Re^2: Why won't this basic Parse::RecDescent example work?

[Athanasius]

I’m not familiar with Parse::RecDescent, but by reference to the docs plus a bit of trial-and-error I got this to work by adjusting the regex and assigning $1 to a local variable (see the section “Start-up Actions” in Parse::RecDescent):

#! perl
use strict;
use warnings;
use 5.012;

use Data::Dumper;
use Parse::RecDescent;

$::RD_ERRORS = 1;   # Parser dies when it encounters an error
$::RD_WARN   = 1;   # Enable warnings- warn on unused rules &c.
$::RD_HINT   = 1;   # Give out hints to help fix problems.
#$::RD_TRACE  = 1;   # if defined, also trace parsers' behaviour

our %HASH;

my $grammar = <<'END_OF_GRAMMAR';

    {
        my $directory;
    }

    startrule: from_clause 

    from_clause: 'from' dir(s)
    {
        print "-->$directory<--\n";
        $main::HASH{dir} = $directory;
    }

    dir: m{ ^ ( .*? / .* ) $ }x
    {
        $directory = $1;
    }

END_OF_GRAMMAR

my $parser = Parse::RecDescent->new($grammar) or die   "Bad grammar!\n
+";
defined $parser->startrule("from ./foo")      or print "Bad text!\n";

say Dumper(\%HASH);

Output:

18:38 >perl 505_SoPW.pl
-->./foo<--
$VAR1 = {
          'dir' => './foo'
        };


18:51 >

Note that setting $::RD_TRACE = 1; is useful for understanding what the parser is doing.

Hope that helps,

[7stud]

I’m not familiar with Parse::RecDescent,

Then thanks for being brave enough to take a look!

by reference to the docs plus a bit of trial-and-error I got this to work by adjusting the regex and assigning $1 to a local variable.

Nice going! After reading the "Start up Action" section in the docs, I reread the "Action" section, and I noticed this statement:

The results of named subrules are stored in the hash under each subrule's name (including the repetition specifier, if any)

So the key I was using in the %item hash, 'dir', was wrong. The key should be 'dir(s)'. So now I can get this output:

use strict; 
use warnings; 
use 5.012;

use Parse::RecDescent;


$::RD_ERRORS = 1; #Parser dies when it encounters an error
$::RD_WARN   = 1; #Enable warnings - warn on unused rules &c.
$::RD_HINT   = 1; # Give out hints to help fix problems.

our %HASH;

my $grammar = <<'END_OF_GRAMMAR';

    startrule : from_clause 

    from_clause : 'from' dir(s) 
                {
                    print "-->@{$item{'dir(s)'}}<-- \n";
                    $main::HASH{dirs} = @{$item{'dir(s)'}}; 
                }

    dir : 'hello'

END_OF_GRAMMAR


my $parser = Parse::RecDescent->new($grammar);
$parser->startrule("from hello world"); 
use Data::Dumper;
say Dumper(\%HASH);

--output:--
-->hello<-- 
$VAR1 = {
          'dirs' => 1
        };

Partial success! Note the dereference of $item{'dir(s)'}. Now, what is that '1'? The return value from print()? But print() isn't the last statement of the action. If I change the dir rule to:

dir : 'hello'
     | 'world'

I get this output:

-->hello world<-- 
$VAR1 = {
          'dirs' => 2
        };

Is 2 the count of the words matched? What is going on? I am using the exact same array in each of these lines:

          print "-->@{$item{'dir(s)'}}<-- \n";
$main::HASH{dirs} = @{$item{'dir(s)'}};

...yet I am getting different results 'hello world' v. 2! How is that possible? Argghh, of course! Perl doesn't care about giving you the exact same results for any expression you use--because perl determines the result by the context in which the expression appears. In my case, the print() statement supplies list context for the array, and "$main::... =" provides scalar context for the array--and an array provides its length in scalar context.

So now I can get the expected output:

use strict; 
use warnings; 
use 5.012;

use Parse::RecDescent;


$::RD_ERRORS = 1; #Parser dies when it encounters an error
$::RD_WARN   = 1; #Enable warnings - warn on unused rules &c.
$::RD_HINT   = 1; # Give out hints to help fix problems.

our %HASH;

my $grammar = <<'END_OF_GRAMMAR';


    startrule : from_clause 

    from_clause : 'from' dir(s) 
                {
                    print "-->@{$item{'dir(s)'}}<-- \n";
                    $main::HASH{dirs} = $item{'dir(s)'}; 
                }

    dir : 'hello'
        | 'world'

END_OF_GRAMMAR


my $parser = Parse::RecDescent->new($grammar);
$parser->startrule("from hello world hello"); 
use Data::Dumper;
say Dumper(\%HASH);

--output:--
-->hello world<-- 
$VAR1 = {
          'dirs' => [
                      'hello',
                      'world'
                    ]
        };

Next up, the regex problem. This doesn't work:

my $grammar = <<'END_OF_GRAMMAR';
    
    #Start up action(executed in parser namespace):
    { 
        use 5.012;  #So I can use say() 
    }

    startrule : from_clause 

    from_clause : 'from' dir(s) 
                { 
                    say "-->$item[0]<--";
                    say "-->@{$item[-1]}<--"; 
                }

    dir : m{/}

END_OF_GRAMMAR


my $parser = Parse::RecDescent->new($grammar);
$parser->startrule("from ./hello"); 

--output:--
(blank)

Note that I tried using the @item array this time. The first item in @item is the rule name, "from_clause", and the next items should be the matches for the subrules, so $item[2], or equivalently $item[-1], should be the matches for dir(s). But because I am not even seeing the arrows in my print statement, that means the parser isn't finding a match for my rule.

I also notice there are weird rules the parser follows for comments. This does not cause an error:

my $grammar = <<'END_OF_GRAMMAR';
    
    #Start up action(executed in parser namespace):
    { 
        use 5.012;  #So I can use say() 
    }

...

...but this does cause an error:

  my $grammar = <<'END_OF_GRAMMAR';
    
    #Start up action(executed in parser namespace):
    {  use 5.012;  #So I can use say()  }


...

--output:--
Unknown starting rule (Parse::RecDescent::namespace000001::startrule) 
+called
 at 3.pl line 76.

Back to the regex problem. It seems that Parse::RecDescent takes the regex pattern and adds a ^ to the beginning of the pattern and adds $ to the end of the pattern. In other words, the regex you specify has to match all of the text you are interested in examining.

my $grammar = <<'END_OF_GRAMMAR';
    
    #Start up action(executed in parser namespace):
    { 
        use 5.012;  #So I can use say() 
    }

    startrule : from_clause 

    from_clause : 'from' dir(s) 
                { say "-->$_<--" for @{ $item{'dir(s)'} };  }
  

    dir : m{\S* / \S*}xms

END_OF_GRAMMAR

my $parser = Parse::RecDescent->new($grammar);
$parser->startrule("from ./hello hello/world"); 

--output:--
-->./hello<--
-->hello/world<--

And reworking my original example:

use strict; 
use warnings; 
use 5.012;

use Parse::RecDescent;


$::RD_ERRORS = 1; #Parser dies when it encounters an error
$::RD_WARN   = 1; #Enable warnings - warn on unused rules &c.
$::RD_HINT   = 1; # Give out hints to help fix problems.

our %HASH;

my $grammar = <<'END_OF_GRAMMAR';
    
    #Start up action(executed in parser namespace):
    { 
        use 5.012;  #So I can use say() 
    }

    startrule : from_clause 

    from_clause : 'from' dir(s) 
                { 
                  say "-->$_<--" for @{ $item{'dir(s)'} };
                  $main::HASH{target_dirs} = $item{'dir(s)'};  
                }

    dir : m{\S* / \S*}xms

END_OF_GRAMMAR


my $parser = Parse::RecDescent->new($grammar);
$parser->startrule("from ./hello hello/world"); 
use Data::Dumper;
say Dumper(\%HASH);

--output:--
-->./hello<--
-->hello/world<--
$VAR1 = {
          'target_dirs' => [
                             './hello',
                             'hello/world'
                           ]
        };

Success! Thanks.