Plugin functionality to add generic import to Rails Admin from CSV, JSON and XLSX files
Rails Admin Import 3.x is compatible with Rails Admin 3.x.
If you're still using Rails Admin 2.x, use Rails Admin Import 2.3.1
- First, add to Gemfile:
gem "rails_admin_import", "~> 3.0"
- Define configuration in
config/initializers/rails_admin_import.rb
:
RailsAdmin.config do |config|
# REQUIRED:
# Include the import action
# See https://github.com/sferik/rails_admin/wiki/Actions
config.actions do
all
import
end
# Optional:
# Configure global RailsAdminImport options
# Configure pass filename to records hashes
config.configure_with(:import) do |config|
config.logging = true
config.pass_filename = true
end
# Optional:
# Configure model-specific options using standard RailsAdmin DSL
# See https://github.com/sferik/rails_admin/wiki/Railsadmin-DSL
config.model 'User' do
import do
include_all_fields
exclude_fields :secret_token
end
end
end
- If you are using CanCanCan for authorization, add to ability.rb to specify which models can be imported:
cannot :import, :all
can :import, [User, Model1, Model2]
Model instances can be both created and updated from import data. Any fields can be imported as long as they are allowed by the model's configuration. Associated records can be looked up for both singular and plural relationships. Both updating existing records and associating records requires the use of mapping keys.
Every importable class has a mapping key that uniquely identifies its
instances. The mapping key can be one or more fields. The value for
these fields can then be provided in import data, either to update the
existing record or to attach it through an association to another model.
This concept exists because id
s are often not constant when moving
records between data stores.
For example, a User
model may have an email
field. When uploading a set
of users where some already exist in our database, we can select "email" as
our mapping key and then provide that field on each record in our data,
allowing us to update existing records with matching emails.
Using a csv formatted example:
Email,First name,Last name
[email protected],Peter,Gibbons
[email protected],Michael,Bolton
would look for existing users with those emails. If one was found, its name fields would be updated. Otherwise, a new one would be created.
For updating building owners, the mapping key could be street_address
and
zip_code
.
Similarly, if each user has favorite books, we could set the mapping key
for Book
to be isbn
and then include the isbn for their books within each
user record. The syntax for this is to use the name of the associated model as
the field name, no matter what actual mapping key has been selected. So
a user record would have one or more fields named "Book" that include each
associated book's ISBN.
Again using a csv formatted example:
Email, Book, Book, Book
[email protected], 9781119997870, 9780671027032
[email protected], 9780446677479
would look up books with those ISBNs and attach them to those users.
Mapping keys can be selected on the import page. Their defaults can also be globally configured in the config file:
RailsAdmin.config do |config|
config.model 'User' do
import do
mapping_key :email
# for multiple values, use mapping_key [:first_name, :last_name]
mapping_key_list [:email, :some_other_id]
end
end
end
Since in models with large number of fields it doesn't make sense to use
most of them as mapping values, you can add mapping_key_list
to
restrict which fields can be selected as mapping key in the UI during import.
Note that a matched record must exist when attaching associated models, or the imported record will fail and be skipped.
Complex associations (has_one ..., :through
or polymorphic associations)
need to be dealt with via custom logic called by one of the import hooks
(see below for more detail on using hooks). If we wanted to import
Service
s and attach them to a User
, but the user relationship
existed through an intermediary model called ServiceProvider
, we could
provide a user_email
field in our records and handle the actual
association with an import hook:
class Service < ActiveRecord::Base
belongs_to :service_provider
has_one :user, through: :service_provider
def before_import_save(record)
if (email = record[:user_email]) && (user = User.find_by_email(email))
self.service_provider = user.service_provider
end
end
end
Importing new records by id is not recommended since it ignores the sequences of ids in database. That will lead to ERROR: duplicate key value violates unique constraint
in future. You can work around this issue by adding an import_id
column to your model, renaming the id
column in your CSV to import_id
and using import_id
as the update lookup field.
The format is inferred by the extension (.csv, .json or .xlsx).
The first line must contain attribute names. They will be converted to lowercase and underscored (First Name ==> first_name).
For "many" associations, you may include multiple columns with the same header in the CSV file.
The repeated header may be singular or plural. For example, for a "children" association, you may have multiple "child" columns or multiple "children" column, each containing one lookup value for an associated record. Blank values are ignored.
Example
First name,Last name,Team,Team
Peter,Gibbons,IT,Management
Michael,Bolton,IT,
Blank lines will be skipped.
The file must be an array or an object with a root key the same name as the plural model name, i.e. the default Rails JSON output format with include_root_in_json on or off.
The Microsoft Excel XLM format (XLSX) is supported, but not the old binary Microsoft Excel format (XLS).
The expected rows and columns are the same as for the CSV format (first line contains headers, multiple columns for "many" associations).
Blank lines will be skipped.
RailsAdmin.config do |config|
config.actions do
all
import
end
# Default global RailsAdminImport options
config.configure_with(:import) do |config|
config.logging = false
config.line_item_limit = 1000
config.update_if_exists = false
config.pass_filename = false
config.rollback_on_error = false
config.header_converter = lambda do |header|
header.parameterize.underscore if header.present?
end
config.csv_options = {}
end
end
-
logging (default
false
): Save a copy of each imported file to log/import and a detailed import log to log/rails_admin_import.log -
line_item_limit (default
1000
): max number of items that can be imported at one time. -
update_if_exists (default
false
): default value for the "Update if exists" checkbox on the import page. -
rollback_on_error (default
false
): import records in a transaction and rollback if there is one error. Only for ActiveRecord, not Mongoid. -
header_converter (default
lambda { ... }
): a lambda to convert each CSV header text string to a model attribute name. The default header converter converts to lowercase and replaces spaces with underscores. -
csv_options (default
{}
): a hash of options that will be passed to a new CSV instance -
pass_filename (default
false
): Access the uploaded file name in your model actions, for example if set to true, inside each record, there will be an addtional propertyrecord[:filename_importer]
which contains the file name of the currently uploaded file.
Use standard RailsAdmin DSL to add or remove fields.
- To change the default attribute that will be used to find associations on import, set
mapping_key
(default attribute isname
)
RailsAdmin.config do |config|
config.model 'Ball' do
import do
mapping_key :color
end
end
end
- To include a specific list of fields:
RailsAdmin.config do |config|
config.model 'User' do
import do
field :first_name
field :last_name
field :email
end
end
end
- To exclude specific fields:
RailsAdmin.config do |config|
config.model 'User' do
import do
include_all_fields
exclude_fields :secret_token
end
end
end
- To add extra fields that will be set as attributes on the model and that will be passed to the import hook methods:
RailsAdmin.config do |config|
config.model 'User' do
import do
include_all_fields
fields :special_import_token
end
end
end
- To overwrite the default excluded fields and allow matching to
:id
on import
RailsAdmin.config do |config|
config.model 'User' do
import do
default_excluded_fields [:created_at, :updated_at, :deleted_at, :c_at, :u_at]
end
end
end
Define methods on your models to be hooked into the import process, if special/additional processing is required on the data:
# some model
class User < ActiveRecord::Base
def self.before_import
# called on the model class once before importing any individual records
end
def self.before_import_find(record)
# called on the model class before finding or creating the new record
# maybe modify the import record that will be used to find the model
# throw :skip to skip importing this record
throw :skip unless record[:email].ends_with? "@mycompany.com"
end
def before_import_attributes(record)
# called on the blank new model or the found model before fields are imported
# maybe delete fields from the import record that you don't need
# throw :skip to skip importing this record
end
def before_import_associations(record)
# called on the model with attributes but before associations are imported
# do custom import of associations
# make sure to delete association fields from the import record to avoid double import
record.delete(:my_association)
# throw :skip to skip importing this record
end
def before_import_save(record)
# called on the model before it is saved but after all fields and associations have been imported
# make final modifications to the record
# throw :skip to skip importing this record
end
def after_import_save(record)
# called on the model after it is saved
end
def after_import_association_error(record)
# called on the model when an association cannot be found
end
def after_import_error(record)
# called on the model when save fails
end
def self.after_import
# called once on the model class after importing all individual records
end
end
For example, you could
-
Set an attribute on a Devise User model to skip checking for a password when importing a new model.
-
Import an image into Carrierwave via a URL provided in the CSV.
def before_import_save(record)
self.remote_image_url = record[:image] if record[:image].present?
end
- Skip some validations when importing.
class User < ActiveRecord::Base
# Non-persistent attribute to allow creating a new user without a password
# Password will be set by the user by following a link in the invitation email
attr_accessor :allow_blank_password
devise :validatable
# Called by Devise to enable/disable password presence validation
def password_required?
allow_blank_password ? false : super
end
# Don't require a password when importing users
def before_import_save(record)
self.allow_blank_password = true
end
end
The gem is tested to work with ActiveRecord and Mongoid.
Support for Mongoid is early, so if you can suggest improvements (especially around importing embedded models), open an issue.
Since the import functionality is rarely used in many applications, some gems are autoloaded when first used during an import in order to save memory at boot.
If you prefer to eager load all dependecies at boot, use this line in your Gemfile
.
gem "rails_admin_import", "~> 3.0", require: "rails_admin_import/eager_load"
If you get an error like Error during import: MyModel(#70286054976500) expected, got MyModel(#70286114743280)
, you need restart the rails server and redo the import. This is due to the fact that Rails reloads the ActiveRecord model classes in development when you make changes to them and Rails Admin is still using the old class.
Another suggestion is to set config.cache_classes = true
to true in your development.rb
for Rails Admin Import to work around the ActiveRecord model class reloading issue. See this comment for more information.
If you want to hide all the advanced fields from the import UI, you can copy app/views/rails_admin/main/import.html.haml
to your project at the same path. Add hidden
class to elements you want to hide.
For example:
<div class="form-group control-group hidden">
<label class="col-sm-2 control-label">
<%= t("admin.import.update_if_exists") %>
</label>
<div class="col-sm-10 controls">
<%= check_box_tag :update_if_exists, '1', RailsAdminImport.config.update_if_exists, :class => "form-control" %>
<p class="help-block">
<%= t('admin.import.help.update_if_exists') %>
</p>
</div>
</div>
-
Move global config to
config.configure_with(:import)
inconfig/initializers/rails_admin_import.rb
. -
Move the field definitions to
config.model 'User' do; import do; // ...
inconfig/initializers/rails_admin_import.rb
. -
No need to mount RailsAdminImport in
config/routes.rb
(RailsAdmin must still be mounted). -
Update model import hooks to take 1 hash argument instead of 2 arrays with values and headers.
-
Support for importing file attributes was removed since I couldn't understand how it works. It should be possible to reimplement it yourself using post import hooks. Open an issue to discuss how to put back support for importing files into the gem.
-
Spanish translation by Giovanni Alberto
-
French translation by Rodolphe Robles. (I suggest to translate also rails admin.fr and your locales.fr to resolve an issue with DatePicker)
-
Italian translation by Alessandro Profiti
-
Japanese translation by Yuta Hinokuma
-
Brazilian Portuguese translation by Matheo Gracia Pegoraro
-
Clone the repository to your machine
git clone https://github.com/stephskardal/rails_admin_import
-
Run
bundle install
-
Run
bundle exec rspec
The structure of the tests is taken from the Rails Admin gem.
Original author: Steph Skardal
Maintainer (since May 2015): Julien Vanier
- Update
lib/rails_admin_import/version.rb
- Update the install instructions at the top of the readme
- Commit to git
rake release
Everyone is encouraged to help improve this project. Here are a few ways you can help:
- Report bugs
- Fix bugs and submit pull requests
- Write, clarify, or fix documentation
- Suggest or add new features
Copyright (c) 2015 End Point, Steph Skardal and contributors. See LICENSE.txt for further details.