Source file src/bytes/bytes.go

Documentation: bytes

  // Copyright 2009 The Go Authors. All rights reserved.
  // Use of this source code is governed by a BSD-style
  // license that can be found in the LICENSE file.
  
  // Package bytes implements functions for the manipulation of byte slices.
  // It is analogous to the facilities of the strings package.
  package bytes
  
  import (
  	"unicode"
  	"unicode/utf8"
  )
  
  func equalPortable(a, b []byte) bool {
  	if len(a) != len(b) {
  		return false
  	}
  	for i, c := range a {
  		if c != b[i] {
  			return false
  		}
  	}
  	return true
  }
  
  // explode splits s into a slice of UTF-8 sequences, one per Unicode code point (still slices of bytes),
  // up to a maximum of n byte slices. Invalid UTF-8 sequences are chopped into individual bytes.
  func explode(s []byte, n int) [][]byte {
  	if n <= 0 {
  		n = len(s)
  	}
  	a := make([][]byte, n)
  	var size int
  	na := 0
  	for len(s) > 0 {
  		if na+1 >= n {
  			a[na] = s
  			na++
  			break
  		}
  		_, size = utf8.DecodeRune(s)
  		a[na] = s[0:size:size]
  		s = s[size:]
  		na++
  	}
  	return a[0:na]
  }
  
  // countGeneric actually implements Count
  func countGeneric(s, sep []byte) int {
  	// special case
  	if len(sep) == 0 {
  		return utf8.RuneCount(s) + 1
  	}
  	n := 0
  	for {
  		i := Index(s, sep)
  		if i == -1 {
  			return n
  		}
  		n++
  		s = s[i+len(sep):]
  	}
  }
  
  // Contains reports whether subslice is within b.
  func Contains(b, subslice []byte) bool {
  	return Index(b, subslice) != -1
  }
  
  // ContainsAny reports whether any of the UTF-8-encoded code points in chars are within b.
  func ContainsAny(b []byte, chars string) bool {
  	return IndexAny(b, chars) >= 0
  }
  
  // ContainsRune reports whether the rune is contained in the UTF-8-encoded byte slice b.
  func ContainsRune(b []byte, r rune) bool {
  	return IndexRune(b, r) >= 0
  }
  
  func indexBytePortable(s []byte, c byte) int {
  	for i, b := range s {
  		if b == c {
  			return i
  		}
  	}
  	return -1
  }
  
  // LastIndex returns the index of the last instance of sep in s, or -1 if sep is not present in s.
  func LastIndex(s, sep []byte) int {
  	n := len(sep)
  	if n == 0 {
  		return len(s)
  	}
  	c := sep[0]
  	for i := len(s) - n; i >= 0; i-- {
  		if s[i] == c && (n == 1 || Equal(s[i:i+n], sep)) {
  			return i
  		}
  	}
  	return -1
  }
  
  // LastIndexByte returns the index of the last instance of c in s, or -1 if c is not present in s.
  func LastIndexByte(s []byte, c byte) int {
  	for i := len(s) - 1; i >= 0; i-- {
  		if s[i] == c {
  			return i
  		}
  	}
  	return -1
  }
  
  // IndexRune interprets s as a sequence of UTF-8-encoded code points.
  // It returns the byte index of the first occurrence in s of the given rune.
  // It returns -1 if rune is not present in s.
  // If r is utf8.RuneError, it returns the first instance of any
  // invalid UTF-8 byte sequence.
  func IndexRune(s []byte, r rune) int {
  	switch {
  	case 0 <= r && r < utf8.RuneSelf:
  		return IndexByte(s, byte(r))
  	case r == utf8.RuneError:
  		for i := 0; i < len(s); {
  			r1, n := utf8.DecodeRune(s[i:])
  			if r1 == utf8.RuneError {
  				return i
  			}
  			i += n
  		}
  		return -1
  	case !utf8.ValidRune(r):
  		return -1
  	default:
  		var b [utf8.UTFMax]byte
  		n := utf8.EncodeRune(b[:], r)
  		return Index(s, b[:n])
  	}
  }
  
  // IndexAny interprets s as a sequence of UTF-8-encoded Unicode code points.
  // It returns the byte index of the first occurrence in s of any of the Unicode
  // code points in chars. It returns -1 if chars is empty or if there is no code
  // point in common.
  func IndexAny(s []byte, chars string) int {
  	if chars == "" {
  		// Avoid scanning all of s.
  		return -1
  	}
  	if len(s) > 8 {
  		if as, isASCII := makeASCIISet(chars); isASCII {
  			for i, c := range s {
  				if as.contains(c) {
  					return i
  				}
  			}
  			return -1
  		}
  	}
  	var width int
  	for i := 0; i < len(s); i += width {
  		r := rune(s[i])
  		if r < utf8.RuneSelf {
  			width = 1
  		} else {
  			r, width = utf8.DecodeRune(s[i:])
  		}
  		for _, ch := range chars {
  			if r == ch {
  				return i
  			}
  		}
  	}
  	return -1
  }
  
  // LastIndexAny interprets s as a sequence of UTF-8-encoded Unicode code
  // points. It returns the byte index of the last occurrence in s of any of
  // the Unicode code points in chars. It returns -1 if chars is empty or if
  // there is no code point in common.
  func LastIndexAny(s []byte, chars string) int {
  	if chars == "" {
  		// Avoid scanning all of s.
  		return -1
  	}
  	if len(s) > 8 {
  		if as, isASCII := makeASCIISet(chars); isASCII {
  			for i := len(s) - 1; i >= 0; i-- {
  				if as.contains(s[i]) {
  					return i
  				}
  			}
  			return -1
  		}
  	}
  	for i := len(s); i > 0; {
  		r, size := utf8.DecodeLastRune(s[:i])
  		i -= size
  		for _, c := range chars {
  			if r == c {
  				return i
  			}
  		}
  	}
  	return -1
  }
  
  // Generic split: splits after each instance of sep,
  // including sepSave bytes of sep in the subslices.
  func genSplit(s, sep []byte, sepSave, n int) [][]byte {
  	if n == 0 {
  		return nil
  	}
  	if len(sep) == 0 {
  		return explode(s, n)
  	}
  	if n < 0 {
  		n = Count(s, sep) + 1
  	}
  
  	a := make([][]byte, n)
  	n--
  	i := 0
  	for i < n {
  		m := Index(s, sep)
  		if m < 0 {
  			break
  		}
  		a[i] = s[: m+sepSave : m+sepSave]
  		s = s[m+len(sep):]
  		i++
  	}
  	a[i] = s
  	return a[:i+1]
  }
  
  // SplitN slices s into subslices separated by sep and returns a slice of
  // the subslices between those separators.
  // If sep is empty, SplitN splits after each UTF-8 sequence.
  // The count determines the number of subslices to return:
  //   n > 0: at most n subslices; the last subslice will be the unsplit remainder.
  //   n == 0: the result is nil (zero subslices)
  //   n < 0: all subslices
  func SplitN(s, sep []byte, n int) [][]byte { return genSplit(s, sep, 0, n) }
  
  // SplitAfterN slices s into subslices after each instance of sep and
  // returns a slice of those subslices.
  // If sep is empty, SplitAfterN splits after each UTF-8 sequence.
  // The count determines the number of subslices to return:
  //   n > 0: at most n subslices; the last subslice will be the unsplit remainder.
  //   n == 0: the result is nil (zero subslices)
  //   n < 0: all subslices
  func SplitAfterN(s, sep []byte, n int) [][]byte {
  	return genSplit(s, sep, len(sep), n)
  }
  
  // Split slices s into all subslices separated by sep and returns a slice of
  // the subslices between those separators.
  // If sep is empty, Split splits after each UTF-8 sequence.
  // It is equivalent to SplitN with a count of -1.
  func Split(s, sep []byte) [][]byte { return genSplit(s, sep, 0, -1) }
  
  // SplitAfter slices s into all subslices after each instance of sep and
  // returns a slice of those subslices.
  // If sep is empty, SplitAfter splits after each UTF-8 sequence.
  // It is equivalent to SplitAfterN with a count of -1.
  func SplitAfter(s, sep []byte) [][]byte {
  	return genSplit(s, sep, len(sep), -1)
  }
  
  var asciiSpace = [256]uint8{'\t': 1, '\n': 1, '\v': 1, '\f': 1, '\r': 1, ' ': 1}
  
  // Fields interprets s as a sequence of UTF-8-encoded code points.
  // It splits the slice s around each instance of one or more consecutive white space
  // characters, as defined by unicode.IsSpace, returning a slice of subslices of s or an
  // empty slice if s contains only white space.
  func Fields(s []byte) [][]byte {
  	// First count the fields.
  	// This is an exact count if s is ASCII, otherwise it is an approximation.
  	n := 0
  	wasSpace := 1
  	// setBits is used to track which bits are set in the bytes of s.
  	setBits := uint8(0)
  	for i := 0; i < len(s); i++ {
  		r := s[i]
  		setBits |= r
  		isSpace := int(asciiSpace[r])
  		n += wasSpace & ^isSpace
  		wasSpace = isSpace
  	}
  
  	if setBits >= utf8.RuneSelf {
  		// Some runes in the input slice are not ASCII.
  		return FieldsFunc(s, unicode.IsSpace)
  	}
  
  	// ASCII fast path
  	a := make([][]byte, n)
  	na := 0
  	fieldStart := 0
  	i := 0
  	// Skip spaces in the front of the input.
  	for i < len(s) && asciiSpace[s[i]] != 0 {
  		i++
  	}
  	fieldStart = i
  	for i < len(s) {
  		if asciiSpace[s[i]] == 0 {
  			i++
  			continue
  		}
  		a[na] = s[fieldStart:i:i]
  		na++
  		i++
  		// Skip spaces in between fields.
  		for i < len(s) && asciiSpace[s[i]] != 0 {
  			i++
  		}
  		fieldStart = i
  	}
  	if fieldStart < len(s) { // Last field might end at EOF.
  		a[na] = s[fieldStart:len(s):len(s)]
  	}
  	return a
  }
  
  // FieldsFunc interprets s as a sequence of UTF-8-encoded code points.
  // It splits the slice s at each run of code points c satisfying f(c) and
  // returns a slice of subslices of s. If all code points in s satisfy f(c), or
  // len(s) == 0, an empty slice is returned.
  // FieldsFunc makes no guarantees about the order in which it calls f(c).
  // If f does not return consistent results for a given c, FieldsFunc may crash.
  func FieldsFunc(s []byte, f func(rune) bool) [][]byte {
  	// A span is used to record a slice of s of the form s[start:end].
  	// The start index is inclusive and the end index is exclusive.
  	type span struct {
  		start int
  		end   int
  	}
  	spans := make([]span, 0, 32)
  
  	// Find the field start and end indices.
  	wasField := false
  	fromIndex := 0
  	for i := 0; i < len(s); {
  		size := 1
  		r := rune(s[i])
  		if r >= utf8.RuneSelf {
  			r, size = utf8.DecodeRune(s[i:])
  		}
  		if f(r) {
  			if wasField {
  				spans = append(spans, span{start: fromIndex, end: i})
  				wasField = false
  			}
  		} else {
  			if !wasField {
  				fromIndex = i
  				wasField = true
  			}
  		}
  		i += size
  	}
  
  	// Last field might end at EOF.
  	if wasField {
  		spans = append(spans, span{fromIndex, len(s)})
  	}
  
  	// Create subslices from recorded field indices.
  	a := make([][]byte, len(spans))
  	for i, span := range spans {
  		a[i] = s[span.start:span.end:span.end]
  	}
  
  	return a
  }
  
  // Join concatenates the elements of s to create a new byte slice. The separator
  // sep is placed between elements in the resulting slice.
  func Join(s [][]byte, sep []byte) []byte {
  	if len(s) == 0 {
  		return []byte{}
  	}
  	if len(s) == 1 {
  		// Just return a copy.
  		return append([]byte(nil), s[0]...)
  	}
  	n := len(sep) * (len(s) - 1)
  	for _, v := range s {
  		n += len(v)
  	}
  
  	b := make([]byte, n)
  	bp := copy(b, s[0])
  	for _, v := range s[1:] {
  		bp += copy(b[bp:], sep)
  		bp += copy(b[bp:], v)
  	}
  	return b
  }
  
  // HasPrefix tests whether the byte slice s begins with prefix.
  func HasPrefix(s, prefix []byte) bool {
  	return len(s) >= len(prefix) && Equal(s[0:len(prefix)], prefix)
  }
  
  // HasSuffix tests whether the byte slice s ends with suffix.
  func HasSuffix(s, suffix []byte) bool {
  	return len(s) >= len(suffix) && Equal(s[len(s)-len(suffix):], suffix)
  }
  
  // Map returns a copy of the byte slice s with all its characters modified
  // according to the mapping function. If mapping returns a negative value, the character is
  // dropped from the byte slice with no replacement. The characters in s and the
  // output are interpreted as UTF-8-encoded code points.
  func Map(mapping func(r rune) rune, s []byte) []byte {
  	// In the worst case, the slice can grow when mapped, making
  	// things unpleasant. But it's so rare we barge in assuming it's
  	// fine. It could also shrink but that falls out naturally.
  	maxbytes := len(s) // length of b
  	nbytes := 0        // number of bytes encoded in b
  	b := make([]byte, maxbytes)
  	for i := 0; i < len(s); {
  		wid := 1
  		r := rune(s[i])
  		if r >= utf8.RuneSelf {
  			r, wid = utf8.DecodeRune(s[i:])
  		}
  		r = mapping(r)
  		if r >= 0 {
  			rl := utf8.RuneLen(r)
  			if rl < 0 {
  				rl = len(string(utf8.RuneError))
  			}
  			if nbytes+rl > maxbytes {
  				// Grow the buffer.
  				maxbytes = maxbytes*2 + utf8.UTFMax
  				nb := make([]byte, maxbytes)
  				copy(nb, b[0:nbytes])
  				b = nb
  			}
  			nbytes += utf8.EncodeRune(b[nbytes:maxbytes], r)
  		}
  		i += wid
  	}
  	return b[0:nbytes]
  }
  
  // Repeat returns a new byte slice consisting of count copies of b.
  //
  // It panics if count is negative or if
  // the result of (len(b) * count) overflows.
  func Repeat(b []byte, count int) []byte {
  	// Since we cannot return an error on overflow,
  	// we should panic if the repeat will generate
  	// an overflow.
  	// See Issue golang.org/issue/16237.
  	if count < 0 {
  		panic("bytes: negative Repeat count")
  	} else if count > 0 && len(b)*count/count != len(b) {
  		panic("bytes: Repeat count causes overflow")
  	}
  
  	nb := make([]byte, len(b)*count)
  	bp := copy(nb, b)
  	for bp < len(nb) {
  		copy(nb[bp:], nb[:bp])
  		bp *= 2
  	}
  	return nb
  }
  
  // ToUpper treats s as UTF-8-encoded bytes and returns a copy with all the Unicode letters within it mapped to their upper case.
  func ToUpper(s []byte) []byte { return Map(unicode.ToUpper, s) }
  
  // ToLower treats s as UTF-8-encoded bytes and returns a copy with all the Unicode letters mapped to their lower case.
  func ToLower(s []byte) []byte { return Map(unicode.ToLower, s) }
  
  // ToTitle treats s as UTF-8-encoded bytes and returns a copy with all the Unicode letters mapped to their title case.
  func ToTitle(s []byte) []byte { return Map(unicode.ToTitle, s) }
  
  // ToUpperSpecial treats s as UTF-8-encoded bytes and returns a copy with all the Unicode letters mapped to their
  // upper case, giving priority to the special casing rules.
  func ToUpperSpecial(c unicode.SpecialCase, s []byte) []byte {
  	return Map(func(r rune) rune { return c.ToUpper(r) }, s)
  }
  
  // ToLowerSpecial treats s as UTF-8-encoded bytes and returns a copy with all the Unicode letters mapped to their
  // lower case, giving priority to the special casing rules.
  func ToLowerSpecial(c unicode.SpecialCase, s []byte) []byte {
  	return Map(func(r rune) rune { return c.ToLower(r) }, s)
  }
  
  // ToTitleSpecial treats s as UTF-8-encoded bytes and returns a copy with all the Unicode letters mapped to their
  // title case, giving priority to the special casing rules.
  func ToTitleSpecial(c unicode.SpecialCase, s []byte) []byte {
  	return Map(func(r rune) rune { return c.ToTitle(r) }, s)
  }
  
  // isSeparator reports whether the rune could mark a word boundary.
  // TODO: update when package unicode captures more of the properties.
  func isSeparator(r rune) bool {
  	// ASCII alphanumerics and underscore are not separators
  	if r <= 0x7F {
  		switch {
  		case '0' <= r && r <= '9':
  			return false
  		case 'a' <= r && r <= 'z':
  			return false
  		case 'A' <= r && r <= 'Z':
  			return false
  		case r == '_':
  			return false
  		}
  		return true
  	}
  	// Letters and digits are not separators
  	if unicode.IsLetter(r) || unicode.IsDigit(r) {
  		return false
  	}
  	// Otherwise, all we can do for now is treat spaces as separators.
  	return unicode.IsSpace(r)
  }
  
  // Title treats s as UTF-8-encoded bytes and returns a copy with all Unicode letters that begin
  // words mapped to their title case.
  //
  // BUG(rsc): The rule Title uses for word boundaries does not handle Unicode punctuation properly.
  func Title(s []byte) []byte {
  	// Use a closure here to remember state.
  	// Hackish but effective. Depends on Map scanning in order and calling
  	// the closure once per rune.
  	prev := ' '
  	return Map(
  		func(r rune) rune {
  			if isSeparator(prev) {
  				prev = r
  				return unicode.ToTitle(r)
  			}
  			prev = r
  			return r
  		},
  		s)
  }
  
  // TrimLeftFunc treats s as UTF-8-encoded bytes and returns a subslice of s by slicing off
  // all leading UTF-8-encoded code points c that satisfy f(c).
  func TrimLeftFunc(s []byte, f func(r rune) bool) []byte {
  	i := indexFunc(s, f, false)
  	if i == -1 {
  		return nil
  	}
  	return s[i:]
  }
  
  // TrimRightFunc returns a subslice of s by slicing off all trailing
  // UTF-8-encoded code points c that satisfy f(c).
  func TrimRightFunc(s []byte, f func(r rune) bool) []byte {
  	i := lastIndexFunc(s, f, false)
  	if i >= 0 && s[i] >= utf8.RuneSelf {
  		_, wid := utf8.DecodeRune(s[i:])
  		i += wid
  	} else {
  		i++
  	}
  	return s[0:i]
  }
  
  // TrimFunc returns a subslice of s by slicing off all leading and trailing
  // UTF-8-encoded code points c that satisfy f(c).
  func TrimFunc(s []byte, f func(r rune) bool) []byte {
  	return TrimRightFunc(TrimLeftFunc(s, f), f)
  }
  
  // TrimPrefix returns s without the provided leading prefix string.
  // If s doesn't start with prefix, s is returned unchanged.
  func TrimPrefix(s, prefix []byte) []byte {
  	if HasPrefix(s, prefix) {
  		return s[len(prefix):]
  	}
  	return s
  }
  
  // TrimSuffix returns s without the provided trailing suffix string.
  // If s doesn't end with suffix, s is returned unchanged.
  func TrimSuffix(s, suffix []byte) []byte {
  	if HasSuffix(s, suffix) {
  		return s[:len(s)-len(suffix)]
  	}
  	return s
  }
  
  // IndexFunc interprets s as a sequence of UTF-8-encoded code points.
  // It returns the byte index in s of the first Unicode
  // code point satisfying f(c), or -1 if none do.
  func IndexFunc(s []byte, f func(r rune) bool) int {
  	return indexFunc(s, f, true)
  }
  
  // LastIndexFunc interprets s as a sequence of UTF-8-encoded code points.
  // It returns the byte index in s of the last Unicode
  // code point satisfying f(c), or -1 if none do.
  func LastIndexFunc(s []byte, f func(r rune) bool) int {
  	return lastIndexFunc(s, f, true)
  }
  
  // indexFunc is the same as IndexFunc except that if
  // truth==false, the sense of the predicate function is
  // inverted.
  func indexFunc(s []byte, f func(r rune) bool, truth bool) int {
  	start := 0
  	for start < len(s) {
  		wid := 1
  		r := rune(s[start])
  		if r >= utf8.RuneSelf {
  			r, wid = utf8.DecodeRune(s[start:])
  		}
  		if f(r) == truth {
  			return start
  		}
  		start += wid
  	}
  	return -1
  }
  
  // lastIndexFunc is the same as LastIndexFunc except that if
  // truth==false, the sense of the predicate function is
  // inverted.
  func lastIndexFunc(s []byte, f func(r rune) bool, truth bool) int {
  	for i := len(s); i > 0; {
  		r, size := rune(s[i-1]), 1
  		if r >= utf8.RuneSelf {
  			r, size = utf8.DecodeLastRune(s[0:i])
  		}
  		i -= size
  		if f(r) == truth {
  			return i
  		}
  	}
  	return -1
  }
  
  // asciiSet is a 32-byte value, where each bit represents the presence of a
  // given ASCII character in the set. The 128-bits of the lower 16 bytes,
  // starting with the least-significant bit of the lowest word to the
  // most-significant bit of the highest word, map to the full range of all
  // 128 ASCII characters. The 128-bits of the upper 16 bytes will be zeroed,
  // ensuring that any non-ASCII character will be reported as not in the set.
  type asciiSet [8]uint32
  
  // makeASCIISet creates a set of ASCII characters and reports whether all
  // characters in chars are ASCII.
  func makeASCIISet(chars string) (as asciiSet, ok bool) {
  	for i := 0; i < len(chars); i++ {
  		c := chars[i]
  		if c >= utf8.RuneSelf {
  			return as, false
  		}
  		as[c>>5] |= 1 << uint(c&31)
  	}
  	return as, true
  }
  
  // contains reports whether c is inside the set.
  func (as *asciiSet) contains(c byte) bool {
  	return (as[c>>5] & (1 << uint(c&31))) != 0
  }
  
  func makeCutsetFunc(cutset string) func(r rune) bool {
  	if len(cutset) == 1 && cutset[0] < utf8.RuneSelf {
  		return func(r rune) bool {
  			return r == rune(cutset[0])
  		}
  	}
  	if as, isASCII := makeASCIISet(cutset); isASCII {
  		return func(r rune) bool {
  			return r < utf8.RuneSelf && as.contains(byte(r))
  		}
  	}
  	return func(r rune) bool {
  		for _, c := range cutset {
  			if c == r {
  				return true
  			}
  		}
  		return false
  	}
  }
  
  // Trim returns a subslice of s by slicing off all leading and
  // trailing UTF-8-encoded code points contained in cutset.
  func Trim(s []byte, cutset string) []byte {
  	return TrimFunc(s, makeCutsetFunc(cutset))
  }
  
  // TrimLeft returns a subslice of s by slicing off all leading
  // UTF-8-encoded code points contained in cutset.
  func TrimLeft(s []byte, cutset string) []byte {
  	return TrimLeftFunc(s, makeCutsetFunc(cutset))
  }
  
  // TrimRight returns a subslice of s by slicing off all trailing
  // UTF-8-encoded code points that are contained in cutset.
  func TrimRight(s []byte, cutset string) []byte {
  	return TrimRightFunc(s, makeCutsetFunc(cutset))
  }
  
  // TrimSpace returns a subslice of s by slicing off all leading and
  // trailing white space, as defined by Unicode.
  func TrimSpace(s []byte) []byte {
  	return TrimFunc(s, unicode.IsSpace)
  }
  
  // Runes interprets s as a sequence of UTF-8-encoded code points.
  // It returns a slice of runes (Unicode code points) equivalent to s.
  func Runes(s []byte) []rune {
  	t := make([]rune, utf8.RuneCount(s))
  	i := 0
  	for len(s) > 0 {
  		r, l := utf8.DecodeRune(s)
  		t[i] = r
  		i++
  		s = s[l:]
  	}
  	return t
  }
  
  // Replace returns a copy of the slice s with the first n
  // non-overlapping instances of old replaced by new.
  // If old is empty, it matches at the beginning of the slice
  // and after each UTF-8 sequence, yielding up to k+1 replacements
  // for a k-rune slice.
  // If n < 0, there is no limit on the number of replacements.
  func Replace(s, old, new []byte, n int) []byte {
  	m := 0
  	if n != 0 {
  		// Compute number of replacements.
  		m = Count(s, old)
  	}
  	if m == 0 {
  		// Just return a copy.
  		return append([]byte(nil), s...)
  	}
  	if n < 0 || m < n {
  		n = m
  	}
  
  	// Apply replacements to buffer.
  	t := make([]byte, len(s)+n*(len(new)-len(old)))
  	w := 0
  	start := 0
  	for i := 0; i < n; i++ {
  		j := start
  		if len(old) == 0 {
  			if i > 0 {
  				_, wid := utf8.DecodeRune(s[start:])
  				j += wid
  			}
  		} else {
  			j += Index(s[start:], old)
  		}
  		w += copy(t[w:], s[start:j])
  		w += copy(t[w:], new)
  		start = j + len(old)
  	}
  	w += copy(t[w:], s[start:])
  	return t[0:w]
  }
  
  // EqualFold reports whether s and t, interpreted as UTF-8 strings,
  // are equal under Unicode case-folding.
  func EqualFold(s, t []byte) bool {
  	for len(s) != 0 && len(t) != 0 {
  		// Extract first rune from each.
  		var sr, tr rune
  		if s[0] < utf8.RuneSelf {
  			sr, s = rune(s[0]), s[1:]
  		} else {
  			r, size := utf8.DecodeRune(s)
  			sr, s = r, s[size:]
  		}
  		if t[0] < utf8.RuneSelf {
  			tr, t = rune(t[0]), t[1:]
  		} else {
  			r, size := utf8.DecodeRune(t)
  			tr, t = r, t[size:]
  		}
  
  		// If they match, keep going; if not, return false.
  
  		// Easy case.
  		if tr == sr {
  			continue
  		}
  
  		// Make sr < tr to simplify what follows.
  		if tr < sr {
  			tr, sr = sr, tr
  		}
  		// Fast check for ASCII.
  		if tr < utf8.RuneSelf && 'A' <= sr && sr <= 'Z' {
  			// ASCII, and sr is upper case.  tr must be lower case.
  			if tr == sr+'a'-'A' {
  				continue
  			}
  			return false
  		}
  
  		// General case. SimpleFold(x) returns the next equivalent rune > x
  		// or wraps around to smaller values.
  		r := unicode.SimpleFold(sr)
  		for r != sr && r < tr {
  			r = unicode.SimpleFold(r)
  		}
  		if r == tr {
  			continue
  		}
  		return false
  	}
  
  	// One string is empty. Are both?
  	return len(s) == len(t)
  }
  
  func indexRabinKarp(s, sep []byte) int {
  	// Rabin-Karp search
  	hashsep, pow := hashStr(sep)
  	n := len(sep)
  	var h uint32
  	for i := 0; i < n; i++ {
  		h = h*primeRK + uint32(s[i])
  	}
  	if h == hashsep && Equal(s[:n], sep) {
  		return 0
  	}
  	for i := n; i < len(s); {
  		h *= primeRK
  		h += uint32(s[i])
  		h -= pow * uint32(s[i-n])
  		i++
  		if h == hashsep && Equal(s[i-n:i], sep) {
  			return i - n
  		}
  	}
  	return -1
  }
  
  // primeRK is the prime base used in Rabin-Karp algorithm.
  const primeRK = 16777619
  
  // hashStr returns the hash and the appropriate multiplicative
  // factor for use in Rabin-Karp algorithm.
  func hashStr(sep []byte) (uint32, uint32) {
  	hash := uint32(0)
  	for i := 0; i < len(sep); i++ {
  		hash = hash*primeRK + uint32(sep[i])
  	}
  	var pow, sq uint32 = 1, primeRK
  	for i := len(sep); i > 0; i >>= 1 {
  		if i&1 != 0 {
  			pow *= sq
  		}
  		sq *= sq
  	}
  	return hash, pow
  }
  

View as plain text