How to build parser with autostubbing

[Yuri Shtil]

The documentation is a little unclear

I use perl -MParse::RecDescent - grammar NewMakepp::Grammar to create a parser module.

What do I do in order to generate stubs for undefined rules?
How do I pass RD_AUTOSTUB?

-----------------------------------------------------------------------------------
This email message is for the sole use of the intended recipient(s) and may contain
confidential information. Any unauthorized review, use, disclosure or distribution
is prohibited. If you are not the intended recipient, please contact the sender by
reply email and destroy all copies of the original message.
-----------------------------------------------------------------------------------

[jtb...@cpan.org]

Consult the section on '-s' at http://perldoc.perl.org/perlrun.html:

-s

enables rudimentary switch parsing for switches on the command line
after the program name but before any filename arguments (or before
an argument of --). Any switch found there is removed from @ARGV and
sets the corresponding variable in the Perl program. The following
program prints "1" if the program is invoked with a -xyz switch, and
"abc" if it is invoked with -xyz=abc.
#!/usr/bin/perl -s
if ($xyz) { print "$xyz\n" }

Do note that a switch like --help creates the variable ${-help} ,
which is not compliant with use strict "refs" . Also, when using
this option on a script with warnings enabled you may get a lot of
spurious "used only once" warnings.


So this is probably what you want:

perl -MParse::RecDescent -s -RD_AUTOSTUB - grammar NewMakepp::Grammar

Untested, you may need to move -s, '-RD_AUTOSTUB and the '- grammar
...' arguments around to make it work.

[Yuri Shtil]

Well, I found as much myself, tried all I could and it did not work.

[jtb...@cpan.org]

So, this works to set RD_TRACE:
perl -MParse::RecDescent -s - -RD_TRACE grammar NewMakepp::Grammar

Unfortunately, it looks like the path taken by the precompilation fails to actually check for undefined rules. This may be a bug, or may have been intentional, I'll have to follow up with Damian.

Even if this were fixed, I'm not sure how useful the '-s' command line option is, as -RD_AUTOSTUB will set $::RD_AUTOSTUB to '1'. An undocumented feature of $::RD_AUTOSTUB is that if it has a boolean-true value, an undefined rule named 'foo' will not get the production
foo: 'foo'
but instead get the production:
foo: $::RD_AUTOSTUB
Which, for -s -RD_AUTOSTUB, will generate:
foo: '1'
1: '1'
Which probably isn't what you wanted in the first place.

Long story short, RD_AUTOSTUB doesn't currently work with precompilation of parsers. You'll have to either not precompile, or define your own stub productions.

Please file a ticket for this issue. I'm out of town for the next week, but will try to get this into a bugfix release next week. I've included details on the issue below, feel free to include them in the bug. You can find the ticket tracker for Parse::RecDescent here:
https://rt.cpan.org/Public/Dist/Display.html?Name=Parse-RecDescent



Note that this isn't unique to the command line, but appears to be a problem with the Precompile method. The callstack is something like:
script: Parse::RecDescent->Precompile($grammar, $class, $sourcefile)
Precompile: $self = Parse::RecDescent->new($grammar,1,$class)
new: $self->Replace($grammar, 1, $class)
Replace: _generate($self, $grammar, 1(fromsplice), 1, $class);
_generate: my ($self, $grammar, $replace, $isimplicit, $isleftop) = (@_, 0);

So, the '1' added to @_ from Precompile in the call to new() ends up becoming $isimplicit.

$isimplicit prevents $self->_check_grammar() from being called, which is the function that does the RD_AUTOSTUB checking/generation. I'll have to look into why _check_grammar is skipped for $isimplicit cases.

There also appears to be another bug here, where the $class parameter gets incorrectly propagated from Precompile into $isleftop inside of _generate, which is probably not correct. Unless I've misread the code somewhere.

[jtb...@cpan.org]

Yuri,

Thank you for the bug report. I've release version 1.967001 on CPAN,
which should address these issues for you.

Thanks again!
Jeremy