...

Source file src/encoding/asn1/asn1.go

Documentation: encoding/asn1

     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 asn1 implements parsing of DER-encoded ASN.1 data structures,
     6  // as defined in ITU-T Rec X.690.
     7  //
     8  // See also “A Layman's Guide to a Subset of ASN.1, BER, and DER,”
     9  // http://luca.ntop.org/Teaching/Appunti/asn1.html.
    10  package asn1
    11  
    12  // ASN.1 is a syntax for specifying abstract objects and BER, DER, PER, XER etc
    13  // are different encoding formats for those objects. Here, we'll be dealing
    14  // with DER, the Distinguished Encoding Rules. DER is used in X.509 because
    15  // it's fast to parse and, unlike BER, has a unique encoding for every object.
    16  // When calculating hashes over objects, it's important that the resulting
    17  // bytes be the same at both ends and DER removes this margin of error.
    18  //
    19  // ASN.1 is very complex and this package doesn't attempt to implement
    20  // everything by any means.
    21  
    22  import (
    23  	"errors"
    24  	"fmt"
    25  	"math"
    26  	"math/big"
    27  	"reflect"
    28  	"strconv"
    29  	"strings"
    30  	"time"
    31  	"unicode/utf16"
    32  	"unicode/utf8"
    33  )
    34  
    35  // A StructuralError suggests that the ASN.1 data is valid, but the Go type
    36  // which is receiving it doesn't match.
    37  type StructuralError struct {
    38  	Msg string
    39  }
    40  
    41  func (e StructuralError) Error() string { return "asn1: structure error: " + e.Msg }
    42  
    43  // A SyntaxError suggests that the ASN.1 data is invalid.
    44  type SyntaxError struct {
    45  	Msg string
    46  }
    47  
    48  func (e SyntaxError) Error() string { return "asn1: syntax error: " + e.Msg }
    49  
    50  // We start by dealing with each of the primitive types in turn.
    51  
    52  // BOOLEAN
    53  
    54  func parseBool(bytes []byte) (ret bool, err error) {
    55  	if len(bytes) != 1 {
    56  		err = SyntaxError{"invalid boolean"}
    57  		return
    58  	}
    59  
    60  	// DER demands that "If the encoding represents the boolean value TRUE,
    61  	// its single contents octet shall have all eight bits set to one."
    62  	// Thus only 0 and 255 are valid encoded values.
    63  	switch bytes[0] {
    64  	case 0:
    65  		ret = false
    66  	case 0xff:
    67  		ret = true
    68  	default:
    69  		err = SyntaxError{"invalid boolean"}
    70  	}
    71  
    72  	return
    73  }
    74  
    75  // INTEGER
    76  
    77  // checkInteger returns nil if the given bytes are a valid DER-encoded
    78  // INTEGER and an error otherwise.
    79  func checkInteger(bytes []byte) error {
    80  	if len(bytes) == 0 {
    81  		return StructuralError{"empty integer"}
    82  	}
    83  	if len(bytes) == 1 {
    84  		return nil
    85  	}
    86  	if (bytes[0] == 0 && bytes[1]&0x80 == 0) || (bytes[0] == 0xff && bytes[1]&0x80 == 0x80) {
    87  		return StructuralError{"integer not minimally-encoded"}
    88  	}
    89  	return nil
    90  }
    91  
    92  // parseInt64 treats the given bytes as a big-endian, signed integer and
    93  // returns the result.
    94  func parseInt64(bytes []byte) (ret int64, err error) {
    95  	err = checkInteger(bytes)
    96  	if err != nil {
    97  		return
    98  	}
    99  	if len(bytes) > 8 {
   100  		// We'll overflow an int64 in this case.
   101  		err = StructuralError{"integer too large"}
   102  		return
   103  	}
   104  	for bytesRead := 0; bytesRead < len(bytes); bytesRead++ {
   105  		ret <<= 8
   106  		ret |= int64(bytes[bytesRead])
   107  	}
   108  
   109  	// Shift up and down in order to sign extend the result.
   110  	ret <<= 64 - uint8(len(bytes))*8
   111  	ret >>= 64 - uint8(len(bytes))*8
   112  	return
   113  }
   114  
   115  // parseInt32 treats the given bytes as a big-endian, signed integer and returns
   116  // the result.
   117  func parseInt32(bytes []byte) (int32, error) {
   118  	if err := checkInteger(bytes); err != nil {
   119  		return 0, err
   120  	}
   121  	ret64, err := parseInt64(bytes)
   122  	if err != nil {
   123  		return 0, err
   124  	}
   125  	if ret64 != int64(int32(ret64)) {
   126  		return 0, StructuralError{"integer too large"}
   127  	}
   128  	return int32(ret64), nil
   129  }
   130  
   131  var bigOne = big.NewInt(1)
   132  
   133  // parseBigInt treats the given bytes as a big-endian, signed integer and returns
   134  // the result.
   135  func parseBigInt(bytes []byte) (*big.Int, error) {
   136  	if err := checkInteger(bytes); err != nil {
   137  		return nil, err
   138  	}
   139  	ret := new(big.Int)
   140  	if len(bytes) > 0 && bytes[0]&0x80 == 0x80 {
   141  		// This is a negative number.
   142  		notBytes := make([]byte, len(bytes))
   143  		for i := range notBytes {
   144  			notBytes[i] = ^bytes[i]
   145  		}
   146  		ret.SetBytes(notBytes)
   147  		ret.Add(ret, bigOne)
   148  		ret.Neg(ret)
   149  		return ret, nil
   150  	}
   151  	ret.SetBytes(bytes)
   152  	return ret, nil
   153  }
   154  
   155  // BIT STRING
   156  
   157  // BitString is the structure to use when you want an ASN.1 BIT STRING type. A
   158  // bit string is padded up to the nearest byte in memory and the number of
   159  // valid bits is recorded. Padding bits will be zero.
   160  type BitString struct {
   161  	Bytes     []byte // bits packed into bytes.
   162  	BitLength int    // length in bits.
   163  }
   164  
   165  // At returns the bit at the given index. If the index is out of range it
   166  // returns 0.
   167  func (b BitString) At(i int) int {
   168  	if i < 0 || i >= b.BitLength {
   169  		return 0
   170  	}
   171  	x := i / 8
   172  	y := 7 - uint(i%8)
   173  	return int(b.Bytes[x]>>y) & 1
   174  }
   175  
   176  // RightAlign returns a slice where the padding bits are at the beginning. The
   177  // slice may share memory with the BitString.
   178  func (b BitString) RightAlign() []byte {
   179  	shift := uint(8 - (b.BitLength % 8))
   180  	if shift == 8 || len(b.Bytes) == 0 {
   181  		return b.Bytes
   182  	}
   183  
   184  	a := make([]byte, len(b.Bytes))
   185  	a[0] = b.Bytes[0] >> shift
   186  	for i := 1; i < len(b.Bytes); i++ {
   187  		a[i] = b.Bytes[i-1] << (8 - shift)
   188  		a[i] |= b.Bytes[i] >> shift
   189  	}
   190  
   191  	return a
   192  }
   193  
   194  // parseBitString parses an ASN.1 bit string from the given byte slice and returns it.
   195  func parseBitString(bytes []byte) (ret BitString, err error) {
   196  	if len(bytes) == 0 {
   197  		err = SyntaxError{"zero length BIT STRING"}
   198  		return
   199  	}
   200  	paddingBits := int(bytes[0])
   201  	if paddingBits > 7 ||
   202  		len(bytes) == 1 && paddingBits > 0 ||
   203  		bytes[len(bytes)-1]&((1<<bytes[0])-1) != 0 {
   204  		err = SyntaxError{"invalid padding bits in BIT STRING"}
   205  		return
   206  	}
   207  	ret.BitLength = (len(bytes)-1)*8 - paddingBits
   208  	ret.Bytes = bytes[1:]
   209  	return
   210  }
   211  
   212  // NULL
   213  
   214  // NullRawValue is a [RawValue] with its Tag set to the ASN.1 NULL type tag (5).
   215  var NullRawValue = RawValue{Tag: TagNull}
   216  
   217  // NullBytes contains bytes representing the DER-encoded ASN.1 NULL type.
   218  var NullBytes = []byte{TagNull, 0}
   219  
   220  // OBJECT IDENTIFIER
   221  
   222  // An ObjectIdentifier represents an ASN.1 OBJECT IDENTIFIER.
   223  type ObjectIdentifier []int
   224  
   225  // Equal reports whether oi and other represent the same identifier.
   226  func (oi ObjectIdentifier) Equal(other ObjectIdentifier) bool {
   227  	if len(oi) != len(other) {
   228  		return false
   229  	}
   230  	for i := 0; i < len(oi); i++ {
   231  		if oi[i] != other[i] {
   232  			return false
   233  		}
   234  	}
   235  
   236  	return true
   237  }
   238  
   239  func (oi ObjectIdentifier) String() string {
   240  	var s strings.Builder
   241  	s.Grow(32)
   242  
   243  	buf := make([]byte, 0, 19)
   244  	for i, v := range oi {
   245  		if i > 0 {
   246  			s.WriteByte('.')
   247  		}
   248  		s.Write(strconv.AppendInt(buf, int64(v), 10))
   249  	}
   250  
   251  	return s.String()
   252  }
   253  
   254  // parseObjectIdentifier parses an OBJECT IDENTIFIER from the given bytes and
   255  // returns it. An object identifier is a sequence of variable length integers
   256  // that are assigned in a hierarchy.
   257  func parseObjectIdentifier(bytes []byte) (s ObjectIdentifier, err error) {
   258  	if len(bytes) == 0 {
   259  		err = SyntaxError{"zero length OBJECT IDENTIFIER"}
   260  		return
   261  	}
   262  
   263  	// In the worst case, we get two elements from the first byte (which is
   264  	// encoded differently) and then every varint is a single byte long.
   265  	s = make([]int, len(bytes)+1)
   266  
   267  	// The first varint is 40*value1 + value2:
   268  	// According to this packing, value1 can take the values 0, 1 and 2 only.
   269  	// When value1 = 0 or value1 = 1, then value2 is <= 39. When value1 = 2,
   270  	// then there are no restrictions on value2.
   271  	v, offset, err := parseBase128Int(bytes, 0)
   272  	if err != nil {
   273  		return
   274  	}
   275  	if v < 80 {
   276  		s[0] = v / 40
   277  		s[1] = v % 40
   278  	} else {
   279  		s[0] = 2
   280  		s[1] = v - 80
   281  	}
   282  
   283  	i := 2
   284  	for ; offset < len(bytes); i++ {
   285  		v, offset, err = parseBase128Int(bytes, offset)
   286  		if err != nil {
   287  			return
   288  		}
   289  		s[i] = v
   290  	}
   291  	s = s[0:i]
   292  	return
   293  }
   294  
   295  // ENUMERATED
   296  
   297  // An Enumerated is represented as a plain int.
   298  type Enumerated int
   299  
   300  // FLAG
   301  
   302  // A Flag accepts any data and is set to true if present.
   303  type Flag bool
   304  
   305  // parseBase128Int parses a base-128 encoded int from the given offset in the
   306  // given byte slice. It returns the value and the new offset.
   307  func parseBase128Int(bytes []byte, initOffset int) (ret, offset int, err error) {
   308  	offset = initOffset
   309  	var ret64 int64
   310  	for shifted := 0; offset < len(bytes); shifted++ {
   311  		// 5 * 7 bits per byte == 35 bits of data
   312  		// Thus the representation is either non-minimal or too large for an int32
   313  		if shifted == 5 {
   314  			err = StructuralError{"base 128 integer too large"}
   315  			return
   316  		}
   317  		ret64 <<= 7
   318  		b := bytes[offset]
   319  		// integers should be minimally encoded, so the leading octet should
   320  		// never be 0x80
   321  		if shifted == 0 && b == 0x80 {
   322  			err = SyntaxError{"integer is not minimally encoded"}
   323  			return
   324  		}
   325  		ret64 |= int64(b & 0x7f)
   326  		offset++
   327  		if b&0x80 == 0 {
   328  			ret = int(ret64)
   329  			// Ensure that the returned value fits in an int on all platforms
   330  			if ret64 > math.MaxInt32 {
   331  				err = StructuralError{"base 128 integer too large"}
   332  			}
   333  			return
   334  		}
   335  	}
   336  	err = SyntaxError{"truncated base 128 integer"}
   337  	return
   338  }
   339  
   340  // UTCTime
   341  
   342  func parseUTCTime(bytes []byte) (ret time.Time, err error) {
   343  	s := string(bytes)
   344  
   345  	formatStr := "0601021504Z0700"
   346  	ret, err = time.Parse(formatStr, s)
   347  	if err != nil {
   348  		formatStr = "060102150405Z0700"
   349  		ret, err = time.Parse(formatStr, s)
   350  	}
   351  	if err != nil {
   352  		return
   353  	}
   354  
   355  	if serialized := ret.Format(formatStr); serialized != s {
   356  		err = fmt.Errorf("asn1: time did not serialize back to the original value and may be invalid: given %q, but serialized as %q", s, serialized)
   357  		return
   358  	}
   359  
   360  	if ret.Year() >= 2050 {
   361  		// UTCTime only encodes times prior to 2050. See https://tools.ietf.org/html/rfc5280#section-4.1.2.5.1
   362  		ret = ret.AddDate(-100, 0, 0)
   363  	}
   364  
   365  	return
   366  }
   367  
   368  // parseGeneralizedTime parses the GeneralizedTime from the given byte slice
   369  // and returns the resulting time.
   370  func parseGeneralizedTime(bytes []byte) (ret time.Time, err error) {
   371  	const formatStr = "20060102150405.999999999Z0700"
   372  	s := string(bytes)
   373  
   374  	if ret, err = time.Parse(formatStr, s); err != nil {
   375  		return
   376  	}
   377  
   378  	if serialized := ret.Format(formatStr); serialized != s {
   379  		err = fmt.Errorf("asn1: time did not serialize back to the original value and may be invalid: given %q, but serialized as %q", s, serialized)
   380  	}
   381  
   382  	return
   383  }
   384  
   385  // NumericString
   386  
   387  // parseNumericString parses an ASN.1 NumericString from the given byte array
   388  // and returns it.
   389  func parseNumericString(bytes []byte) (ret string, err error) {
   390  	for _, b := range bytes {
   391  		if !isNumeric(b) {
   392  			return "", SyntaxError{"NumericString contains invalid character"}
   393  		}
   394  	}
   395  	return string(bytes), nil
   396  }
   397  
   398  // isNumeric reports whether the given b is in the ASN.1 NumericString set.
   399  func isNumeric(b byte) bool {
   400  	return '0' <= b && b <= '9' ||
   401  		b == ' '
   402  }
   403  
   404  // PrintableString
   405  
   406  // parsePrintableString parses an ASN.1 PrintableString from the given byte
   407  // array and returns it.
   408  func parsePrintableString(bytes []byte) (ret string, err error) {
   409  	for _, b := range bytes {
   410  		if !isPrintable(b, allowAsterisk, allowAmpersand) {
   411  			err = SyntaxError{"PrintableString contains invalid character"}
   412  			return
   413  		}
   414  	}
   415  	ret = string(bytes)
   416  	return
   417  }
   418  
   419  type asteriskFlag bool
   420  type ampersandFlag bool
   421  
   422  const (
   423  	allowAsterisk  asteriskFlag = true
   424  	rejectAsterisk asteriskFlag = false
   425  
   426  	allowAmpersand  ampersandFlag = true
   427  	rejectAmpersand ampersandFlag = false
   428  )
   429  
   430  // isPrintable reports whether the given b is in the ASN.1 PrintableString set.
   431  // If asterisk is allowAsterisk then '*' is also allowed, reflecting existing
   432  // practice. If ampersand is allowAmpersand then '&' is allowed as well.
   433  func isPrintable(b byte, asterisk asteriskFlag, ampersand ampersandFlag) bool {
   434  	return 'a' <= b && b <= 'z' ||
   435  		'A' <= b && b <= 'Z' ||
   436  		'0' <= b && b <= '9' ||
   437  		'\'' <= b && b <= ')' ||
   438  		'+' <= b && b <= '/' ||
   439  		b == ' ' ||
   440  		b == ':' ||
   441  		b == '=' ||
   442  		b == '?' ||
   443  		// This is technically not allowed in a PrintableString.
   444  		// However, x509 certificates with wildcard strings don't
   445  		// always use the correct string type so we permit it.
   446  		(bool(asterisk) && b == '*') ||
   447  		// This is not technically allowed either. However, not
   448  		// only is it relatively common, but there are also a
   449  		// handful of CA certificates that contain it. At least
   450  		// one of which will not expire until 2027.
   451  		(bool(ampersand) && b == '&')
   452  }
   453  
   454  // IA5String
   455  
   456  // parseIA5String parses an ASN.1 IA5String (ASCII string) from the given
   457  // byte slice and returns it.
   458  func parseIA5String(bytes []byte) (ret string, err error) {
   459  	for _, b := range bytes {
   460  		if b >= utf8.RuneSelf {
   461  			err = SyntaxError{"IA5String contains invalid character"}
   462  			return
   463  		}
   464  	}
   465  	ret = string(bytes)
   466  	return
   467  }
   468  
   469  // T61String
   470  
   471  // parseT61String parses an ASN.1 T61String (8-bit clean string) from the given
   472  // byte slice and returns it.
   473  func parseT61String(bytes []byte) (ret string, err error) {
   474  	return string(bytes), nil
   475  }
   476  
   477  // UTF8String
   478  
   479  // parseUTF8String parses an ASN.1 UTF8String (raw UTF-8) from the given byte
   480  // array and returns it.
   481  func parseUTF8String(bytes []byte) (ret string, err error) {
   482  	if !utf8.Valid(bytes) {
   483  		return "", errors.New("asn1: invalid UTF-8 string")
   484  	}
   485  	return string(bytes), nil
   486  }
   487  
   488  // BMPString
   489  
   490  // parseBMPString parses an ASN.1 BMPString (Basic Multilingual Plane of
   491  // ISO/IEC/ITU 10646-1) from the given byte slice and returns it.
   492  func parseBMPString(bmpString []byte) (string, error) {
   493  	if len(bmpString)%2 != 0 {
   494  		return "", errors.New("pkcs12: odd-length BMP string")
   495  	}
   496  
   497  	// Strip terminator if present.
   498  	if l := len(bmpString); l >= 2 && bmpString[l-1] == 0 && bmpString[l-2] == 0 {
   499  		bmpString = bmpString[:l-2]
   500  	}
   501  
   502  	s := make([]uint16, 0, len(bmpString)/2)
   503  	for len(bmpString) > 0 {
   504  		s = append(s, uint16(bmpString[0])<<8+uint16(bmpString[1]))
   505  		bmpString = bmpString[2:]
   506  	}
   507  
   508  	return string(utf16.Decode(s)), nil
   509  }
   510  
   511  // A RawValue represents an undecoded ASN.1 object.
   512  type RawValue struct {
   513  	Class, Tag int
   514  	IsCompound bool
   515  	Bytes      []byte
   516  	FullBytes  []byte // includes the tag and length
   517  }
   518  
   519  // RawContent is used to signal that the undecoded, DER data needs to be
   520  // preserved for a struct. To use it, the first field of the struct must have
   521  // this type. It's an error for any of the other fields to have this type.
   522  type RawContent []byte
   523  
   524  // Tagging
   525  
   526  // parseTagAndLength parses an ASN.1 tag and length pair from the given offset
   527  // into a byte slice. It returns the parsed data and the new offset. SET and
   528  // SET OF (tag 17) are mapped to SEQUENCE and SEQUENCE OF (tag 16) since we
   529  // don't distinguish between ordered and unordered objects in this code.
   530  func parseTagAndLength(bytes []byte, initOffset int) (ret tagAndLength, offset int, err error) {
   531  	offset = initOffset
   532  	// parseTagAndLength should not be called without at least a single
   533  	// byte to read. Thus this check is for robustness:
   534  	if offset >= len(bytes) {
   535  		err = errors.New("asn1: internal error in parseTagAndLength")
   536  		return
   537  	}
   538  	b := bytes[offset]
   539  	offset++
   540  	ret.class = int(b >> 6)
   541  	ret.isCompound = b&0x20 == 0x20
   542  	ret.tag = int(b & 0x1f)
   543  
   544  	// If the bottom five bits are set, then the tag number is actually base 128
   545  	// encoded afterwards
   546  	if ret.tag == 0x1f {
   547  		ret.tag, offset, err = parseBase128Int(bytes, offset)
   548  		if err != nil {
   549  			return
   550  		}
   551  		// Tags should be encoded in minimal form.
   552  		if ret.tag < 0x1f {
   553  			err = SyntaxError{"non-minimal tag"}
   554  			return
   555  		}
   556  	}
   557  	if offset >= len(bytes) {
   558  		err = SyntaxError{"truncated tag or length"}
   559  		return
   560  	}
   561  	b = bytes[offset]
   562  	offset++
   563  	if b&0x80 == 0 {
   564  		// The length is encoded in the bottom 7 bits.
   565  		ret.length = int(b & 0x7f)
   566  	} else {
   567  		// Bottom 7 bits give the number of length bytes to follow.
   568  		numBytes := int(b & 0x7f)
   569  		if numBytes == 0 {
   570  			err = SyntaxError{"indefinite length found (not DER)"}
   571  			return
   572  		}
   573  		ret.length = 0
   574  		for i := 0; i < numBytes; i++ {
   575  			if offset >= len(bytes) {
   576  				err = SyntaxError{"truncated tag or length"}
   577  				return
   578  			}
   579  			b = bytes[offset]
   580  			offset++
   581  			if ret.length >= 1<<23 {
   582  				// We can't shift ret.length up without
   583  				// overflowing.
   584  				err = StructuralError{"length too large"}
   585  				return
   586  			}
   587  			ret.length <<= 8
   588  			ret.length |= int(b)
   589  			if ret.length == 0 {
   590  				// DER requires that lengths be minimal.
   591  				err = StructuralError{"superfluous leading zeros in length"}
   592  				return
   593  			}
   594  		}
   595  		// Short lengths must be encoded in short form.
   596  		if ret.length < 0x80 {
   597  			err = StructuralError{"non-minimal length"}
   598  			return
   599  		}
   600  	}
   601  
   602  	return
   603  }
   604  
   605  // parseSequenceOf is used for SEQUENCE OF and SET OF values. It tries to parse
   606  // a number of ASN.1 values from the given byte slice and returns them as a
   607  // slice of Go values of the given type.
   608  func parseSequenceOf(bytes []byte, sliceType reflect.Type, elemType reflect.Type) (ret reflect.Value, err error) {
   609  	matchAny, expectedTag, compoundType, ok := getUniversalType(elemType)
   610  	if !ok {
   611  		err = StructuralError{"unknown Go type for slice"}
   612  		return
   613  	}
   614  
   615  	// First we iterate over the input and count the number of elements,
   616  	// checking that the types are correct in each case.
   617  	numElements := 0
   618  	for offset := 0; offset < len(bytes); {
   619  		var t tagAndLength
   620  		t, offset, err = parseTagAndLength(bytes, offset)
   621  		if err != nil {
   622  			return
   623  		}
   624  		switch t.tag {
   625  		case TagIA5String, TagGeneralString, TagT61String, TagUTF8String, TagNumericString, TagBMPString:
   626  			// We pretend that various other string types are
   627  			// PRINTABLE STRINGs so that a sequence of them can be
   628  			// parsed into a []string.
   629  			t.tag = TagPrintableString
   630  		case TagGeneralizedTime, TagUTCTime:
   631  			// Likewise, both time types are treated the same.
   632  			t.tag = TagUTCTime
   633  		}
   634  
   635  		if !matchAny && (t.class != ClassUniversal || t.isCompound != compoundType || t.tag != expectedTag) {
   636  			err = StructuralError{"sequence tag mismatch"}
   637  			return
   638  		}
   639  		if invalidLength(offset, t.length, len(bytes)) {
   640  			err = SyntaxError{"truncated sequence"}
   641  			return
   642  		}
   643  		offset += t.length
   644  		numElements++
   645  	}
   646  	ret = reflect.MakeSlice(sliceType, numElements, numElements)
   647  	params := fieldParameters{}
   648  	offset := 0
   649  	for i := 0; i < numElements; i++ {
   650  		offset, err = parseField(ret.Index(i), bytes, offset, params)
   651  		if err != nil {
   652  			return
   653  		}
   654  	}
   655  	return
   656  }
   657  
   658  var (
   659  	bitStringType        = reflect.TypeFor[BitString]()
   660  	objectIdentifierType = reflect.TypeFor[ObjectIdentifier]()
   661  	enumeratedType       = reflect.TypeFor[Enumerated]()
   662  	flagType             = reflect.TypeFor[Flag]()
   663  	timeType             = reflect.TypeFor[time.Time]()
   664  	rawValueType         = reflect.TypeFor[RawValue]()
   665  	rawContentsType      = reflect.TypeFor[RawContent]()
   666  	bigIntType           = reflect.TypeFor[*big.Int]()
   667  )
   668  
   669  // invalidLength reports whether offset + length > sliceLength, or if the
   670  // addition would overflow.
   671  func invalidLength(offset, length, sliceLength int) bool {
   672  	return offset+length < offset || offset+length > sliceLength
   673  }
   674  
   675  // parseField is the main parsing function. Given a byte slice and an offset
   676  // into the array, it will try to parse a suitable ASN.1 value out and store it
   677  // in the given Value.
   678  func parseField(v reflect.Value, bytes []byte, initOffset int, params fieldParameters) (offset int, err error) {
   679  	offset = initOffset
   680  	fieldType := v.Type()
   681  
   682  	// If we have run out of data, it may be that there are optional elements at the end.
   683  	if offset == len(bytes) {
   684  		if !setDefaultValue(v, params) {
   685  			err = SyntaxError{"sequence truncated"}
   686  		}
   687  		return
   688  	}
   689  
   690  	// Deal with the ANY type.
   691  	if ifaceType := fieldType; ifaceType.Kind() == reflect.Interface && ifaceType.NumMethod() == 0 {
   692  		var t tagAndLength
   693  		t, offset, err = parseTagAndLength(bytes, offset)
   694  		if err != nil {
   695  			return
   696  		}
   697  		if invalidLength(offset, t.length, len(bytes)) {
   698  			err = SyntaxError{"data truncated"}
   699  			return
   700  		}
   701  		var result any
   702  		if !t.isCompound && t.class == ClassUniversal {
   703  			innerBytes := bytes[offset : offset+t.length]
   704  			switch t.tag {
   705  			case TagPrintableString:
   706  				result, err = parsePrintableString(innerBytes)
   707  			case TagNumericString:
   708  				result, err = parseNumericString(innerBytes)
   709  			case TagIA5String:
   710  				result, err = parseIA5String(innerBytes)
   711  			case TagT61String:
   712  				result, err = parseT61String(innerBytes)
   713  			case TagUTF8String:
   714  				result, err = parseUTF8String(innerBytes)
   715  			case TagInteger:
   716  				result, err = parseInt64(innerBytes)
   717  			case TagBitString:
   718  				result, err = parseBitString(innerBytes)
   719  			case TagOID:
   720  				result, err = parseObjectIdentifier(innerBytes)
   721  			case TagUTCTime:
   722  				result, err = parseUTCTime(innerBytes)
   723  			case TagGeneralizedTime:
   724  				result, err = parseGeneralizedTime(innerBytes)
   725  			case TagOctetString:
   726  				result = innerBytes
   727  			case TagBMPString:
   728  				result, err = parseBMPString(innerBytes)
   729  			default:
   730  				// If we don't know how to handle the type, we just leave Value as nil.
   731  			}
   732  		}
   733  		offset += t.length
   734  		if err != nil {
   735  			return
   736  		}
   737  		if result != nil {
   738  			v.Set(reflect.ValueOf(result))
   739  		}
   740  		return
   741  	}
   742  
   743  	t, offset, err := parseTagAndLength(bytes, offset)
   744  	if err != nil {
   745  		return
   746  	}
   747  	if params.explicit {
   748  		expectedClass := ClassContextSpecific
   749  		if params.application {
   750  			expectedClass = ClassApplication
   751  		}
   752  		if offset == len(bytes) {
   753  			err = StructuralError{"explicit tag has no child"}
   754  			return
   755  		}
   756  		if t.class == expectedClass && t.tag == *params.tag && (t.length == 0 || t.isCompound) {
   757  			if fieldType == rawValueType {
   758  				// The inner element should not be parsed for RawValues.
   759  			} else if t.length > 0 {
   760  				t, offset, err = parseTagAndLength(bytes, offset)
   761  				if err != nil {
   762  					return
   763  				}
   764  			} else {
   765  				if fieldType != flagType {
   766  					err = StructuralError{"zero length explicit tag was not an asn1.Flag"}
   767  					return
   768  				}
   769  				v.SetBool(true)
   770  				return
   771  			}
   772  		} else {
   773  			// The tags didn't match, it might be an optional element.
   774  			ok := setDefaultValue(v, params)
   775  			if ok {
   776  				offset = initOffset
   777  			} else {
   778  				err = StructuralError{"explicitly tagged member didn't match"}
   779  			}
   780  			return
   781  		}
   782  	}
   783  
   784  	matchAny, universalTag, compoundType, ok1 := getUniversalType(fieldType)
   785  	if !ok1 {
   786  		err = StructuralError{fmt.Sprintf("unknown Go type: %v", fieldType)}
   787  		return
   788  	}
   789  
   790  	// Special case for strings: all the ASN.1 string types map to the Go
   791  	// type string. getUniversalType returns the tag for PrintableString
   792  	// when it sees a string, so if we see a different string type on the
   793  	// wire, we change the universal type to match.
   794  	if universalTag == TagPrintableString {
   795  		if t.class == ClassUniversal {
   796  			switch t.tag {
   797  			case TagIA5String, TagGeneralString, TagT61String, TagUTF8String, TagNumericString, TagBMPString:
   798  				universalTag = t.tag
   799  			}
   800  		} else if params.stringType != 0 {
   801  			universalTag = params.stringType
   802  		}
   803  	}
   804  
   805  	// Special case for time: UTCTime and GeneralizedTime both map to the
   806  	// Go type time.Time.
   807  	if universalTag == TagUTCTime && t.tag == TagGeneralizedTime && t.class == ClassUniversal {
   808  		universalTag = TagGeneralizedTime
   809  	}
   810  
   811  	if params.set {
   812  		universalTag = TagSet
   813  	}
   814  
   815  	matchAnyClassAndTag := matchAny
   816  	expectedClass := ClassUniversal
   817  	expectedTag := universalTag
   818  
   819  	if !params.explicit && params.tag != nil {
   820  		expectedClass = ClassContextSpecific
   821  		expectedTag = *params.tag
   822  		matchAnyClassAndTag = false
   823  	}
   824  
   825  	if !params.explicit && params.application && params.tag != nil {
   826  		expectedClass = ClassApplication
   827  		expectedTag = *params.tag
   828  		matchAnyClassAndTag = false
   829  	}
   830  
   831  	if !params.explicit && params.private && params.tag != nil {
   832  		expectedClass = ClassPrivate
   833  		expectedTag = *params.tag
   834  		matchAnyClassAndTag = false
   835  	}
   836  
   837  	// We have unwrapped any explicit tagging at this point.
   838  	if !matchAnyClassAndTag && (t.class != expectedClass || t.tag != expectedTag) ||
   839  		(!matchAny && t.isCompound != compoundType) {
   840  		// Tags don't match. Again, it could be an optional element.
   841  		ok := setDefaultValue(v, params)
   842  		if ok {
   843  			offset = initOffset
   844  		} else {
   845  			err = StructuralError{fmt.Sprintf("tags don't match (%d vs %+v) %+v %s @%d", expectedTag, t, params, fieldType.Name(), offset)}
   846  		}
   847  		return
   848  	}
   849  	if invalidLength(offset, t.length, len(bytes)) {
   850  		err = SyntaxError{"data truncated"}
   851  		return
   852  	}
   853  	innerBytes := bytes[offset : offset+t.length]
   854  	offset += t.length
   855  
   856  	// We deal with the structures defined in this package first.
   857  	switch v := v.Addr().Interface().(type) {
   858  	case *RawValue:
   859  		*v = RawValue{t.class, t.tag, t.isCompound, innerBytes, bytes[initOffset:offset]}
   860  		return
   861  	case *ObjectIdentifier:
   862  		*v, err = parseObjectIdentifier(innerBytes)
   863  		return
   864  	case *BitString:
   865  		*v, err = parseBitString(innerBytes)
   866  		return
   867  	case *time.Time:
   868  		if universalTag == TagUTCTime {
   869  			*v, err = parseUTCTime(innerBytes)
   870  			return
   871  		}
   872  		*v, err = parseGeneralizedTime(innerBytes)
   873  		return
   874  	case *Enumerated:
   875  		parsedInt, err1 := parseInt32(innerBytes)
   876  		if err1 == nil {
   877  			*v = Enumerated(parsedInt)
   878  		}
   879  		err = err1
   880  		return
   881  	case *Flag:
   882  		*v = true
   883  		return
   884  	case **big.Int:
   885  		parsedInt, err1 := parseBigInt(innerBytes)
   886  		if err1 == nil {
   887  			*v = parsedInt
   888  		}
   889  		err = err1
   890  		return
   891  	}
   892  	switch val := v; val.Kind() {
   893  	case reflect.Bool:
   894  		parsedBool, err1 := parseBool(innerBytes)
   895  		if err1 == nil {
   896  			val.SetBool(parsedBool)
   897  		}
   898  		err = err1
   899  		return
   900  	case reflect.Int, reflect.Int32, reflect.Int64:
   901  		if val.Type().Size() == 4 {
   902  			parsedInt, err1 := parseInt32(innerBytes)
   903  			if err1 == nil {
   904  				val.SetInt(int64(parsedInt))
   905  			}
   906  			err = err1
   907  		} else {
   908  			parsedInt, err1 := parseInt64(innerBytes)
   909  			if err1 == nil {
   910  				val.SetInt(parsedInt)
   911  			}
   912  			err = err1
   913  		}
   914  		return
   915  	// TODO(dfc) Add support for the remaining integer types
   916  	case reflect.Struct:
   917  		structType := fieldType
   918  
   919  		for i := 0; i < structType.NumField(); i++ {
   920  			if !structType.Field(i).IsExported() {
   921  				err = StructuralError{"struct contains unexported fields"}
   922  				return
   923  			}
   924  		}
   925  
   926  		if structType.NumField() > 0 &&
   927  			structType.Field(0).Type == rawContentsType {
   928  			bytes := bytes[initOffset:offset]
   929  			val.Field(0).Set(reflect.ValueOf(RawContent(bytes)))
   930  		}
   931  
   932  		innerOffset := 0
   933  		for i := 0; i < structType.NumField(); i++ {
   934  			field := structType.Field(i)
   935  			if i == 0 && field.Type == rawContentsType {
   936  				continue
   937  			}
   938  			innerOffset, err = parseField(val.Field(i), innerBytes, innerOffset, parseFieldParameters(field.Tag.Get("asn1")))
   939  			if err != nil {
   940  				return
   941  			}
   942  		}
   943  		// We allow extra bytes at the end of the SEQUENCE because
   944  		// adding elements to the end has been used in X.509 as the
   945  		// version numbers have increased.
   946  		return
   947  	case reflect.Slice:
   948  		sliceType := fieldType
   949  		if sliceType.Elem().Kind() == reflect.Uint8 {
   950  			val.Set(reflect.MakeSlice(sliceType, len(innerBytes), len(innerBytes)))
   951  			reflect.Copy(val, reflect.ValueOf(innerBytes))
   952  			return
   953  		}
   954  		newSlice, err1 := parseSequenceOf(innerBytes, sliceType, sliceType.Elem())
   955  		if err1 == nil {
   956  			val.Set(newSlice)
   957  		}
   958  		err = err1
   959  		return
   960  	case reflect.String:
   961  		var v string
   962  		switch universalTag {
   963  		case TagPrintableString:
   964  			v, err = parsePrintableString(innerBytes)
   965  		case TagNumericString:
   966  			v, err = parseNumericString(innerBytes)
   967  		case TagIA5String:
   968  			v, err = parseIA5String(innerBytes)
   969  		case TagT61String:
   970  			v, err = parseT61String(innerBytes)
   971  		case TagUTF8String:
   972  			v, err = parseUTF8String(innerBytes)
   973  		case TagGeneralString:
   974  			// GeneralString is specified in ISO-2022/ECMA-35,
   975  			// A brief review suggests that it includes structures
   976  			// that allow the encoding to change midstring and
   977  			// such. We give up and pass it as an 8-bit string.
   978  			v, err = parseT61String(innerBytes)
   979  		case TagBMPString:
   980  			v, err = parseBMPString(innerBytes)
   981  
   982  		default:
   983  			err = SyntaxError{fmt.Sprintf("internal error: unknown string type %d", universalTag)}
   984  		}
   985  		if err == nil {
   986  			val.SetString(v)
   987  		}
   988  		return
   989  	}
   990  	err = StructuralError{"unsupported: " + v.Type().String()}
   991  	return
   992  }
   993  
   994  // canHaveDefaultValue reports whether k is a Kind that we will set a default
   995  // value for. (A signed integer, essentially.)
   996  func canHaveDefaultValue(k reflect.Kind) bool {
   997  	switch k {
   998  	case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
   999  		return true
  1000  	}
  1001  
  1002  	return false
  1003  }
  1004  
  1005  // setDefaultValue is used to install a default value, from a tag string, into
  1006  // a Value. It is successful if the field was optional, even if a default value
  1007  // wasn't provided or it failed to install it into the Value.
  1008  func setDefaultValue(v reflect.Value, params fieldParameters) (ok bool) {
  1009  	if !params.optional {
  1010  		return
  1011  	}
  1012  	ok = true
  1013  	if params.defaultValue == nil {
  1014  		return
  1015  	}
  1016  	if canHaveDefaultValue(v.Kind()) {
  1017  		v.SetInt(*params.defaultValue)
  1018  	}
  1019  	return
  1020  }
  1021  
  1022  // Unmarshal parses the DER-encoded ASN.1 data structure b
  1023  // and uses the reflect package to fill in an arbitrary value pointed at by val.
  1024  // Because Unmarshal uses the reflect package, the structs
  1025  // being written to must use upper case field names. If val
  1026  // is nil or not a pointer, Unmarshal returns an error.
  1027  //
  1028  // After parsing b, any bytes that were leftover and not used to fill
  1029  // val will be returned in rest. When parsing a SEQUENCE into a struct,
  1030  // any trailing elements of the SEQUENCE that do not have matching
  1031  // fields in val will not be included in rest, as these are considered
  1032  // valid elements of the SEQUENCE and not trailing data.
  1033  //
  1034  //   - An ASN.1 INTEGER can be written to an int, int32, int64,
  1035  //     or *[big.Int].
  1036  //     If the encoded value does not fit in the Go type,
  1037  //     Unmarshal returns a parse error.
  1038  //
  1039  //   - An ASN.1 BIT STRING can be written to a [BitString].
  1040  //
  1041  //   - An ASN.1 OCTET STRING can be written to a []byte.
  1042  //
  1043  //   - An ASN.1 OBJECT IDENTIFIER can be written to an [ObjectIdentifier].
  1044  //
  1045  //   - An ASN.1 ENUMERATED can be written to an [Enumerated].
  1046  //
  1047  //   - An ASN.1 UTCTIME or GENERALIZEDTIME can be written to a [time.Time].
  1048  //
  1049  //   - An ASN.1 PrintableString, IA5String, or NumericString can be written to a string.
  1050  //
  1051  //   - Any of the above ASN.1 values can be written to an interface{}.
  1052  //     The value stored in the interface has the corresponding Go type.
  1053  //     For integers, that type is int64.
  1054  //
  1055  //   - An ASN.1 SEQUENCE OF x or SET OF x can be written
  1056  //     to a slice if an x can be written to the slice's element type.
  1057  //
  1058  //   - An ASN.1 SEQUENCE or SET can be written to a struct
  1059  //     if each of the elements in the sequence can be
  1060  //     written to the corresponding element in the struct.
  1061  //
  1062  // The following tags on struct fields have special meaning to Unmarshal:
  1063  //
  1064  //	application specifies that an APPLICATION tag is used
  1065  //	private     specifies that a PRIVATE tag is used
  1066  //	default:x   sets the default value for optional integer fields (only used if optional is also present)
  1067  //	explicit    specifies that an additional, explicit tag wraps the implicit one
  1068  //	optional    marks the field as ASN.1 OPTIONAL
  1069  //	set         causes a SET, rather than a SEQUENCE type to be expected
  1070  //	tag:x       specifies the ASN.1 tag number; implies ASN.1 CONTEXT SPECIFIC
  1071  //
  1072  // When decoding an ASN.1 value with an IMPLICIT tag into a string field,
  1073  // Unmarshal will default to a PrintableString, which doesn't support
  1074  // characters such as '@' and '&'. To force other encodings, use the following
  1075  // tags:
  1076  //
  1077  //	ia5     causes strings to be unmarshaled as ASN.1 IA5String values
  1078  //	numeric causes strings to be unmarshaled as ASN.1 NumericString values
  1079  //	utf8    causes strings to be unmarshaled as ASN.1 UTF8String values
  1080  //
  1081  // If the type of the first field of a structure is RawContent then the raw
  1082  // ASN1 contents of the struct will be stored in it.
  1083  //
  1084  // If the name of a slice type ends with "SET" then it's treated as if
  1085  // the "set" tag was set on it. This results in interpreting the type as a
  1086  // SET OF x rather than a SEQUENCE OF x. This can be used with nested slices
  1087  // where a struct tag cannot be given.
  1088  //
  1089  // Other ASN.1 types are not supported; if it encounters them,
  1090  // Unmarshal returns a parse error.
  1091  func Unmarshal(b []byte, val any) (rest []byte, err error) {
  1092  	return UnmarshalWithParams(b, val, "")
  1093  }
  1094  
  1095  // An invalidUnmarshalError describes an invalid argument passed to Unmarshal.
  1096  // (The argument to Unmarshal must be a non-nil pointer.)
  1097  type invalidUnmarshalError struct {
  1098  	Type reflect.Type
  1099  }
  1100  
  1101  func (e *invalidUnmarshalError) Error() string {
  1102  	if e.Type == nil {
  1103  		return "asn1: Unmarshal recipient value is nil"
  1104  	}
  1105  
  1106  	if e.Type.Kind() != reflect.Pointer {
  1107  		return "asn1: Unmarshal recipient value is non-pointer " + e.Type.String()
  1108  	}
  1109  	return "asn1: Unmarshal recipient value is nil " + e.Type.String()
  1110  }
  1111  
  1112  // UnmarshalWithParams allows field parameters to be specified for the
  1113  // top-level element. The form of the params is the same as the field tags.
  1114  func UnmarshalWithParams(b []byte, val any, params string) (rest []byte, err error) {
  1115  	v := reflect.ValueOf(val)
  1116  	if v.Kind() != reflect.Pointer || v.IsNil() {
  1117  		return nil, &invalidUnmarshalError{reflect.TypeOf(val)}
  1118  	}
  1119  	offset, err := parseField(v.Elem(), b, 0, parseFieldParameters(params))
  1120  	if err != nil {
  1121  		return nil, err
  1122  	}
  1123  	return b[offset:], nil
  1124  }
  1125  

View as plain text