DeeplyEnumerable and DataLeafWalker

Traverse and map values of deeply nested data structures: Provides a map_deeply/2 function for Maps and Lists and Keyword Lists.

rroouugghhllyy based on ideas from Perl's Data::Leaf::Walker

Synopsis

data = [1, 2, 3, [{:a, 2}, {:b, 4}], %{ a: 2 } ]
DeeplyEnumerable.map_deeply map, &( &1 + 1 )

# yields
[2, 3, 4,[a: 3, b: 5], %{a: 3}]

Usage

DeeplyEnumerable.map_deeply( data, fn( value ) -> ...transform value... end )

Where data can ba a map, list or keyword list.

Direct usage of DataLeafWalker

Or, you may DataLeafWalker's map/2 directly (discouraged):

DataLeafWalker.map( data, fn( value ) -> ...transform value... end )

The difference beetween the module DataLeafWalker and the protocol DeeplyEnumerable is that by implementing the latter for your structs you may add in extra functionality to the transformation or make it conditionally transform or whatever. I needed this feature, that's why it's here. For Lists and Maps, DeeplyEnumerable is already implemented (simply calling to DataLeafWalker.map/2). Though it seems overkill at first glance, having an extension entry point to your transformations of complex structs is usually a safe thing.

Structs

In order to support structs you may either use DataLeafWalker.map/2 or (better) implement DeeplyEnumerable for you struct type, most often like this:

defimpl DeeplyEnumerable, for: MyStruct do
  def map_deeply( struct, fn_transform ) do
    DataLeafWalker.map( struct, fn_transform )
  end
end

What I use this for

I have to migrate ancient projects that use YAML-serialized in database columns. Unfortunately they are doubly UTF-8 encoded so I convert these deeply-nested maps into maps of the same structure with all values decoded into proper UTF-8.

Installation

The package can be installed as:

  1. Add data_leaf_walker to your list of dependencies in mix.exs:

    def deps do 

     [{:data_leaf_walker, "~> 0.1.0"}]

    end

Todo