of the string s with all Unicode letters mapped to
   681  // their Unicode title case.
   682  func ToTitle(s string) string { return Map(unicode.ToTitle, s) }
   683  
   684  // ToUpperSpecial returns a copy of the string s with all Unicode letters mapped to their
   685  // upper case using the case mapping specified by c.
   686  func ToUpperSpecial(c unicode.SpecialCase, s string) string {
   687  	return Map(c.ToUpper, s)
   688  }
   689  
   690  // ToLowerSpecial returns a copy of the string s with all Unicode letters mapped to their
   691  // lower case using the case mapping specified by c.
   692  func ToLowerSpecial(c unicode.SpecialCase, s string) string {
   693  	return Map(c.ToLower, s)
   694  }
   695  
   696  // ToTitleSpecial returns a copy of the string s with all Unicode letters mapped to their
   697  // Unicode title case, giving priority to the special casing rules.
   698  func ToTitleSpecial(c unicode.SpecialCase, s string) string {
   699  	return Map(c.ToTitle, s)
   700  }
   701  
   702  // ToValidUTF8 returns a copy of the string s with each run of invalid UTF-8 byte sequences
   703  // replaced by the replacement string, which may be empty.
   704  func ToValidUTF8(s, replacement string) string {
   705  	var b Builder
   706  
   707  	for i, c := range s {
   708  		if c != utf8.RuneError {
   709  			continue
   710  		}
   711  
   712  		_, wid := utf8.DecodeRuneInString(s[i:])
   713  		if wid == 1 {
   714  			b.Grow(len(s) + len(replacement))
   715  			b.WriteString(s[:i])
   716  			s = s[i:]
   717  			break
   718  		}
   719  	}
   720  
   721  	// Fast path for unchanged input
   722  	if b.Cap() == 0 { // didn't call b.Grow above
   723  		return s
   724  	}
   725  
   726  	invalid := false // previous byte was from an invalid UTF-8 sequence
   727  	for i := 0; i < len(s); {
   728  		c := s[i]
   729  		if c < utf8.RuneSelf {
   730  			i++
   731  			invalid = false
   732  			b.WriteByte(c)
   733  			continue
   734  		}
   735  		_, wid := utf8.DecodeRuneInString(s[i:])
   736  		if wid == 1 {
   737  			i++
   738  			if !invalid {
   739  				invalid = true
   740  				b.WriteString(replacement)
   741  			}
   742  			continue
   743  		}
   744  		invalid = false
   745  		b.WriteString(s[i : i+wid])
   746  		i += wid
   747  	}
   748  
   749  	return b.String()
   750  }
   751  
   752  // isSeparator reports whether the rune could mark a word boundary.
   753  // TODO: update when package unicode captures more of the properties.
   754  func isSeparator(r rune) bool {
   755  	// ASCII alphanumerics and underscore are not separators
   756  	if r <= 0x7F {
   757  		switch {
   758  		case '0' <= r && r <= '9':
   759  			return false
   760  		case 'a' <= r && r <= 'z':
   761  			return false
   762  		case 'A' <= r && r <= 'Z':
   763  			return false
   764  		case r == '_':
   765  			return false
   766  		}
   767  		return true
   768  	}
   769  	// Letters and digits are not separators
   770  	if unicode.IsLetter(r) || unicode.IsDigit(r) {
   771  		return false
   772  	}
   773  	// Otherwise, all we can do for now is treat spaces as separators.
   774  	return unicode.IsSpace(r)
   775  }
   776  
   777  // Title returns a copy of the string s with all Unicode letters that begin words
   778  // mapped to their Unicode title case.
   779  //
   780  // Deprecated: The rule Title uses for word boundaries does not handle Unicode
   781  // punctuation properly. Use golang.org/x/text/cases instead.
   782  func Title(s string) string {
   783  	// Use a closure here to remember state.
   784  	// Hackish but effective. Depends on Map scanning in order and calling
   785  	// the closure once per rune.
   786  	prev := ' '
   787  	return Map(
   788  		func(r rune) rune {
   789  			if isSeparator(prev) {
   790  				prev = r
   791  				return unicode.ToTitle(r)
   792  			}
   793  			prev = r
   794  			return r
   795  		},
   796  		s)
   797  }
   798  
   799  // TrimLeftFunc returns a slice of the string s with all leading
   800  // Unicode code points c satisfying f(c) removed.
   801  func TrimLeftFunc(s string, f func(rune) bool) string {
   802  	i := indexFunc(s, f, false)
   803  	if i == -1 {
   804  		return ""
   805  	}
   806  	return s[i:]
   807  }
   808  
   809  // TrimRightFunc returns a slice of the string s with all trailing
   810  // Unicode code points c satisfying f(c) removed.
   811  func TrimRightFunc(s string, f func(rune) bool) string {
   812  	i := lastIndexFunc(s, f, false)
   813  	if i >= 0 && s[i] >= utf8.RuneSelf {
   814  		_, wid := utf8.DecodeRuneInString(s[i:])
   815  		i += wid
   816  	} else {
   817  		i++
   818  	}
   819  	return s[0:i]
   820  }
   821  
   822  // TrimFunc returns a slice of the string s with all leading
   823  // and trailing Unicode code points c satisfying f(c) removed.
   824  func TrimFunc(s string, f func(rune) bool) string {
   825  	return TrimRightFunc(TrimLeftFunc(s, f), f)
   826  }
   827  
   828  // IndexFunc returns the index into s of the first Unicode
   829  // code point satisfying f(c), or -1 if none do.
   830  func IndexFunc(s string, f func(rune) bool) int {
   831  	return indexFunc(s, f, true)
   832  }
   833  
   834  // LastIndexFunc returns the index into s of the last
   835  // Unicode code point satisfying f(c), or -1 if none do.
   836  func LastIndexFunc(s string, f func(rune) bool) int {
   837  	return lastIndexFunc(s, f, true)
   838  }
   839  
   840  // indexFunc is the same as IndexFunc except that if
   841  // truth==false, the sense of the predicate function is
   842  // inverted.
   843  func indexFunc(s string, f func(rune) bool, truth bool) int {
   844  	for i, r := range s {
   845  		if f(r) == truth {
   846  			return i
   847  		}
   848  	}
   849  	return -1
   850  }
   851  
   852  // lastIndexFunc is the same as LastIndexFunc except that if
   853  // truth==false, the sense of the predicate function is
   854  // inverted.
   855  func lastIndexFunc(s string, f func(rune) bool, truth bool) int {
   856  	for i := len(s); i > 0; {
   857  		r, size := utf8.DecodeLastRuneInString(s[0:i])
   858  		i -= size
   859  		if f(r) == truth {
   860  			return i
   861  		}
   862  	}
   863  	return -1
   864  }
   865  
   866  // asciiSet is a 32-byte value, where each bit represents the presence of a
   867  // given ASCII character in the set. The 128-bits of the lower 16 bytes,
   868  // starting with the least-significant bit of the lowest word to the
   869  // most-significant bit of the highest word, map to the full range of all
   870  // 128 ASCII characters. The 128-bits of the upper 16 bytes will be zeroed,
   871  // ensuring that any non-ASCII character will be reported as not in the set.
   872  // This allocates a total of 32 bytes even though the upper half
   873  // is unused to avoid bounds checks in asciiSet.contains.
   874  type asciiSet [8]uint32
   875  
   876  // makeASCIISet creates a set of ASCII characters and reports whether all
   877  // characters in chars are ASCII.
   878  func makeASCIISet(chars string) (as asciiSet, ok bool) {
   879  	for i := 0; i < len(chars); i++ {
   880  		c := chars[i]
   881  		if c >= utf8.RuneSelf {
   882  			return as, false
   883  		}
   884  		as[c/32] |= 1 << (c % 32)
   885  	}
   886  	return as, true
   887  }
   888  
   889  // contains reports whether c is inside the set.
   890  func (as *asciiSet) contains(c byte) bool {
   891  	return (as[c/32] & (1 << (c % 32))) != 0
   892  }
   893  
   894  // Trim returns a slice of the string s with all leading and
   895  // trailing Unicode code points contained in cutset removed.
   896  func Trim(s, cutset string) string {
   897  	if s == "" || cutset == "" {
   898  		return s
   899  	}
   900  	if len(cutset) == 1 && cutset[0] < utf8.RuneSelf {
   901  		return trimLeftByte(trimRightByte(s, cutset[0]), cutset[0])
   902  	}
   903  	if as, ok := makeASCIISet(cutset); ok {
   904  		return trimLeftASCII(trimRightASCII(s, &as), &as)
   905  	}
   906  	return trimLeftUnicode(trimRightUnicode(s, cutset), cutset)
   907  }
   908  
   909  // TrimLeft returns a slice of the string s with all leading
   910  // Unicode code points contained in cutset removed.
   911  //
   912  // To remove a prefix, use TrimPrefix instead.
   913  func TrimLeft(s, cutset string) string {
   914  	if s == "" || cutset == "" {
   915  		return s
   916  	}
   917  	if len(cutset) == 1 && cutset[0] < utf8.RuneSelf {
   918  		return trimLeftByte(s, cutset[0])
   919  	}
   920  	if as, ok := makeASCIISet(cutset); ok {
   921  		return trimLeftASCII(s, &as)
   922  	}
   923  	return trimLeftUnicode(s, cutset)
   924  }
   925  
   926  func trimLeftByte(s string, c byte) string {
   927  	for len(s) > 0 && s[0] == c {
   928  		s = s[1:]
   929  	}
   930  	return s
   931  }
   932  
   933  func trimLeftASCII(s string, as *asciiSet) string {
   934  	for len(s) > 0 {
   935  		if !as.contains(s[0]) {
   936  			break
   937  		}
   938  		s = s[1:]
   939  	}
   940  	return s
   941  }
   942  
   943  func trimLeftUnicode(s, cutset string) string {
   944  	for len(s) > 0 {
   945  		r, n := rune(s[0]), 1
   946  		if r >= utf8.RuneSelf {
   947  			r, n = utf8.DecodeRuneInString(s)
   948  		}
   949  		if !ContainsRune(cutset, r) {
   950  			break
   951  		}
   952  		s = s[n:]
   953  	}
   954  	return s
   955  }
   956  
   957  // TrimRight returns a slice of the string s, with all trailing
   958  // Unicode code points contained in cutset removed.
   959  //
   960  // To remove a suffix, use TrimSuffix instead.
   961  func TrimRight(s, cutset string) string {
   962  	if s == "" || cutset == "" {
   963  		return s
   964  	}
   965  	if len(cutset) == 1 && cutset[0] < utf8.RuneSelf {
   966  		return trimRightByte(s, cutset[0])
   967  	}
   968  	if as, ok := makeASCIISet(cutset); ok {
   969  		return trimRightASCII(s, &as)
   970  	}
   971  	return trimRightUnicode(s, cutset)
   972  }
   973  
   974  func trimRightByte(s string, c byte) string {
   975  	for len(s) > 0 && s[len(s)-1] == c {
   976  		s = s[:len(s)-1]
   977  	}
   978  	return s
   979  }
   980  
   981  func trimRightASCII(s string, as *asciiSet) string {
   982  	for len(s) > 0 {
   983  		if !as.contains(s[len(s)-1]) {
   984  			break
   985  		}
   986  		s = s[:len(s)-1]
   987  	}
   988  	return s
   989  }
   990  
   991  func trimRightUnicode(s, cutset string) string {
   992  	for len(s) > 0 {
   993  		r, n := rune(s[len(s)-1]), 1
   994  		if r >= utf8.RuneSelf {
   995  			r, n = utf8.DecodeLastRuneInString(s)
   996  		}
   997  		if !ContainsRune(cutset, r) {
   998  			break
   999  		}
  1000  		s = s[:len(s)-n]
  1001  	}
  1002  	return s
  1003  }
  1004  
  1005  // TrimSpace returns a slice of the string s, with all leading
  1006  // and trailing white space removed, as defined by Unicode.
  1007  func TrimSpace(s string) string {
  1008  	// Fast path for ASCII: look for the first ASCII non-space byte
  1009  	start := 0
  1010  	for ; start < len(s); start++ {
  1011  		c := s[start]
  1012  		if c >= utf8.RuneSelf {
  1013  			// If we run into a non-ASCII byte, fall back to the
  1014  			// slower unicode-aware method on the remaining bytes
  1015  			return TrimFunc(s[start:], unicode.IsSpace)
  1016  		}
  1017  		if asciiSpace[c] == 0 {
  1018  			break
  1019  		}
  1020  	}
  1021  
  1022  	// Now look for the first ASCII non-space byte from the end
  1023  	stop := len(s)
  1024  	for ; stop > start; stop-- {
  1025  		c := s[stop-1]
  1026  		if c >= utf8.RuneSelf {
  1027  			// start has been already trimmed above, should trim end only
  1028  			return TrimRightFunc(s[start:stop], unicode.IsSpace)
  1029  		}
  1030  		if asciiSpace[c] == 0 {
  1031  			break
  1032  		}
  1033  	}
  1034  
  1035  	// At this point s[start:stop] starts and ends with an ASCII
  1036  	// non-space bytes, so we're done. Non-ASCII cases have already
  1037  	// been handled above.
  1038  	return s[start:stop]
  1039  }
  1040  
  1041  // TrimPrefix returns s without the provided leading prefix string.
  1042  // If s doesn't start with prefix, s is returned unchanged.
  1043  func TrimPrefix(s, prefix string) string {
  1044  	if HasPrefix(s, prefix) {
  1045  		return s[len(prefix):]
  1046  	}
  1047  	return s
  1048  }
  1049  
  1050  // TrimSuffix returns s without the provided trailing suffix string.
  1051  // If s doesn't end with suffix, s is returned unchanged.
  1052  func TrimSuffix(s, suffix string) string {
  1053  	if HasSuffix(s, suffix) {
  1054  		return s[:len(s)-len(suffix)]
  1055  	}
  1056  	return s
  1057  }
  1058  
  1059  // Replace returns a copy of the string s with the first n
  1060  // non-overlapping instances of old replaced by new.
  1061  // If old is empty, it matches at the beginning of the string
  1062  // and after each UTF-8 sequence, yielding up to k+1 replacements
  1063  // for a k-rune string.
  1064  // If n < 0, there is no limit on the number of replacements.
  1065  func Replace(s, old, new string, n int) string {
  1066  	if old == new || n == 0 {
  1067  		return s // avoid allocation
  1068  	}
  1069  
  1070  	// Compute number of replacements.
  1071  	if m := Count(s, old); m == 0 {
  1072  		return s // avoid allocation
  1073  	} else if n < 0 || m < n {
  1074  		n = m
  1075  	}
  1076  
  1077  	// Apply replacements to buffer.
  1078  	var b Builder
  1079  	b.Grow(len(s) + n*(len(new)-len(old)))
  1080  	start := 0
  1081  	for i := 0; i < n; i++ {
  1082  		j := start
  1083  		if len(old) == 0 {
  1084  			if i > 0 {
  1085  				_, wid := utf8.DecodeRuneInString(s[start:])
  1086  				j += wid
  1087  			}
  1088  		} else {
  1089  			j += Index(s[start:], old)
  1090  		}
  1091  		b.WriteString(s[start:j])
  1092  		b.WriteString(new)
  1093  		start = j + len(old)
  1094  	}
  1095  	b.WriteString(s[start:])
  1096  	return b.String()
  1097  }
  1098  
  1099  // ReplaceAll returns a copy of the string s with all
  1100  // non-overlapping instances of old replaced by new.
  1101  // If old is empty, it matches at the beginning of the string
  1102  // and after each UTF-8 sequence, yielding up to k+1 replacements
  1103  // for a k-rune string.
  1104  func ReplaceAll(s, old, new string) string {
  1105  	return Replace(s, old, new, -1)
  1106  }
  1107  
  1108  // EqualFold reports whether s and t, interpreted as UTF-8 strings,
  1109  // are equal under simple Unicode case-folding, which is a more general
  1110  // form of case-insensitivity.
  1111  func EqualFold(s, t string) bool {
  1112  	// ASCII fast path
  1113  	i := 0
  1114  	for ; i < len(s) && i < len(t); i++ {
  1115  		sr := s[i]
  1116  		tr := t[i]
  1117  		if sr|tr >= utf8.RuneSelf {
  1118  			goto hasUnicode
  1119  		}
  1120  
  1121  		// Easy case.
  1122  		if tr == sr {
  1123  			continue
  1124  		}
  1125  
  1126  		// Make sr < tr to simplify what follows.
  1127  		if tr < sr {
  1128  			tr, sr = sr, tr
  1129  		}
  1130  		// ASCII only, sr/tr must be upper/lower case
  1131  		if 'A' <= sr && sr <= 'Z' && tr == sr+'a'-'A' {
  1132  			continue
  1133  		}
  1134  		return false
  1135  	}
  1136  	// Check if we've exhausted both strings.
  1137  	return len(s) == len(t)
  1138  
  1139  hasUnicode:
  1140  	s = s[i:]
  1141  	t = t[i:]
  1142  	for _, sr := range s {
  1143  		// If t is exhausted the strings are not equal.
  1144  		if len(t) == 0 {
  1145  			return false
  1146  		}
  1147  
  1148  		// Extract first rune from second string.
  1149  		var tr rune
  1150  		if t[0] < utf8.RuneSelf {
  1151  			tr, t = rune(t[0]), t[1:]
  1152  		} else {
  1153  			r, size := utf8.DecodeRuneInString(t)
  1154  			tr, t = r, t[size:]
  1155  		}
  1156  
  1157  		// If they match, keep going; if not, return false.
  1158  
  1159  		// Easy case.
  1160  		if tr == sr {
  1161  			continue
  1162  		}
  1163  
  1164  		// Make sr < tr to simplify what follows.
  1165  		if tr < sr {
  1166  			tr, sr = sr, tr
  1167  		}
  1168  		// Fast check for ASCII.
  1169  		if tr < utf8.RuneSelf {
  1170  			// ASCII only, sr/tr must be upper/lower case
  1171  			if 'A' <= sr && sr <= 'Z' && tr == sr+'a'-'A' {
  1172  				continue
  1173  			}
  1174  			return false
  1175  		}
  1176  
  1177  		// General case. SimpleFold(x) returns the next equivalent rune > x
  1178  		// or wraps around to smaller values.
  1179  		r := unicode.SimpleFold(sr)
  1180  		for r != sr && r < tr {
  1181  			r = unicode.SimpleFold(r)
  1182  		}
  1183  		if r == tr {
  1184  			continue
  1185  		}
  1186  		return false
  1187  	}
  1188  
  1189  	// First string is empty, so check if the second one is also empty.
  1190  	return len(t) == 0
  1191  }
  1192  
  1193  // Index returns the index of the first instance of substr in s, or -1 if substr is not present in s.
  1194  func Index(s, substr string) int {
  1195  	n := len(substr)
  1196  	switch {
  1197  	case n == 0:
  1198  		return 0
  1199  	case n == 1:
  1200  		return IndexByte(s, substr[0])
  1201  	case n == len(s):
  1202  		if substr == s {
  1203  			return 0
  1204  		}
  1205  		return -1
  1206  	case n > len(s):
  1207  		return -1
  1208  	case n <= bytealg.MaxLen:
  1209  		// Use brute force when s and substr both are small
  1210  		if len(s) <= bytealg.MaxBruteForce {
  1211  			return bytealg.IndexString(s, substr)
  1212  		}
  1213  		c0 := substr[0]
  1214  		c1 := substr[1]
  1215  		i := 0
  1216  		t := len(s) - n + 1
  1217  		fails := 0
  1218  		for i < t {
  1219  			if s[i] != c0 {
  1220  				// IndexByte is faster than bytealg.IndexString, so use it as long as
  1221  				// we're not getting lots of false positives.
  1222  				o := IndexByte(s[i+1:t], c0)
  1223  				if o < 0 {
  1224  					return -1
  1225  				}
  1226  				i += o + 1
  1227  			}
  1228  			if s[i+1] == c1 && s[i:i+n] == substr {
  1229  				return i
  1230  			}
  1231  			fails++
  1232  			i++
  1233  			// Switch to bytealg.IndexString when IndexByte produces too many false positives.
  1234  			if fails > bytealg.Cutover(i) {
  1235  				r := bytealg.IndexString(s[i:], substr)
  1236  				if r >= 0 {
  1237  					return r + i
  1238  				}
  1239  				return -1
  1240  			}
  1241  		}
  1242  		return -1
  1243  	}
  1244  	c0 := substr[0]
  1245  	c1 := substr[1]
  1246  	i := 0
  1247  	t := len(s) - n + 1
  1248  	fails := 0
  1249  	for i < t {
  1250  		if s[i] != c0 {
  1251  			o := IndexByte(s[i+1:t], c0)
  1252  			if o < 0 {
  1253  				return -1
  1254  			}
  1255  			i += o + 1
  1256  		}
  1257  		if s[i+1] == c1 && s[i:i+n] == substr {
  1258  			return i
  1259  		}
  1260  		i++
  1261  		fails++
  1262  		if fails >= 4+i>>4 && i < t {
  1263  			// See comment in ../bytes/bytes.go.
  1264  			j := bytealg.IndexRabinKarp(s[i:], substr)
  1265  			if j < 0 {
  1266  				return -1
  1267  			}
  1268  			return i + j
  1269  		}
  1270  	}
  1271  	return -1
  1272  }
  1273  
  1274  // Cut slices s around the first instance of sep,
  1275  // returning the text before and after sep.
  1276  // The found result reports whether sep appears in s.
  1277  // If sep does not appear in s, cut returns s, "", false.
  1278  func Cut(s, sep string) (before, after string, found bool) {
  1279  	if i := Index(s, sep); i >= 0 {
  1280  		return s[:i], s[i+len(sep):], true
  1281  	}
  1282  	return s, "", false
  1283  }
  1284  
  1285  // CutPrefix returns s without the provided leading prefix string
  1286  // and reports whether it found the prefix.
  1287  // If s doesn't start with prefix, CutPrefix returns s, false.
  1288  // If prefix is the empty string, CutPrefix returns s, true.
  1289  func CutPrefix(s, prefix string) (after string, found bool) {
  1290  	if !HasPrefix(s, prefix) {
  1291  		return s, false
  1292  	}
  1293  	return s[len(prefix):], true
  1294  }
  1295  
  1296  // CutSuffix returns s without the provided ending suffix string
  1297  // and reports whether it found the suffix.
  1298  // If s doesn't end with suffix, CutSuffix returns s, false.
  1299  // If suffix is the empty string, CutSuffix returns s, true.
  1300  func CutSuffix(s, suffix string) (before string, found bool) {
  1301  	if !HasSuffix(s, suffix) {
  1302  		return s, false
  1303  	}
  1304  	return s[:len(s)-len(suffix)], true
  1305  }
  1306  

View as plain text