SexpressionFormats

(public-key

 (rsa

  (e #010001#)

  (n

   |AOQjBj+wSG/BlAlir8Xuz62Hv3xAfAxJQeMl0kf93oWFzEcbK03h0kP3ueX4FaMMvsBYEqT

   uCK7h1CQHvuZrsRmjZmoP08zTOfrYYstU9wHW0QrPvTPrWlh52YXygS3NE8fHLOQkjwdCVf1

   CHubDxTnovrO7j7xBOsbeMgJArrvv|)))

The precise S-expression format used by SPKI/SDSI is described in the [SPKI certificate structure document] and on the homepage of [Ron Rivest's "sexp" library].

Any given S-expressions can be expressed in three formats:

1. advanced form: the pretty-printed, human readable form shown in the example above;
2. canonical form: a compact form, possibly containing non-printable characters, used to derive hash values; and
3. transport form: essentially a Base64-encoded canonical form.

To understand how S-expressions are hashed, it's important to know how their canonical form is derived.

Basically, an advanced S-expression consists of lists of atoms delimited by ( ). Atoms can be anything: strings such as public-key or "long string with spaces", or streams of raw bytes. Raw bytes are usually delimited by vertical bars and encoded in Base64, e.g., |aGlkZGVuIG1lc3NhZ2U=|, or can be represented as bit strings, as in #010001# above.

In canonical form, S-expressions are still delimited by parentheses, but atoms are encoded as a length in readable ASCII, followed by a colon, followed by the string or stream of bytes itself. For example, the advanced S-expression

(not-after "2004-07-18_17:58:18")

(9:not-after19:2004-07-18_17:58:18)

The sexp utility

The sexp utility can be gotten from [Ron Rivest's "sexp" library] page and compiled with (I think):

$ gcc -o sexp sexp-*.c

It's particularly useful for converting from ugly canonical S-expressions, which Greenpass uses a lot, to advanced ones, like so:

$ sexp -i Example.spkicert -a

If I recall, it actually gets used inside the Python scripts that make up the Web apps for just this purpose.