I'm trying to figure out an inconsistency between what's happening in a functional test and what is happening in my development environment. I have a custom validation method unique_entry
that is essentially a specialized version of validates_uniqueness_of
. It looks like this:
def unique_entry
matched_entry = Entry.first(:conditions => ['LOWER(field_one) = LOWER(?) AND LOWER(field_two) = LOWER(?)', self.field_one, self.field_two])
errors.add_to_base('Duplicate detected') if matched_entry && (matched_entry.id != self.id)
end
The update action in the controller is very basic:
def update
if @entry.update_attributes(params[:entry])
flash.now[:success] = 'Success'
render :action => 'show'
else
flash.now[:error] = 'Error'
render :action => 'edit'
end
end
This works just fine when I'm creating a new record. When I update a record, however, I get inconsistent behavior. If I test it from a browser in my development environment, it correctly renders the edit
action with an error message, but in my functional test, it accepts the update as successful. Here is the test:
test "should not update entry and should render edit view if invalid update" do
put :update, { :id => 1, :field_one => 'new_value', :field_two => 'new_value' } # 'new values' are the same as another existing record to trigger the duplication check
assert_template :edit
assert_not_nil flash[:error]
end
I looked at the test log and discovered that the values unique_entry
is using are the record's original values instead of the values it should be attempting to update with. That is, the first line of unique_entry
generates an SQL query like this:
SELECT * FROM "entries" WHERE (LOWER(field_one) = LOWER('original_value_of_field_one') AND LOWER(field_two) = LOWER('original_value_of_field_two')) LIMIT 1
What am I missing here? Why do my validations seem to be running against the original record instead of the new values only in the test environment?