Grammar building blocks

Haka grammar is made of basic blocks and coumpound blocks. The former enables parsing of basic elements such as booleans (flag()), bytes (bytes()), number (number()), regular expression (token()), etc. The latter allows to form complex blocks by combining basic and compound blocks. For instance, the record block (record()) is used to define a structure of elements.

Creating the grammar

Smtp protocol grammar is defined through the following skeleton where we define protocol messages syntax. Then, we can export the root elements of our grammar in order to compile them and make them ready to be parsed:

SmtpDissector.grammar ="smtp", function ()


    smtp_command = entity{

    smtp_reponse = entity{

    smtp_data = entity{

    export(smtp_command, smtp_response, smtp_data)

Specifying terminal enitites

Here, we give the definition of terminal entities that will be used to specify smtp messages structure:

WS = token('[[:blank:]]+')
CRLF = token('[%r]?[%n]')

SEP = token('[- ]'))
COMMAND = token('[[:alpha:]]+'))
MESSAGE = token('[^%r%n]*'))
CODE = token('[0-9]{3}'))

PARAM = record{
    field('parameter', MESSAGE)

The first ones are self explanatory. PARAM is defined using the record keyword wich represents consecutive elements.

Specifying protocol message syntax

We distinguish three SMTP messages: commands, responses and data (mail content). Hereafter, we give their specification using the Haka grammar.

SMTP commands

Following RFC 2821, smtp command messages are alphabetic chars terminated by a CRLF. Command themselves are followed by white space(s) if paramters are present. These parameters could be required (MAIL, RCPT), optional (HELP) or not permitted at all (DATA, QUIT). These information is stored in the following table which will be used later by the grammar to adapt dissection according to parsed smtp commands:

local CMD = {
      ['HELO'] = 'required',
      ['EHLO'] = 'required',
      ['MAIL'] = 'required',
      ['RCPT'] = 'required',
      ['DATA'] = 'none',
     ['RESET'] = 'none',
    ['VERIFY'] = 'required',
    ['EXPAND'] = 'required',
      ['HELP'] = 'optional',
      ['NOOP'] = 'optional',
      ['QUIT'] = 'none'

The syntax of smtp command messages is defined as a record starting with a command name (defined previusouly as terminal token) and ending with a CRLF. We use the branch entity to distinguish between the three configuration cases:

  • Parameters must follow.
  • Parameters may be present.
  • No parameters follow.

The branch entity is endowed with a selection function allowing to select the branch to follow depending on the command name. Note that the grammar has a special element optional allowing to handle cases where messages may be present or not. In our case, we detect if paramteres are present by looking one byte further if CRLF is present. This is done thanks to the lookahead function (see Grammar)

smtp_command = record {
    field('command', COMMAND),
            required = PARAM,
            optional = optional(PARAM,
                function(self, ctx)
                    local la = ctx:lookahead()
                    return not (la == 0xa or la == 0xd)
            none = empty()
        function (self, ctx)
            return CMD[self.command]

SMTP responses

A smtp response message is defined as a status code followed by a separator, a comprehensive message and a trailing CRLF:

smtp_response = record {
    field('code', CODE),
    field('sep', SEP),
    field('parameter', MESSAGE),

Smtp server may respond by a sequence of response messages which are captured in Haka grammar using an array entity. The array size is determined thanks to the untilcond option (i.e. hyphen is missing in the last response message):

smtp_responses = field('responses',
                :untilcond(function (elem, ctx)
                        return elem and elem.sep == ' '


untilcond is an array option that returns true to indicate that we have reached the end of the array. See Grammar to get the list of available options.

SMTP data

Finally, data content is defined as following:

smtp_data = record {
    field('data', bytes()