=head1 NAME

Convert::BER - ASN.1 Basic Encoding Rules

=head1 SYNOPSIS

    use Convert::BER;
    
    $ber = new Convert::BER;
    
    $ber->encode(
	INTEGER => 1,
	SEQUENCE => [
            BOOLEAN => 0,
	    STRING => "Hello",
	],
        REAL => 3.7,
    );
    
    $ber->decode(
	INTEGER => \$i,
	SEQUENCE => [
            BOOLEAN => \$b,
	    STRING => \$s,
	],
        REAL => \$r,
    );

=head1 DESCRIPTION

B<WARNING> this module is no longer supported, See L<Convert::ASN1>

C<Convert::BER> provides an OO interface to encoding and decoding data
using the ASN.1 Basic Encoding Rules (BER), a platform independent way
of encoding structured binary data together with the structure.

=head1 METHODS

=over 4

=item new

=item new ( BUFFER )

=item new ( opList )

C<new> creates a new C<Convert::BER> object.

=item encode ( opList )

Encode data in I<opList> appending to the data in the buffer.

=item decode ( opList )

Decode the data in the buffer as described by I<opList>, starting
where the last decode finished or position set by C<pos>.

=item buffer ( [ BUFFER ] )

Return the buffer contents. If I<BUFFER> is specified set the buffer
contents and reset pos to zero.

=item pos ( [ POS ] )

Without any arguments C<pos> returns the offset where the last decode
finished, or the last offset set by C<pos>. If I<POS> is specified
then I<POS> will be where the next decode starts.

=item tag ( )

Returns the tag at the current position in the buffer.

=item length ( )

Returns the length of the buffer.

=item error ( )

Returns the error message associated with the last method, if
any. This value is not automatically reset. If C<encode> or
C<decode> returns undef, check this. 

=item dump ( [ FH ] )

Dump the buffer to the filehandle C<FH>, or STDERR if not specified. The
output contains the hex dump of each element, and an ASN.1-like text
representation of that element.

=item hexdump  ( [ FH ] )

Dump the buffer to the filehandle C<FH>, or STDERR if not specified. The
output is hex with the possibly-printable text alongside.

=back

=head1 IO METHODS

=over 4

=item read ( IO )

=item write ( IO )

=item recv ( SOCK )

=item send ( SOCK [, ADDR ] )

=back

=head1 OPLIST

An I<opList> is a list of I<operator>-I<value> pairs. An operator can
be any of those defined below, or any defined by sub-classing
C<Convert::BER>, which will probably be derived from the primitives
given here.

The I<value>s depend on whether BER is being encoded or decoded:

=over 4

=item Encoding

If the I<value> is a scalar, just encode it. If the I<value> is a
reference to a list, then encode each item in the list in turn. If the
I<value> is a code reference, then execute the code. If the returned
value is a scalar, encode that value. If the returned value is a
reference to a list, encode each item in the list in turn.

=item Decoding

If the I<value> is a reference to a scalar, decode the value into the
scalar. If the I<value> is a reference to a list, then decode all the
items of this type into the list. Note that there must be at least one
item to decode, otherwise the decode will fail. If the I<value> is a
code reference, then execute the code and decode the value into the
reference returned from the evaluated code.

=back

=head1 PRIMITIVE OPERATORS

These operators encode and decode the basic primitive types defined by
BER.

=head2 BOOLEAN

A BOOLEAN value is either true or false.

=over 4

=item Encoding

The I<value> is tested for boolean truth, and encoded appropriately.

    # Encode a TRUE value
    $ber->encode(
        BOOLEAN => 1,
    ) or die;

=item Decoding

The decoded I<value>s will be either 1 or 0.

    # Decode a boolean value into $bval
    $ber->decode(
        BOOLEAN => \$bval,
    ) or die;

=back

=head2 INTEGER

An INTEGER value is either a positive whole number, or a negative
whole number, or zero. Numbers can either be native perl integers, or
values of the C<Math::BigInt> class.

=over 4

=item Encoding

The I<value> is the integer value to be encoded.

    $ber->encode(
        INTEGER => -123456,
    ) or die;

=item Decoding

The I<value> will be the decoded integer value.

    $ber->decode(
        INTEGER => \$ival,
    ) or die;

=back

=head2 STRING

This is an OCTET STRING, which is an arbitrarily long binary value.

=over 4

=item Encoding

The I<value> contains the binary value to be encoded.

    $ber->encode(
        STRING => "\xC0First character is hex C0",
    ) or die;

=item Decoding

The I<value> will be the binary bytes.

    $ber->decode(
        STRING => \$sval,
    ) or die;

=back

=head2 NULL

There is no value for NULL. You often use NULL in ASN.1 when you want
to denote that something else is absent rather than just not encoding
the 'something else'.

=over 4

=item Encoding

The I<value>s are ignored, but must be present.

    $ber->encode(
        NULL => undef,
    ) or die;

=item Decoding

Dummy values are stored in the returned I<value>s, as though they were
present in the encoding.

    $ber->decode(
        NULL => \$nval,
    ) or die;

=back

=head2 OBJECT_ID

An OBJECT_ID value is an OBJECT IDENTIFIER (also called an OID). This
is a hierarchically structured value that is used in protocols to
uniquely identify something. For example, SNMP (the Simple Network
Management Protocol) uses OIDs to denote the information being
requested, and LDAP (the Lightweight Directory Access Protocol, RFC
2251) uses OIDs to denote each attribute in a directory entry.

Each level of the OID hierarchy is either zero or a positive integer.

=over 4

=item Encoding

The I<value> should be a dotted-decimal representation of the OID.

    $ber->encode(
        OBJECT_ID => '2.5.4.0', # LDAP objectClass
    ) or die;

=item Decoding

The I<value> will be the dotted-decimal representation of the OID.

    $ber->decode(
        OBJECT_ID => \$oval,
    ) or die;

=back

=head2 ENUM

The ENUMERATED type is effectively the same as the INTEGER type. It
exists so that friendly names can be assigned to certain integer
values. To be useful, you should sub-class this operator.

=head2 BIT_STRING

The BIT STRING type is an arbitrarily long string of bits - C<0>'s and
C<1>'s.

=over 4

=item Encoding

The I<value> is a string of arbitrary C<0> and C<1> characters. As
these are packed into 8-bit octets when encoding and there may not be
a multiple of 8 bits to be encoded, trailing padding bits are added in
the encoding.

    $ber->encode(
        BIT_STRING => '0011',
    ) or die;

=item Decoding

The I<value> will be a string of C<0> and C<1> characters. The string
will have the same number of bits as were encoded (the padding bits
are ignored.)

    $ber->decode(
        BIT_STRING => \$bval,
    ) or die;

=back

=head2 BIT_STRING8

This is a variation of the BIT_STRING operator, which is optimized for
writing bit strings which are multiples of 8-bits in length. You can
use the BIT_STRING operator to decode BER encoded with the BIT_STRING8
operator (and vice-versa.)

=over 4

=item Encoding

The I<value> should be the packed bits to encode, B<not> a string of
C<0> and C<1> characters.

    $ber->encode(
        BIT_STRING8 => pack('B8', '10110101'),
    ) or die;

=item Decoding

The I<value> will be the decoded packed bits.

    $ber->decode(
        BIT_STRING8 => \$bval,
    ) or die;

=back

=head2 REAL

The REAL type encodes an floating-point number. It requires the POSIX
module.

=over 4

=item Encoding

The I<value> should be the number to encode.

    $ber->encode(
        REAL => 3.14159265358979,
    ) or die;

=item Decoding

The I<value> will be the decoded floating-point value.

    $ber->decode(
        REAL => \$rval,
    );

=head2 ObjectDescriptor

The ObjectDescriptor type encodes an ObjectDescriptor string. It is a
sub-class of C<STRING>.

=head2 UTF8String

The UTF8String type encodes a string encoded in UTF-8. It is a
sub-class of C<STRING>.

=head2 NumericString

The NumericString type encodes a NumericString, which is defined to
only contain the characters 0-9 and space. It is a sub-class of
C<STRING>.

=head2 PrintableString

The PrintableString type encodes a PrintableString, which is defined
to only contain the characters A-Z, a-z, 0-9, space, and the
punctuation characters ()-+=:',./?. It is a sub-class of C<STRING>.

=head2 TeletexString/T61String

The TeletexString type encodes a TeletexString, which is a string
containing characters according to the T.61 character set. Each T.61
character may be one or more bytes wide. It is a sub-class of
C<STRING>.

T61String is an alternative name for TeletexString.

=head2 VideotexString

The VideotexString type encodes a VideotexString, which is a
string. It is a sub-class of C<STRING>.

=head2 IA5String

The IA5String type encodes an IA5String. IA5 (International Alphabet
5) is equivalent to US-ASCII. It is a sub-class of C<STRING>.

=head2 UTCTime

The UTCTime type encodes a UTCTime value. Note this value only
represents years using two digits, so it is not recommended in
Y2K-compliant applications. It is a sub-class of C<STRING>.

UTCTime values must be strings like:

    yymmddHHMM[SS]Z
or:
    yymmddHHMM[SS]sHHMM

Where yy is the year, mm is the month (01-12), dd is the day (01-31),
HH is the hour (00-23), MM is the minutes (00-60). SS is the optional
seconds (00-61).

The time is either terminated by the literal character Z, or a
timezone offset. The "Z" character indicates Zulu time or UTC. The
timezone offset specifies the sign s, which is + or -, and the
difference in hours and minutes.

=head2 GeneralizedTime

The GeneralizedTime type encodes a GeneralizedTime value. Unlike
C<UTCTime> it represents years using 4 digits, so is Y2K-compliant. It
is a sub-class of C<STRING>.

GeneralizedTime values must be strings like:

    yyyymmddHHMM[SS][.U][Z]
or:
    yyyymmddHHMM[SS][.U]sHHMM

Where yyyy is the year, mm is the month (01-12), dd is the day
(01-31), HH is the hour (00-23), MM is the minutes (00-60). SS is the
optional seconds (00-61). U is the optional fractional seconds value;
a comma is permitted instead of a dot before this value.

The time may be terminated by the literal character Z, or a timezone
offset. The "Z" character indicates Zulu time or UTC. The timezone
offset specifies the sign s, which is + or -, and the difference in
hours and minutes. If there is timezone specified UTC is assumed.

=head2 GraphicString

The GraphicString type encodes a GraphicString value. It is a
sub-class of C<STRING>.

=head2 VisibleString/ISO646String

The VisibleString type encodes a VisibleString value, which is a value
using the ISO646 character set. It is a sub-class of C<STRING>.

ISO646String is an alternative name for VisibleString.

=head2 GeneralString

The GeneralString type encodes a GeneralString value. It is a
sub-class of C<STRING>.

=head2 UniversalString/CharacterString

The UniveralString type encodes a UniveralString value, which is a
value using the ISO10646 character set. Each character in ISO10646 is
4-bytes wide. It is a sub-class of C<STRING>.

CharacterString is an alternative name for UniversalString.

=head2 BMPString

The BMPString type encodes a BMPString value, which is a value using
the Unicode character set. Each character in the Unicode character set
is 2-bytes wide. It is a sub-class of C<STRING>.

=head1 CONSTRUCTED OPERATORS

These operators are used to build constructed types, which contain
values in different types, like a C structure.

=head2 SEQUENCE

A SEQUENCE is a complex type that contains other types, a bit like a C
structure. Elements inside a SEQUENCE are encoded and decoded in the
order given.

=over 4

=item Encoding

The I<value> should be a reference to an array containing another
I<opList> which defines the elements inside the SEQUENCE.

    $ber->encode(
        SEQUENCE => [
            INTEGER => 123,
            BOOLEAN => [ 1, 0 ],
        ]
    ) or die;

=item Decoding

The I<value> should a reference to an array that contains the
I<opList> which decodes the contents of the SEQUENCE.

    $ber->decode(
        SEQUENCE => [
            INTEGER => \$ival,
            BOOLEAN => \@bvals,
        ]
    ) or die;

=back

=head2 SET

A SET is an complex type that contains other types, rather like a
SEQUENCE. Elements inside a SET may be present in any order.

=over 4

=item Encoding

The I<value> is the same as for the SEQUENCE operator.

    $ber->encode(
        SET => [
            INTEGER => 13,
            STRING => 'Hello',
        ]
    ) or die;

=item Decoding

The I<value> should be a reference to an B<equivalent> I<opList> to
that used to encode the SET. The ordering of the I<opList> should not
matter.

    $ber->decode(
        SET => [
            STRING => \$sval,
            INTEGER => \$ival,
        ]
    ) or die;

=back

=head2 SEQUENCE_OF

A SEQUENCE_OF is an ordered list of other types.

=over 4

=item Encoding

The I<value> is a I<ref> followed by an I<opList>. The I<ref> must be
a reference to a list or a hash: if it is to a list, then the
I<opList> will be repeated once for every element in the list. If it
is to a hash, then the I<opList> will be repeated once for every key
in the hash (note that ordering of keys in a hash is not guaranteed by
perl.)

The remaining I<opList> will then usually contain I<value>s which are
code references. If the I<ref> is to a list, then the contents of that
item in the list are passed as the only argument to the code
reference. If the I<ref> is to a hash, then only the key is passed to
the code.

    @vals = ( [ 10, 'Foo' ], [ 20, 'Bar' ] ); # List of refs to lists
    $ber->encode(
        SEQUENCE_OF => [ \@vals,
            SEQUENCE => [
                INTEGER => sub { $_[0][0] }, # Passed a ref to the inner list
                STRING => sub { $_[0][1] }, # Passed a ref to the inner list
            ]
        ]
    ) or die;
    %hash = ( 40 => 'Baz', 30 => 'Bletch' ); # Just a hash
    $ber->decode(
        SEQUENCE_OF => [ \%hash,
            SEQUENCE => [
                INTEGER => sub { $_[0] }, # Passed the key
                STRING => sub { $hash{$_[0]} }, # Passed the key
            ]
        ]
    );

=item Decoding

The I<value> must be a reference to a list containing a I<ref> and an
I<opList>. The I<ref> must always be a reference to a scalar. Each
value in the <opList> is usually a code reference. The code referenced
is called with the value of the I<ref> (dereferenced); the value of
the I<ref> is incremented for each item in the SEQUENCE_OF.

    $ber->decode(
        SEQUENCE_OF => [ \$count,
            # In the following subs, make space at the end of an array, and
            # return a reference to that newly created space.
            SEQUENCE => [
                INTEGER => sub { $ival[$_[0]] = undef; \$ival[-1] },
                STRING => sub { $sval[$_[0]] = undef; \$sval[-1] },
            ]
        ]
    ) or die;

=back

=head2 SET_OF

A SET_OF is an unordered list. This is treated in an identical way to
a SEQUENCE_OF, except that no ordering should be inferred from the
list passed or returned.

=head1 SPECIAL OPERATORS

=head2 BER

It is sometimes useful to construct or deconstruct BER encodings in
several pieces. The BER operator lets you do this.

=over 4

=item Encoding

The I<value> should be another C<Convert::BER> object, which will be
inserted into the buffer. If I<value> is undefined then nothing is
added.

    $tmp->encode(
        SEQUENCE => [
            INTEGER => 20,
            STRING => 'Foo',
        ]
    );
    $ber->encode(
        BER => $tmp,
        BOOLEAN => 1
    );

=item Decoding

I<value> should be a reference to a scalar, which will contain a
C<Convert::BER> object. This object will contain the remainder of the
current sequence or set being decoded.

    # After this, ber2 will contain the encoded INTEGER B<and> STRING.
    # sval will be ignored and left undefined, but bval will be decoded. The
    # decode of ber2 will return the integer and string values.
    $ber->decode(
        SEQUENCE => [
            BER => \$ber2,
            STRING => \$sval,
        ],
        BOOLEAN => \$bval,
    );
    $ber2->decode(
        INTEGER => \$ival,
        STRING => \$sval2,
    );

=back

=head2 ANY

This is like the C<BER> operator except that when decoding only the
next item is decoded and placed into the C<Convert::BER> object
returned. There is no difference when encoding.

=over 4

=item Decoding

I<value> should be a reference to a scalar, which will contain a
C<Convert::BER> object. This object will only contain the next single
item in the current sequence being decoded.

    # After this, ber2 will decode further, and ival and sval
    # will be decoded.
    $ber->decode(
        INTEGER = \$ival,
        ANY => \$ber2,
        STRING => \$sval,
    );

=back

=head2 OPTIONAL

This operator allows you to specify that an element is absent from the
encoding.

=over 4

=item Encoding

The I<value> should be a reference to another list with another
I<opList>. If all of the values of the inner I<opList> are defined,
the entire OPTIONAL I<value> will be encoded, otherwise it will be
omitted.

    $ber->encode(
        SEQUENCE => [
	    INTEGER => 16, # Will be encoded
            OPTIONAL => [
                INTEGER => undef, # Will not be encoded
            ],
            STRING => 'Foo', # Will be encoded
        ]
    );

=item Decoding

The contents of I<value> are decoded if possible, if not then decode
continues at the next I<operator>-I<value> pair.

    $ber->decode(
        SEQUENCE => [
            INTEGER => \$ival1,
            OPTIONAL => [
                INTEGER => \$ival2,
            ],
            STRING => \$sval,
        ]
    );

=back

=head2 CHOICE

The I<opList> is a list of alternate I<operator>-I<value> pairs. Only
one will be encoded, and only one will be decoded.

=over 4

=item Encoding

A scalar at the start of the I<opList> identifies which I<opList>
alternative to use for encoding the value. A value of 0 means the
first one is used, 1 means the second one, etc.

    # Encode the BMPString alternate of the CHOICE
    $ber->encode(
        CHOICE => [ 2,
            PrintableString => 'Printable',
            TeletexString   => 'Teletex/T61',
            BMPString       => 'BMP/Unicode',
            UniversalString => 'Universal/ISO10646',
        ]
    ) or die;

=item Decoding

A reference to a scalar at the start of the I<opList> is used to store
which alternative is decoded (0 for the first one, 1 for the second
one, etc.) Pass undef instead of the ref if you don't care about this,
or you store all the alternate values in different variables.

    # Decode the above.
    # Afterwards, $alt will be set to 2, $str will be set to 'BMP/Unicode'.
    $ber->decode(
        CHOICE => [ \$alt,
            PrintableString => \$str,
            TeletexString   => \$str,
            BMPString       => \$str,
            UniversalString => \$str,
        ]
    ) or die;

=back

=head1 TAGS

In BER everything being encoded has a tag, a length, and a
value. Normally the tag is derived from the operator - so INTEGER has
a different tag from a BOOLEAN, for instance.

In some applications it is necessary to change the tags used. For
example, a SET may need to contain two different INTEGER values. Tags
may be changed in two ways, either IMPLICITly or EXPLICITly. With
IMPLICIT tagging, the new tag completely replaces the old tag. With
EXPLICIT tagging, the new tag is used B<as well as> the old tag.

C<Convert::BER> supports two ways of using IMPLICIT tagging. One
method is to sub-class C<Convert::BER>, which is described in the next
section. For small applications or those that think sub-classing is
just too much then the operator may be passed an arrayref. The array
must contain two elements, the first is the usual operator name and
the second is the tag value to use, as shown below.

    $ber->encode(
	[ SEQUENCE => 0x34 ] => [
	    INTEGER => 10,
	    STRING  => "A"
	]
    ) or die;

This will encode a sequence, with a tag value of C<0x34>, which will
contain and integer and a string which will have their default tag
values.

You may wish to construct your tags using some pre-defined functions
such as C<&Convert::BER::BER_APPLICATION>,
C<&Convert::BER::BER_CONTEXT>, etc, instead of calculating the tag
values yourself.

To use EXPLICIT tagging, enclose the original element in a SEQUENCE,
and just override the SEQUENCE's tag as above. Don't forget to set the
constructed bit using C<&Convert::BER::BER_CONSTRUCTOR>. For example,
the ASN.1 definition:

    Foo ::= SEQUENCE {
        [0] EXPLICIT INTEGER,
        INTEGER
    }

might be encoded using this:

    $ber->encode(
        SEQUENCE => [
            [ SEQUENCE => &Convert::BER::BER_CONTEXT |
			  &Convert::BER::BER_CONSTRUCTOR | 0 ] => [
                INTEGER => 10,
            ],
            INTEGER => 11,
        ],
    ) or die;

=head1 SUB-CLASSING

For large applications where operators with non default tags are used
a lot the above mechanism can be very error-prone. For this reason,
C<Convert::BER> may be sub-classed.

To do this the sub-class must call a static method C<define>. The
arguments to C<define> is a list of arrayrefs. Each arrayref will
define one new operator. Each arrayref contains three values, the
first is the name of the operator, the second is how the data is
encoded and the third is the tag value. To aid with the creation of
these arguments C<Convert::BER> exports some variables and constant
subroutines.

For each operator defined by C<Convert::BER>, or a C<Convert::BER>
sub-class, a scalar variable with the same name is available for
import, for example C<$INTEGER> is available from C<Convert::BER>. And
any operators defined by a new sub-class will be available for import
from that class.  One of these variables may be used as the second
element of each arrayref.

C<Convert::BER> also exports some constant subroutines that can be
used to create the tag value. The subroutines exported are:

	BER_BOOLEAN
	BER_INTEGER
	BER_BIT_STR
	BER_OCTET_STR
	BER_NULL
	BER_OBJECT_ID
	BER_SEQUENCE
	BER_SET
    
	BER_UNIVERSAL
	BER_APPLICATION
	BER_CONTEXT
	BER_PRIVATE
	BER_PRIMITIVE
	BER_CONSTRUCTOR

C<Convert::BER> also provides a subroutine called C<ber_tag> to calculate
an integer value that will be used to represent a tag. For tags with
values less than 30 this is not needed, but for tags >= 30 then tag
value passed for an operator definition must be the result of C<ber_tag>

C<ber_tag> takes two arguments, the first is the tag class and the second
is the tag value.

Using this information a sub-class of Convert::BER can be created as
shown below.

    package Net::LDAP::BER;

    use Convert::BER qw(/^(\$|BER_)/);

    use strict;
    use vars qw($VERSION @ISA);

    @ISA = qw(Convert::BER);
    $VERSION = "1.00";

    Net::LDAP::BER->define(

      # Name		Type      Tag
      ########################################

      [ REQ_UNBIND     => $NULL,
			  BER_APPLICATION  	 	    | 0x02 ],
    
      [ REQ_COMPARE    => $SEQUENCE,
			  BER_APPLICATION | BER_CONSTRUCTOR | 0x0E ],
    
      [ REQ_ABANDON    => $INTEGER,
			  ber_tag(BER_APPLICATION, 0x10) ],
    );

This will create a new class C<Net::LDAP::BER> which has three new operators
available. This class then may be used as follows

    $ber = new Net::LDAP::BER;
    
    $ber->encode(
	REQ_UNBIND => 0,
	REQ_COMPARE => [
	    REQ_ABANDON => 123,
	]
    );

    $ber->decode(
	REQ_UNBIND => \$var,
	REQ_COMPARE => [
	    REQ_ABANDON => \$num,
	]
    );

Which will encode or decode the data using the formats and tags
defined in the C<Net::LDAP::BER> sub-class. It also helps to make the
code more readable.

=head2 DEFINING NEW PACKING OPERATORS

As well as defining new operators which inherit from existing
operators it is also possible to define a new operator and how data is
encoded and decoded. The interface for doing this is still changing
but will be documented here when it is done. To be continued ...

=head1 LIMITATIONS

Convert::BER cannot support tags that contain more bits than can be
stored in a scalar variable, typically this is 32 bits.

Convert::BER cannot support items that have a packed length which
cannot be stored in 32 bits.

=head1 BUGS

The C<SET> decode method fails if the encoded order is different to
the I<opList> order.

=head1 AUTHOR

Graham Barr <gbarr@pobox.com>

Significant POD updates from
Chris Ridd <Chris.Ridd@messagingdirect.com>

=head1 COPYRIGHT

Copyright (c) 1995-2000 Graham Barr. All rights reserved.
This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.

=cut