No release in over 3 years
Low commit activity in last 3 years
Comprehensive import/export tools between Excel/CSV & ActiveRecord Databases, Rails apps, and any Ruby project.


>= 4.8.2, ~> 4.8
~> 3
~> 11
~> 3.4
~> 0.38


>= 2.7.0, ~> 2.7
<= 6.0.0, >= 4.3
<= 5.2, >= 4.2
<= 1.2.1, >= 0.9.9
<= 0.20, >= 0.18
>= 0.3, ~> 0.3
 Project Readme


Build Status Code Climate Test Coverage

Datashift is a suite of tools to help you import or export data from a Rails application, including all association data.

Formats currently supported are .xls files (Excel/OpenOffice/LibraOffice) and CSV files.

It's not the fastest, but a key feature of the import is that unlike say a pure DB load, inbound data is validated by your Rail's business logic, as defined by your validations, associations etc.

Paperclip bulk import tools for attaching uploads from filesystem, to Rails model instances.

Table of Contents
  • Installation
  • Features
  • Usage
  • Import / Export
  • Associations
  • Testing
  • Wiki

Add gem 'datashift' to your Gemfile

gem 'datashift'

Direct install via usual gem install datashift

There are also specific import/export loaders for Spree E-Commerce here @ datashift_spree

  • Import and Export direct to .xls files (Excel/OpenOffice etc) - Win OLE and MS Excel are NOT required.

  • Import and Export direct to CSV files.

  • Bulk upload Paperclip supported filetypes, from the filesystem, such as images, documents, mp3s, files.

    • Auto attach the uploaded assets to associated instances of the parent model, using the file name to find and attach to DB models. For example :

      - Upload the image to Rails storage
      - Looks up a product by it's '''SKU''', which **is present in the image filename** - my_sku_2017.jpg
      - Attaches new image to the Product `my_sku_2017` '''images''' association
  • Smart import - Datashift will try its best to automatically map the headers in your import data to your ActiveRecord model attributes and associations

  • Easy to configure and map columns to your database when automatic mapping doesn't quite cut it.

  • Fast mapping - Generate configuration and mapping documents automatically, to speed up mapping data to the destination target.

  • Transform the data during import or export with defaults, substitutions etc.

  • Associations supported, with ability to define lookup column, and to find existing associated models to attach to the main Upload model,

  • Association types to include/exclude can be set in configuration as well as specific columns to exclude.

  • Easily exclude Rails standard columns such as id, created_at, updated_at etc.

It's simple to use the facilities in standard Ruby, for example'db/seeds/permit.csv', PermitModel)

    music_to_export = MP3.where(style: 'banging techno').all {|d| d.export('/tmp/mp3_dump.xls', music_to_export) }

In Rails, generally you would drive this via a Controller Action

For example

class CategoriesController < ApplicationController

  def index
     @categories = Category.all
     respond_to do |format|
         format.xls do
              contents =
    , @categories)
              send_data contents.string.force_encoding('binary'), type: 'application/xls'

N.B You need to have registered the xls format as mime type, somewhere like config/initializers/mime_types.rb

Mime::Type.register "application/xls", :xls

Multiple apps are provided through command line tasks via Thor.

To use the command line applications, pull in the tasks.

Create or add to a high level .thor file e.g mysite.thor in your lib/tasks or Rails root directory

Add the following lines, to pull in the datashift thor commands :

    require 'thor'
    require 'datashift'


To keep the availability to only development mode you can use

DataShift::load_commands if(Rails.env.development?)

To check the available tasks run

bundle exec thor list datashift

To get usage information use thor help , for example

bundle exec thor help datashift:generate:excel

Please use thor list and thor help ` to get latest command lines

thor datashift:config:generate:import -m, --model=MODEL -r, --result=RESULT                                                                                                     ...
thor datashift:export:csv -m, --model=MODEL -r, --result=RESULT                                                                                                                 ...
thor datashift:export:db -p, --path=PATH                                                                                                                                        ...
thor datashift:export:excel -m, --model=MODEL -r, --result=RESULT                                                                                                               ...
thor datashift:generate:csv -m, --model=MODEL -r, --result=RESULT                                                                                                               ...
thor datashift:generate:db -p, --path=PATH                                                                                                                                      ...
thor datashift:generate:excel -m, --model=MODEL -r, --result=RESULT                                                                                                             ...
thor datashift:import:csv -i, --input=INPUT -m, --model=MODEL                                                                                                                   ...
thor datashift:import:excel -i, --input=INPUT -m, --model=MODEL                                                                                                                 ...
thor datashift:import:load -i, --input=INPUT -m, --model=MODEL 

Exports are currently based around a single fundamental DB model represented in a single Worksheet, but can include all the associations of that model.

Column headings will normally simply reflect the database columns names and association names, but this is configurable.

A mapping configuration can be used for both imports and exports, to explicitly map between headers and database names, when automatic mapping not suitable.

On Import, a main DB model is supplied, for which a dictionary of all possible attributes and associations is created.

The Import is then based on column headings with Semi-Smart Name Lookup, managing white space, pluralisation, under_scores etc.

So the user supplied name (column heading) need only be an approximation of the actual name.

For Example given column heading 'Product Properties', will still find real association 'product_properties'

Datashift can populate your main model's associations, searching for matching objects and assigning them to the load object.

To achieve this it supports a syntax within either column headings or individual cells to specify :

  • Search Attribute - the field on the association model to search on.
  • Search Value - the specific value(s) to attempt to find.

So in the following example our main Project object has an has_many association with Category.

Category has a field called reference. We want to attach existing Categories to our newly created Project, based on values of this reference.

Header or Cell based lookup

Specifying the lookup field in the column headings, our Excel (or CSV) file might look like this :

project name Categories:reference
aphex category_001
boc category_003
autechre category_001,category_002

Note: Lookup can be specified at the individual cell level You can specify different lookup fields in different columns or cells.

project name categories
aphex reference:category_001,category_002
boc reference:category_003
autechre reference:category_001,category_002,category_003
Lookup Syntax
  • : - Seperates lookup field name, here 'reference', from the values.
  • , - Seperates multiple lookup values for has_many relationships

So in our example, datashift will perform searches like :

Category.where("reference IN (?)", [category_001,category_002,category_003])

The resulting DB objects, will be assigned to the Project.categories

When specify has_many relationships multiple file columns can also be used. The following would lead to exactly the same end result, as the first example.

project name categories categories
aphex reference:category_001 reference:category_002
boc reference:category_003
autechre reference:category_001,category_002 reference:category_003


Configuration of datashift can be done through a typical Rails initialisation code block, a YAML configuration file provided at run time, or both in which case run time options over ride global ones.

The easiest way to create a global configuration file, loaded during server start, is to run our rail's install generator :

rails g datashift:install

You can create a model specific file at anytime via

    thor datashift:config:generate:import

To create a Rails tyle config block manually, create an initialisation file within config/initializers, and see lib/datashift/configuration.rb for details of all possible options, for example do |c|
  c.verbose = false
  c.remove_columns = [:milestones, :versions]
  c.remove_rails = true
  c.with = :all


You can use a YAML file or snippet to configure the column headers, and set of columns to include in an export

In this code based example, we only want 4 columns from our model.

DataFlowSchema is the class that represents a schema for the data flows, that is it directs an import or export.

We can use the presentation keyword to over ride rhe normal header (default is simply the column name).

yaml= <<EOS
       - id:
       - status_str:
           presentation: "Status Str"
       - user:
           presentation: "User Name"
       - status:

         data_flow_schema = { |dfs| dfs.prepare_from_string(yaml) }
  do |d|
           d.data_flow_schema = data_flow_schema
           d.export(File.join('tmp', 'conversion_for_removal.xls'), conversions)

In this example we stored the YAML in memory, but you can also use a file, in which case the only change, is that the call to create a DataFlowSchema becomes :

  data_flow_schema = { |dfs| dfs.prepare_from_file(file_name) }


Individual Imports/Export runs can also be directed from YAML configuration file.

This allows more fine grained control of the process, and can include column mappings instructions, data transformations and custom processing methods for columns/data that require non standard processing, as well as global configuration parameters.

There is another generator, to create a skeleton configuration file, based on the model to be imported :

thor help datashift:generate:config:import -m <MyModelToImport> -r config/datashift.rb


Transform the data during an import in various ways.

  • Defaults - Where your column is empty, provide a default value to be used.

  • Overrides - When you want to provide a set value in all cases, or when you have no inbound data.

  • Prefixes/Postfixes - Amend data on the fly. e.g if you wish to prepend a string id to a reference type field.

        DataShift::Transformation.factory do |factory|
          factory.set_default_on(Project, 'value_as_string',  'some default text' )
          factory.set_default_on(Project, 'value_as_double',   45.467 )
          factory.set_default_on(Project, 'value_as_boolean',  true )
          factory.set_default_on(Project, 'value_as_datetime', )

N.B The operator/column (2nd parameter) must match the inbound HEADER

For example given a header SKU, for a class with real operator sku=, even though we know assignment will eventually use sku= this will not work : factory.set_prefix_on(Spree::Product, 'sku', 'SPEC_')

But this will set the right prefix, because the header in the FILE is SKU

  `factory.set_prefix_on(Spree::Product, 'SKU', 'SPEC_')`

Paperclip Import

Bulk upload from filesystem usign paperclip.

The general usage of paperclip is to define a model, which has associated attachments, for example

   class Product < ActiveRecord::Base
     has_attached_file :image, styles: { medium: "300x300>", thumb: "100x100>" }, default_url: "/images/:style/missing.png"
     validates_attachment_content_type :image, content_type: /\Aimage\/.*\z/

Where datashift shines is when you want to bulk upload a hole load of images and attach them to existing Products.

The loaded content is automatically attached to the model - containing the has_attached_file directive - by matching the filename to a column of the model, in the case of our Product, perhaps the SKU or name.

The database field to match on, and the filename matching pattern are all configurable.

So in this example, to fix a set of products without images, the setup required would be :

  • Create a directory of images, with the SKU of the product in the filename
  • Configure datashift the upload with model Product and column matching on 'SKU'
  • Run datashift:paperclip:attach --input /tmp/images --attach-to-klass Product --attach-to-find-by-field SKU --attachment-klass Image --attach-to-field image

See wiki

Specs need to run against a Rails sandbox app.

A sandbox will be generated in spec/dummy if no such directory exists.

There are spec helpers to build the dummy app, via shelling out to rails new

The rails version used will be based on the latest you have installed, via the gemspec.

Changing Versions

To test different versions update the gemspec and run bundle update rails

**N.B Manual Step**
When changing versions you should **delete this whole directory**  `spec/dummy`
Next time you run rspec it will auto generate a new dummy app using latest Rails versions

Run the Tests

** N.B You should run the specs from within the specs directory. **
        bundle exec rspec -c .

A datashift log will be written within spec/logs, which hooks into the standard active record logger


Thomas Statter - Initial idea and dev

Thanks to all contributors who have participated in this project.


Copyright:: (c) Autotelik Media Ltd 2016

This project is licensed under the MIT License - see the file for details