MySQL has a very nice option for INSERT
statement, which is particularly helpful for join tables without the id
column. It inserts a record, but, instead of throwing an error if its key clashed with the existing one, that record is updated. Here's an example:
INSERT INTO table (key1,key2,data) VALUES (1,2,3)
ON DUPLICATE KEY UPDATE data=3;
How to achieve the same with Active Record? The resultant code would then look like this:
class Model < ActiveRecord::Base
belongs_to :key1
belongs_to :key2
end
record = Model.new
record.key1 = key1
record.key2 = key2
record.data = 'new data'
record.WHAT? #Inserts or updates `data` for the exist开发者_StackOverflowing record
The answer by j is the right idea, but you probably want to go with find_or_initialize_by to avoid multiple saves on the object. E.g.
record = Model.find_or_initialize_by_key1_and_key2 new
record.data = 'new data'
record.save
By the sounds of it, you also want a validation to check that the model doesn't allow duplicate joins too:
class Model < ActiveRecord::Base
belongs_to :key1
belongs_to :key2
validates_uniqueness_of :key1, :scope => :key2
end
You can do the following:
record = Model.find_or_create_by_key1_and_key2(:key1 => key1, :key2 => key2)
record.update_attribute(:data, "new data")
EDIT
zaius idea looks better, you should use find_or_initialize_by
to avoid multiple saves, as he said :]
You may want to take a look at composite primary keys. It's a plugin for ActiveRecord.
First of that's a MySQL-specific solution, as far as I know that syntax doesn't work with any other SQL-servers. Usually called Upsert in discussions.
I'd go with making a new method called save_or_update
and then if the save fails, with a duplicate key exception, load and update_params
.
精彩评论