NAME

Data::Edit::Struct - Edit a Perl structure addressed with a Data::DPath path

VERSION

version 0.07

SYNOPSIS

 use Data::Edit::Struct qw[ edit ];
 
 
 my $src  = { foo => 9, bar => 2 };
 my $dest = { foo => 1, bar => [22] };
 
 edit(
     replace => {
         src   => $src,
         spath => '/foo',
         dest  => $dest,
         dpath => '/foo'
     } );
 
 edit(
     insert => {
         src   => $src,
         spath => '/bar',
         dest  => $dest,
         dpath => '/bar'
     } );
 
 # $dest = { foo => 9, bar => [ 2, 22 ] }

DESCRIPTION

Data::Edit::Struct provides a high-level interface for editing data within complex data structures. Edit and source points are specified via Data::DPath paths.

The destination structure is the structure to be edited. If data are to be inserted into the structure, they are extracted from the source structure. See "Data Copying" for the copying policy.

The following actions may be performed on the destination structure:

  • shift - remove one or more elements from the front of an array

  • pop - remove one or more elements from the end of an array

  • splice - invoke splice on an array

  • insert - insert elements into an array or a hash

  • delete - delete array or hash elements

  • replace - replace array or hash elements (and in the latter case keys)

  • transform - transform elements

Elements vs. Containers

Data::Edit::Struct operates on elements in the destination structure by following a Data::DPath path. For example, if

 $src  = { dogs => 'rule' };
 $dest = { bar  => [ 2, { cats => 'rule' }, 4 ] };

then a data path of

 /bar/*[0]

identifies the first element in the bar array. That element may be treated either as a container or as an element (this is specified by the "dtype" option).

In the above example, $dest->{bar}[0] resolves to a scalar, so by default it is treated as an element. However $dest->{bar[1]} resolves to a hashref. When operating on it, should it be treated as an opaque object, or as container? For example,

 edit(
     insert => {
         src   => $src,
         dest  => $dest,
         dpath => '/bar/*[1]',
     } );

Should $src be inserted into element 2, as in

 $dest = { bar => [2, { cats => "rule", dogs => "rule" }, 4] };

or should it be inserted before element 2 in bar, as in?

 $dest = { bar => [2, "dogs", "rule", { cats => "rule" }, 4] };

The first behavior treats it as a container, the second as an element. By default destination paths which resolve to hash or array references are treated as containers, so the above code generates the first behavior. To explicitly indicate how a path should be treated, use the dtype option. For example,

 edit(
     insert => {
         src   => $src,
         dest  => $dest,
         dpath => '/bar/*[1]',
         dtype => 'element',
     } );

results in

 $dest = { bar => [2, "dogs", "rule", { cats => "rule" }, 4] };

Source structures may have the same ambiguity. In the above example, note that the contents of the hash in the source path are inserted, not the reference itself. This is because non-blessed references in sources are by default considered to be containers, and their contents are copied. To treat a source reference as an opaque element, use the "stype" option to specify it as such:

 edit(
     insert => {
         src   => $src,
         stype => 'element',
         dest  => $dest,
         dpath => '/bar/*[1]',
         dtype => 'element',
     } );

which results in

 $dest = { bar => [2, { dogs => "rule" }, { cats => "rule" }, 4] };

Note that dpath was set to element, otherwise edit would have attempted to insert the source hashref (not its contents) into the destination hash, which would have failed, as insertion into a hash requires a multiple of two elements (i.e., $key, $value).

Source Transformations

Data extracted from the source structure may undergo transformations prior to being inserted into the destination structure. There are several predefined transformations and the caller may specify a callback to perform their own.

Most of the transformations have to do with multiple values being returned by the source path. For example,

 $src   = { foo => [1], bar => [5], baz => [5] };
 $spath = '/*/*[value == 5]';

would result in multiple extracted values:

 (5, 5)

By default multiple values are not allowed, but a source transformation (specified by the sxfrm option ) may be used to change that behavior. The provided transforms are:

array

The values are assembled into an array. The stype parameter is used to determine whether that array is treated as a container or an element.

hash

The items are assembled into a hash. The stype parameter is used to determine whether that hash is treated as a container or an element. Keys are derived from the data:

  • Keys for hash values will be their hash keys

  • Keys for array values will be their array indices

If there is a single value, a hash key may be specified via the key option to the sxfrm_args option.

iterate

The edit action is applied independently to each source value in turn.

coderef

If sxfrm is a code reference, it will be called to generate the source values. See "Source Callbacks" for more information.

Source Callbacks

If the sxfrm option is a code reference, it is called to generate the source values. It must return an array which contains references to the values (even if they are already references). For example, to return a hash:

  my %src = ( foo => 1 );
  return [ \\%hash ];

It is called with the arguments

$ctx

A "Data::DPath::Context" object representing the source structure.

$spath

The source path. Unless otherwise specified, this defaults to /, except when the source is not a plain array or plain hash, in which case the source is embedded in an array, and spath is set to /*[0].

This is because "Data::DPath" requires a container to be at the root of the source structure, and anything other than a plain array or hash is most likely a blessed object or a scalar, both of which should be treated as elements.

$args

The value of the sxfrm_args option.

Data Copying

By default, copying of data from the source structure is done shallowly, e.g. references to arrays or hashes are not copied recursively. This may cause problems if further modifications are made to the destination structure which may, through references, alter the source structure.

For example, given the following input structures:

 $src  = { dogs => { say => 'bark' } };
 $dest = { cats => { say => 'meow' } };

and this edit operation:

 edit(
     insert => {
         src  => $src,
         dest => $dest,
     } );

We get a destination structure that looks like this:

 $dest = { cats => { say => "meow" }, dogs => { say => "bark" } };

But if later we change $dest,

 # dogs are more excited now
 $dest->{dogs}{say} = 'howl';

the source structure is also changed:

 $src = { dogs => { say => "howl" } };

To avoid this possible problem, Data::Edit::Struct can be passed the clone option, which will instruct it how to copy data.

SUBROUTINES

edit ( $action, $params )

Edit a data structure. The available actions are discussed below.

Parameters

Destination structure parameters are:

dest

A reference to a structure or a Data::DPath::Context object.

dpath

A string representing the data path. This may result in multiple extracted values from the structure; the action will be applied to each in turn.

dtype

May be auto, element or container, to treat the extracted values either as elements or containers. If auto, non-blessed arrays and hashes are treated as containers.

Some actions require a source structure; parameters related to that are:

src

A reference to a structure or a Data::DPath::Context object.

spath

A string representing the source path. This may result in multiple extracted values from the structure; the sxfrm option provides the context for how to interpret these values.

stype

May be auto, element or container, to treat the extracted values either as elements or containers. If auto, non-blessed arrays and hashes are treated as containers.

sxfrm

A transformation to be applied to the data extracted from the source. The available values are

array
hash
iterate
coderef

See "Source Transformations" for more information.

clone

This may be a boolean or a code reference. If a boolean, and true, "dclone" in Storable is used to clone the source structure. If set to a code reference, it is called with a reference to the structure to be cloned. It should return a reference to the cloned structure.

Actions

Actions may have additional parameters

pop

Remove one or more elements from the end of an array. The destination structure must be an array. Additional parameters are:

length

The number of elements to remove. Defaults to 1.

shift

Remove one or more elements from the front of an array. The destination structure must be an array. Additional parameters are:

length

The number of elements to remove. Defaults to 1.

splice

Perform a splice operation on an array, e.g.

  splice( @$dest, $offset, $length, @$src );

The $offset and $length parameters are provided by the offset and length options.

The destination structure may be an array or an array element. In the latter case, the actual offset passed to splice is the sum of the index of the array element and the value provided by the offset option.

A source structure is optional, and may be an array or a hash.

insert

Insert a source structure into the destination structure. The result depends upon whether the point at which to insert is to be treated as a container or an element.

container
Hash

If the container is a hash, the source must be a container (either array or hash), and must contain an even number of elements. Each sequential pair of values is treated as a key, value pair.

Array

If the container is an array, the source may be either a container or an element. The following options are available:

offset

The offset into the array of the insertion point. Defaults to 0. See "anchor".

anchor

Indicates which end of the array the offset parameter is relative to. May be first or last. It defaults to first.

pad

If the array must be enlarged to accommodate the specified insertion point, fill the new values with this value. Defaults to undef.

insert

Indicates which side of the insertion point data will be inserted. May be either before or after. It defaults to before.

element

The insertion point must be an array value. The source may be either a container or an element. The following options are available:

offset

Move the insertion point by this value.

pad

If the array must be enlarged to accommodate the specified insertion point, fill the new values with this value. Defaults to undef.

insert

Indicates which side of the insertion point data will be inserted. May be either before or after. It defaults to before.

delete

Remove an array or hash value. If an array, the length option specifies the number of elements to remove, starting at that element. it defaults to 1.

replace

Replace an array or hash element, or a hash key. The source data is always treated as an element. It takes the following options:

replace

Indicates which part of a hash element to replace, either key or value. Defaults to value. If replacing the key and the source value is a reference, the value returned by Scalar::Util::refaddr will be used.

transform

Transform an element via a user provided subroutine. It takes the following options:

callback_data

Arbitrary data to be passed to the callback.

callback

A code reference which will be called for each element. It is invoked as

  $callback->( $point, $callback_data );

where $point is the Data::DPath::Point object representing the element, and $data is the value of the callback_data option.

Here's a silly example showing how to use the element object's attributes for a hash element:

 $dest = { a => '1', 'b' => 2 };
 
 edit(
     transform => {
         dest     => $dest,
         dpath    => '/*',
         callback => sub {
             my ( $point, $data ) = @_;
             ${ $point->ref } .= $point->attrs->key;
         },
     },
 );

With the result:

 $dest = { a => "1a", b => "2b" };

AUTHOR

Diab Jerius <djerius@cpan.org>

COPYRIGHT AND LICENSE

This software is Copyright (c) 2017 by Smithsonian Astrophysical Observatory.

This is free software, licensed under:

  The GNU General Public License, Version 3, June 2007