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

View as plain text