Author image Curtis "Ovid" Poe


TAPx::Parser::Grammar - A grammar for the original TAP version.


Version 0.50_07


TAPx::Parser::Gramamr is actually just a means for identifying individual chunks (usually lines) of TAP.

Do not attempt to use this class directly. It won't make sense. It's mainly here to ensure that we will be able to have pluggable grammars when TAP is expanded at some future date (plus, this stuff was really cluttering the parser).

Note that currently all methods are class methods. It's intended that this will eventually support TAP 2.0 and beyond which will necessitate actual instance data, but for now, we don't need this. Hence, the curious decision to use a class where one doesn't apparently need one.

Class Methods


  my $grammar = TAPx::Grammar->new;

Returns TAP grammar object. Future versions may accept a version number.


  my $token = $grammar->tokenize($string);

Passed a line of TAP, this method will return a data structure representing a 'token' matching that line of TAP input. Designed to be passed to TAPx::Parser::Result to create a result object.

This is really the only method you need to worry about for the grammar. The methods below are merely for convenience, if needed.

Class methods


  my @types = $grammar->token_types;

Returns the different types of tokens which this grammar can parse.


  my $syntax = $grammar->syntax_for($token_type);

Returns a pre-compiled regular expression which will match a chunk of TAP corresponding to the token type. For example (not that you should really pay attention to this, $grammar->syntax_for('comment') will return qr/^#(.*)/.


  my $handler = $grammar->handler_for($token_type);

Returns a code reference which, when passed an appropriate line of TAP, returns the lexed token corresponding to that line. As a result, the basic TAP parsing loop looks similar to the following:

 my @tokens;
 my $grammar = TAPx::Grammar->new;
 LINE: while ( defined( my $line = $parser->_next_chunk_of_tap ) ) {
     foreach my $type ( $grammar->token_types ) {
         my $syntax  = $grammar->syntax_for($type);
         if ( $line =~ $syntax ) {
             my $handler = $grammar->handler_for($type);
             push @tokens => $grammar->$handler($line);
             next LINE;
     push @tokens => $grammar->_make_unknown_token($line);


NOTE: This grammar is slightly out of date. There's still some discussion about it and a new one will be provided when we have things better defined.

The TAPx::Parser does not use a formal grammar because TAP is essentially a stream-based protocol. In fact, it's quite legal to have an infinite stream. For the same reason that we don't apply regexes to streams, we're not using a formal grammar here. Instead, we parse the TAP in lines.

For purposes for forward compatability, any result which does not match the following grammar is currently referred to as TAPx::Parser::Result::Unknown. It is not a parse error.

A formal grammar would look similar to the following:

     For the time being, I'm cheating on the EBNF by allowing 
     certain terms to be defined by POSIX character classes by
     using the following syntax:
       digit ::= [:digit:]
     As far as I am aware, that's not valid EBNF.  Sue me.  I
     didn't know how to write "char" otherwise (Unicode issues).  
     Suggestions welcome.
 (* POSIX character classes and other terminals *)
 digit              ::= [:digit:]
 character          ::= ([:print:] - "\n")
 positiveInteger    ::= ( digit - '0' ) {digit}
 nonNegativeInteger ::= digit {digit}
 tap            ::= { comment | unknown } leading_plan lines 
                    lines trailing_plan {comment}
 leading_plan   ::= plan skip_directive? "\n"

 trailing_plan  ::= plan "\n"

 plan           ::= '1..' nonNegativeInteger
 lines          ::= line {line}

 line           ::= (comment | test | unknown | bailout ) "\n"
 test           ::= status positiveInteger? description? directive?
 status         ::= 'not '? 'ok '
 description    ::= (character - (digit | '#')) {character - '#'}
 directive      ::= todo_directive | skip_directive

 todo_directive ::= hash_mark 'TODO' ' ' {character}

 skip_directive ::= hash_mark 'SKIP' ' ' {character}

 comment        ::= hash_mark {character}

 hash_mark      ::= '#' {' '}

 bailout        ::= 'Bail out!' {character}

 unknown        ::= { (character - "\n") }