ArcFurnace

ArcFurnace melts, melds, and transforms your scrap data into perfectly crafted data for ingest into applications, analysis, or whatnot. ArcFurnace simplifies simple ETL (Extract, Transform, Load) tasks for small to medium sets of data using a programmatic DSL interface. Here's an example:

class Transform < ArcFurnace::Pipeline

    source :marketing_info_csv, type: ArcFurnace::CSVSource, params: { filename: :marketing_filename }

    transform :marketing_info_source, params: { source: :marketing_info_csv } do |row|
      row.delete('Name')
      row
    end

    source :product_attributes,
           type: ArcFurnace::MultiCSVSource,
           params: { filenames: :product_attribute_filenames }

    hash_node :marketing_info,
              params: {
                  key_column: :primary_key,
                  source: :marketing_info_source
              }

    outer_join :join_results,
               params: {
                   source: :product_attributes,
                   hash: :marketing_info
               }

    sink type: ArcFurnace::AllFieldsCSVSink,
         source: :join_results,
         params: { filename: :destination_name }

end

Installation

Add this line to your application's Gemfile:

gem 'arc-furnace', github: 'salsify/arc-furnace'

And then execute:

$ bundle

Usage

ArcFurnace provides a few concepts useful to extracting and transforming data.

Node Types Available

Pipelines

Pipelines define a a complete transformation and define a directed, acyclic graph of operations that define how data is transformed. Each type of node in a Pipeline is defined below, but a Pipelines defines the network of nodes that transform data.

Sources

A Source provides values to a Pipeline. A Pipeline may have many sources. Essentially, any nodes that require a stream of data (Hash, Transform, Join, Sink) will have one.

Hashes

A Hash provides indexed access to a Source but pre-computing the index based on a key. The processing happens during the prepare stage of pipeline processing. Hashes have a simple interface, #get(primary_key), to requesting data. Hashes are almost exclusively used as inputs to one side of joins.

Joins

An InnerJoin or an OuterJoin join two sources of data (one must be a Hash) based upon a key. By default the join key is the key that the hash was rolled-up on, however, the key_column option on both InnerJoin and OuterJoin may override this. Note the default join is an inner join, which will drop source rows if the hash does not contain a matching row.

Filters

A Filter acts as a source, however, takes a source as an input and determines whether to pass each row to the next downstream node by calling the #filter method on itself. There is an associated BlockFilter and sugar on Pipeline to make this easy.

Transforms

A Transform acts as a source, however, takes a source as an input and transforms each input. The BlockTransform and associated sugar in the transform method of Pipeline make this very easy (see the example above).

Unfolds

An Unfold acts as a source, however, takes a source as an input and produces multiple rows for that source as an output. A common case for this is splitting rows into multiple rows depending upon their keys. The BlockTransform and associated sugar in the unfold method of Pipeline make this fairly easy (see pipeline_spec.rb).

Observers

An Observer acts as a source and takes a source as an input and serves as a pass-through for a stream. Observers are used to observe the data stream--record data for use elsewhere.

Sinks

Each Pipeline has a single sink. Pipelines must produce data somewhere, and that data goes to a sink. Sinks subscribe to the #row(hash) interace--each output row is passed to this method for handling.

General pipeline development process

1. Define a source. Choose an existing Source implementation in this library (CSVSource or ExcelSource), extend the EnumeratorSource, or implement the row() method for a new source.
2. Define any transformations, or joins. This may cause you to revisit #1.
3. Define the sink. This is generally custom, or, may be one of the provided CSVSink types.
4. Roll it together in a Pipeline.

Development

After checking out the repo, run bin/setup to install dependencies. Then, run bin/console for an interactive prompt that will allow you to experiment.

To install this gem onto your local machine, run bundle exec rake install. To release a new version, update the version number in version.rb, and then run bundle exec rake release to create a git tag for the version, push git commits and tags, and push the .gem file to rubygems.org.

TODOs

1. Add examples for ErrorHandler interface.
2. Add sugar to define a BlockTransform on a Source definition in a Pipeline.

Contributing

1. Fork it ( https://github.com/[my-github-username]/arc-furnace/fork )
2. Create your feature branch (git checkout -b my-new-feature)
3. Commit your changes (git commit -am 'Add some feature')
4. Push to the branch (git push origin my-new-feature)
5. Create a new Pull Request