SYNOPSIS

       use Data::Crumber; # imports `crumbr`
    
       # some data to work with
       my $data = { what => 'ever', hey => 'you' };
    
       # crumbr provides an anonyous sub back. This has defaults
       my $csub = crumbr();
    
       # use it to encode the data
       my $encoded = $crumbr->($data);
       # {"here"}{"hey"}:"you"
       # {"here"}{"what"}:"ever"
    
       # URI profile simplifies things but loses something
       $encoded = crumbr(profile => 'URI')->($data);
       # here/hey "you"
       # here/what "ever"
    
       # JSON profile produces valid JSON "slices"
       $encoded = crumbr(profile => 'JSON')->($data);
       # {"here":{"hey":"you"}}
       # {"here":{"what":"ever"}}
    
       # Object Oriented Interface
       my $crobj = Data::Crumbr->new();
       $encoded = $crobj->encode($data); # same as default

DESCRIPTION

    Data::Crumbr lets you render data structures in a way that can then be
    easily searched and parsed in "slices". The basic idea is that data
    shaped in this way will then be easily filtered in the shell for
    extracting interesting parts.

    The input data structure is traversed is if it is a tree (so no
    circular structures please!), and a record is generated for each leaf
    in the tree. Depending on the backend and the configurations, the full
    path from the root to the parent of the leaf is represented as a
    sequence of keys (which can be hash keys or array indexes) followed by
    the value. This should make your life easier e.g. in the shell, so that
    you can specify the full path to the data structure part you're
    interested into with common Unix tools like grep and/or sed.

 Example

    Suppose you have the following data structure in Perl:

       my $data = {
          one => '1',
          two => 2,
          three => 3.1,
          four => '4.0',
          true => \1,
          false => \0,
          array => [
             qw< what ever >,
             { inner => 'part', empty => [] }
           ],
          hash => {
             'with ♜' => {},
             ar => [ 1..3 ],
             something => "funny \x{263A} ☻",
          },
       };

    If you encode this e.g. in JSON, it will be easy to parse with the
    right program, but not from the shell, even if you pretty print it:

       {
          "hash" : {
             "something" : "funny ☺ ☻",
             "with ♜" : {},
             "ar" : [
                1,
                2,
                3
             ]
          },
          "one" : "1",
          "array" : [
             "what",
             "ever",
             {
                "inner" : "part",
                "empty" : []
             }
          ],
          "four" : "4.0",
          "true" : true,
          "two" : 2,
          "three" : 3.1,
          "false" : false
       }

    How do you get the second item in the array ari inside the hash hash?
    Would you do better with YAML instead?

       ---
       array:
         - what
         - ever
         - empty: []
           inner: part
       false: !!perl/ref
         =: 0
       four: 4.0
       hash:
         ar:
           - 1
           - 2
           - 3
         something: funny ☺ ☻
         with ♜: {}
       one: 1
       three: 3.1
       true: !!perl/ref
         =: 1
       two: 2

    Not really. Data::Crumbr lets you represent the data in a more verbose
    but easily consumable way for the shell. Hence, this:

       use Data::Crumbr;
       print crumbr()->($data), "\n";

    will give you this:

       {"array"}[0]:"what"
       {"array"}[1]:"ever"
       {"array"}[2]{"empty"}:[]
       {"array"}[2]{"inner"}:"part"
       {"false"}:false
       {"four"}:"4.0"
       {"hash"}{"ar"}[0]:1
       {"hash"}{"ar"}[1]:2
       {"hash"}{"ar"}[2]:3
       {"hash"}{"something"}:"funny \u263A \u263B"
       {"hash"}{"with \u265C"}:{}
       {"one"}:"1"
       {"three"}:3.1
       {"true"}:true
       {"two"}:2

    Now it should pretty easy for a shell program to get at the data, e.g.
    with this sed substitution:

       sed -ne 's/^{"hash"}{"ar"}\[2\]://p'

 Profiles

    If you don't like the default encoding, you can get a different one by
    using a profile. This is a set of configurations for
    Data::Crumbr::Default, which is a pretty generic class for representing
    a wide class of possible record-oriented encodings.

    A Data::Crumbr::Default encoder is defined in terms of the following
    parameters:

    array_open

      sequence to put when an array is opened

    array_close

      sequence to put when an array is closed

    array_key_prefix

      sequence to put before an array's index

    array_key_suffix

      sequence to put after an array's index

    array_key_encoder

      a reference to a function that encodes an array's index

    hash_open

      sequence to put when a hash is opened

    hash_close

      sequence to put when a hash is closed

    hash_key_prefix

      sequence to put before a hash's key

    hash_key_suffix

      sequence to put after a hash's key

    hash_key_encoder

      a reference to a function that encodes a hash's key

    value_encoder

      a reference to a function that encodes a leaf value

    keys_separator

      sequence to separate the keys breadcrumb

    value_separator

      sequence to separate the keys from the value

    By default, Data::Crumbr ships with the following profiles:

    Default

      i.e. the profile you get by default, and what you saw in action in
      the example above. It has the following settings:

	* no openers and closers:

           array_open  => ''
           array_close => ''
           hash_open   => ''
           hash_close  => ''

	* array keys are printed verbatim, surrounded by square brackets:

           array_key_prefix  => '['
           array_key_suffix  => ']'
           array_key_encoder => Data::Crumbr::Util::id_encoder

	* hash keys encoded as JSON strings, surrounded by curly brackets:

           hash_key_prefix   => '['
           hash_key_suffix   => ']'
           hash_key_encoder  => Data::Crumbr::Util::json_leaf_encoder

	* no separator between keys (because they already stand out very
	clearly, but a colon to separate the sequence of keys from the
	value:

           keys_separator  => ''
           value_separator => ':'

	* leaf values encoded as JSON scalars:

           value_encoder => Data::Crumbr::Util::json_leaf_encoder

      This is quite verbose, but lets you specify very precisely what you
      are looking for because the hash keys stand out clearly with respect
      to array identifiers, i.e. there's no chance that you will mistake an
      array index for a hash key (because they are embedded in different
      bracket types).

    JSON

      this profile always provides you compact JSON-compliant string
      representations that contain only one single leaf value.

      It has the following characteristics:

	* openers and closers are what you would expect for JSON objects
	and arrays:

           array_open  => '['
           array_close => ']'
           hash_open   => '{'
           hash_close  => '}'

	* there is only one non-empty suffix, i.e. the hash key suffix, so
	that we can separate the hash key from the value with : according
	to JSON:

           array_key_prefix => ''
           array_key_suffix => ''
           hash_key_prefix  => ''
           hash_key_suffix  => ':'

	* array keys are not printed:

           array_key_encoder => sub { }

	* hash keys are JSON encoded:

           hash_key_encoder  => Data::Crumbr::Util::json_encoder()

	* no separators are needed:

           keys_separator  => ''
           value_separator => ''

	* leaf values encoded as JSON scalars:

           value_encoder => Data::Crumbr::Util::json_leaf_encoder

    URI

      this is the simplest of the profiles, and sacrifices the possibility
      to distinguish between hash and array keys to the altar of
      simplicity.

      It has the following characteristics:

	* no openers, closers, prefixes or suffixes:

           array_open  => ''
           array_close => ''
           array_key_prefix => ''
           array_key_suffix => ''
        
           hash_open   => ''
           hash_close  => ''
           hash_key_prefix => ''
           hash_key_suffix => ''

	* array keys are printed verbatim

	* hash keys are URI encoded

           hash_key_encoder  => Data::Crumbr::Util::uri_encoder

	* keys are separated by a slash character / and values are
	separated by a single space  :

           keys_separator  => '/'
           value_separator => ' '

	* leaf values encoded as JSON scalars:

           value_encoder => Data::Crumbr::Util::json_leaf_encoder

INTERFACE

    There are two ways to use Data::Crumber: a function crumbr, that is
    exported by default, and the object-oriented interface.

    crumbr

         $subref = crumbr(%args); # OR
         $subref = crumbr(\%args);

      get a crumbr generator based on provided %args.

      Returns a reference to a sub, which can then be called upon a data
      structure in order to get the crumbed version.

      The input arguments can be:

      encoder

	details about the encoder, see "Profiles" for the available
	key-value pairs. In addition, you can also set the following:

	output

	  the output channel to use for sending encoded data. This can be:

	    * filename

	    this will be opened in raw mode and used to send the output

	    * filehandle

	    used directly

	    * array reference

	    each output line will be pushed as a new element in the array

	    * object reference

	    which is assumed to support the print() method, that will be
	    called with each generated line

	    * sub reference

	    which will be called with each generated line

      profile

	the name of a profile to use as a base - see "Profiles". Settings
	in the profile are always overridden by corresponding ones in the
	provided encoder, if any.

    encode

         $dc->encode($data_structure);

      generate the encoding for the provided $data_structure. Output is
      generated depending on how it is specified, see "crumbr" above.

    new

         my $dc = Data::Crumber->new(encoder => \%args);

      create a new instance of Data::Crumbr. Data provided for the encoder
      parameter (i.e. %args) are those discussed in "Profiles".

      The new instance can then be used to encode data using the /encode
      method.