chahn has asked for the wisdom of the Perl Monks concerning the following question:

Hello all,

I need to parse property files that might have multiline entries (otherwise a simple regexp would do) and I need to process comments and maintain order (otherwise Config::Properties would do)

I thought to use a positive lookahead to define a var=val pair to start with ^var= and end just before the next ^var=

I am probably needing a nudge in using either the skip directive, or the proper pattern modifier....but here is sample code: 

use strict;
use Parse::RecDescent;

$::TestGrammar = <<'TG';

  Output: PropLine(s) /\Z/

  PropLine: CommentLine | SimpleProp | LastProp

  CommentLine: /\#.*\n/ { print "RULE: $item{__RULE__}\n"; 
                          print "MATCH: $item{__PATTERN1__}\n";
                        }

  SimpleProp: VAR EQ VAL ...LastProp { print "RULE: $item{__RULE__}\n"
+;
                                       print "VAR: $item{VAR}\n";
                                       print "EQ: $item{EQ}\n";
                                       print "VAL: $item{VAL}\n\n"; }

  LastProp: VAR EQ VAL { print "RULE: $item{__RULE__}\n";
                         print "VAR: $item{VAR}\n";
                         print "EQ: $item{EQ}\n";
                         print "VAL: $item{VAL}\n\n"; }

  VAR: /[^=]+/

  EQ: '='

  VAL: /.*/

TG

undef $/;
my $foo = <>;

my $parser = Parse::RecDescent->new($::TestGrammar);

defined $parser->Output($foo) or die "FAILURE";

__END__

When I run this using this file:
===================
# Comment Line
# Comment #2
foo=this is property one but 
bar=does it grab this one too?
baz=snark
===================

I see this, as expected:
===================
RULE: CommentLine
MATCH: # Comment Line

RULE: CommentLine
MATCH: # Comment #2

RULE: SimpleProp
VAR: foo
EQ: =
VAL: this is property one but

RULE: SimpleProp
VAR: bar
EQ: =
VAL: does it grab this one too?

RULE: LastProp
VAR: baz
EQ: =
VAL: snark
===================

However, if I make one of the props multiline:
===================
# Comment Line
# Comment #2
foo=this is property 
one but 
bar=does it grab this one too?
baz=snark
===================

I see this:
===================
RULE: CommentLine
MATCH: # Comment Line

RULE: CommentLine
MATCH: # Comment #2

RULE: SimpleProp
VAR: foo
EQ: =
VAL: this is property   <----the rest of this VAL becomes
                             part of the next VAR.  no joy.
RULE: SimpleProp
VAR: one but
bar
EQ: =
VAL: does it grab this one too?

RULE: LastProp
VAR: baz
EQ: =
VAL: snark
===================

[download]

I am trying things like changing VAR: to /^[^=]+/m and such but have yet to find the right combination.
Thank you in advance for any comments you can make.

Replies are listed 'Best First'.
Re: Modifying Parse::RecDescent Grammar to deal with multiline property file entries
by suaveant (Parson) on Aug 15, 2007 at 20:33 UTC
    This seems to work... (your LastProp, in this context, was superflous)

    Update Kind of works, but I realized it allows continuation lines before a prop 

    use strict;
use Parse::RecDescent;

$::TestGrammar = <<'TG';

Output: PropLine(s) /\Z/

PropLine: CommentLine | SimpleProp | ContinuationLine

CommentLine: /\#.*\n/ { print "RULE: $item{__RULE__}\n";
                        print "MATCH: $item{__PATTERN1__}\n";
                    }

SimpleProp: VAR EQ VAL { print "RULE: $item{__RULE__}\n";
                         print "VAR: $item{VAR}\n";
                         print "EQ: $item{EQ}\n";
                         print "VAL: $item{VAL}\n\n"; }


VAR: /^[^=\n]+/

EQ: '='

VAL: /.+/

ContinuationLine: VAL { print "RULE: $item{__RULE__}\n";
                        print "VAL: $item{VAL}\n\n"; }

TG

    undef $/;
my $foo = <DATA>;

my $parser = Parse::RecDescent->new($::TestGrammar);

defined $parser->Output($foo) or die "FAILURE";

__DATA__
# Comment Line
# Comment #2
foo=this is property
one but
bar=does it grab this one too?
baz=snark

    [download]

                    - Ant
                    - Some of my best work - (1 2 3)

[reply]
[d/l]
      Thank you for the suggestion.

      I need to work on each property as a whole, but could still piece together your ContinuationLine with its corresponding SimpleProp....

      In any case, what I was hoping to express was that a property is everything from a line beginning with Var=Val and continuing up to, but not including, the next line beginning with Var=Val. This seems a likely spot for a look-ahead.

      I am going to brute force the situation, using regexps and flags to indicate when in a Property, but do hope to poke on this when I can. I will certainly post back any useful conclusions.
[reply]
        yeah you could do something like 
        /.*(?=(^[^=]+=|\Z))/ # not tested, but should be about right

        [download]
        Or you could use the code... set a flag when you find a propname and unset only except continuation lines when it is set. If you need to look at the prop as one piece just build a parse tree... that usually is the better way to go unless you are working on a huge dataset or a real time stream. And, as you say, you could pop it back together in simpleprop.

        Of course.. unless you need them it is often easier to pre-process out the comments.

                        - Ant
                        - Some of my best work - (1 2 3)

[reply]
[d/l]
Re: Modifying Parse::RecDescent Grammar to deal with multiline property file entries
by princepawn (Parson) on Aug 17, 2007 at 07:51 UTC
[reply]

Back to Seekers of Perl Wisdom