fix bug on added? method

[coorasse]

Summary

This Pull Requests fixes an issue with the added? method of ActiveModel::Errors.
When looking for an error by symbol, the method is returning true even if the error is not present. This happens because another error, with a different symbol, but same message, is present.

[rails-bot]

Thanks for the pull request, and welcome! The Rails team is excited to review your changes, and you should hear from @kamipo (or someone else) soon.

If any changes to this PR are deemed necessary, please add them as extra commits. This ensures that the reviewer can see what has changed since they last reviewed the code. Due to the way GitHub handles out-of-date commits, this should also make it reasonably obvious what issues have or haven't been addressed. Large or tricky changes may require several passes of review and changes.

This repository is being automatically checked for code quality issues using Code Climate. You can see results for this analysis in the PR status below. Newly introduced issues should be fixed before a Pull Request is considered ready to review.

Please see the contribution instructions for more information.

[rafaelfranca]

I think we can remove this conditional and always read from the details hash.

[coorasse]

This would result in breaking changes, right?
The details hash doesn't contain the message strings (normalised messages) and such a change wouldn't allow to call added? using the message string anymore.
I would expect, in my example test, that a call to person.errors.added?(:name, 'is wrong') is still possible and returns true.

[rafaelfranca]

Ah yeah, in the case for asserting for the string correspondent to the symbol that would break.

[coorasse]

r? @rafaelfranca

[rafaelfranca]

Could you fix the rubocop violations?

[coorasse]

Rubocop fixed and commits squashed 👍

[jhubert]

@coorasse @rafaelfranca Bumped into an issue with this today. A minor one but it took me a solid chunk of time to figure out what was going wrong.

I have the following test:

  %w[alpha beta charlie].each do |attr|
    test "should be invalid without #{attr}" do
      package = Package.new
      package[attr] = nil

      assert_invalid package, attr => :blank
    end
  end

and the following test helper:

  def assert_invalid(record, options = {})
    assert_predicate record, :invalid?
    options.each do |attribute, message|
      assert record.errors.added?(attribute, message), "Expected #{attribute} to have the following error: #{message}"
    end
  end

These started failing when upgrading to 5.2.0. I finally tracked it down to this change.

The issue is that when attr => :blank is passed to the helper the value of attr is a string.

This causes self.details[attribute] to return nothing, compared to self.details[attribute.to_sym] which returns the expected error.

The fix was pretty simple:

  %w[name brand_name theme_color].each do |attr|
    test "should be invalid without #{attr}" do
      package = BrandingPackage.new
      package[attr] = nil

      assert_invalid package, attr.to_sym => :blank
    end
  end

But, it raised the question of what the expected behavior was for handling string and symbol differences in the details hash? Should package.added?("name", :blank) not be expected to work? It does in 5.1 seemingly because self[attribute.to_s] and self[attribute.to_sym] are equivalent.

Is this a bug? If so, I'm happy to submit a PR to fix it. If it's not a bug, it may be worth noting that it's a breaking change.

[edit] Fwiw, it looks like details is always a symbol, so it's a very simple fix.

[kamipo]

Yeah, it is a bug, we look forward to coming a PR to fix it from you.

[jhubert]

Thanks @kamipo! PR #33361 submitted.