parse-wormedit: seperate parsing module Parse::Binary::Nested

Make the package a (mostly) stand-alone include, potientially reusable by
other projects, but primarily keeping it more maintainable (uncluttered
code, testable, and allowing futher modularisation of file parsers).

diff --git a/Parse/Binary/Nested.pm b/Parse/Binary/Nested.pm
new file mode 100644 (file)
index 0000000..db5e208
--- /dev/null
+++ b/Parse/Binary/Nested.pm
@@ -0,0 +1,94 @@
+package Parse::Binary::Nested;
+
+use strict;
+use warnings;
+
+use Carp;
+
+our $VERSION = '1.00';
+
+sub new {
+       my ($class, $format) = @_;
+       ref $format eq 'ARRAY'
+               or croak "Invalid Parse::Binary::Nested format: should be an array ref";
+       bless $format, $class;
+}
+
+sub template {
+       my ($self, $format) = @_;
+       # total (flattened) unpack template from nested format definitions
+       return join '', map {
+               my $value = $format->[-($_ << 1) - 1];
+               if (ref $value eq 'ARRAY') {
+                       my $count = $value->[0];
+                       $value = $self->template($value);
+                       $value = $count =~ s/^([*\d]+)// ? "$count($value)$1"
+                               : $count."X[$count]$count/($value)";
+               }
+               else {
+                       $value =~ s/^C(a)(\d+)/$1 . ($2 + 1)/e;  # length prefix
+               }
+               $value;
+       } reverse 0 .. ($#$format - 1) >> 1;
+}
+
+sub convert {
+       my ($self, $format, $data) = @_;
+       # map flat results into a named and nested hash
+       my %res;
+       while (my ($field, $template) = splice @$format, 0, 2) {
+               if (ref $template eq 'ARRAY') {
+                       my ($count, @subformat) = @$template;
+                       my $max = $count =~ s/^(\d+)// ? $1 : 0;
+                       $count = !$count ? $max
+                               : $count eq '*' ? $res{levelcount}->{total} : shift @$data;
+                       $res{$field}->[$_] = $self->convert([@subformat], $data) for 0 .. ($max || $count)-1;
+                       splice @{ $res{$field} }, $count if $max > $count;
+                       $res{$field} = $res{$field}->[0] if $max == 1;
+                       next;
+               }
+               elsif ($template =~ /^Ca/) {
+                       $data->[0] = unpack 'C/a', $data->[0];
+               }
+               $res{$field} = shift @$data;
+       }
+       return \%res;
+}
+
+sub unpackf {
+       my ($self, $input) = @_;
+       my @data = unpack $self->template($self), $input;
+       return $self->convert([@$self], \@data);
+}
+
+1;
+
+=head1 NAME
+
+Parse::Binary::Nested - Structured unpack
+
+=head1 SYNOPSIS
+
+       use Parse::Binary::Nested;
+       my $parser = Parser::Binary::Nested->new([
+               foos => [
+                       'C', # count
+                       message => 'Z*',
+                       period  => 'C',
+               ],
+               trail => 'a*',
+       ]);
+       
+       my $data = $parser->unpackf("\1foo\0.rest");
+       print $data->{foos}->[0]->{message};
+
+=head1 DESCRIPTION
+
+=head1 AUTHOR
+
+Mischa POSLAWSKY <perl@shiar.org>
+
+=head1 LICENSE
+
+GPL version 3.
+

diff --git a/parse-wormedit b/parse-wormedit
index 84e4e31c0140810983205359b0b05df2d3a54283..3811863103ea42ea4719a1310eca52e4852b89b1 100755 (executable)
--- a/parse-wormedit
+++ b/parse-wormedit
@@ -6,7 +6,7 @@ use 5.010;
 use Data::Dumper;
 use Getopt::Long 2.33 qw(HelpMessage :config bundling);
 
 use Data::Dumper;
 use Getopt::Long 2.33 qw(HelpMessage :config bundling);
 

-our $VERSION = '1.03';
+our $VERSION = '1.04';

 
 GetOptions(\my %opt,
        'raw|r',  # full output
 
 GetOptions(\my %opt,
        'raw|r',  # full output


@@ -19,6 +19,8 @@ package Shiar_Parse::WormEdit;
 use strict;
 use warnings;
 
 use strict;
 use warnings;
 

+use Parse::Binary::Nested;
+

 our %MAGICID = (
        "WormEdit053\000LVL" => 53,
        "WormEdit\34195\000LVL" => 95,
 our %MAGICID = (
        "WormEdit053\000LVL" => 53,
        "WormEdit\34195\000LVL" => 95,


@@ -149,9 +151,9 @@ sub read {
        }
 
        # convert to an easily accessible hash
        }
 
        # convert to an easily accessible hash

-       my @values = unpack Shiar_Parse::Nested->template(\@FORMAT).'a*', $input;
-       my $data = Shiar_Parse::Nested->convert(\@FORMAT, \@values);
-       warn "Trailing data left unparsed\n" if grep {length} @values;
+       push @FORMAT, -trail => 'a*';
+       my $data = Parse::Binary::Nested->new(\@FORMAT)->unpackf($input);
+       warn "Trailing data left unparsed\n" if length delete $data->{-trail};

        $data->{format} = 'WormEdit';
        return $data;
 }
        $data->{format} = 'WormEdit';
        return $data;
 }


@@ -164,6 +166,7 @@ use warnings;
 
 use List::Util qw(sum min max);
 use Data::Dumper;
 
 use List::Util qw(sum min max);
 use Data::Dumper;

+use Parse::Binary::Nested;

 
 sub read {
        my ($self, $input) = @_;
 
 sub read {
        my ($self, $input) = @_;


@@ -288,7 +291,7 @@ sub read {
                }
        }
 
                }
        }
 

-       my $data = Shiar_Parse::Nested->unpack(\@FORMAT, $input);
+       my $data = Parse::Binary::Nested->new(\@FORMAT)->unpackf($input);

        my $offset = 0;
        $offsetbase += 1 + @{ $data->{sprite} } if $data->{sprite};
        $data->{moderef}->{offset}->{single} == $offsetbase
        my $offset = 0;
        $offsetbase += 1 + @{ $data->{sprite} } if $data->{sprite};
        $data->{moderef}->{offset}->{single} == $offsetbase


@@ -326,6 +329,7 @@ sub read {
                unshift @varform, name => 'Z*' unless $variant eq 'single' or $version <= 91;
                $varform[-1]->[0] = 1 if $variant eq 'race' and $version > 91;
                $varform[-1]->[0] = 2 if $variant eq 'ctf';
                unshift @varform, name => 'Z*' unless $variant eq 'single' or $version <= 91;
                $varform[-1]->[0] = 1 if $variant eq 'race' and $version > 91;
                $varform[-1]->[0] = 2 if $variant eq 'ctf';

+               my $parselevel = Parse::Binary::Nested->new(\@varform);

 
                while ($offset < length $data->{leveldata}) {
                        last if substr($data->{leveldata}, $offset, 1) eq chr(255);
 
                while ($offset < length $data->{leveldata}) {
                        last if substr($data->{leveldata}, $offset, 1) eq chr(255);


@@ -336,9 +340,7 @@ sub read {
                                $data->{moderef}->{start}->{$mode} = 1 + scalar @{ $data->{levels} };
                        }
 
                                $data->{moderef}->{start}->{$mode} = 1 + scalar @{ $data->{levels} };
                        }
 

-                       my $level = Shiar_Parse::Nested->unpack(
-                               [@varform], substr $data->{leveldata}, $offset
-                       );
+                       my $level = $parselevel->unpackf(substr $data->{leveldata}, $offset);

                        my $size = 8  # unpack length (ugh, ugly recalculation)
                                + (defined $level->{name} ? 1 + length $level->{name} : 0)
                                + 3 * (ref $level->{worms} eq 'ARRAY' ? scalar @{$level->{worms}} : 1)
                        my $size = 8  # unpack length (ugh, ugly recalculation)
                                + (defined $level->{name} ? 1 + length $level->{name} : 0)
                                + 3 * (ref $level->{worms} eq 'ARRAY' ? scalar @{$level->{worms}} : 1)


@@ -357,8 +359,8 @@ sub read {
                }
                else {
                        while (my $object = ord substr($data->{leveldata}, $offset+$size, 1)) {
                }
                else {
                        while (my $object = ord substr($data->{leveldata}, $offset+$size, 1)) {

-                               push @{ $level->{objects} }, Shiar_Parse::Nested->unpack(
-                                       [@OBJECTFORM], substr($data->{leveldata}, $offset+$size, 5)
+                               push @{ $level->{objects} }, Parse::Binary::Nested->new([@OBJECTFORM])->unpackf(
+                                       substr $data->{leveldata}, $offset+$size, 5

                                );
                                $size += 5;
                        }
                                );
                                $size += 5;
                        }


@@ -411,56 +413,6 @@ sub read {
 }
 
 
 }
 
 

-package Shiar_Parse::Nested;
-
-sub template {
-       my ($self, $format) = @_;
-       # total (flattened) unpack template from nested format definitions
-       return join '', map {
-               my $value = $format->[-($_ << 1) - 1];
-               if (ref $value eq 'ARRAY') {
-                       my $count = $value->[0];
-                       $value = $self->template($value);
-                       $value = $count =~ s/^([*\d]+)// ? "$count($value)$1"
-                               : $count."X[$count]$count/($value)";
-               }
-               else {
-                       $value =~ s/^C(a)(\d+)/$1 . ($2 + 1)/e;  # length prefix
-               }
-               $value;
-       } reverse 0 .. ($#$format - 1) >> 1;
-}
-
-sub convert {
-       my ($self, $format, $data) = @_;
-       # map flat results into a named and nested hash
-       my %res;
-       while (my ($field, $template) = splice @$format, 0, 2) {
-               if (ref $template eq 'ARRAY') {
-                       my ($count, @subformat) = @$template;
-                       my $max = $count =~ s/^(\d+)// ? $1 : 0;
-                       $count = !$count ? $max
-                               : $count eq '*' ? $res{levelcount}->{total} : shift @$data;
-                       $res{$field}->[$_] = $self->convert([@subformat], $data) for 0 .. ($max || $count)-1;
-                       splice @{ $res{$field} }, $count if $max > $count;
-                       $res{$field} = $res{$field}->[0] if $max == 1;
-                       next;
-               }
-               elsif ($template =~ /^Ca/) {
-                       $data->[0] = CORE::unpack 'C/a', $data->[0];
-               }
-               $res{$field} = shift @$data;
-       }
-       return \%res;
-}
-
-sub unpack {
-       my ($self, $format, $input) = @_;
-       my @data = CORE::unpack $self->template($format), $input;
-       return $self->convert($format, \@data);
-}
-
-

 package main;
 
 my @OBJTYPE = ('none', 'line', 'fat line', 'bar', 'circle');
 package main;
 
 my @OBJTYPE = ('none', 'line', 'fat line', 'bar', 'circle');

diff --git a/t/parser.t b/t/parser.t
new file mode 100644 (file)
index 0000000..cebd099
--- /dev/null
+++ b/t/parser.t
@@ -0,0 +1,33 @@
+#!/usr/bin/env perl
+
+use strict;
+use warnings;
+
+use Test::More;
+use Data::Dumper;
+
+plan tests => 5;
+
+use_ok('Parse::Binary::Nested');
+
+my $example = Parse::Binary::Nested->new([
+       foos => [
+               'C',
+               message => 'Z*',
+               period  => 'C',
+       ],
+       trail => 'a*',
+]);
+ok($example, 'example parser');
+my $data = $example->unpackf("\2foo\0!\0.rest");
+is(ref $data, 'HASH', 'output structure');
+is($data->{foos}->[1]->{period}, ord '.', 'sample element');
+
+is_deeply(
+       Parse::Binary::Nested->new(
+               [ lstr => 'Ca3', rest => 'a*' ]
+       )->unpackf("\2quux"),
+       { lstr => 'qu', rest => 'x' },
+       'length string'
+);
+