Source file src/regexp/regexp.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 regexp implements regular expression search.
     6  //
     7  // The syntax of the regular expressions accepted is the same
     8  // general syntax used by Perl, Python, and other languages.
     9  // More precisely, it is the syntax accepted by RE2 and described at
    10  // https://golang.org/s/re2syntax, except for \C.
    11  // For an overview of the syntax, see the [regexp/syntax] package.
    12  //
    13  // The regexp implementation provided by this package is
    14  // guaranteed to run in time linear in the size of the input.
    15  // (This is a property not guaranteed by most open source
    16  // implementations of regular expressions.) For more information
    17  // about this property, see https://swtch.com/~rsc/regexp/regexp1.html
    18  // or any book about automata theory.
    19  //
    20  // All characters are UTF-8-encoded code points.
    21  // Following [utf8.DecodeRune], each byte of an invalid UTF-8 sequence
    22  // is treated as if it encoded utf8.RuneError (U+FFFD).
    23  //
    24  // There are 16 methods of [Regexp] that match a regular expression and identify
    25  // the matched text. Their names are matched by this regular expression:
    26  //
    27  //	Find(All)?(String)?(Submatch)?(Index)?
    28  //
    29  // If 'All' is present, the routine matches successive non-overlapping
    30  // matches of the entire expression. Empty matches abutting a preceding
    31  // match are ignored. The return value is a slice containing the successive
    32  // return values of the corresponding non-'All' routine. These routines take
    33  // an extra integer argument, n. If n >= 0, the function returns at most n
    34  // matches/submatches; otherwise, it returns all of them.
    35  //
    36  // If 'String' is present, the argument is a string; otherwise it is a slice
    37  // of bytes; return values are adjusted as appropriate.
    38  //
    39  // If 'Submatch' is present, the return value is a slice identifying the
    40  // successive submatches of the expression. Submatches are matches of
    41  // parenthesized subexpressions (also known as capturing groups) within the
    42  // regular expression, numbered from left to right in order of opening
    43  // parenthesis. Submatch 0 is the match of the entire expression, submatch 1 is
    44  // the match of the first parenthesized subexpression, and so on.
    45  //
    46  // If 'Index' is present, matches and submatches are identified by byte index
    47  // pairs within the input string: result[2*n:2*n+2] identifies the indexes of
    48  // the nth submatch. The pair for n==0 identifies the match of the entire
    49  // expression. If 'Index' is not present, the match is identified by the text
    50  // of the match/submatch. If an index is negative or text is nil, it means that
    51  // subexpression did not match any string in the input. For 'String' versions
    52  // an empty string means either no match or an empty match.
    53  //
    54  // There is also a subset of the methods that can be applied to text read from
    55  // an [io.RuneReader]: [Regexp.MatchReader], [Regexp.FindReaderIndex],
    56  // [Regexp.FindReaderSubmatchIndex].
    57  //
    58  // This set may grow. Note that regular expression matches may need to
    59  // examine text beyond the text returned by a match, so the methods that
    60  // match text from an [io.RuneReader] may read arbitrarily far into the input
    61  // before returning.
    62  //
    63  // (There are a few other methods that do not match this pattern.)
    64  package regexp
    65  
    66  import (
    67  	"bytes"
    68  	"io"
    69  	"regexp/syntax"
    70  	"strconv"
    71  	"strings"
    72  	"sync"
    73  	"unicode"
    74  	"unicode/utf8"
    75  )
    76  
    77  // Regexp is the representation of a compiled regular expression.
    78  // A Regexp is safe for concurrent use by multiple goroutines,
    79  // except for configuration methods, such as [Regexp.Longest].
    80  type Regexp struct {
    81  	expr           string       // as passed to Compile
    82  	prog           *syntax.Prog // compiled program
    83  	onepass        *onePassProg // onepass program or nil
    84  	numSubexp      int
    85  	maxBitStateLen int
    86  	subexpNames    []string
    87  	prefix         string         // required prefix in unanchored matches
    88  	prefixBytes    []byte         // prefix, as a []byte
    89  	prefixRune     rune           // first rune in prefix
    90  	prefixEnd      uint32         // pc for last rune in prefix
    91  	mpool          int            // pool for machines
    92  	matchcap       int            // size of recorded match lengths
    93  	prefixComplete bool           // prefix is the entire regexp
    94  	cond           syntax.EmptyOp // empty-width conditions required at start of match
    95  	minInputLen    int            // minimum length of the input in bytes
    96  
    97  	// This field can be modified by the Longest method,
    98  	// but it is otherwise read-only.
    99  	longest bool // whether regexp prefers leftmost-longest match
   100  }
   101  
   102  // String returns the source text used to compile the regular expression.
   103  func (re *Regexp) String() string {
   104  	return re.expr
   105  }
   106  
   107  // Copy returns a new [Regexp] object copied from re.
   108  // Calling [Regexp.Longest] on one copy does not affect another.
   109  //
   110  // Deprecated: In earlier releases, when using a [Regexp] in multiple goroutines,
   111  // giving each goroutine its own copy helped to avoid lock contention.
   112  // As of Go 1.12, using Copy is no longer necessary to avoid lock contention.
   113  // Copy may still be appropriate if the reason for its use is to make
   114  // two copies with different [Regexp.Longest] settings.
   115  func (re *Regexp) Copy() *Regexp {
   116  	re2 := *re
   117  	return &re2
   118  }
   119  
   120  // Compile parses a regular expression and returns, if successful,
   121  // a [Regexp] object that can be used to match against text.
   122  //
   123  // When matching against text, the regexp returns a match that
   124  // begins as early as possible in the input (leftmost), and among those
   125  // it chooses the one that a backtracking search would have found first.
   126  // This so-called leftmost-first matching is the same semantics
   127  // that Perl, Python, and other implementations use, although this
   128  // package implements it without the expense of backtracking.
   129  // For POSIX leftmost-longest matching, see [CompilePOSIX].
   130  func Compile(expr string) (*Regexp, error) {
   131  	return compile(expr, syntax.Perl, false)
   132  }
   133  
   134  // CompilePOSIX is like [Compile] but restricts the regular expression
   135  // to POSIX ERE (egrep) syntax and changes the match semantics to
   136  // leftmost-longest.
   137  //
   138  // That is, when matching against text, the regexp returns a match that
   139  // begins as early as possible in the input (leftmost), and among those
   140  // it chooses a match that is as long as possible.
   141  // This so-called leftmost-longest matching is the same semantics
   142  // that early regular expression implementations used and that POSIX
   143  // specifies.
   144  //
   145  // However, there can be multiple leftmost-longest matches, with different
   146  // submatch choices, and here this package diverges from POSIX.
   147  // Among the possible leftmost-longest matches, this package chooses
   148  // the one that a backtracking search would have found first, while POSIX
   149  // specifies that the match be chosen to maximize the length of the first
   150  // subexpression, then the second, and so on from left to right.
   151  // The POSIX rule is computationally prohibitive and not even well-defined.
   152  // See https://swtch.com/~rsc/regexp/regexp2.html#posix for details.
   153  func CompilePOSIX(expr string) (*Regexp, error) {
   154  	return compile(expr, syntax.POSIX, true)
   155  }
   156  
   157  // Longest makes future searches prefer the leftmost-longest match.
   158  // That is, when matching against text, the regexp returns a match that
   159  // begins as early as possible in the input (leftmost), and among those
   160  // it chooses a match that is as long as possible.
   161  // This method modifies the [Regexp] and may not be called concurrently
   162  // with any other methods.
   163  func (re *Regexp) Longest() {
   164  	re.longest = true
   165  }
   166  
   167  func compile(expr string, mode syntax.Flags, longest bool) (*Regexp, error) {
   168  	re, err := syntax.Parse(expr, mode)
   169  	if err != nil {
   170  		return nil, err
   171  	}
   172  	maxCap := re.MaxCap()
   173  	capNames := re.CapNames()
   174  
   175  	re = re.Simplify()
   176  	prog, err := syntax.Compile(re)
   177  	if err != nil {
   178  		return nil, err
   179  	}
   180  	matchcap := prog.NumCap
   181  	if matchcap < 2 {
   182  		matchcap = 2
   183  	}
   184  	regexp := &Regexp{
   185  		expr:        expr,
   186  		prog:        prog,
   187  		onepass:     compileOnePass(prog),
   188  		numSubexp:   maxCap,
   189  		subexpNames: capNames,
   190  		cond:        prog.StartCond(),
   191  		longest:     longest,
   192  		matchcap:    matchcap,
   193  		minInputLen: minInputLen(re),
   194  	}
   195  	if regexp.onepass == nil {
   196  		regexp.prefix, regexp.prefixComplete = prog.Prefix()
   197  		regexp.maxBitStateLen = maxBitStateLen(prog)
   198  	} else {
   199  		regexp.prefix, regexp.prefixComplete, regexp.prefixEnd = onePassPrefix(prog)
   200  	}
   201  	if regexp.prefix != "" {
   202  		// TODO(rsc): Remove this allocation by adding
   203  		// IndexString to package bytes.
   204  		regexp.prefixBytes = []byte(regexp.prefix)
   205  		regexp.prefixRune, _ = utf8.DecodeRuneInString(regexp.prefix)
   206  	}
   207  
   208  	n := len(prog.Inst)
   209  	i := 0
   210  	for matchSize[i] != 0 && matchSize[i] < n {
   211  		i++
   212  	}
   213  	regexp.mpool = i
   214  
   215  	return regexp, nil
   216  }
   217  
   218  // Pools of *machine for use during (*Regexp).doExecute,
   219  // split up by the size of the execution queues.
   220  // matchPool[i] machines have queue size matchSize[i].
   221  // On a 64-bit system each queue entry is 16 bytes,
   222  // so matchPool[0] has 16*2*128 = 4kB queues, etc.
   223  // The final matchPool is a catch-all for very large queues.
   224  var (
   225  	matchSize = [...]int{128, 512, 2048, 16384, 0}
   226  	matchPool [len(matchSize)]sync.Pool
   227  )
   228  
   229  // get returns a machine to use for matching re.
   230  // It uses the re's machine cache if possible, to avoid
   231  // unnecessary allocation.
   232  func (re *Regexp) get() *machine {
   233  	m, ok := matchPool[re.mpool].Get().(*machine)
   234  	if !ok {
   235  		m = new(machine)
   236  	}
   237  	m.re = re
   238  	m.p = re.prog
   239  	if cap(m.matchcap) < re.matchcap {
   240  		m.matchcap = make([]int, re.matchcap)
   241  		for _, t := range m.pool {
   242  			t.cap = make([]int, re.matchcap)
   243  		}
   244  	}
   245  
   246  	// Allocate queues if needed.
   247  	// Or reallocate, for "large" match pool.
   248  	n := matchSize[re.mpool]
   249  	if n == 0 { // large pool
   250  		n = len(re.prog.Inst)
   251  	}
   252  	if len(m.q0.sparse) < n {
   253  		m.q0 = queue{make([]uint32, n), make([]entry, 0, n)}
   254  		m.q1 = queue{make([]uint32, n), make([]entry, 0, n)}
   255  	}
   256  	return m
   257  }
   258  
   259  // put returns a machine to the correct machine pool.
   260  func (re *Regexp) put(m *machine) {
   261  	m.re = nil
   262  	m.p = nil
   263  	m.inputs.clear()
   264  	matchPool[re.mpool].Put(m)
   265  }
   266  
   267  // minInputLen walks the regexp to find the minimum length of any matchable input.
   268  func minInputLen(re *syntax.Regexp) int {
   269  	switch re.Op {
   270  	default:
   271  		return 0
   272  	case syntax.OpAnyChar, syntax.OpAnyCharNotNL, syntax.OpCharClass:
   273  		return 1
   274  	case syntax.OpLiteral:
   275  		l := 0
   276  		for _, r := range re.Rune {
   277  			if r == utf8.RuneError {
   278  				l++
   279  			} else {
   280  				l += utf8.RuneLen(r)
   281  			}
   282  		}
   283  		return l
   284  	case syntax.OpCapture, syntax.OpPlus:
   285  		return minInputLen(re.Sub[0])
   286  	case syntax.OpRepeat:
   287  		return re.Min * minInputLen(re.Sub[0])
   288  	case syntax.OpConcat:
   289  		l := 0
   290  		for _, sub := range re.Sub {
   291  			l += minInputLen(sub)
   292  		}
   293  		return l
   294  	case syntax.OpAlternate:
   295  		l := minInputLen(re.Sub[0])
   296  		var lnext int
   297  		for _, sub := range re.Sub[1:] {
   298  			lnext = minInputLen(sub)
   299  			if lnext < l {
   300  				l = lnext
   301  			}
   302  		}
   303  		return l
   304  	}
   305  }
   306  
   307  // MustCompile is like [Compile] but panics if the expression cannot be parsed.
   308  // It simplifies safe initialization of global variables holding compiled regular
   309  // expressions.
   310  func MustCompile(str string) *Regexp {
   311  	regexp, err := Compile(str)
   312  	if err != nil {
   313  		panic(`regexp: Compile(` + quote(str) + `): ` + err.Error())
   314  	}
   315  	return regexp
   316  }
   317  
   318  // MustCompilePOSIX is like [CompilePOSIX] but panics if the expression cannot be parsed.
   319  // It simplifies safe initialization of global variables holding compiled regular
   320  // expressions.
   321  func MustCompilePOSIX(str string) *Regexp {
   322  	regexp, err := CompilePOSIX(str)
   323  	if err != nil {
   324  		panic(`regexp: CompilePOSIX(` + quote(str) + `): ` + err.Error())
   325  	}
   326  	return regexp
   327  }
   328  
   329  func quote(s string) string {
   330  	if strconv.CanBackquote(s) {
   331  		return "`" + s + "`"
   332  	}
   333  	return strconv.Quote(s)
   334  }
   335  
   336  // NumSubexp returns the number of parenthesized subexpressions in this [Regexp].
   337  func (re *Regexp) NumSubexp() int {
   338  	return re.numSubexp
   339  }
   340  
   341  // SubexpNames returns the names of the parenthesized subexpressions
   342  // in this [Regexp]. The name for the first sub-expression is names[1],
   343  // so that if m is a match slice, the name for m[i] is SubexpNames()[i].
   344  // Since the Regexp as a whole cannot be named, names[0] is always
   345  // the empty string. The slice should not be modified.
   346  func (re *Regexp) SubexpNames() []string {
   347  	return re.subexpNames
   348  }
   349  
   350  // SubexpIndex returns the index of the first subexpression with the given name,
   351  // or -1 if there is no subexpression with that name.
   352  //
   353  // Note that multiple subexpressions can be written using the same name, as in
   354  // (?P<bob>a+)(?P<bob>b+), which declares two subexpressions named "bob".
   355  // In this case, SubexpIndex returns the index of the leftmost such subexpression
   356  // in the regular expression.
   357  func (re *Regexp) SubexpIndex(name string) int {
   358  	if name != "" {
   359  		for i, s := range re.subexpNames {
   360  			if name == s {
   361  				return i
   362  			}
   363  		}
   364  	}
   365  	return -1
   366  }
   367  
   368  const endOfText rune = -1
   369  
   370  // input abstracts different representations of the input text. It provides
   371  // one-character lookahead.
   372  type input interface {
   373  	step(pos int) (r rune, width int) // advance one rune
   374  	canCheckPrefix() bool             // can we look ahead without losing info?
   375  	hasPrefix(re *Regexp) bool
   376  	index(re *Regexp, pos int) int
   377  	context(pos int) lazyFlag
   378  }
   379  
   380  // inputString scans a string.
   381  type inputString struct {
   382  	str string
   383  }
   384  
   385  func (i *inputString) step(pos int) (rune, int) {
   386  	if pos < len(i.str) {
   387  		return utf8.DecodeRuneInString(i.str[pos:])
   388  	}
   389  	return endOfText, 0
   390  }
   391  
   392  func (i *inputString) canCheckPrefix() bool {
   393  	return true
   394  }
   395  
   396  func (i *inputString) hasPrefix(re *Regexp) bool {
   397  	return strings.HasPrefix(i.str, re.prefix)
   398  }
   399  
   400  func (i *inputString) index(re *Regexp, pos int) int {
   401  	return strings.Index(i.str[pos:], re.prefix)
   402  }
   403  
   404  func (i *inputString) context(pos int) lazyFlag {
   405  	r1, r2 := endOfText, endOfText
   406  	// 0 < pos && pos <= len(i.str)
   407  	if uint(pos-1) < uint(len(i.str)) {
   408  		r1, _ = utf8.DecodeLastRuneInString(i.str[:pos])
   409  	}
   410  	// 0 <= pos && pos < len(i.str)
   411  	if uint(pos) < uint(len(i.str)) {
   412  		r2, _ = utf8.DecodeRuneInString(i.str[pos:])
   413  	}
   414  	return newLazyFlag(r1, r2)
   415  }
   416  
   417  // inputBytes scans a byte slice.
   418  type inputBytes struct {
   419  	str []byte
   420  }
   421  
   422  func (i *inputBytes) step(pos int) (rune, int) {
   423  	if pos < len(i.str) {
   424  		return utf8.DecodeRune(i.str[pos:])
   425  	}
   426  	return endOfText, 0
   427  }
   428  
   429  func (i *inputBytes) canCheckPrefix() bool {
   430  	return true
   431  }
   432  
   433  func (i *inputBytes) hasPrefix(re *Regexp) bool {
   434  	return bytes.HasPrefix(i.str, re.prefixBytes)
   435  }
   436  
   437  func (i *inputBytes) index(re *Regexp, pos int) int {
   438  	return bytes.Index(i.str[pos:], re.prefixBytes)
   439  }
   440  
   441  func (i *inputBytes) context(pos int) lazyFlag {
   442  	r1, r2 := endOfText, endOfText
   443  	// 0 < pos && pos <= len(i.str)
   444  	if uint(pos-1) < uint(len(i.str)) {
   445  		r1, _ = utf8.DecodeLastRune(i.str[:pos])
   446  	}
   447  	// 0 <= pos && pos < len(i.str)
   448  	if uint(pos) < uint(len(i.str)) {
   449  		r2, _ = utf8.DecodeRune(i.str[pos:])
   450  	}
   451  	return newLazyFlag(r1, r2)
   452  }
   453  
   454  // inputReader scans a RuneReader.
   455  type inputReader struct {
   456  	r     io.RuneReader
   457  	atEOT bool
   458  	pos   int
   459  }
   460  
   461  func (i *inputReader) step(pos int) (rune, int) {
   462  	if !i.atEOT && pos != i.pos {
   463  		return endOfText, 0
   464  
   465  	}
   466  	r, w, err := i.r.ReadRune()
   467  	if err != nil {
   468  		i.atEOT = true
   469  		return endOfText, 0
   470  	}
   471  	i.pos += w
   472  	return r, w
   473  }
   474  
   475  func (i *inputReader) canCheckPrefix() bool {
   476  	return false
   477  }
   478  
   479  func (i *inputReader) hasPrefix(re *Regexp) bool {
   480  	return false
   481  }
   482  
   483  func (i *inputReader) index(re *Regexp, pos int) int {
   484  	return -1
   485  }
   486  
   487  func (i *inputReader) context(pos int) lazyFlag {
   488  	return 0 // not used
   489  }
   490  
   491  // LiteralPrefix returns a literal string that must begin any match
   492  // of the regular expression re. It returns the boolean true if the
   493  // literal string comprises the entire regular expression.
   494  func (re *Regexp) LiteralPrefix() (prefix string, complete bool) {
   495  	return re.prefix, re.prefixComplete
   496  }
   497  
   498  // MatchReader reports whether the text returned by the [io.RuneReader]
   499  // contains any match of the regular expression re.
   500  func (re *Regexp) MatchReader(r io.RuneReader) bool {
   501  	return re.doMatch(r, nil, "")
   502  }
   503  
   504  // MatchString reports whether the string s
   505  // contains any match of the regular expression re.
   506  func (re *Regexp) MatchString(s string) bool {
   507  	return re.doMatch(nil, nil, s)
   508  }
   509  
   510  // Match reports whether the byte slice b
   511  // contains any match of the regular expression re.
   512  func (re *Regexp) Match(b []byte) bool {
   513  	return re.doMatch(nil, b, "")
   514  }
   515  
   516  // MatchReader reports whether the text returned by the [io.RuneReader]
   517  // contains any match of the regular expression pattern.
   518  // More complicated queries need to use [Compile] and the full [Regexp] interface.
   519  func MatchReader(pattern string, r io.RuneReader) (matched bool, err error) {
   520  	re, err := Compile(pattern)
   521  	if err != nil {
   522  		return false, err
   523  	}
   524  	return re.MatchReader(r), nil
   525  }
   526  
   527  // MatchString reports whether the string s
   528  // contains any match of the regular expression pattern.
   529  // More complicated queries need to use [Compile] and the full [Regexp] interface.
   530  func MatchString(pattern string, s string) (matched bool, err error) {
   531  	re, err := Compile(pattern)
   532  	if err != nil {
   533  		return false, err
   534  	}
   535  	return re.MatchString(s), nil
   536  }
   537  
   538  // Match reports whether the byte slice b
   539  // contains any match of the regular expression pattern.
   540  // More complicated queries need to use [Compile] and the full [Regexp] interface.
   541  func Match(pattern string, b []byte) (matched bool, err error) {
   542  	re, err := Compile(pattern)
   543  	if err != nil {
   544  		return false, err
   545  	}
   546  	return re.Match(b), nil
   547  }
   548  
   549  // ReplaceAllString returns a copy of src, replacing matches of the [Regexp]
   550  // with the replacement string repl.
   551  // Inside repl, $ signs are interpreted as in [Regexp.Expand].
   552  func (re *Regexp) ReplaceAllString(src, repl string) string {
   553  	n := 2
   554  	if strings.Contains(repl, "$") {
   555  		n = 2 * (re.numSubexp + 1)
   556  	}
   557  	b := re.replaceAll(nil, src, n, func(dst []byte, match []int) []byte {
   558  		return re.expand(dst, repl, nil, src, match)
   559  	})
   560  	return string(b)
   561  }
   562  
   563  // ReplaceAllLiteralString returns a copy of src, replacing matches of the [Regexp]
   564  // with the replacement string repl. The replacement repl is substituted directly,
   565  // without using [Regexp.Expand].
   566  func (re *Regexp) ReplaceAllLiteralString(src, repl string) string {
   567  	return string(re.replaceAll(nil, src, 2, func(dst []byte, match []int) []byte {
   568  		return append(dst, repl...)
   569  	}))
   570  }
   571  
   572  // ReplaceAllStringFunc returns a copy of src in which all matches of the
   573  // [Regexp] have been replaced by the return value of function repl applied
   574  // to the matched substring. The replacement returned by repl is substituted
   575  // directly, without using [Regexp.Expand].
   576  func (re *Regexp) ReplaceAllStringFunc(src string, repl func(string) string) string {
   577  	b := re.replaceAll(nil, src, 2, func(dst []byte, match []int) []byte {
   578  		return append(dst, repl(src[match[0]:match[1]])...)
   579  	})
   580  	return string(b)
   581  }
   582  
   583  func (re *Regexp) replaceAll(bsrc []byte, src string, nmatch int, repl func(dst []byte, m []int) []byte) []byte {
   584  	lastMatchEnd := 0 // end position of the most recent match
   585  	searchPos := 0    // position where we next look for a match
   586  	var buf []byte
   587  	var endPos int
   588  	if bsrc != nil {
   589  		endPos = len(bsrc)
   590  	} else {
   591  		endPos = len(src)
   592  	}
   593  	if nmatch > re.prog.NumCap {
   594  		nmatch = re.prog.NumCap
   595  	}
   596  
   597  	var dstCap [2]int
   598  	for searchPos <= endPos {
   599  		a := re.doExecute(nil, bsrc, src, searchPos, nmatch, dstCap[:0])
   600  		if len(a) == 0 {
   601  			break // no more matches
   602  		}
   603  
   604  		// Copy the unmatched characters before this match.
   605  		if bsrc != nil {
   606  			buf = append(buf, bsrc[lastMatchEnd:a[0]]...)
   607  		} else {
   608  			buf = append(buf, src[lastMatchEnd:a[0]]...)
   609  		}
   610  
   611  		// Now insert a copy of the replacement string, but not for a
   612  		// match of the empty string immediately after another match.
   613  		// (Otherwise, we get double replacement for patterns that
   614  		// match both empty and nonempty strings.)
   615  		if a[1] > lastMatchEnd || a[0] == 0 {
   616  			buf = repl(buf, a)
   617  		}
   618  		lastMatchEnd = a[1]
   619  
   620  		// Advance past this match; always advance at least one character.
   621  		var width int
   622  		if bsrc != nil {
   623  			_, width = utf8.DecodeRune(bsrc[searchPos:])
   624  		} else {
   625  			_, width = utf8.DecodeRuneInString(src[searchPos:])
   626  		}
   627  		if searchPos+width > a[1] {
   628  			searchPos += width
   629  		} else if searchPos+1 > a[1] {
   630  			// This clause is only needed at the end of the input
   631  			// string. In that case, DecodeRuneInString returns width=0.
   632  			searchPos++
   633  		} else {
   634  			searchPos = a[1]
   635  		}
   636  	}
   637  
   638  	// Copy the unmatched characters after the last match.
   639  	if bsrc != nil {
   640  		buf = append(buf, bsrc[lastMatchEnd:]...)
   641  	} else {
   642  		buf = append(buf, src[lastMatchEnd:]...)
   643  	}
   644  
   645  	return buf
   646  }
   647  
   648  // ReplaceAll returns a copy of src, replacing matches of the [Regexp]
   649  // with the replacement text repl.
   650  // Inside repl, $ signs are interpreted as in [Regexp.Expand].
   651  func (re *Regexp) ReplaceAll(src, repl []byte) []byte {
   652  	n := 2
   653  	if bytes.IndexByte(repl, '$') >= 0 {
   654  		n = 2 * (re.numSubexp + 1)
   655  	}
   656  	srepl := ""
   657  	b := re.replaceAll(src, "", n, func(dst []byte, match []int) []byte {
   658  		if len(srepl) != len(repl) {
   659  			srepl = string(repl)
   660  		}
   661  		return re.expand(dst, srepl, src, "", match)
   662  	})
   663  	return b
   664  }
   665  
   666  // ReplaceAllLiteral returns a copy of src, replacing matches of the [Regexp]
   667  // with the replacement bytes repl. The replacement repl is substituted directly,
   668  // without using [Regexp.Expand].
   669  func (re *Regexp) ReplaceAllLiteral(src, repl []byte) []byte {
   670  	return re.replaceAll(src, "", 2, func(dst []byte, match []int) []byte {
   671  		return append(dst, repl...)
   672  	})
   673  }
   674  
   675  // ReplaceAllFunc returns a copy of src in which all matches of the
   676  // [Regexp] have been replaced by the return value of function repl applied
   677  // to the matched byte slice. The replacement returned by repl is substituted
   678  // directly, without using [Regexp.Expand].
   679  func (re *Regexp) ReplaceAllFunc(src []byte, repl func([]byte) []byte) []byte {
   680  	return re.replaceAll(src, "", 2, func(dst []byte, match []int) []byte {
   681  		return append(dst, repl(src[match[0]:match[1]])...)
   682  	})
   683  }
   684  
   685  // Bitmap used by func special to check whether a character needs to be escaped.
   686  var specialBytes [16]byte
   687  
   688  // special reports whether byte b needs to be escaped by QuoteMeta.
   689  func special(b byte) bool {
   690  	return b < utf8.RuneSelf && specialBytes[b%16]&(1<<(b/16)) != 0
   691  }
   692  
   693  func init() {
   694  	for _, b := range []byte(`\.+*?()|[]{}^$`) {
   695  		specialBytes[b%16] |= 1 << (b / 16)
   696  	}
   697  }
   698  
   699  // QuoteMeta returns a string that escapes all regular expression metacharacters
   700  // inside the argument text; the returned string is a regular expression matching
   701  // the literal text.
   702  func QuoteMeta(s string) string {
   703  	// A byte loop is correct because all metacharacters are ASCII.
   704  	var i int
   705  	for i = 0; i < len(s); i++ {
   706  		if special(s[i]) {
   707  			break
   708  		}
   709  	}
   710  	// No meta characters found, so return original string.
   711  	if i >= len(s) {
   712  		return s
   713  	}
   714  
   715  	b := make([]byte, 2*len(s)-i)
   716  	copy(b, s[:i])
   717  	j := i
   718  	for ; i < len(s); i++ {
   719  		if special(s[i]) {
   720  			b[j] = '\\'
   721  			j++
   722  		}
   723  		b[j] = s[i]
   724  		j++
   725  	}
   726  	return string(b[:j])
   727  }
   728  
   729  // The number of capture values in the program may correspond
   730  // to fewer capturing expressions than are in the regexp.
   731  // For example, "(a){0}" turns into an empty program, so the
   732  // maximum capture in the program is 0 but we need to return
   733  // an expression for \1.  Pad appends -1s to the slice a as needed.
   734  func (re *Regexp) pad(a []int) []int {
   735  	if a == nil {
   736  		// No match.
   737  		return nil
   738  	}
   739  	n := (1 + re.numSubexp) * 2
   740  	for len(a) < n {
   741  		a = append(a, -1)
   742  	}
   743  	return a
   744  }
   745  
   746  // allMatches calls deliver at most n times
   747  // with the location of successive matches in the input text.
   748  // The input text is b if non-nil, otherwise s.
   749  func (re *Regexp) allMatches(s string, b []byte, n int, deliver func([]int)) {
   750  	var end int
   751  	if b == nil {
   752  		end = len(s)
   753  	} else {
   754  		end = len(b)
   755  	}
   756  
   757  	for pos, i, prevMatchEnd := 0, 0, -1; i < n && pos <= end; {
   758  		matches := re.doExecute(nil, b, s, pos, re.prog.NumCap, nil)
   759  		if len(matches) == 0 {
   760  			break
   761  		}
   762  
   763  		accept := true
   764  		if matches[1] == pos {
   765  			// We've found an empty match.
   766  			if matches[0] == prevMatchEnd {
   767  				// We don't allow an empty match right
   768  				// after a previous match, so ignore it.
   769  				accept = false
   770  			}
   771  			var width int
   772  			if b == nil {
   773  				is := inputString{str: s}
   774  				_, width = is.step(pos)
   775  			} else {
   776  				ib := inputBytes{str: b}
   777  				_, width = ib.step(pos)
   778  			}
   779  			if width > 0 {
   780  				pos += width
   781  			} else {
   782  				pos = end + 1
   783  			}
   784  		} else {
   785  			pos = matches[1]
   786  		}
   787  		prevMatchEnd = matches[1]
   788  
   789  		if accept {
   790  			deliver(re.pad(matches))
   791  			i++
   792  		}
   793  	}
   794  }
   795  
   796  // Find returns a slice holding the text of the leftmost match in b of the regular expression.
   797  // A return value of nil indicates no match.
   798  func (re *Regexp) Find(b []byte) []byte {
   799  	var dstCap [2]int
   800  	a := re.doExecute(nil, b, "", 0, 2, dstCap[:0])
   801  	if a == nil {
   802  		return nil
   803  	}
   804  	return b[a[0]:a[1]:a[1]]
   805  }
   806  
   807  // FindIndex returns a two-element slice of integers defining the location of
   808  // the leftmost match in b of the regular expression. The match itself is at
   809  // b[loc[0]:loc[1]].
   810  // A return value of nil indicates no match.
   811  func (re *Regexp) FindIndex(b []byte) (loc []int) {
   812  	a := re.doExecute(nil, b, "", 0, 2, nil)
   813  	if a == nil {
   814  		return nil
   815  	}
   816  	return a[0:2]
   817  }
   818  
   819  // FindString returns a string holding the text of the leftmost match in s of the regular
   820  // expression. If there is no match, the return value is an empty string,
   821  // but it will also be empty if the regular expression successfully matches
   822  // an empty string. Use [Regexp.FindStringIndex] or [Regexp.FindStringSubmatch] if it is
   823  // necessary to distinguish these cases.
   824  func (re *Regexp) FindString(s string) string {
   825  	var dstCap [2]int
   826  	a := re.doExecute(nil, nil, s, 0, 2, dstCap[:0])
   827  	if a == nil {
   828  		return ""
   829  	}
   830  	return s[a[0]:a[1]]
   831  }
   832  
   833  // FindStringIndex returns a two-element slice of integers defining the
   834  // location of the leftmost match in s of the regular expression. The match
   835  // itself is at s[loc[0]:loc[1]].
   836  // A return value of nil indicates no match.
   837  func (re *Regexp) FindStringIndex(s string) (loc []int) {
   838  	a := re.doExecute(nil, nil, s, 0, 2, nil)
   839  	if a == nil {
   840  		return nil
   841  	}
   842  	return a[0:2]
   843  }
   844  
   845  // FindReaderIndex returns a two-element slice of integers defining the
   846  // location of the leftmost match of the regular expression in text read from
   847  // the [io.RuneReader]. The match text was found in the input stream at
   848  // byte offset loc[0] through loc[1]-1.
   849  // A return value of nil indicates no match.
   850  func (re *Regexp) FindReaderIndex(r io.RuneReader) (loc []int) {
   851  	a := re.doExecute(r, nil, "", 0, 2, nil)
   852  	if a == nil {
   853  		return nil
   854  	}
   855  	return a[0:2]
   856  }
   857  
   858  // FindSubmatch returns a slice of slices holding the text of the leftmost
   859  // match of the regular expression in b and the matches, if any, of its
   860  // subexpressions, as defined by the 'Submatch' descriptions in the package
   861  // comment.
   862  // A return value of nil indicates no match.
   863  func (re *Regexp) FindSubmatch(b []byte) [][]byte {
   864  	var dstCap [4]int
   865  	a := re.doExecute(nil, b, "", 0, re.prog.NumCap, dstCap[:0])
   866  	if a == nil {
   867  		return nil
   868  	}
   869  	ret := make([][]byte, 1+re.numSubexp)
   870  	for i := range ret {
   871  		if 2*i < len(a) && a[2*i] >= 0 {
   872  			ret[i] = b[a[2*i]:a[2*i+1]:a[2*i+1]]
   873  		}
   874  	}
   875  	return ret
   876  }
   877  
   878  // Expand appends template to dst and returns the result; during the
   879  // append, Expand replaces variables in the template with corresponding
   880  // matches drawn from src. The match slice should have been returned by
   881  // [Regexp.FindSubmatchIndex].
   882  //
   883  // In the template, a variable is denoted by a substring of the form
   884  // $name or ${name}, where name is a non-empty sequence of letters,
   885  // digits, and underscores. A purely numeric name like $1 refers to
   886  // the submatch with the corresponding index; other names refer to
   887  // capturing parentheses named with the (?P<name>...) syntax. A
   888  // reference to an out of range or unmatched index or a name that is not
   889  // present in the regular expression is replaced with an empty slice.
   890  //
   891  // In the $name form, name is taken to be as long as possible: $1x is
   892  // equivalent to ${1x}, not ${1}x, and, $10 is equivalent to ${10}, not ${1}0.
   893  //
   894  // To insert a literal $ in the output, use $$ in the template.
   895  func (re *Regexp) Expand(dst []byte, template []byte, src []byte, match []int) []byte {
   896  	return re.expand(dst, string(template), src, "", match)
   897  }
   898  
   899  // ExpandString is like [Regexp.Expand] but the template and source are strings.
   900  // It appends to and returns a byte slice in order to give the calling
   901  // code control over allocation.
   902  func (re *Regexp) ExpandString(dst []byte, template string, src string, match []int) []byte {
   903  	return re.expand(dst, template, nil, src, match)
   904  }
   905  
   906  func (re *Regexp) expand(dst []byte, template string, bsrc []byte, src string, match []int) []byte {
   907  	for len(template) > 0 {
   908  		before, after, ok := strings.Cut(template, "$")
   909  		if !ok {
   910  			break
   911  		}
   912  		dst = append(dst, before...)
   913  		template = after
   914  		if template != "" && template[0] == '$' {
   915  			// Treat $$ as $.
   916  			dst = append(dst, '$')
   917  			template = template[1:]
   918  			continue
   919  		}
   920  		name, num, rest, ok := extract(template)
   921  		if !ok {
   922  			// Malformed; treat $ as raw text.
   923  			dst = append(dst, '$')
   924  			continue
   925  		}
   926  		template = rest
   927  		if num >= 0 {
   928  			if 2*num+1 < len(match) && match[2*num] >= 0 {
   929  				if bsrc != nil {
   930  					dst = append(dst, bsrc[match[2*num]:match[2*num+1]]...)
   931  				} else {
   932  					dst = append(dst, src[match[2*num]:match[2*num+1]]...)
   933  				}
   934  			}
   935  		} else {
   936  			for i, namei := range re.subexpNames {
   937  				if name == namei && 2*i+1 < len(match) && match[2*i] >= 0 {
   938  					if bsrc != nil {
   939  						dst = append(dst, bsrc[match[2*i]:match[2*i+1]]...)
   940  					} else {
   941  						dst = append(dst, src[match[2*i]:match[2*i+1]]...)
   942  					}
   943  					break
   944  				}
   945  			}
   946  		}
   947  	}
   948  	dst = append(dst, template...)
   949  	return dst
   950  }
   951  
   952  // extract returns the name from a leading "name" or "{name}" in str.
   953  // (The $ has already been removed by the caller.)
   954  // If it is a number, extract returns num set to that number; otherwise num = -1.
   955  func extract(str string) (name string, num int, rest string, ok bool) {
   956  	if str == "" {
   957  		return
   958  	}
   959  	brace := false
   960  	if str[0] == '{' {
   961  		brace = true
   962  		str = str[1:]
   963  	}
   964  	i := 0
   965  	for i < len(str) {
   966  		rune, size := utf8.DecodeRuneInString(str[i:])
   967  		if !unicode.IsLetter(rune) && !unicode.IsDigit(rune) && rune != '_' {
   968  			break
   969  		}
   970  		i += size
   971  	}
   972  	if i == 0 {
   973  		// empty name is not okay
   974  		return
   975  	}
   976  	name = str[:i]
   977  	if brace {
   978  		if i >= len(str) || str[i] != '}' {
   979  			// missing closing brace
   980  			return
   981  		}
   982  		i++
   983  	}
   984  
   985  	// Parse number.
   986  	num = 0
   987  	for i := 0; i < len(name); i++ {
   988  		if name[i] < '0' || '9' < name[i] || num >= 1e8 {
   989  			num = -1
   990  			break
   991  		}
   992  		num = num*10 + int(name[i]) - '0'
   993  	}
   994  	// Disallow leading zeros.
   995  	if name[0] == '0' && len(name) > 1 {
   996  		num = -1
   997  	}
   998  
   999  	rest = str[i:]
  1000  	ok = true
  1001  	return
  1002  }
  1003  
  1004  // FindSubmatchIndex returns a slice holding the index pairs identifying the
  1005  // leftmost match of the regular expression in b and the matches, if any, of
  1006  // its subexpressions, as defined by the 'Submatch' and 'Index' descriptions
  1007  // in the package comment.
  1008  // A return value of nil indicates no match.
  1009  func (re *Regexp) FindSubmatchIndex(b []byte) []int {
  1010  	return re.pad(re.doExecute(nil, b, "", 0, re.prog.NumCap, nil))
  1011  }
  1012  
  1013  // FindStringSubmatch returns a slice of strings holding the text of the
  1014  // leftmost match of the regular expression in s and the matches, if any, of
  1015  // its subexpressions, as defined by the 'Submatch' description in the
  1016  // package comment.
  1017  // A return value of nil indicates no match.
  1018  func (re *Regexp) FindStringSubmatch(s string) []string {
  1019  	var dstCap [4]int
  1020  	a := re.doExecute(nil, nil, s, 0, re.prog.NumCap, dstCap[:0])
  1021  	if a == nil {
  1022  		return nil
  1023  	}
  1024  	ret := make([]string, 1+re.numSubexp)
  1025  	for i := range ret {
  1026  		if 2*i < len(a) && a[2*i] >= 0 {
  1027  			ret[i] = s[a[2*i]:a[2*i+1]]
  1028  		}
  1029  	}
  1030  	return ret
  1031  }
  1032  
  1033  // FindStringSubmatchIndex returns a slice holding the index pairs
  1034  // identifying the leftmost match of the regular expression in s and the
  1035  // matches, if any, of its subexpressions, as defined by the 'Submatch' and
  1036  // 'Index' descriptions in the package comment.
  1037  // A return value of nil indicates no match.
  1038  func (re *Regexp) FindStringSubmatchIndex(s string) []int {
  1039  	return re.pad(re.doExecute(nil, nil, s, 0, re.prog.NumCap, nil))
  1040  }
  1041  
  1042  // FindReaderSubmatchIndex returns a slice holding the index pairs
  1043  // identifying the leftmost match of the regular expression of text read by
  1044  // the [io.RuneReader], and the matches, if any, of its subexpressions, as defined
  1045  // by the 'Submatch' and 'Index' descriptions in the package comment. A
  1046  // return value of nil indicates no match.
  1047  func (re *Regexp) FindReaderSubmatchIndex(r io.RuneReader) []int {
  1048  	return re.pad(re.doExecute(r, nil, "", 0, re.prog.NumCap, nil))
  1049  }
  1050  
  1051  const startSize = 10 // The size at which to start a slice in the 'All' routines.
  1052  
  1053  // FindAll is the 'All' version of [Regexp.Find]; it returns a slice of all successive
  1054  // matches of the expression, as defined by the 'All' description in the
  1055  // package comment.
  1056  // A return value of nil indicates no match.
  1057  func (re *Regexp) FindAll(b []byte, n int) [][]byte {
  1058  	if n < 0 {
  1059  		n = len(b) + 1
  1060  	}
  1061  	var result [][]byte
  1062  	re.allMatches("", b, n, func(match []int) {
  1063  		if result == nil {
  1064  			result = make([][]byte, 0, startSize)
  1065  		}
  1066  		result = append(result, b[match[0]:match[1]:match[1]])
  1067  	})
  1068  	return result
  1069  }
  1070  
  1071  // FindAllIndex is the 'All' version of [Regexp.FindIndex]; it returns a slice of all
  1072  // successive matches of the expression, as defined by the 'All' description
  1073  // in the package comment.
  1074  // A return value of nil indicates no match.
  1075  func (re *Regexp) FindAllIndex(b []byte, n int) [][]int {
  1076  	if n < 0 {
  1077  		n = len(b) + 1
  1078  	}
  1079  	var result [][]int
  1080  	re.allMatches("", b, n, func(match []int) {
  1081  		if result == nil {
  1082  			result = make([][]int, 0, startSize)
  1083  		}
  1084  		result = append(result, match[0:2])
  1085  	})
  1086  	return result
  1087  }
  1088  
  1089  // FindAllString is the 'All' version of [Regexp.FindString]; it returns a slice of all
  1090  // successive matches of the expression, as defined by the 'All' description
  1091  // in the package comment.
  1092  // A return value of nil indicates no match.
  1093  func (re *Regexp) FindAllString(s string, n int) []string {
  1094  	if n < 0 {
  1095  		n = len(s) + 1
  1096  	}
  1097  	var result []string
  1098  	re.allMatches(s, nil, n, func(match []int) {
  1099  		if result == nil {
  1100  			result = make([]string, 0, startSize)
  1101  		}
  1102  		result = append(result, s[match[0]:match[1]])
  1103  	})
  1104  	return result
  1105  }
  1106  
  1107  // FindAllStringIndex is the 'All' version of [Regexp.FindStringIndex]; it returns a
  1108  // slice of all successive matches of the expression, as defined by the 'All'
  1109  // description in the package comment.
  1110  // A return value of nil indicates no match.
  1111  func (re *Regexp) FindAllStringIndex(s string, n int) [][]int {
  1112  	if n < 0 {
  1113  		n = len(s) + 1
  1114  	}
  1115  	var result [][]int
  1116  	re.allMatches(s, nil, n, func(match []int) {
  1117  		if result == nil {
  1118  			result = make([][]int, 0, startSize)
  1119  		}
  1120  		result = append(result, match[0:2])
  1121  	})
  1122  	return result
  1123  }
  1124  
  1125  // FindAllSubmatch is the 'All' version of [Regexp.FindSubmatch]; it returns a slice
  1126  // of all successive matches of the expression, as defined by the 'All'
  1127  // description in the package comment.
  1128  // A return value of nil indicates no match.
  1129  func (re *Regexp) FindAllSubmatch(b []byte, n int) [][][]byte {
  1130  	if n < 0 {
  1131  		n = len(b) + 1
  1132  	}
  1133  	var result [][][]byte
  1134  	re.allMatches("", b, n, func(match []int) {
  1135  		if result == nil {
  1136  			result = make([][][]byte, 0, startSize)
  1137  		}
  1138  		slice := make([][]byte, len(match)/2)
  1139  		for j := range slice {
  1140  			if match[2*j] >= 0 {
  1141  				slice[j] = b[match[2*j]:match[2*j+1]:match[2*j+1]]
  1142  			}
  1143  		}
  1144  		result = append(result, slice)
  1145  	})
  1146  	return result
  1147  }
  1148  
  1149  // FindAllSubmatchIndex is the 'All' version of [Regexp.FindSubmatchIndex]; it returns
  1150  // a slice of all successive matches of the expression, as defined by the
  1151  // 'All' description in the package comment.
  1152  // A return value of nil indicates no match.
  1153  func (re *Regexp) FindAllSubmatchIndex(b []byte, n int) [][]int {
  1154  	if n < 0 {
  1155  		n = len(b) + 1
  1156  	}
  1157  	var result [][]int
  1158  	re.allMatches("", b, n, func(match []int) {
  1159  		if result == nil {
  1160  			result = make([][]int, 0, startSize)
  1161  		}
  1162  		result = append(result, match)
  1163  	})
  1164  	return result
  1165  }
  1166  
  1167  // FindAllStringSubmatch is the 'All' version of [Regexp.FindStringSubmatch]; it
  1168  // returns a slice of all successive matches of the expression, as defined by
  1169  // the 'All' description in the package comment.
  1170  // A return value of nil indicates no match.
  1171  func (re *Regexp) FindAllStringSubmatch(s string, n int) [][]string {
  1172  	if n < 0 {
  1173  		n = len(s) + 1
  1174  	}
  1175  	var result [][]string
  1176  	re.allMatches(s, nil, n, func(match []int) {
  1177  		if result == nil {
  1178  			result = make([][]string, 0, startSize)
  1179  		}
  1180  		slice := make([]string, len(match)/2)
  1181  		for j := range slice {
  1182  			if match[2*j] >= 0 {
  1183  				slice[j] = s[match[2*j]:match[2*j+1]]
  1184  			}
  1185  		}
  1186  		result = append(result, slice)
  1187  	})
  1188  	return result
  1189  }
  1190  
  1191  // FindAllStringSubmatchIndex is the 'All' version of
  1192  // [Regexp.FindStringSubmatchIndex]; it returns a slice of all successive matches of
  1193  // the expression, as defined by the 'All' description in the package
  1194  // comment.
  1195  // A return value of nil indicates no match.
  1196  func (re *Regexp) FindAllStringSubmatchIndex(s string, n int) [][]int {
  1197  	if n < 0 {
  1198  		n = len(s) + 1
  1199  	}
  1200  	var result [][]int
  1201  	re.allMatches(s, nil, n, func(match []int) {
  1202  		if result == nil {
  1203  			result = make([][]int, 0, startSize)
  1204  		}
  1205  		result = append(result, match)
  1206  	})
  1207  	return result
  1208  }
  1209  
  1210  // Split slices s into substrings separated by the expression and returns a slice of
  1211  // the substrings between those expression matches.
  1212  //
  1213  // The slice returned by this method consists of all the substrings of s
  1214  // not contained in the slice returned by [Regexp.FindAllString]. When called on an expression
  1215  // that contains no metacharacters, it is equivalent to [strings.SplitN].
  1216  //
  1217  // Example:
  1218  //
  1219  //	s := regexp.MustCompile("a*").Split("abaabaccadaaae", 5)
  1220  //	// s: ["", "b", "b", "c", "cadaaae"]
  1221  //
  1222  // The count determines the number of substrings to return:
  1223  //   - n > 0: at most n substrings; the last substring will be the unsplit remainder;
  1224  //   - n == 0: the result is nil (zero substrings);
  1225  //   - n < 0: all substrings.
  1226  func (re *Regexp) Split(s string, n int) []string {
  1227  
  1228  	if n == 0 {
  1229  		return nil
  1230  	}
  1231  
  1232  	if len(re.expr) > 0 && len(s) == 0 {
  1233  		return []string{""}
  1234  	}
  1235  
  1236  	matches := re.FindAllStringIndex(s, n)
  1237  	strings := make([]string, 0, len(matches))
  1238  
  1239  	beg := 0
  1240  	end := 0
  1241  	for _, match := range matches {
  1242  		if n > 0 && len(strings) >= n-1 {
  1243  			break
  1244  		}
  1245  
  1246  		end = match[0]
  1247  		if match[1] != 0 {
  1248  			strings = append(strings, s[beg:end])
  1249  		}
  1250  		beg = match[1]
  1251  	}
  1252  
  1253  	if end != len(s) {
  1254  		strings = append(strings, s[beg:])
  1255  	}
  1256  
  1257  	return strings
  1258  }
  1259  
  1260  // AppendText implements [encoding.TextAppender]. The output
  1261  // matches that of calling the [Regexp.String] method.
  1262  //
  1263  // Note that the output is lossy in some cases: This method does not indicate
  1264  // POSIX regular expressions (i.e. those compiled by calling [CompilePOSIX]), or
  1265  // those for which the [Regexp.Longest] method has been called.
  1266  func (re *Regexp) AppendText(b []byte) ([]byte, error) {
  1267  	return append(b, re.String()...), nil
  1268  }
  1269  
  1270  // MarshalText implements [encoding.TextMarshaler]. The output
  1271  // matches that of calling the [Regexp.AppendText] method.
  1272  //
  1273  // See [Regexp.AppendText] for more information.
  1274  func (re *Regexp) MarshalText() ([]byte, error) {
  1275  	return re.AppendText(nil)
  1276  }
  1277  
  1278  // UnmarshalText implements [encoding.TextUnmarshaler] by calling
  1279  // [Compile] on the encoded value.
  1280  func (re *Regexp) UnmarshalText(text []byte) error {
  1281  	newRE, err := Compile(string(text))
  1282  	if err != nil {
  1283  		return err
  1284  	}
  1285  	*re = *newRE
  1286  	return nil
  1287  }
  1288  

View as plain text