rust.go

     1  // Copyright 2021 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 demangle
     6  
     7  import (
     8  	"fmt"
     9  	"math"
    10  	"math/bits"
    11  	"strings"
    12  	"unicode/utf8"
    13  )
    14  
    15  // rustToString demangles a Rust symbol.
    16  func rustToString(name string, options []Option) (ret string, err error) {
    17  	if !strings.HasPrefix(name, "_R") {
    18  		return "", ErrNotMangledName
    19  	}
    20  
    21  	// When the demangling routines encounter an error, they panic
    22  	// with a value of type demangleErr.
    23  	defer func() {
    24  		if r := recover(); r != nil {
    25  			if de, ok := r.(demangleErr); ok {
    26  				ret = ""
    27  				err = de
    28  				return
    29  			}
    30  			panic(r)
    31  		}
    32  	}()
    33  
    34  	suffix := ""
    35  	dot := strings.Index(name, ".")
    36  	if dot >= 0 {
    37  		suffix = name[dot:]
    38  		name = name[:dot]
    39  	}
    40  
    41  	name = name[2:]
    42  	rst := &rustState{orig: name, str: name}
    43  
    44  	for _, o := range options {
    45  		if o == NoTemplateParams {
    46  			rst.noGenericArgs = true
    47  		} else if isMaxLength(o) {
    48  			rst.max = maxLength(o)
    49  		}
    50  	}
    51  
    52  	rst.symbolName()
    53  
    54  	if len(rst.str) > 0 {
    55  		rst.fail("unparsed characters at end of mangled name")
    56  	}
    57  
    58  	if suffix != "" {
    59  		llvmStyle := false
    60  		for _, o := range options {
    61  			if o == LLVMStyle {
    62  				llvmStyle = true
    63  				break
    64  			}
    65  		}
    66  		if llvmStyle {
    67  			rst.skip = false
    68  			rst.writeString(" (")
    69  			rst.writeString(suffix)
    70  			rst.writeByte(')')
    71  		}
    72  	}
    73  
    74  	s := rst.buf.String()
    75  	if rst.max > 0 && len(s) > rst.max {
    76  		s = s[:rst.max]
    77  	}
    78  	return s, nil
    79  }
    80  
    81  // A rustState holds the current state of demangling a Rust string.
    82  type rustState struct {
    83  	orig          string          // the original string being demangled
    84  	str           string          // remainder of string to demangle
    85  	off           int             // offset of str within original string
    86  	buf           strings.Builder // demangled string being built
    87  	skip          bool            // don't print, just skip
    88  	lifetimes     int64           // number of bound lifetimes
    89  	last          byte            // last byte written to buffer
    90  	noGenericArgs bool            // don't demangle generic arguments
    91  	max           int             // maximum output length
    92  }
    93  
    94  // fail panics with demangleErr, to be caught in rustToString.
    95  func (rst *rustState) fail(err string) {
    96  	panic(demangleErr{err: err, off: rst.off})
    97  }
    98  
    99  // advance advances the current string offset.
   100  func (rst *rustState) advance(add int) {
   101  	if len(rst.str) < add {
   102  		panic("internal error")
   103  	}
   104  	rst.str = rst.str[add:]
   105  	rst.off += add
   106  }
   107  
   108  // checkChar requires that the next character in the string be c,
   109  // and advances past it.
   110  func (rst *rustState) checkChar(c byte) {
   111  	if len(rst.str) == 0 || rst.str[0] != c {
   112  		rst.fail("expected " + string(c))
   113  	}
   114  	rst.advance(1)
   115  }
   116  
   117  // writeByte writes a byte to the buffer.
   118  func (rst *rustState) writeByte(c byte) {
   119  	if rst.skip {
   120  		return
   121  	}
   122  	if rst.max > 0 && rst.buf.Len() > rst.max {
   123  		rst.skip = true
   124  		return
   125  	}
   126  	rst.last = c
   127  	rst.buf.WriteByte(c)
   128  }
   129  
   130  // writeString writes a string to the buffer.
   131  func (rst *rustState) writeString(s string) {
   132  	if rst.skip {
   133  		return
   134  	}
   135  	if rst.max > 0 && rst.buf.Len() > rst.max {
   136  		rst.skip = true
   137  		return
   138  	}
   139  	if len(s) > 0 {
   140  		rst.last = s[len(s)-1]
   141  		rst.buf.WriteString(s)
   142  	}
   143  }
   144  
   145  // symbolName parses:
   146  //
   147  //	<symbol-name> = "_R" [<decimal-number>] <path> [<instantiating-crate>]
   148  //	<instantiating-crate> = <path>
   149  //
   150  // We've already skipped the "_R".
   151  func (rst *rustState) symbolName() {
   152  	if len(rst.str) < 1 {
   153  		rst.fail("expected symbol-name")
   154  	}
   155  
   156  	if isDigit(rst.str[0]) {
   157  		rst.fail("unsupported Rust encoding version")
   158  	}
   159  
   160  	rst.path(true)
   161  
   162  	if len(rst.str) > 0 {
   163  		rst.skip = true
   164  		rst.path(false)
   165  	}
   166  }
   167  
   168  // path parses:
   169  //
   170  //	<path> = "C" <identifier>                    // crate root
   171  //	       | "M" <impl-path> <type>              // <T> (inherent impl)
   172  //	       | "X" <impl-path> <type> <path>       // <T as Trait> (trait impl)
   173  //	       | "Y" <type> <path>                   // <T as Trait> (trait definition)
   174  //	       | "N" <namespace> <path> <identifier> // ...::ident (nested path)
   175  //	       | "I" <path> {<generic-arg>} "E"      // ...<T, U> (generic args)
   176  //	       | <backref>
   177  //	<namespace> = "C"      // closure
   178  //	            | "S"      // shim
   179  //	            | <A-Z>    // other special namespaces
   180  //	            | <a-z>    // internal namespaces
   181  //
   182  // needsSeparator is true if we need to write out :: for a generic;
   183  // it is passed as false if we are in the middle of a type.
   184  func (rst *rustState) path(needsSeparator bool) {
   185  	if len(rst.str) < 1 {
   186  		rst.fail("expected path")
   187  	}
   188  	switch c := rst.str[0]; c {
   189  	case 'C':
   190  		rst.advance(1)
   191  		_, ident := rst.identifier()
   192  		rst.writeString(ident)
   193  	case 'M', 'X':
   194  		rst.advance(1)
   195  		rst.implPath()
   196  		rst.writeByte('<')
   197  		rst.demangleType()
   198  		if c == 'X' {
   199  			rst.writeString(" as ")
   200  			rst.path(false)
   201  		}
   202  		rst.writeByte('>')
   203  	case 'Y':
   204  		rst.advance(1)
   205  		rst.writeByte('<')
   206  		rst.demangleType()
   207  		rst.writeString(" as ")
   208  		rst.path(false)
   209  		rst.writeByte('>')
   210  	case 'N':
   211  		rst.advance(1)
   212  
   213  		if len(rst.str) < 1 {
   214  			rst.fail("expected namespace")
   215  		}
   216  		ns := rst.str[0]
   217  		switch {
   218  		case ns >= 'a' && ns <= 'z':
   219  		case ns >= 'A' && ns <= 'Z':
   220  		default:
   221  			rst.fail("invalid namespace character")
   222  		}
   223  		rst.advance(1)
   224  
   225  		rst.path(needsSeparator)
   226  
   227  		dis, ident := rst.identifier()
   228  
   229  		if ns >= 'A' && ns <= 'Z' {
   230  			rst.writeString("::{")
   231  			switch ns {
   232  			case 'C':
   233  				rst.writeString("closure")
   234  			case 'S':
   235  				rst.writeString("shim")
   236  			default:
   237  				rst.writeByte(ns)
   238  			}
   239  			if len(ident) > 0 {
   240  				rst.writeByte(':')
   241  				rst.writeString(ident)
   242  			}
   243  			if !rst.skip {
   244  				fmt.Fprintf(&rst.buf, "#%d}", dis)
   245  				rst.last = '}'
   246  			}
   247  		} else {
   248  			rst.writeString("::")
   249  			rst.writeString(ident)
   250  		}
   251  	case 'I':
   252  		rst.advance(1)
   253  		rst.path(needsSeparator)
   254  		if needsSeparator {
   255  			rst.writeString("::")
   256  		}
   257  		rst.writeByte('<')
   258  		rst.genericArgs()
   259  		rst.writeByte('>')
   260  		rst.checkChar('E')
   261  	case 'B':
   262  		rst.backref(func() { rst.path(needsSeparator) })
   263  	default:
   264  		rst.fail("unrecognized letter in path")
   265  	}
   266  }
   267  
   268  // implPath parses:
   269  //
   270  //	<impl-path> = [<disambiguator>] <path>
   271  func (rst *rustState) implPath() {
   272  	// This path is not part of the demangled string.
   273  	hold := rst.skip
   274  	rst.skip = true
   275  	defer func() {
   276  		rst.skip = hold
   277  	}()
   278  
   279  	rst.disambiguator()
   280  	rst.path(false)
   281  }
   282  
   283  // identifier parses:
   284  //
   285  //	<identifier> = [<disambiguator>] <undisambiguated-identifier>
   286  //
   287  // It returns the disambiguator and the identifier.
   288  func (rst *rustState) identifier() (int64, string) {
   289  	dis := rst.disambiguator()
   290  	ident, _ := rst.undisambiguatedIdentifier()
   291  	return dis, ident
   292  }
   293  
   294  // disambiguator parses an optional:
   295  //
   296  //	<disambiguator> = "s" <base-62-number>
   297  func (rst *rustState) disambiguator() int64 {
   298  	if len(rst.str) == 0 || rst.str[0] != 's' {
   299  		return 0
   300  	}
   301  	rst.advance(1)
   302  	return rst.base62Number() + 1
   303  }
   304  
   305  // undisambiguatedIdentifier parses:
   306  //
   307  //	<undisambiguated-identifier> = ["u"] <decimal-number> ["_"] <bytes>
   308  func (rst *rustState) undisambiguatedIdentifier() (id string, isPunycode bool) {
   309  	isPunycode = false
   310  	if len(rst.str) > 0 && rst.str[0] == 'u' {
   311  		rst.advance(1)
   312  		isPunycode = true
   313  	}
   314  
   315  	val := rst.decimalNumber()
   316  
   317  	if len(rst.str) > 0 && rst.str[0] == '_' {
   318  		rst.advance(1)
   319  	}
   320  
   321  	if len(rst.str) < val {
   322  		rst.fail("not enough characters for identifier")
   323  	}
   324  	id = rst.str[:val]
   325  	rst.advance(val)
   326  
   327  	for i := 0; i < len(id); i++ {
   328  		c := id[i]
   329  		switch {
   330  		case c >= '0' && c <= '9':
   331  		case c >= 'A' && c <= 'Z':
   332  		case c >= 'a' && c <= 'z':
   333  		case c == '_':
   334  		default:
   335  			rst.fail("invalid character in identifier")
   336  		}
   337  	}
   338  
   339  	if isPunycode {
   340  		id = rst.expandPunycode(id)
   341  	}
   342  
   343  	return id, isPunycode
   344  }
   345  
   346  // expandPunycode decodes the Rust version of punycode.
   347  // This algorithm is taken from RFC 3492 section 6.2.
   348  func (rst *rustState) expandPunycode(s string) string {
   349  	const (
   350  		base        = 36
   351  		tmin        = 1
   352  		tmax        = 26
   353  		skew        = 38
   354  		damp        = 700
   355  		initialBias = 72
   356  		initialN    = 128
   357  	)
   358  
   359  	var (
   360  		output   []rune
   361  		encoding string
   362  	)
   363  	idx := strings.LastIndex(s, "_")
   364  	if idx >= 0 {
   365  		output = []rune(s[:idx])
   366  		encoding = s[idx+1:]
   367  	} else {
   368  		encoding = s
   369  	}
   370  
   371  	i := 0
   372  	n := initialN
   373  	bias := initialBias
   374  
   375  	pos := 0
   376  	for pos < len(encoding) {
   377  		oldI := i
   378  		w := 1
   379  		for k := base; ; k += base {
   380  			if pos == len(encoding) {
   381  				rst.fail("unterminated punycode")
   382  			}
   383  
   384  			var digit byte
   385  			d := encoding[pos]
   386  			pos++
   387  			switch {
   388  			case '0' <= d && d <= '9':
   389  				digit = d - '0' + 26
   390  			case 'A' <= d && d <= 'Z':
   391  				digit = d - 'A'
   392  			case 'a' <= d && d <= 'z':
   393  				digit = d - 'a'
   394  			default:
   395  				rst.fail("invalid punycode digit")
   396  			}
   397  
   398  			i += int(digit) * w
   399  			if i < 0 {
   400  				rst.fail("punycode number overflow")
   401  			}
   402  
   403  			var t int
   404  			if k <= bias {
   405  				t = tmin
   406  			} else if k > bias+tmax {
   407  				t = tmax
   408  			} else {
   409  				t = k - bias
   410  			}
   411  
   412  			if int(digit) < t {
   413  				break
   414  			}
   415  
   416  			if w >= math.MaxInt32/base {
   417  				rst.fail("punycode number overflow")
   418  			}
   419  			w *= base - t
   420  		}
   421  
   422  		delta := i - oldI
   423  		numPoints := len(output) + 1
   424  		firstTime := oldI == 0
   425  		if firstTime {
   426  			delta /= damp
   427  		} else {
   428  			delta /= 2
   429  		}
   430  		delta += delta / numPoints
   431  		k := 0
   432  		for delta > ((base-tmin)*tmax)/2 {
   433  			delta /= base - tmin
   434  			k += base
   435  		}
   436  		bias = k + ((base-tmin+1)*delta)/(delta+skew)
   437  
   438  		n += i / (len(output) + 1)
   439  		if n > utf8.MaxRune {
   440  			rst.fail("punycode rune overflow")
   441  		} else if !utf8.ValidRune(rune(n)) {
   442  			rst.fail("punycode invalid code point")
   443  		}
   444  		i %= len(output) + 1
   445  		output = append(output, 0)
   446  		copy(output[i+1:], output[i:])
   447  		output[i] = rune(n)
   448  		i++
   449  	}
   450  
   451  	return string(output)
   452  }
   453  
   454  // genericArgs prints a list of generic arguments, without angle brackets.
   455  func (rst *rustState) genericArgs() {
   456  	if rst.noGenericArgs {
   457  		hold := rst.skip
   458  		rst.skip = true
   459  		defer func() {
   460  			rst.skip = hold
   461  		}()
   462  	}
   463  
   464  	first := true
   465  	for len(rst.str) > 0 && rst.str[0] != 'E' {
   466  		if first {
   467  			first = false
   468  		} else {
   469  			rst.writeString(", ")
   470  		}
   471  		rst.genericArg()
   472  	}
   473  }
   474  
   475  // genericArg parses:
   476  //
   477  //	<generic-arg> = <lifetime>
   478  //	              | <type>
   479  //	              | "K" <const> // forward-compat for const generics
   480  //	<lifetime> = "L" <base-62-number>
   481  func (rst *rustState) genericArg() {
   482  	if len(rst.str) < 1 {
   483  		rst.fail("expected generic-arg")
   484  	}
   485  	if rst.str[0] == 'L' {
   486  		rst.advance(1)
   487  		rst.writeLifetime(rst.base62Number())
   488  	} else if rst.str[0] == 'K' {
   489  		rst.advance(1)
   490  		rst.demangleConst()
   491  	} else {
   492  		rst.demangleType()
   493  	}
   494  }
   495  
   496  // binder parses an optional:
   497  //
   498  //	<binder> = "G" <base-62-number>
   499  func (rst *rustState) binder() {
   500  	if len(rst.str) < 1 || rst.str[0] != 'G' {
   501  		return
   502  	}
   503  	rst.advance(1)
   504  
   505  	binderLifetimes := rst.base62Number() + 1
   506  
   507  	// Every bound lifetime should be referenced later.
   508  	if binderLifetimes >= int64(len(rst.str))-rst.lifetimes {
   509  		rst.fail("binder lifetimes overflow")
   510  	}
   511  
   512  	rst.writeString("for<")
   513  	for i := int64(0); i < binderLifetimes; i++ {
   514  		if i > 0 {
   515  			rst.writeString(", ")
   516  		}
   517  		rst.lifetimes++
   518  		rst.writeLifetime(1)
   519  	}
   520  	rst.writeString("> ")
   521  }
   522  
   523  // demangleType parses:
   524  //
   525  //	<type> = <basic-type>
   526  //	       | <path>                      // named type
   527  //	       | "A" <type> <const>          // [T; N]
   528  //	       | "S" <type>                  // [T]
   529  //	       | "T" {<type>} "E"            // (T1, T2, T3, ...)
   530  //	       | "R" [<lifetime>] <type>     // &T
   531  //	       | "Q" [<lifetime>] <type>     // &mut T
   532  //	       | "P" <type>                  // *const T
   533  //	       | "O" <type>                  // *mut T
   534  //	       | "F" <fn-sig>                // fn(...) -> ...
   535  //	       | "D" <dyn-bounds> <lifetime> // dyn Trait<Assoc = X> + Send + 'a
   536  //	       | <backref>
   537  func (rst *rustState) demangleType() {
   538  	if len(rst.str) < 1 {
   539  		rst.fail("expected type")
   540  	}
   541  	c := rst.str[0]
   542  	if c >= 'a' && c <= 'z' {
   543  		rst.basicType()
   544  		return
   545  	}
   546  	switch c {
   547  	case 'C', 'M', 'X', 'Y', 'N', 'I':
   548  		rst.path(false)
   549  	case 'A', 'S':
   550  		rst.advance(1)
   551  		rst.writeByte('[')
   552  		rst.demangleType()
   553  		if c == 'A' {
   554  			rst.writeString("; ")
   555  			rst.demangleConst()
   556  		}
   557  		rst.writeByte(']')
   558  	case 'T':
   559  		rst.advance(1)
   560  		rst.writeByte('(')
   561  		c := 0
   562  		for len(rst.str) > 0 && rst.str[0] != 'E' {
   563  			if c > 0 {
   564  				rst.writeString(", ")
   565  			}
   566  			c++
   567  			rst.demangleType()
   568  		}
   569  		if c == 1 {
   570  			rst.writeByte(',')
   571  		}
   572  		rst.writeByte(')')
   573  		rst.checkChar('E')
   574  	case 'R', 'Q':
   575  		rst.advance(1)
   576  		rst.writeByte('&')
   577  		if len(rst.str) > 0 && rst.str[0] == 'L' {
   578  			rst.advance(1)
   579  			if lifetime := rst.base62Number(); lifetime > 0 {
   580  				rst.writeLifetime(lifetime)
   581  				rst.writeByte(' ')
   582  			}
   583  		}
   584  		if c == 'Q' {
   585  			rst.writeString("mut ")
   586  		}
   587  		rst.demangleType()
   588  	case 'P':
   589  		rst.advance(1)
   590  		rst.writeString("*const ")
   591  		rst.demangleType()
   592  	case 'O':
   593  		rst.advance(1)
   594  		rst.writeString("*mut ")
   595  		rst.demangleType()
   596  	case 'F':
   597  		rst.advance(1)
   598  		hold := rst.lifetimes
   599  		rst.fnSig()
   600  		rst.lifetimes = hold
   601  	case 'D':
   602  		rst.advance(1)
   603  		hold := rst.lifetimes
   604  		rst.dynBounds()
   605  		rst.lifetimes = hold
   606  		if len(rst.str) == 0 || rst.str[0] != 'L' {
   607  			rst.fail("expected L")
   608  		}
   609  		rst.advance(1)
   610  		if lifetime := rst.base62Number(); lifetime > 0 {
   611  			if rst.last != ' ' {
   612  				rst.writeByte(' ')
   613  			}
   614  			rst.writeString("+ ")
   615  			rst.writeLifetime(lifetime)
   616  		}
   617  	case 'B':
   618  		rst.backref(rst.demangleType)
   619  	default:
   620  		rst.fail("unrecognized character in type")
   621  	}
   622  }
   623  
   624  var rustBasicTypes = map[byte]string{
   625  	'a': "i8",
   626  	'b': "bool",
   627  	'c': "char",
   628  	'd': "f64",
   629  	'e': "str",
   630  	'f': "f32",
   631  	'h': "u8",
   632  	'i': "isize",
   633  	'j': "usize",
   634  	'l': "i32",
   635  	'm': "u32",
   636  	'n': "i128",
   637  	'o': "u128",
   638  	'p': "_",
   639  	's': "i16",
   640  	't': "u16",
   641  	'u': "()",
   642  	'v': "...",
   643  	'x': "i64",
   644  	'y': "u64",
   645  	'z': "!",
   646  }
   647  
   648  // basicType parses:
   649  //
   650  //	<basic-type>
   651  func (rst *rustState) basicType() {
   652  	if len(rst.str) < 1 {
   653  		rst.fail("expected basic type")
   654  	}
   655  	str, ok := rustBasicTypes[rst.str[0]]
   656  	if !ok {
   657  		rst.fail("unrecognized basic type character")
   658  	}
   659  	rst.advance(1)
   660  	rst.writeString(str)
   661  }
   662  
   663  // fnSig parses:
   664  //
   665  //	<fn-sig> = [<binder>] ["U"] ["K" <abi>] {<type>} "E" <type>
   666  //	<abi> = "C"
   667  //	      | <undisambiguated-identifier>
   668  func (rst *rustState) fnSig() {
   669  	rst.binder()
   670  	if len(rst.str) > 0 && rst.str[0] == 'U' {
   671  		rst.advance(1)
   672  		rst.writeString("unsafe ")
   673  	}
   674  	if len(rst.str) > 0 && rst.str[0] == 'K' {
   675  		rst.advance(1)
   676  		if len(rst.str) > 0 && rst.str[0] == 'C' {
   677  			rst.advance(1)
   678  			rst.writeString(`extern "C" `)
   679  		} else {
   680  			rst.writeString(`extern "`)
   681  			id, isPunycode := rst.undisambiguatedIdentifier()
   682  			if isPunycode {
   683  				rst.fail("punycode used in ABI string")
   684  			}
   685  			id = strings.ReplaceAll(id, "_", "-")
   686  			rst.writeString(id)
   687  			rst.writeString(`" `)
   688  		}
   689  	}
   690  	rst.writeString("fn(")
   691  	first := true
   692  	for len(rst.str) > 0 && rst.str[0] != 'E' {
   693  		if first {
   694  			first = false
   695  		} else {
   696  			rst.writeString(", ")
   697  		}
   698  		rst.demangleType()
   699  	}
   700  	rst.checkChar('E')
   701  	rst.writeByte(')')
   702  	if len(rst.str) > 0 && rst.str[0] == 'u' {
   703  		rst.advance(1)
   704  	} else {
   705  		rst.writeString(" -> ")
   706  		rst.demangleType()
   707  	}
   708  }
   709  
   710  // dynBounds parses:
   711  //
   712  //	<dyn-bounds> = [<binder>] {<dyn-trait>} "E"
   713  func (rst *rustState) dynBounds() {
   714  	rst.writeString("dyn ")
   715  	rst.binder()
   716  	first := true
   717  	for len(rst.str) > 0 && rst.str[0] != 'E' {
   718  		if first {
   719  			first = false
   720  		} else {
   721  			rst.writeString(" + ")
   722  		}
   723  		rst.dynTrait()
   724  	}
   725  	rst.checkChar('E')
   726  }
   727  
   728  // dynTrait parses:
   729  //
   730  //	<dyn-trait> = <path> {<dyn-trait-assoc-binding>}
   731  //	<dyn-trait-assoc-binding> = "p" <undisambiguated-identifier> <type>
   732  func (rst *rustState) dynTrait() {
   733  	started := rst.pathStartGenerics()
   734  	for len(rst.str) > 0 && rst.str[0] == 'p' {
   735  		rst.advance(1)
   736  		if started {
   737  			rst.writeString(", ")
   738  		} else {
   739  			rst.writeByte('<')
   740  			started = true
   741  		}
   742  		id, _ := rst.undisambiguatedIdentifier()
   743  		rst.writeString(id)
   744  		rst.writeString(" = ")
   745  		rst.demangleType()
   746  	}
   747  	if started {
   748  		rst.writeByte('>')
   749  	}
   750  }
   751  
   752  // pathStartGenerics is like path but if it sees an I to start generic
   753  // arguments it won't close them. It reports whether it started generics.
   754  func (rst *rustState) pathStartGenerics() bool {
   755  	if len(rst.str) < 1 {
   756  		rst.fail("expected path")
   757  	}
   758  	switch rst.str[0] {
   759  	case 'I':
   760  		rst.advance(1)
   761  		rst.path(false)
   762  		rst.writeByte('<')
   763  		rst.genericArgs()
   764  		rst.checkChar('E')
   765  		return true
   766  	case 'B':
   767  		var started bool
   768  		rst.backref(func() { started = rst.pathStartGenerics() })
   769  		return started
   770  	default:
   771  		rst.path(false)
   772  		return false
   773  	}
   774  }
   775  
   776  // writeLifetime writes out a lifetime binding.
   777  func (rst *rustState) writeLifetime(lifetime int64) {
   778  	rst.writeByte('\'')
   779  	if lifetime == 0 {
   780  		rst.writeByte('_')
   781  		return
   782  	}
   783  	depth := rst.lifetimes - lifetime
   784  	if depth < 0 {
   785  		rst.fail("invalid lifetime")
   786  	} else if depth < 26 {
   787  		rst.writeByte('a' + byte(depth))
   788  	} else {
   789  		rst.writeByte('z')
   790  		if !rst.skip {
   791  			fmt.Fprintf(&rst.buf, "%d", depth-26+1)
   792  			rst.last = '0'
   793  		}
   794  	}
   795  }
   796  
   797  // demangleConst parses:
   798  //
   799  //	<const> = <type> <const-data>
   800  //	        | "p" // placeholder, shown as _
   801  //	        | <backref>
   802  //	<const-data> = ["n"] {<hex-digit>} "_"
   803  func (rst *rustState) demangleConst() {
   804  	if len(rst.str) < 1 {
   805  		rst.fail("expected constant")
   806  	}
   807  
   808  	if rst.str[0] == 'B' {
   809  		rst.backref(rst.demangleConst)
   810  		return
   811  	}
   812  
   813  	if rst.str[0] == 'p' {
   814  		rst.advance(1)
   815  		rst.writeByte('_')
   816  		return
   817  	}
   818  
   819  	typ := rst.str[0]
   820  
   821  	const (
   822  		invalid = iota
   823  		signedInt
   824  		unsignedInt
   825  		boolean
   826  		character
   827  	)
   828  
   829  	var kind int
   830  	switch typ {
   831  	case 'a', 's', 'l', 'x', 'n', 'i':
   832  		kind = signedInt
   833  	case 'h', 't', 'm', 'y', 'o', 'j':
   834  		kind = unsignedInt
   835  	case 'b':
   836  		kind = boolean
   837  	case 'c':
   838  		kind = character
   839  	default:
   840  		rst.fail("unrecognized constant type")
   841  	}
   842  
   843  	rst.advance(1)
   844  
   845  	if kind == signedInt && len(rst.str) > 0 && rst.str[0] == 'n' {
   846  		rst.advance(1)
   847  		rst.writeByte('-')
   848  	}
   849  
   850  	start := rst.str
   851  	digits := 0
   852  	val := uint64(0)
   853  digitLoop:
   854  	for len(rst.str) > 0 {
   855  		c := rst.str[0]
   856  		var digit uint64
   857  		switch {
   858  		case c >= '0' && c <= '9':
   859  			digit = uint64(c - '0')
   860  		case c >= 'a' && c <= 'f':
   861  			digit = uint64(c - 'a' + 10)
   862  		case c == '_':
   863  			rst.advance(1)
   864  			break digitLoop
   865  		default:
   866  			rst.fail("expected hex digit or _")
   867  		}
   868  		rst.advance(1)
   869  		if val == 0 && digit == 0 && (len(rst.str) == 0 || rst.str[0] != '_') {
   870  			rst.fail("invalid leading 0 in constant")
   871  		}
   872  		val *= 16
   873  		val += digit
   874  		digits++
   875  	}
   876  
   877  	if digits == 0 {
   878  		rst.fail("expected constant")
   879  	}
   880  
   881  	switch kind {
   882  	case signedInt, unsignedInt:
   883  		if digits > 16 {
   884  			// Value too big, just write out the string.
   885  			rst.writeString("0x")
   886  			rst.writeString(start[:digits])
   887  		} else {
   888  			if !rst.skip {
   889  				fmt.Fprintf(&rst.buf, "%d", val)
   890  				rst.last = '0'
   891  			}
   892  		}
   893  	case boolean:
   894  		if digits > 1 {
   895  			rst.fail("boolean value too large")
   896  		} else if val == 0 {
   897  			rst.writeString("false")
   898  		} else if val == 1 {
   899  			rst.writeString("true")
   900  		} else {
   901  			rst.fail("invalid boolean value")
   902  		}
   903  	case character:
   904  		if digits > 6 {
   905  			rst.fail("character value too large")
   906  		}
   907  		rst.writeByte('\'')
   908  		if val == '\t' {
   909  			rst.writeString(`\t`)
   910  		} else if val == '\r' {
   911  			rst.writeString(`\r`)
   912  		} else if val == '\n' {
   913  			rst.writeString(`\n`)
   914  		} else if val == '\\' {
   915  			rst.writeString(`\\`)
   916  		} else if val == '\'' {
   917  			rst.writeString(`\'`)
   918  		} else if val >= ' ' && val <= '~' {
   919  			// printable ASCII character
   920  			rst.writeByte(byte(val))
   921  		} else {
   922  			if !rst.skip {
   923  				fmt.Fprintf(&rst.buf, `\u{%x}`, val)
   924  				rst.last = '}'
   925  			}
   926  		}
   927  		rst.writeByte('\'')
   928  	default:
   929  		panic("internal error")
   930  	}
   931  }
   932  
   933  // base62Number parses:
   934  //
   935  //	<base-62-number> = {<0-9a-zA-Z>} "_"
   936  func (rst *rustState) base62Number() int64 {
   937  	if len(rst.str) > 0 && rst.str[0] == '_' {
   938  		rst.advance(1)
   939  		return 0
   940  	}
   941  	val := int64(0)
   942  	for len(rst.str) > 0 {
   943  		c := rst.str[0]
   944  		rst.advance(1)
   945  		if c == '_' {
   946  			return val + 1
   947  		}
   948  		val *= 62
   949  		if c >= '0' && c <= '9' {
   950  			val += int64(c - '0')
   951  		} else if c >= 'a' && c <= 'z' {
   952  			val += int64(c - 'a' + 10)
   953  		} else if c >= 'A' && c <= 'Z' {
   954  			val += int64(c - 'A' + 36)
   955  		} else {
   956  			rst.fail("invalid digit in base 62 number")
   957  		}
   958  	}
   959  	rst.fail("expected _ after base 62 number")
   960  	return 0
   961  }
   962  
   963  // backref parses:
   964  //
   965  //	<backref> = "B" <base-62-number>
   966  func (rst *rustState) backref(demangle func()) {
   967  	backoff := rst.off
   968  
   969  	rst.checkChar('B')
   970  	idx64 := rst.base62Number()
   971  
   972  	if rst.skip {
   973  		return
   974  	}
   975  	if rst.max > 0 && rst.buf.Len() > rst.max {
   976  		return
   977  	}
   978  
   979  	idx := int(idx64)
   980  	if int64(idx) != idx64 {
   981  		rst.fail("backref index overflow")
   982  	}
   983  	if idx < 0 || idx >= backoff {
   984  		rst.fail("invalid backref index")
   985  	}
   986  
   987  	holdStr := rst.str
   988  	holdOff := rst.off
   989  	rst.str = rst.orig[idx:backoff]
   990  	rst.off = idx
   991  	defer func() {
   992  		rst.str = holdStr
   993  		rst.off = holdOff
   994  	}()
   995  
   996  	demangle()
   997  }
   998  
   999  func (rst *rustState) decimalNumber() int {
  1000  	if len(rst.str) == 0 {
  1001  		rst.fail("expected number")
  1002  	}
  1003  
  1004  	val := 0
  1005  	for len(rst.str) > 0 && isDigit(rst.str[0]) {
  1006  		add := int(rst.str[0] - '0')
  1007  		if val >= math.MaxInt32/10-add {
  1008  			rst.fail("decimal number overflow")
  1009  		}
  1010  		val *= 10
  1011  		val += add
  1012  		rst.advance(1)
  1013  	}
  1014  	return val
  1015  }
  1016  
  1017  // oldRustToString demangles a Rust symbol using the old demangling.
  1018  // The second result reports whether this is a valid Rust mangled name.
  1019  func oldRustToString(name string, options []Option) (string, bool) {
  1020  	max := 0
  1021  	for _, o := range options {
  1022  		if isMaxLength(o) {
  1023  			max = maxLength(o)
  1024  		}
  1025  	}
  1026  
  1027  	// We know that the string starts with _ZN.
  1028  	name = name[3:]
  1029  
  1030  	hexDigit := func(c byte) (byte, bool) {
  1031  		switch {
  1032  		case c >= '0' && c <= '9':
  1033  			return c - '0', true
  1034  		case c >= 'a' && c <= 'f':
  1035  			return c - 'a' + 10, true
  1036  		default:
  1037  			return 0, false
  1038  		}
  1039  	}
  1040  
  1041  	// We know that the strings end with "17h" followed by 16 characters
  1042  	// followed by "E". We check that the 16 characters are all hex digits.
  1043  	// Also the hex digits must contain at least 5 distinct digits.
  1044  	seen := uint16(0)
  1045  	for i := len(name) - 17; i < len(name)-1; i++ {
  1046  		digit, ok := hexDigit(name[i])
  1047  		if !ok {
  1048  			return "", false
  1049  		}
  1050  		seen |= 1 << digit
  1051  	}
  1052  	if bits.OnesCount16(seen) < 5 {
  1053  		return "", false
  1054  	}
  1055  	name = name[:len(name)-20]
  1056  
  1057  	// The name is a sequence of length-preceded identifiers.
  1058  	var sb strings.Builder
  1059  	for len(name) > 0 {
  1060  		if max > 0 && sb.Len() > max {
  1061  			break
  1062  		}
  1063  
  1064  		if !isDigit(name[0]) {
  1065  			return "", false
  1066  		}
  1067  
  1068  		val := 0
  1069  		for len(name) > 0 && isDigit(name[0]) {
  1070  			add := int(name[0] - '0')
  1071  			if val >= math.MaxInt32/10-add {
  1072  				return "", false
  1073  			}
  1074  			val *= 10
  1075  			val += add
  1076  			name = name[1:]
  1077  		}
  1078  
  1079  		// An optional trailing underscore can separate the
  1080  		// length from the identifier.
  1081  		if len(name) > 0 && name[0] == '_' {
  1082  			name = name[1:]
  1083  			val--
  1084  		}
  1085  
  1086  		if len(name) < val {
  1087  			return "", false
  1088  		}
  1089  
  1090  		id := name[:val]
  1091  		name = name[val:]
  1092  
  1093  		if sb.Len() > 0 {
  1094  			sb.WriteString("::")
  1095  		}
  1096  
  1097  		// Ignore leading underscores preceding escape sequences.
  1098  		if strings.HasPrefix(id, "_$") {
  1099  			id = id[1:]
  1100  		}
  1101  
  1102  		// The identifier can have escape sequences.
  1103  	escape:
  1104  		for len(id) > 0 {
  1105  			switch c := id[0]; c {
  1106  			case '$':
  1107  				codes := map[string]byte{
  1108  					"SP": '@',
  1109  					"BP": '*',
  1110  					"RF": '&',
  1111  					"LT": '<',
  1112  					"GT": '>',
  1113  					"LP": '(',
  1114  					"RP": ')',
  1115  				}
  1116  
  1117  				valid := true
  1118  				if len(id) > 2 && id[1] == 'C' && id[2] == '$' {
  1119  					sb.WriteByte(',')
  1120  					id = id[3:]
  1121  				} else if len(id) > 4 && id[1] == 'u' && id[4] == '$' {
  1122  					dig1, ok1 := hexDigit(id[2])
  1123  					dig2, ok2 := hexDigit(id[3])
  1124  					val := (dig1 << 4) | dig2
  1125  					if !ok1 || !ok2 || dig1 > 7 || val < ' ' {
  1126  						valid = false
  1127  					} else {
  1128  						sb.WriteByte(val)
  1129  						id = id[5:]
  1130  					}
  1131  				} else if len(id) > 3 && id[3] == '$' {
  1132  					if code, ok := codes[id[1:3]]; !ok {
  1133  						valid = false
  1134  					} else {
  1135  						sb.WriteByte(code)
  1136  						id = id[4:]
  1137  					}
  1138  				} else {
  1139  					valid = false
  1140  				}
  1141  				if !valid {
  1142  					sb.WriteString(id)
  1143  					break escape
  1144  				}
  1145  			case '.':
  1146  				if strings.HasPrefix(id, "..") {
  1147  					sb.WriteString("::")
  1148  					id = id[2:]
  1149  				} else {
  1150  					sb.WriteByte(c)
  1151  					id = id[1:]
  1152  				}
  1153  			default:
  1154  				sb.WriteByte(c)
  1155  				id = id[1:]
  1156  			}
  1157  		}
  1158  	}
  1159  
  1160  	s := sb.String()
  1161  	if max > 0 && len(s) > max {
  1162  		s = s[:max]
  1163  	}
  1164  	return s, true
  1165  }
  1166
View as plain text