lib/WebPAC/Parser.pm

package WebPAC::Parser;

use warnings;
use strict;

use base qw/WebPAC::Common WebPAC::Normalize/;

use PPI;
use PPI::Dumper;
use Data::Dump qw/dump/;
use File::Slurp;


=head1 NAME

WebPAC::Parser - parse perl normalization configuration files and mungle it

=head1 VERSION

Version 0.01

=cut

our $VERSION = '0.01';

=head1 SYNOPSIS

FIXME

=head1 FUNCTIONS

=head2 new

Create new parser object.

  my $parser = new WebPAC::Parser(
        config => new WebPAC::Config(),
        base_path => '/optional/path/to/conf',
  );

=cut

sub new {
        my $class = shift;
        my $self = {@_};
        bless($self, $class);

        my $log = $self->_get_logger();

        $log->logdie("config isn't WebPAC::Config") unless ($self->{config} && $self->{config}->isa('WebPAC::Config'));

        $log->logdie("can't iterate_inputs over this config object") unless ($self->{config}->can('iterate_inputs'));

        my $source;

        $self->{config}->iterate_inputs( sub {
                my $input = shift;
                my $path = $input->{normalize}->{path} || return;
                my $full = $self->{base_path} ? $self->{base_path} . '/' . $path : $path;
                $log->logdie("normalization input $full doesn't exist") unless (-e $full);
                my $s = read_file( $full ) || $log->logdie("can't read $full: $!");
                $log->debug("adding $path to parser [",length($s)," bytes]");
                $source .= $s;
        } );

        $log->debug("collected ", length($source), " bytes of source");

        $self->{source} = $source;

        $self ? return $self : return undef;
}

=head2 parse

=cut

sub parse {
        my $self = shift;

        my $log = $self->_get_logger();

        $log->logdie('no source found in object') unless ($self->{source});

        my $Document = PPI::Document->new( \$self->{source} ) || $log->logdie("can't parse source:\n", $self->{source});

        $Document->prune('PPI::Token::Whitespace');
        #$Document->prune('PPI::Token::Operator');

        # Find all the named subroutines

        my $eval_create;

        $Document->find( sub {
                        my ($Document,$Element) = @_;

                        $Element->isa('PPI::Token::Word') or return '';
                        $Element->content eq 'lookup' or return '';

                        $log->debug("expansion: ", $Element->snext_sibling);

                        my $args = $Element->snext_sibling;
                
                        my @e = $args->child(0)->elements;
                        $log->logdie("hum, expect at least 8 elements, got ", scalar @e, " in $args") if ($#e < 8);

                        if ($log->is_debug) {
                                my $report = "found " . scalar @e . " elements:\n";

                                foreach my $i ( 0 .. $#e ) {
                                        $report .= sprintf("# %-2d: %-30s %s\n", $i, ( $e[$i] || 'undef' ), $e[$i]->class );
                                }

                                $log->debug($report);
                        }

                        my $key_element = $e[8]->clone;

                        $log->logdie("key element must be PPI::Structure::Block") unless $key_element->isa('PPI::Structure::Block');

                        $log->debug("key part: ", $key_element);

                        my @key;

                        $key_element->find( sub {
                                my $e = $_[1] || die "no element?";
                                $e->isa('PPI::Token::Word') or return '';
                                $e->content eq 'rec' or return '';

                                my $kf = $e->snext_sibling;

                                $log->debug("key fragment = $kf");

                                push @key, eval $kf;
                                $log->logdie("can't eval { $kf }: $@") if ($@);

                                return 1;
                        });

                        my $key = join('-', @key ) || $log->logdie("no key found!");

                        $log->debug("key = $key");

                        my $create = ' 
                                $coderef = ' . $e[7] . $e[8] . ';
                                foreach my $v ($coderef->()) {
                                        next unless (defined($v) && $v ne \'\');
                                        push @{ $lookup->{\'' . $key . '\'}->{$v} }, $mfn;
                                }
                        ';

                        $log->debug("create: $create");

                        $create =~ s/\s+/ /gs;
                        $eval_create->{ $e[3] }->{ $e[5] } .= $create;

                        if ($#e < 10) {
                                $e[8]->insert_after( $e[8]->clone );
                                $e[8]->insert_after( $e[7]->clone );
                                $e[8]->insert_after( $e[6]->clone );
                        }

                        $e[7]->remove;
                        $e[8]->insert_before( new PPI::Token::Quote::Single( "'$key'" ) );
                        $e[8]->remove;


                        $log->debug(">>> ", $Element->snext_sibling);
        });

        $log->info("create: ", dump($eval_create) );
        $log->info("lookup: ", $Document->serialize );

        if ($self->{debug}) {
                my $Dumper = PPI::Dumper->new( $Document );
                $Dumper->print;
        }

        return 1;
}

=head1 AUTHOR

Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>

=head1 COPYRIGHT & LICENSE

Copyright 2006 Dobrica Pavlinusic, All Rights Reserved.

This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.

=cut

1; # End of WebPAC::Parser
1	dpavlin	686	package WebPAC::Parser;
2
3			use warnings;
4			use strict;
5
6			use base qw/WebPAC::Common WebPAC::Normalize/;
7
8			use PPI;
9			use PPI::Dumper;
10			use Data::Dump qw/dump/;
11			use File::Slurp;
12
13
14			=head1 NAME
15
16			WebPAC::Parser - parse perl normalization configuration files and mungle it
17
18			=head1 VERSION
19
20			Version 0.01
21
22			=cut
23
24			our $VERSION = '0.01';
25
26			=head1 SYNOPSIS
27
28			FIXME
29
30			=head1 FUNCTIONS
31
32			=head2 new
33
34			Create new parser object.
35
36			my $parser = new WebPAC::Parser(
37			config => new WebPAC::Config(),
38			base_path => '/optional/path/to/conf',
39			);
40
41			=cut
42
43			sub new {
44			my $class = shift;
45			my $self = {@_};
46			bless($self, $class);
47
48			my $log = $self->_get_logger();
49
50			$log->logdie("config isn't WebPAC::Config") unless ($self->{config} && $self->{config}->isa('WebPAC::Config'));
51
52			$log->logdie("can't iterate_inputs over this config object") unless ($self->{config}->can('iterate_inputs'));
53
54			my $source;
55
56			$self->{config}->iterate_inputs( sub {
57			my $input = shift;
58			my $path = $input->{normalize}->{path} \|\| return;
59			my $full = $self->{base_path} ? $self->{base_path} . '/' . $path : $path;
60			$log->logdie("normalization input $full doesn't exist") unless (-e $full);
61			my $s = read_file( $full ) \|\| $log->logdie("can't read $full: $!");
62			$log->debug("adding $path to parser [",length($s)," bytes]");
63			$source .= $s;
64			} );
65
66			$log->debug("collected ", length($source), " bytes of source");
67
68			$self->{source} = $source;
69
70			$self ? return $self : return undef;
71			}
72
73			=head2 parse
74
75			=cut
76
77			sub parse {
78			my $self = shift;
79
80			my $log = $self->_get_logger();
81
82			$log->logdie('no source found in object') unless ($self->{source});
83
84			my $Document = PPI::Document->new( \$self->{source} ) \|\| $log->logdie("can't parse source:\n", $self->{source});
85
86			$Document->prune('PPI::Token::Whitespace');
87			#$Document->prune('PPI::Token::Operator');
88
89			# Find all the named subroutines
90
91			my $eval_create;
92
93			$Document->find( sub {
94			my ($Document,$Element) = @_;
95
96			$Element->isa('PPI::Token::Word') or return '';
97			$Element->content eq 'lookup' or return '';
98
99	dpavlin	687	$log->debug("expansion: ", $Element->snext_sibling);
100	dpavlin	686
101			my $args = $Element->snext_sibling;
102
103			my @e = $args->child(0)->elements;
104	dpavlin	687	$log->logdie("hum, expect at least 8 elements, got ", scalar @e, " in $args") if ($#e < 8);
105	dpavlin	686
106	dpavlin	687	if ($log->is_debug) {
107			my $report = "found " . scalar @e . " elements:\n";
108	dpavlin	686
109	dpavlin	687	foreach my $i ( 0 .. $#e ) {
110			$report .= sprintf("# %-2d: %-30s %s\n", $i, ( $e[$i] \|\| 'undef' ), $e[$i]->class );
111			}
112
113			$log->debug($report);
114	dpavlin	686	}
115
116			my $key_element = $e[8]->clone;
117
118	dpavlin	687	$log->logdie("key element must be PPI::Structure::Block") unless $key_element->isa('PPI::Structure::Block');
119	dpavlin	686
120	dpavlin	687	$log->debug("key part: ", $key_element);
121	dpavlin	686
122			my @key;
123
124			$key_element->find( sub {
125			my $e = $_[1] \|\| die "no element?";
126			$e->isa('PPI::Token::Word') or return '';
127			$e->content eq 'rec' or return '';
128
129			my $kf = $e->snext_sibling;
130
131	dpavlin	687	$log->debug("key fragment = $kf");
132	dpavlin	686
133			push @key, eval $kf;
134	dpavlin	687	$log->logdie("can't eval { $kf }: $@") if ($@);
135	dpavlin	686
136			return 1;
137			});
138
139	dpavlin	687	my $key = join('-', @key ) \|\| $log->logdie("no key found!");
140	dpavlin	686
141	dpavlin	687	$log->debug("key = $key");
142	dpavlin	686
143			my $create = '
144			$coderef = ' . $e[7] . $e[8] . ';
145			foreach my $v ($coderef->()) {
146			next unless (defined($v) && $v ne \'\');
147			push @{ $lookup->{\'' . $key . '\'}->{$v} }, $mfn;
148			}
149			';
150
151	dpavlin	687	$log->debug("create: $create");
152	dpavlin	686
153			$create =~ s/\s+/ /gs;
154			$eval_create->{ $e[3] }->{ $e[5] } .= $create;
155
156			if ($#e < 10) {
157			$e[8]->insert_after( $e[8]->clone );
158			$e[8]->insert_after( $e[7]->clone );
159			$e[8]->insert_after( $e[6]->clone );
160			}
161
162			$e[7]->remove;
163			$e[8]->insert_before( new PPI::Token::Quote::Single( "'$key'" ) );
164			$e[8]->remove;
165
166
167	dpavlin	687	$log->debug(">>> ", $Element->snext_sibling);
168	dpavlin	686	});
169
170	dpavlin	687	$log->info("create: ", dump($eval_create) );
171			$log->info("lookup: ", $Document->serialize );
172	dpavlin	686
173	dpavlin	687	if ($self->{debug}) {
174			my $Dumper = PPI::Dumper->new( $Document );
175			$Dumper->print;
176			}
177	dpavlin	686
178	dpavlin	687	return 1;
179	dpavlin	686	}
180
181			=head1 AUTHOR
182
183			Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
184
185			=head1 COPYRIGHT & LICENSE
186
187			Copyright 2006 Dobrica Pavlinusic, All Rights Reserved.
188
189			This program is free software; you can redistribute it and/or modify it
190			under the same terms as Perl itself.
191
192			=cut
193
194			1; # End of WebPAC::Parser