Updating record with join table creates duplicate entries - ruby-on-rails

Question

I have 3 models as follows:
class Document < ActiveRecord::Base
has_many :documents_tasks, inverse_of: :document
has_many :tasks, through: :documents_tasks, dependent: :destroy
end
class Task < ActiveRecord::Base
has_many :documents_tasks, inverse_of: :task
has_many :documents, through: :documents_tasks, dependent: :destroy
end
class DocumentsTask < ActiveRecord::Base
belongs_to :task, inverse_of: :documents_tasks
belongs_to :document, inverse_of: :documents_tasks
validates_uniqueness_of :document_id, scope: :task_id
end
In the above when I try to update the record for a Task it throws a validation error for duplicate entries on the DocumentsTask model if I keep the validation or directly inserts duplicates if remove the validation.
My code to update the Task record is:
def update
#task = #coach.tasks.find(params[:id])
#task.update(:name => task_params[:name], :description => task_params[:description] )
#task.documents << Document.find(task_params[:documents])
if #task.save
render 'show'
else
render status: 500, json: {
error: true,
reason: #task.errors.full_messages.to_sentence
}
end
end
I know I can add unique index to the db to automatically prevent duplicate entries but is there some way I can prevent the controller from updating the join table values when they're the same?
So when I attempt to update the associated documents, ex:
I had document 5 initially
Now I add document 6 and call the update function
It attempts to re-add both documents 5 and 6 to the db so I get the error:
Completed 422 Unprocessable Entity in 9176ms
ActiveRecord::RecordInvalid (Validation failed: Document has already been taken)
This is because I added the following validation:
validates_uniqueness_of :document_id, scope: :task_id
in my DocumentsTask model as shown above. The issue is how can I prevent it from attempting to re-add existing records

Answer 1

Assuming that task_params[:documents] is an array of document ids (based on how you're using it with find now), you should be able to do something like this as a quick fix:
#task.documents << Document.where(id: task_params[:documents]).where.not(task_id: #task.id)
Basically this just filters out the documents that are already associated to the given task before assigning them to the task.
That said, I'd suggest something more robust as a long term solution. A couple of options (among many) would be extracting the responsibility of task creation out into it's own class (so you can more easily test it and make that functionality more portable), or you could look into overriding the setter method(s) for documents in your task model similar to what this answer describes: https://stackoverflow.com/a/2891245/456673