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

NAME

Net::DAAP::DMAP - Perl module for reading and writing DAAP structures

SYNOPSIS

  use Net::DAAP::DMAP qw(:all);

  $hash_ref = dmap_to_hash_ref($dmap);       # crude
  $array_ref = dmap_to_array_ref($dmap);     # crude

  $array_ref = dmap_unpack($dmap);           # knows about data types
  $node      = dmap_seek($array_ref, $path);

  $flattened = dmap_flatten($array_ref);     # convert to path = data formta
  $flat_list = dmap_flat_list($array_ref);   # convert to [ path, data ] format
  $xml       = dmap_to_xml($dmap);           # convert to XML fragment
  $dmap      = dmap_pack($dmap);             # convert to DMAP packet
  update_content_codes($unpacked_content_codes_response);

DESCRIPTION

WARNING!

Until 2.0, I reserve the right to change the interface. In particular, I think dmap_flatten, dmap_to_hash_ref, and dmap_to_array_ref are likely to disappear. And I suspect the hive brain of Perl can come up with a better data structure than I have.

Back to the Description

A DMAP structure is a binary record used in Apple's DAAP protocol. A DMAP structure may contain other DMAP structures. Fields in a DMAP structure are identified by a short name ("msdc"). The short name is what's in the binary record, but a content codes list gives a long name ("dmap.databasescount") and a data type for the record (32-bit integer).

A parsed DMAP structure is built out of arrays. For example:

  [
    [
      'dmap.loginresponse',
      [
         [
           'dmap.status',
           200
         ],
         [
           'dmap.sessionid',
            2393
         ]
      ]
    ]
  ]

(dmap_unpack returns this kind of structure)

There are two rules here: a field is wrapped in an array, and a container's values are wrapped in an array. So the structure is programmatically built as:

  $status_field = [ 'dmap.status', 200 ];
  $session_id_field = [ 'dmap.sessionid', 2393 ];
  $response_value = [ $status_field, $session_id_field ];
  $login_response_field = [ 'dmap.loginresponse', $response_value ];
  $entire_response = [ $login_response_field ];

The outer array is necessary because not every response has only one top-level container as this does.

In XML you'd write the response as:

  <dmap.loginresponse>
      <dmap.status>200</dmap.status>
      <dmap.sessionid>2393</dmap.sessionid>
  </dmap.loginresponse>

This is what dmap_to_xml returns.

A much more convenient structure for representing this data would be:

  {
    'dmap.loginresponse' => {
      { 'dmap.status' => 200,
        'dmap.sessionid' => 2393,
      },
  }

This is the output of dmap_to_hash_ref, but beware! This isn't suitable for every response. The hash is indexed by field name and a structure may contain many elements of the same name. For example, requesting the content codes list gives you a list of records that have the field name dmap.dictionary.

The array structure returned by dmap_to_array_ref is complex, but the dmap_seek function makes it easier. This takes a structure and a path expressed as a slash-separated list of field names:

  dmap.loginresponse/dmap.sessionid

The return value is the the value of the first dmap.sessionid found in the first dmap.loginresponse structure. In the case of the sample record above, it would be 2393.

Another way to handle these complex arrays is to dmap_flatten them. This returns an array of "path = value" lines, where path is a slash-separated path. For example:

  [
    '/dmap.loginresponse/dmap.status = 200',
    '/dmap.loginresponse/dmap.sessionid = 2393'
  ]

You can use grep and regexps to find data if that's the way your mind works.

dmap_flatten has a similar looking cousin called dmap_flat_list, which returns an array of "path => value" pairs. For example:

  [
    '/dmap.loginresponse/dmap.status' => 200,
    '/dmap.loginresponse/dmap.sessionid' => 2393,
  ]

You can then turn this into a hash (which may of course lose you the first elements), or iterate over it in pairs, if that's easier.

You can, but don't have to, update the tables of field names ("content codes") and data types. DAAP offers a request that returns a packet of content codes. Feed that packet to update_content_codes.

Implementation Details

It's all implementation details. Here are the various data types.

 1, 3, 5, 7 = ints, size 8,16,32,64 bit
 9 = string, 10 = time_t-style time
 11 = version (two 16-bit ints, I think)
 12 = container

This uses Math::BigInt for 64-bit quantities, as not every platform has 64-bit int support available.

There's no support for types 2, 4, 6, 8 yet because nobody'd found examples of them in the field: are they endian changes, or signedness changes. The assumption is that all numbers are unsigned (why allow the possibility of a negative number of songs?).

AUTHOR

Nathan Torkington, <nathan AT torkington.com>. For support, join the DAAP developers mailing list by sending mail to <daap-dev-subscribe AT develooper.com>.

Richard Clamp <richardc@unixbeard.net> is the current maintainer, and took over in July 2004.