Black Lives Matter. Support the Equal Justice Initiative.

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  	"time"
    30  	"unicode/utf16"
    31  	"unicode/utf8"
    32  )
    33  
    34  // A StructuralError suggests that the ASN.1 data is valid, but the Go type
    35  // which is receiving it doesn't match.
    36  type StructuralError struct {
    37  	Msg string
    38  }
    39  
    40  func (e StructuralError) Error() string { return "asn1: structure error: " + e.Msg }
    41  
    42  // A SyntaxError suggests that the ASN.1 data is invalid.
    43  type SyntaxError struct {
    44  	Msg string
    45  }
    46  
    47  func (e SyntaxError) Error() string { return "asn1: syntax error: " + e.Msg }
    48  
    49  // We start by dealing with each of the primitive types in turn.
    50  
    51  // BOOLEAN
    52  
    53  func parseBool(bytes []byte) (ret bool, err error) {
    54  	if len(bytes) != 1 {
    55  		err = SyntaxError{"invalid boolean"}
    56  		return
    57  	}
    58  
    59  	// DER demands that "If the encoding represents the boolean value TRUE,
    60  	// its single contents octet shall have all eight bits set to one."
    61  	// Thus only 0 and 255 are valid encoded values.
    62  	switch bytes[0] {
    63  	case 0:
    64  		ret = false
    65  	case 0xff:
    66  		ret = true
    67  	default:
    68  		err = SyntaxError{"invalid boolean"}
    69  	}
    70  
    71  	return
    72  }
    73  
    74  // INTEGER
    75  
    76  // checkInteger returns nil if the given bytes are a valid DER-encoded
    77  // INTEGER and an error otherwise.
    78  func checkInteger(bytes []byte) error {
    79  	if len(bytes) == 0 {
    80  		return StructuralError{"empty integer"}
    81  	}
    82  	if len(bytes) == 1 {
    83  		return nil
    84  	}
    85  	if (bytes[0] == 0 && bytes[1]&0x80 == 0) || (bytes[0] == 0xff && bytes[1]&0x80 == 0x80) {
    86  		return StructuralError{"integer not minimally-encoded"}
    87  	}
    88  	return nil
    89  }
    90  
    91  // parseInt64 treats the given bytes as a big-endian, signed integer and
    92  // returns the result.
    93  func parseInt64(bytes []byte) (ret int64, err error) {
    94  	err = checkInteger(bytes)
    95  	if err != nil {
    96  		return
    97  	}
    98  	if len(bytes) > 8 {
    99  		// We'll overflow an int64 in this case.
   100  		err = StructuralError{"integer too large"}
   101  		return
   102  	}
   103  	for bytesRead := 0; bytesRead < len(bytes); bytesRead++ {
   104  		ret <<= 8
   105  		ret |= int64(bytes[bytesRead])
   106  	}
   107  
   108  	// Shift up and down in order to sign extend the result.
   109  	ret <<= 64 - uint8(len(bytes))*8
   110  	ret >>= 64 - uint8(len(bytes))*8
   111  	return
   112  }
   113  
   114  // parseInt treats the given bytes as a big-endian, signed integer and returns
   115  // the result.
   116  func parseInt32(bytes []byte) (int32, error) {
   117  	if err := checkInteger(bytes); err != nil {
   118  		return 0, err
   119  	}
   120  	ret64, err := parseInt64(bytes)
   121  	if err != nil {
   122  		return 0, err
   123  	}
   124  	if ret64 != int64(int32(ret64)) {
   125  		return 0, StructuralError{"integer too large"}
   126  	}
   127  	return int32(ret64), nil
   128  }
   129  
   130  var bigOne = big.NewInt(1)
   131  
   132  // parseBigInt treats the given bytes as a big-endian, signed integer and returns
   133  // the result.
   134  func parseBigInt(bytes []byte) (*big.Int, error) {
   135  	if err := checkInteger(bytes); err != nil {
   136  		return nil, err
   137  	}
   138  	ret := new(big.Int)
   139  	if len(bytes) > 0 && bytes[0]&0x80 == 0x80 {
   140  		// This is a negative number.
   141  		notBytes := make([]byte, len(bytes))
   142  		for i := range notBytes {
   143  			notBytes[i] = ^bytes[i]
   144  		}
   145  		ret.SetBytes(notBytes)
   146  		ret.Add(ret, bigOne)
   147  		ret.Neg(ret)
   148  		return ret, nil
   149  	}
   150  	ret.SetBytes(bytes)
   151  	return ret, nil
   152  }
   153  
   154  // BIT STRING
   155  
   156  // BitString is the structure to use when you want an ASN.1 BIT STRING type. A
   157  // bit string is padded up to the nearest byte in memory and the number of
   158  // valid bits is recorded. Padding bits will be zero.
   159  type BitString struct {
   160  	Bytes     []byte // bits packed into bytes.
   161  	BitLength int    // length in bits.
   162  }
   163  
   164  // At returns the bit at the given index. If the index is out of range it
   165  // returns false.
   166  func (b BitString) At(i int) int {
   167  	if i < 0 || i >= b.BitLength {
   168  		return 0
   169  	}
   170  	x := i / 8
   171  	y := 7 - uint(i%8)
   172  	return int(b.Bytes[x]>>y) & 1
   173  }
   174  
   175  // RightAlign returns a slice where the padding bits are at the beginning. The
   176  // slice may share memory with the BitString.
   177  func (b BitString) RightAlign() []byte {
   178  	shift := uint(8 - (b.BitLength % 8))
   179  	if shift == 8 || len(b.Bytes) == 0 {
   180  		return b.Bytes
   181  	}
   182  
   183  	a := make([]byte, len(b.Bytes))
   184  	a[0] = b.Bytes[0] >> shift
   185  	for i := 1; i < len(b.Bytes); i++ {
   186  		a[i] = b.Bytes[i-1] << (8 - shift)
   187  		a[i] |= b.Bytes[i] >> shift
   188  	}
   189  
   190  	return a
   191  }
   192  
   193  // parseBitString parses an ASN.1 bit string from the given byte slice and returns it.
   194  func parseBitString(bytes []byte) (ret BitString, err error) {
   195  	if len(bytes) == 0 {
   196  		err = SyntaxError{"zero length BIT STRING"}
   197  		return
   198  	}
   199  	paddingBits := int(bytes[0])
   200  	if paddingBits > 7 ||
   201  		len(bytes) == 1 && paddingBits > 0 ||
   202  		bytes[len(bytes)-1]&((1<<bytes[0])-1) != 0 {
   203  		err = SyntaxError{"invalid padding bits in BIT STRING"}
   204  		return
   205  	}
   206  	ret.BitLength = (len(bytes)-1)*8 - paddingBits
   207  	ret.Bytes = bytes[1:]
   208  	return
   209  }
   210  
   211  // NULL
   212  
   213  // NullRawValue is a RawValue with its Tag set to the ASN.1 NULL type tag (5).
   214  var NullRawValue = RawValue{Tag: TagNull}
   215  
   216  // NullBytes contains bytes representing the DER-encoded ASN.1 NULL type.
   217  var NullBytes = []byte{TagNull, 0}
   218  
   219  // OBJECT IDENTIFIER
   220  
   221  // An ObjectIdentifier represents an ASN.1 OBJECT IDENTIFIER.
   222  type ObjectIdentifier []int
   223  
   224  // Equal reports whether oi and other represent the same identifier.
   225  func (oi ObjectIdentifier) Equal(other ObjectIdentifier) bool {
   226  	if len(oi) != len(other) {
   227  		return false
   228  	}
   229  	for i := 0; i < len(oi); i++ {
   230  		if oi[i] != other[i] {
   231  			return false
   232  		}
   233  	}
   234  
   235  	return true
   236  }
   237  
   238  func (oi ObjectIdentifier) String() string {
   239  	var s string
   240  
   241  	for i, v := range oi {
   242  		if i > 0 {
   243  			s += "."
   244  		}
   245  		s += strconv.Itoa(v)
   246  	}
   247  
   248  	return s
   249  }
   250  
   251  // parseObjectIdentifier parses an OBJECT IDENTIFIER from the given bytes and
   252  // returns it. An object identifier is a sequence of variable length integers
   253  // that are assigned in a hierarchy.
   254  func parseObjectIdentifier(bytes []byte) (s ObjectIdentifier, err error) {
   255  	if len(bytes) == 0 {
   256  		err = SyntaxError{"zero length OBJECT IDENTIFIER"}
   257  		return
   258  	}
   259  
   260  	// In the worst case, we get two elements from the first byte (which is
   261  	// encoded differently) and then every varint is a single byte long.
   262  	s = make([]int, len(bytes)+1)
   263  
   264  	// The first varint is 40*value1 + value2:
   265  	// According to this packing, value1 can take the values 0, 1 and 2 only.
   266  	// When value1 = 0 or value1 = 1, then value2 is <= 39. When value1 = 2,
   267  	// then there are no restrictions on value2.
   268  	v, offset, err := parseBase128Int(bytes, 0)
   269  	if err != nil {
   270  		return
   271  	}
   272  	if v < 80 {
   273  		s[0] = v / 40
   274  		s[1] = v % 40
   275  	} else {
   276  		s[0] = 2
   277  		s[1] = v - 80
   278  	}
   279  
   280  	i := 2
   281  	for ; offset < len(bytes); i++ {
   282  		v, offset, err = parseBase128Int(bytes, offset)
   283  		if err != nil {
   284  			return
   285  		}
   286  		s[i] = v
   287  	}
   288  	s = s[0:i]
   289  	return
   290  }
   291  
   292  // ENUMERATED
   293  
   294  // An Enumerated is represented as a plain int.
   295  type Enumerated int
   296  
   297  // FLAG
   298  
   299  // A Flag accepts any data and is set to true if present.
   300  type Flag bool
   301  
   302  // parseBase128Int parses a base-128 encoded int from the given offset in the
   303  // given byte slice. It returns the value and the new offset.
   304  func parseBase128Int(bytes []byte, initOffset int) (ret, offset int, err error) {
   305  	offset = initOffset
   306  	var ret64 int64
   307  	for shifted := 0; offset < len(bytes); shifted++ {
   308  		// 5 * 7 bits per byte == 35 bits of data
   309  		// Thus the representation is either non-minimal or too large for an int32
   310  		if shifted == 5 {
   311  			err = StructuralError{"base 128 integer too large"}
   312  			return
   313  		}
   314  		ret64 <<= 7
   315  		b := bytes[offset]
   316  		// integers should be minimally encoded, so the leading octet should
   317  		// never be 0x80
   318  		if shifted == 0 && b == 0x80 {
   319  			err = SyntaxError{"integer is not minimally encoded"}
   320  			return
   321  		}
   322  		ret64 |= int64(b & 0x7f)
   323  		offset++
   324  		if b&0x80 == 0 {
   325  			ret = int(ret64)
   326  			// Ensure that the returned value fits in an int on all platforms
   327  			if ret64 > math.MaxInt32 {
   328  				err = StructuralError{"base 128 integer too large"}
   329  			}
   330  			return
   331  		}
   332  	}
   333  	err = SyntaxError{"truncated base 128 integer"}
   334  	return
   335  }
   336  
   337  // UTCTime
   338  
   339  func parseUTCTime(bytes []byte) (ret time.Time, err error) {
   340  	s := string(bytes)
   341  
   342  	formatStr := "0601021504Z0700"
   343  	ret, err = time.Parse(formatStr, s)
   344  	if err != nil {
   345  		formatStr = "060102150405Z0700"
   346  		ret, err = time.Parse(formatStr, s)
   347  	}
   348  	if err != nil {
   349  		return
   350  	}
   351  
   352  	if serialized := ret.Format(formatStr); serialized != s {
   353  		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)
   354  		return
   355  	}
   356  
   357  	if ret.Year() >= 2050 {
   358  		// UTCTime only encodes times prior to 2050. See https://tools.ietf.org/html/rfc5280#section-4.1.2.5.1
   359  		ret = ret.AddDate(-100, 0, 0)
   360  	}
   361  
   362  	return
   363  }
   364  
   365  // parseGeneralizedTime parses the GeneralizedTime from the given byte slice
   366  // and returns the resulting time.
   367  func parseGeneralizedTime(bytes []byte) (ret time.Time, err error) {
   368  	const formatStr = "20060102150405Z0700"
   369  	s := string(bytes)
   370  
   371  	if ret, err = time.Parse(formatStr, s); err != nil {
   372  		return
   373  	}
   374  
   375  	if serialized := ret.Format(formatStr); serialized != s {
   376  		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)
   377  	}
   378  
   379  	return
   380  }
   381  
   382  // NumericString
   383  
   384  // parseNumericString parses an ASN.1 NumericString from the given byte array
   385  // and returns it.
   386  func parseNumericString(bytes []byte) (ret string, err error) {
   387  	for _, b := range bytes {
   388  		if !isNumeric(b) {
   389  			return "", SyntaxError{"NumericString contains invalid character"}
   390  		}
   391  	}
   392  	return string(bytes), nil
   393  }
   394  
   395  // isNumeric reports whether the given b is in the ASN.1 NumericString set.
   396  func isNumeric(b byte) bool {
   397  	return '0' <= b && b <= '9' ||
   398  		b == ' '
   399  }
   400  
   401  // PrintableString
   402  
   403  // parsePrintableString parses an ASN.1 PrintableString from the given byte
   404  // array and returns it.
   405  func parsePrintableString(bytes []byte) (ret string, err error) {
   406  	for _, b := range bytes {
   407  		if !isPrintable(b, allowAsterisk, allowAmpersand) {
   408  			err = SyntaxError{"PrintableString contains invalid character"}
   409  			return
   410  		}
   411  	}
   412  	ret = string(bytes)
   413  	return
   414  }
   415  
   416  type asteriskFlag bool
   417  type ampersandFlag bool
   418  
   419  const (
   420  	allowAsterisk  asteriskFlag = true
   421  	rejectAsterisk asteriskFlag = false
   422  
   423  	allowAmpersand  ampersandFlag = true
   424  	rejectAmpersand ampersandFlag = false
   425  )
   426  
   427  // isPrintable reports whether the given b is in the ASN.1 PrintableString set.
   428  // If asterisk is allowAsterisk then '*' is also allowed, reflecting existing
   429  // practice. If ampersand is allowAmpersand then '&' is allowed as well.
   430  func isPrintable(b byte, asterisk asteriskFlag, ampersand ampersandFlag) bool {
   431  	return 'a' <= b && b <= 'z' ||
   432  		'A' <= b && b <= 'Z' ||
   433  		'0' <= b && b <= '9' ||
   434  		'\'' <= b && b <= ')' ||
   435  		'+' <= b && b <= '/' ||
   436  		b == ' ' ||
   437  		b == ':' ||
   438  		b == '=' ||
   439  		b == '?' ||
   440  		// This is technically not allowed in a PrintableString.
   441  		// However, x509 certificates with wildcard strings don't
   442  		// always use the correct string type so we permit it.
   443  		(bool(asterisk) && b == '*') ||
   444  		// This is not technically allowed either. However, not
   445  		// only is it relatively common, but there are also a
   446  		// handful of CA certificates that contain it. At least
   447  		// one of which will not expire until 2027.
   448  		(bool(ampersand) && b == '&')
   449  }
   450  
   451  // IA5String
   452  
   453  // parseIA5String parses an ASN.1 IA5String (ASCII string) from the given
   454  // byte slice and returns it.
   455  func parseIA5String(bytes []byte) (ret string, err error) {
   456  	for _, b := range bytes {
   457  		if b >= utf8.RuneSelf {
   458  			err = SyntaxError{"IA5String contains invalid character"}
   459  			return
   460  		}
   461  	}
   462  	ret = string(bytes)
   463  	return
   464  }
   465  
   466  // T61String
   467  
   468  // parseT61String parses an ASN.1 T61String (8-bit clean string) from the given
   469  // byte slice and returns it.
   470  func parseT61String(bytes []byte) (ret string, err error) {
   471  	return string(bytes), nil
   472  }
   473  
   474  // UTF8String
   475  
   476  // parseUTF8String parses an ASN.1 UTF8String (raw UTF-8) from the given byte
   477  // array and returns it.
   478  func parseUTF8String(bytes []byte) (ret string, err error) {
   479  	if !utf8.Valid(bytes) {
   480  		return "", errors.New("asn1: invalid UTF-8 string")
   481  	}
   482  	return string(bytes), nil
   483  }
   484  
   485  // BMPString
   486  
   487  // parseBMPString parses an ASN.1 BMPString (Basic Multilingual Plane of
   488  // ISO/IEC/ITU 10646-1) from the given byte slice and returns it.
   489  func parseBMPString(bmpString []byte) (string, error) {
   490  	if len(bmpString)%2 != 0 {
   491  		return "", errors.New("pkcs12: odd-length BMP string")
   492  	}
   493  
   494  	// Strip terminator if present.
   495  	if l := len(bmpString); l >= 2 && bmpString[l-1] == 0 && bmpString[l-2] == 0 {
   496  		bmpString = bmpString[:l-2]
   497  	}
   498  
   499  	s := make([]uint16, 0, len(bmpString)/2)
   500  	for len(bmpString) > 0 {
   501  		s = append(s, uint16(bmpString[0])<<8+uint16(bmpString[1]))
   502  		bmpString = bmpString[2:]
   503  	}
   504  
   505  	return string(utf16.Decode(s)), nil
   506  }
   507  
   508  // A RawValue represents an undecoded ASN.1 object.
   509  type RawValue struct {
   510  	Class, Tag int
   511  	IsCompound bool
   512  	Bytes      []byte
   513  	FullBytes  []byte // includes the tag and length
   514  }
   515  
   516  // RawContent is used to signal that the undecoded, DER data needs to be
   517  // preserved for a struct. To use it, the first field of the struct must have
   518  // this type. It's an error for any of the other fields to have this type.
   519  type RawContent []byte
   520  
   521  // Tagging
   522  
   523  // parseTagAndLength parses an ASN.1 tag and length pair from the given offset
   524  // into a byte slice. It returns the parsed data and the new offset. SET and
   525  // SET OF (tag 17) are mapped to SEQUENCE and SEQUENCE OF (tag 16) since we
   526  // don't distinguish between ordered and unordered objects in this code.
   527  func parseTagAndLength(bytes []byte, initOffset int) (ret tagAndLength, offset int, err error) {
   528  	offset = initOffset
   529  	// parseTagAndLength should not be called without at least a single
   530  	// byte to read. Thus this check is for robustness:
   531  	if offset >= len(bytes) {
   532  		err = errors.New("asn1: internal error in parseTagAndLength")
   533  		return
   534  	}
   535  	b := bytes[offset]
   536  	offset++
   537  	ret.class = int(b >> 6)
   538  	ret.isCompound = b&0x20 == 0x20
   539  	ret.tag = int(b & 0x1f)
   540  
   541  	// If the bottom five bits are set, then the tag number is actually base 128
   542  	// encoded afterwards
   543  	if ret.tag == 0x1f {
   544  		ret.tag, offset, err = parseBase128Int(bytes, offset)
   545  		if err != nil {
   546  			return
   547  		}
   548  		// Tags should be encoded in minimal form.
   549  		if ret.tag < 0x1f {
   550  			err = SyntaxError{"non-minimal tag"}
   551  			return
   552  		}
   553  	}
   554  	if offset >= len(bytes) {
   555  		err = SyntaxError{"truncated tag or length"}
   556  		return
   557  	}
   558  	b = bytes[offset]
   559  	offset++
   560  	if b&0x80 == 0 {
   561  		// The length is encoded in the bottom 7 bits.
   562  		ret.length = int(b & 0x7f)
   563  	} else {
   564  		// Bottom 7 bits give the number of length bytes to follow.
   565  		numBytes := int(b & 0x7f)
   566  		if numBytes == 0 {
   567  			err = SyntaxError{"indefinite length found (not DER)"}
   568  			return
   569  		}
   570  		ret.length = 0
   571  		for i := 0; i < numBytes; i++ {
   572  			if offset >= len(bytes) {
   573  				err = SyntaxError{"truncated tag or length"}
   574  				return
   575  			}
   576  			b = bytes[offset]
   577  			offset++
   578  			if ret.length >= 1<<23 {
   579  				// We can't shift ret.length up without
   580  				// overflowing.
   581  				err = StructuralError{"length too large"}
   582  				return
   583  			}
   584  			ret.length <<= 8
   585  			ret.length |= int(b)
   586  			if ret.length == 0 {
   587  				// DER requires that lengths be minimal.
   588  				err = StructuralError{"superfluous leading zeros in length"}
   589  				return
   590  			}
   591  		}
   592  		// Short lengths must be encoded in short form.
   593  		if ret.length < 0x80 {
   594  			err = StructuralError{"non-minimal length"}
   595  			return
   596  		}
   597  	}
   598  
   599  	return
   600  }
   601  
   602  // parseSequenceOf is used for SEQUENCE OF and SET OF values. It tries to parse
   603  // a number of ASN.1 values from the given byte slice and returns them as a
   604  // slice of Go values of the given type.
   605  func parseSequenceOf(bytes []byte, sliceType reflect.Type, elemType reflect.Type) (ret reflect.Value, err error) {
   606  	matchAny, expectedTag, compoundType, ok := getUniversalType(elemType)
   607  	if !ok {
   608  		err = StructuralError{"unknown Go type for slice"}
   609  		return
   610  	}
   611  
   612  	// First we iterate over the input and count the number of elements,
   613  	// checking that the types are correct in each case.
   614  	numElements := 0
   615  	for offset := 0; offset < len(bytes); {
   616  		var t tagAndLength
   617  		t, offset, err = parseTagAndLength(bytes, offset)
   618  		if err != nil {
   619  			return
   620  		}
   621  		switch t.tag {
   622  		case TagIA5String, TagGeneralString, TagT61String, TagUTF8String, TagNumericString, TagBMPString:
   623  			// We pretend that various other string types are
   624  			// PRINTABLE STRINGs so that a sequence of them can be
   625  			// parsed into a []string.
   626  			t.tag = TagPrintableString
   627  		case TagGeneralizedTime, TagUTCTime:
   628  			// Likewise, both time types are treated the same.
   629  			t.tag = TagUTCTime
   630  		}
   631  
   632  		if !matchAny && (t.class != ClassUniversal || t.isCompound != compoundType || t.tag != expectedTag) {
   633  			err = StructuralError{"sequence tag mismatch"}
   634  			return
   635  		}
   636  		if invalidLength(offset, t.length, len(bytes)) {
   637  			err = SyntaxError{"truncated sequence"}
   638  			return
   639  		}
   640  		offset += t.length
   641  		numElements++
   642  	}
   643  	ret = reflect.MakeSlice(sliceType, numElements, numElements)
   644  	params := fieldParameters{}
   645  	offset := 0
   646  	for i := 0; i < numElements; i++ {
   647  		offset, err = parseField(ret.Index(i), bytes, offset, params)
   648  		if err != nil {
   649  			return
   650  		}
   651  	}
   652  	return
   653  }
   654  
   655  var (
   656  	bitStringType        = reflect.TypeOf(BitString{})
   657  	objectIdentifierType = reflect.TypeOf(ObjectIdentifier{})
   658  	enumeratedType       = reflect.TypeOf(Enumerated(0))
   659  	flagType             = reflect.TypeOf(Flag(false))
   660  	timeType             = reflect.TypeOf(time.Time{})
   661  	rawValueType         = reflect.TypeOf(RawValue{})
   662  	rawContentsType      = reflect.TypeOf(RawContent(nil))
   663  	bigIntType           = reflect.TypeOf(new(big.Int))
   664  )
   665  
   666  // invalidLength reports whether offset + length > sliceLength, or if the
   667  // addition would overflow.
   668  func invalidLength(offset, length, sliceLength int) bool {
   669  	return offset+length < offset || offset+length > sliceLength
   670  }
   671  
   672  // parseField is the main parsing function. Given a byte slice and an offset
   673  // into the array, it will try to parse a suitable ASN.1 value out and store it
   674  // in the given Value.
   675  func parseField(v reflect.Value, bytes []byte, initOffset int, params fieldParameters) (offset int, err error) {
   676  	offset = initOffset
   677  	fieldType := v.Type()
   678  
   679  	// If we have run out of data, it may be that there are optional elements at the end.
   680  	if offset == len(bytes) {
   681  		if !setDefaultValue(v, params) {
   682  			err = SyntaxError{"sequence truncated"}
   683  		}
   684  		return
   685  	}
   686  
   687  	// Deal with the ANY type.
   688  	if ifaceType := fieldType; ifaceType.Kind() == reflect.Interface && ifaceType.NumMethod() == 0 {
   689  		var t tagAndLength
   690  		t, offset, err = parseTagAndLength(bytes, offset)
   691  		if err != nil {
   692  			return
   693  		}
   694  		if invalidLength(offset, t.length, len(bytes)) {
   695  			err = SyntaxError{"data truncated"}
   696  			return
   697  		}
   698  		var result interface{}
   699  		if !t.isCompound && t.class == ClassUniversal {
   700  			innerBytes := bytes[offset : offset+t.length]
   701  			switch t.tag {
   702  			case TagPrintableString:
   703  				result, err = parsePrintableString(innerBytes)
   704  			case TagNumericString:
   705  				result, err = parseNumericString(innerBytes)
   706  			case TagIA5String:
   707  				result, err = parseIA5String(innerBytes)
   708  			case TagT61String:
   709  				result, err = parseT61String(innerBytes)
   710  			case TagUTF8String:
   711  				result, err = parseUTF8String(innerBytes)
   712  			case TagInteger:
   713  				result, err = parseInt64(innerBytes)
   714  			case TagBitString:
   715  				result, err = parseBitString(innerBytes)
   716  			case TagOID:
   717  				result, err = parseObjectIdentifier(innerBytes)
   718  			case TagUTCTime:
   719  				result, err = parseUTCTime(innerBytes)
   720  			case TagGeneralizedTime:
   721  				result, err = parseGeneralizedTime(innerBytes)
   722  			case TagOctetString:
   723  				result = innerBytes
   724  			case TagBMPString:
   725  				result, err = parseBMPString(innerBytes)
   726  			default:
   727  				// If we don't know how to handle the type, we just leave Value as nil.
   728  			}
   729  		}
   730  		offset += t.length
   731  		if err != nil {
   732  			return
   733  		}
   734  		if result != nil {
   735  			v.Set(reflect.ValueOf(result))
   736  		}
   737  		return
   738  	}
   739  
   740  	t, offset, err := parseTagAndLength(bytes, offset)
   741  	if err != nil {
   742  		return
   743  	}
   744  	if params.explicit {
   745  		expectedClass := ClassContextSpecific
   746  		if params.application {
   747  			expectedClass = ClassApplication
   748  		}
   749  		if offset == len(bytes) {
   750  			err = StructuralError{"explicit tag has no child"}
   751  			return
   752  		}
   753  		if t.class == expectedClass && t.tag == *params.tag && (t.length == 0 || t.isCompound) {
   754  			if fieldType == rawValueType {
   755  				// The inner element should not be parsed for RawValues.
   756  			} else if t.length > 0 {
   757  				t, offset, err = parseTagAndLength(bytes, offset)
   758  				if err != nil {
   759  					return
   760  				}
   761  			} else {
   762  				if fieldType != flagType {
   763  					err = StructuralError{"zero length explicit tag was not an asn1.Flag"}
   764  					return
   765  				}
   766  				v.SetBool(true)
   767  				return
   768  			}
   769  		} else {
   770  			// The tags didn't match, it might be an optional element.
   771  			ok := setDefaultValue(v, params)
   772  			if ok {
   773  				offset = initOffset
   774  			} else {
   775  				err = StructuralError{"explicitly tagged member didn't match"}
   776  			}
   777  			return
   778  		}
   779  	}
   780  
   781  	matchAny, universalTag, compoundType, ok1 := getUniversalType(fieldType)
   782  	if !ok1 {
   783  		err = StructuralError{fmt.Sprintf("unknown Go type: %v", fieldType)}
   784  		return
   785  	}
   786  
   787  	// Special case for strings: all the ASN.1 string types map to the Go
   788  	// type string. getUniversalType returns the tag for PrintableString
   789  	// when it sees a string, so if we see a different string type on the
   790  	// wire, we change the universal type to match.
   791  	if universalTag == TagPrintableString {
   792  		if t.class == ClassUniversal {
   793  			switch t.tag {
   794  			case TagIA5String, TagGeneralString, TagT61String, TagUTF8String, TagNumericString, TagBMPString:
   795  				universalTag = t.tag
   796  			}
   797  		} else if params.stringType != 0 {
   798  			universalTag = params.stringType
   799  		}
   800  	}
   801  
   802  	// Special case for time: UTCTime and GeneralizedTime both map to the
   803  	// Go type time.Time.
   804  	if universalTag == TagUTCTime && t.tag == TagGeneralizedTime && t.class == ClassUniversal {
   805  		universalTag = TagGeneralizedTime
   806  	}
   807  
   808  	if params.set {
   809  		universalTag = TagSet
   810  	}
   811  
   812  	matchAnyClassAndTag := matchAny
   813  	expectedClass := ClassUniversal
   814  	expectedTag := universalTag
   815  
   816  	if !params.explicit && params.tag != nil {
   817  		expectedClass = ClassContextSpecific
   818  		expectedTag = *params.tag
   819  		matchAnyClassAndTag = false
   820  	}
   821  
   822  	if !params.explicit && params.application && params.tag != nil {
   823  		expectedClass = ClassApplication
   824  		expectedTag = *params.tag
   825  		matchAnyClassAndTag = false
   826  	}
   827  
   828  	if !params.explicit && params.private && params.tag != nil {
   829  		expectedClass = ClassPrivate
   830  		expectedTag = *params.tag
   831  		matchAnyClassAndTag = false
   832  	}
   833  
   834  	// We have unwrapped any explicit tagging at this point.
   835  	if !matchAnyClassAndTag && (t.class != expectedClass || t.tag != expectedTag) ||
   836  		(!matchAny && t.isCompound != compoundType) {
   837  		// Tags don't match. Again, it could be an optional element.
   838  		ok := setDefaultValue(v, params)
   839  		if ok {
   840  			offset = initOffset
   841  		} else {
   842  			err = StructuralError{fmt.Sprintf("tags don't match (%d vs %+v) %+v %s @%d", expectedTag, t, params, fieldType.Name(), offset)}
   843  		}
   844  		return
   845  	}
   846  	if invalidLength(offset, t.length, len(bytes)) {
   847  		err = SyntaxError{"data truncated"}
   848  		return
   849  	}
   850  	innerBytes := bytes[offset : offset+t.length]
   851  	offset += t.length
   852  
   853  	// We deal with the structures defined in this package first.
   854  	switch v := v.Addr().Interface().(type) {
   855  	case *RawValue:
   856  		*v = RawValue{t.class, t.tag, t.isCompound, innerBytes, bytes[initOffset:offset]}
   857  		return
   858  	case *ObjectIdentifier:
   859  		*v, err = parseObjectIdentifier(innerBytes)
   860  		return
   861  	case *BitString:
   862  		*v, err = parseBitString(innerBytes)
   863  		return
   864  	case *time.Time:
   865  		if universalTag == TagUTCTime {
   866  			*v, err = parseUTCTime(innerBytes)
   867  			return
   868  		}
   869  		*v, err = parseGeneralizedTime(innerBytes)
   870  		return
   871  	case *Enumerated:
   872  		parsedInt, err1 := parseInt32(innerBytes)
   873  		if err1 == nil {
   874  			*v = Enumerated(parsedInt)
   875  		}
   876  		err = err1
   877  		return
   878  	case *Flag:
   879  		*v = true
   880  		return
   881  	case **big.Int:
   882  		parsedInt, err1 := parseBigInt(innerBytes)
   883  		if err1 == nil {
   884  			*v = parsedInt
   885  		}
   886  		err = err1
   887  		return
   888  	}
   889  	switch val := v; val.Kind() {
   890  	case reflect.Bool:
   891  		parsedBool, err1 := parseBool(innerBytes)
   892  		if err1 == nil {
   893  			val.SetBool(parsedBool)
   894  		}
   895  		err = err1
   896  		return
   897  	case reflect.Int, reflect.Int32, reflect.Int64:
   898  		if val.Type().Size() == 4 {
   899  			parsedInt, err1 := parseInt32(innerBytes)
   900  			if err1 == nil {
   901  				val.SetInt(int64(parsedInt))
   902  			}
   903  			err = err1
   904  		} else {
   905  			parsedInt, err1 := parseInt64(innerBytes)
   906  			if err1 == nil {
   907  				val.SetInt(parsedInt)
   908  			}
   909  			err = err1
   910  		}
   911  		return
   912  	// TODO(dfc) Add support for the remaining integer types
   913  	case reflect.Struct:
   914  		structType := fieldType
   915  
   916  		for i := 0; i < structType.NumField(); i++ {
   917  			if !structType.Field(i).IsExported() {
   918  				err = StructuralError{"struct contains unexported fields"}
   919  				return
   920  			}
   921  		}
   922  
   923  		if structType.NumField() > 0 &&
   924  			structType.Field(0).Type == rawContentsType {
   925  			bytes := bytes[initOffset:offset]
   926  			val.Field(0).Set(reflect.ValueOf(RawContent(bytes)))
   927  		}
   928  
   929  		innerOffset := 0
   930  		for i := 0; i < structType.NumField(); i++ {
   931  			field := structType.Field(i)
   932  			if i == 0 && field.Type == rawContentsType {
   933  				continue
   934  			}
   935  			innerOffset, err = parseField(val.Field(i), innerBytes, innerOffset, parseFieldParameters(field.Tag.Get("asn1")))
   936  			if err != nil {
   937  				return
   938  			}
   939  		}
   940  		// We allow extra bytes at the end of the SEQUENCE because
   941  		// adding elements to the end has been used in X.509 as the
   942  		// version numbers have increased.
   943  		return
   944  	case reflect.Slice:
   945  		sliceType := fieldType
   946  		if sliceType.Elem().Kind() == reflect.Uint8 {
   947  			val.Set(reflect.MakeSlice(sliceType, len(innerBytes), len(innerBytes)))
   948  			reflect.Copy(val, reflect.ValueOf(innerBytes))
   949  			return
   950  		}
   951  		newSlice, err1 := parseSequenceOf(innerBytes, sliceType, sliceType.Elem())
   952  		if err1 == nil {
   953  			val.Set(newSlice)
   954  		}
   955  		err = err1
   956  		return
   957  	case reflect.String:
   958  		var v string
   959  		switch universalTag {
   960  		case TagPrintableString:
   961  			v, err = parsePrintableString(innerBytes)
   962  		case TagNumericString:
   963  			v, err = parseNumericString(innerBytes)
   964  		case TagIA5String:
   965  			v, err = parseIA5String(innerBytes)
   966  		case TagT61String:
   967  			v, err = parseT61String(innerBytes)
   968  		case TagUTF8String:
   969  			v, err = parseUTF8String(innerBytes)
   970  		case TagGeneralString:
   971  			// GeneralString is specified in ISO-2022/ECMA-35,
   972  			// A brief review suggests that it includes structures
   973  			// that allow the encoding to change midstring and
   974  			// such. We give up and pass it as an 8-bit string.
   975  			v, err = parseT61String(innerBytes)
   976  		case TagBMPString:
   977  			v, err = parseBMPString(innerBytes)
   978  
   979  		default:
   980  			err = SyntaxError{fmt.Sprintf("internal error: unknown string type %d", universalTag)}
   981  		}
   982  		if err == nil {
   983  			val.SetString(v)
   984  		}
   985  		return
   986  	}
   987  	err = StructuralError{"unsupported: " + v.Type().String()}
   988  	return
   989  }
   990  
   991  // canHaveDefaultValue reports whether k is a Kind that we will set a default
   992  // value for. (A signed integer, essentially.)
   993  func canHaveDefaultValue(k reflect.Kind) bool {
   994  	switch k {
   995  	case reflect.Int, reflect.Int8, reflect.Int16, reflect.Int32, reflect.Int64:
   996  		return true
   997  	}
   998  
   999  	return false
  1000  }
  1001  
  1002  // setDefaultValue is used to install a default value, from a tag string, into
  1003  // a Value. It is successful if the field was optional, even if a default value
  1004  // wasn't provided or it failed to install it into the Value.
  1005  func setDefaultValue(v reflect.Value, params fieldParameters) (ok bool) {
  1006  	if !params.optional {
  1007  		return
  1008  	}
  1009  	ok = true
  1010  	if params.defaultValue == nil {
  1011  		return
  1012  	}
  1013  	if canHaveDefaultValue(v.Kind()) {
  1014  		v.SetInt(*params.defaultValue)
  1015  	}
  1016  	return
  1017  }
  1018  
  1019  // Unmarshal parses the DER-encoded ASN.1 data structure b
  1020  // and uses the reflect package to fill in an arbitrary value pointed at by val.
  1021  // Because Unmarshal uses the reflect package, the structs
  1022  // being written to must use upper case field names. If val
  1023  // is nil or not a pointer, Unmarshal returns an error.
  1024  //
  1025  // After parsing b, any bytes that were leftover and not used to fill
  1026  // val will be returned in rest. When parsing a SEQUENCE into a struct,
  1027  // any trailing elements of the SEQUENCE that do not have matching
  1028  // fields in val will not be included in rest, as these are considered
  1029  // valid elements of the SEQUENCE and not trailing data.
  1030  //
  1031  // An ASN.1 INTEGER can be written to an int, int32, int64,
  1032  // or *big.Int (from the math/big package).
  1033  // If the encoded value does not fit in the Go type,
  1034  // Unmarshal returns a parse error.
  1035  //
  1036  // An ASN.1 BIT STRING can be written to a BitString.
  1037  //
  1038  // An ASN.1 OCTET STRING can be written to a []byte.
  1039  //
  1040  // An ASN.1 OBJECT IDENTIFIER can be written to an
  1041  // ObjectIdentifier.
  1042  //
  1043  // An ASN.1 ENUMERATED can be written to an Enumerated.
  1044  //
  1045  // An ASN.1 UTCTIME or GENERALIZEDTIME can be written to a time.Time.
  1046  //
  1047  // An ASN.1 PrintableString, IA5String, or NumericString can be written to a string.
  1048  //
  1049  // Any of the above ASN.1 values can be written to an interface{}.
  1050  // The value stored in the interface has the corresponding Go type.
  1051  // For integers, that type is int64.
  1052  //
  1053  // An ASN.1 SEQUENCE OF x or SET OF x can be written
  1054  // to a slice if an x can be written to the slice's element type.
  1055  //
  1056  // An ASN.1 SEQUENCE or SET can be written to a struct
  1057  // if each of the elements in the sequence can be
  1058  // written to the corresponding element in the struct.
  1059  //
  1060  // The following tags on struct fields have special meaning to Unmarshal:
  1061  //
  1062  //	application specifies that an APPLICATION tag is used
  1063  //	private     specifies that a PRIVATE tag is used
  1064  //	default:x   sets the default value for optional integer fields (only used if optional is also present)
  1065  //	explicit    specifies that an additional, explicit tag wraps the implicit one
  1066  //	optional    marks the field as ASN.1 OPTIONAL
  1067  //	set         causes a SET, rather than a SEQUENCE type to be expected
  1068  //	tag:x       specifies the ASN.1 tag number; implies ASN.1 CONTEXT SPECIFIC
  1069  //
  1070  // When decoding an ASN.1 value with an IMPLICIT tag into a string field,
  1071  // Unmarshal will default to a PrintableString, which doesn't support
  1072  // characters such as '@' and '&'. To force other encodings, use the following
  1073  // tags:
  1074  //
  1075  //	ia5     causes strings to be unmarshaled as ASN.1 IA5String values
  1076  //	numeric causes strings to be unmarshaled as ASN.1 NumericString values
  1077  //	utf8    causes strings to be unmarshaled as ASN.1 UTF8String values
  1078  //
  1079  // If the type of the first field of a structure is RawContent then the raw
  1080  // ASN1 contents of the struct will be stored in it.
  1081  //
  1082  // If the name of a slice type ends with "SET" then it's treated as if
  1083  // the "set" tag was set on it. This results in interpreting the type as a
  1084  // SET OF x rather than a SEQUENCE OF x. This can be used with nested slices
  1085  // where a struct tag cannot be given.
  1086  //
  1087  // Other ASN.1 types are not supported; if it encounters them,
  1088  // Unmarshal returns a parse error.
  1089  func Unmarshal(b []byte, val interface{}) (rest []byte, err error) {
  1090  	return UnmarshalWithParams(b, val, "")
  1091  }
  1092  
  1093  // An invalidUnmarshalError describes an invalid argument passed to Unmarshal.
  1094  // (The argument to Unmarshal must be a non-nil pointer.)
  1095  type invalidUnmarshalError struct {
  1096  	Type reflect.Type
  1097  }
  1098  
  1099  func (e *invalidUnmarshalError) Error() string {
  1100  	if e.Type == nil {
  1101  		return "asn1: Unmarshal recipient value is nil"
  1102  	}
  1103  
  1104  	if e.Type.Kind() != reflect.Ptr {
  1105  		return "asn1: Unmarshal recipient value is non-pointer " + e.Type.String()
  1106  	}
  1107  	return "asn1: Unmarshal recipient value is nil " + e.Type.String()
  1108  }
  1109  
  1110  // UnmarshalWithParams allows field parameters to be specified for the
  1111  // top-level element. The form of the params is the same as the field tags.
  1112  func UnmarshalWithParams(b []byte, val interface{}, params string) (rest []byte, err error) {
  1113  	v := reflect.ValueOf(val)
  1114  	if v.Kind() != reflect.Ptr || v.IsNil() {
  1115  		return nil, &invalidUnmarshalError{reflect.TypeOf(val)}
  1116  	}
  1117  	offset, err := parseField(v.Elem(), b, 0, parseFieldParameters(params))
  1118  	if err != nil {
  1119  		return nil, err
  1120  	}
  1121  	return b[offset:], nil
  1122  }
  1123  

View as plain text