The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.

NAME

Python::Serialise::Marshal - a module for reading and writing marshalled Python files

SYNOPSIS

    use Python::Serialise::Marshal;

    my $pr = Python::Serialise::Marshal->new("file/for/reading");
    while (my $data = $pr->load()) {
          print Dumper $data;
    }

    my $pw = Python::Serialise::Marshal->new(">file/for/writing");
    
    $pw->dump(['a' 'list']);
    $pw->dump("a string");
    $pw->dump(42);
    $pw->dump({'a'=>'hash'});

    $pw->close();  

DESCRIPTION

Marshalling is a method of serialising files in Python (another method, Pickling, is also available). It is the method that Mailman uses to store its config files.

This module is an attempt to write a pure Perl implementation of the algorithm.

METHODS

new <filename>

Open a file for reading or writing. Can take any arguments that IO::File can.

load

Returns the next data structure from the marshalled file or undef.

dump <data structure>

Takes a ref to an array or a hash or a number or string and pickles it.

Structures may be nested.

close

Closes the current file.

NOTES

Complex numbers

Python has inbuilt support form complex numbers whilst Perl provides it through the core module Math::Complex. Unserialising a Python complex number will return a Math::Complex object and, as you'd expect, serialising something that ISA Math::Complex will result in a serialised Python complex number.

None

Python has None objects, similar to Perl's undef.

Because load indictaes "no more objects" by returning undef we have to return Python::Serialise::None objects. However dump can take undef and serialise it as a None object.

BUGS

Much less than my Pickle module because this is a much saner file format.

Tests for None

I can't think of a nice elegant way of doing tests at the moment.

I'm sure I will soon.

Longs

There's no support for longs. I've figured out how to write them in Python but I just can't seem to extract them properly.

Unicode

Not an itch that needs scratching at the moment so there's no support.

Code

Ditto

ALTERNATIVES

You could always dump the data structure out as YAML in Python and then read it back in with YAML in Perl.

I also may look into wrapping the Python source code file in XS.

AUTHOR

Simon Wistow <simon@thegestalt.org>

COPYRIGHT

(c) 2003 Simon Wistow

Distributed under the same terms as Perl itself.

This software is under no warranty and will probably ruin your life, kill your friends, burn your house and bring about the apocalypse.

SEE ALSO

http://www.python.org, YAML, File::Binary, Math::Complex and the RESOURCES file in this distribution.