Style/FormatStringToken false positive

[pegasd]

I came across the following false positive while doing rspec-puppet testing on an sshd module.
Sometimes contents of a file can include valid string tokens that look like unannotated Ruby tokens.
These are usually wrapped inside a RegEx for testing.

For example, sshd_config can include %u which is a valid username reference.

Expected behavior

The following code shouldn't be flagged as offensive:

it {
  is_expected.to contain_file('/etc/ssh/sshd_config')
    .with_content(%r{^    ChrootDirectory /var/chroot/%u$})
}

Actual behavior

RuboCop triggers Style/FormatStringToken error:

spec/classes/sshd_spec.rb:3:55: C: Style/FormatStringToken: Prefer annotated tokens (like %<foo>s) over unannotated tokens (like %s).
    .with_content(%r{^    ChrootDirectory /var/chroot/%u$})
                                                      ^^

Steps to reproduce the problem

The snippet above triggers the described offense.

RuboCop version

$ rubocop -V
0.52.0

[jeremywadsack]

Another instance of this is the use of printf style format strings which I think are pretty common — in my experience much more common than format strings (outside of Rails I18n).

Style/FormatStringToken: Prefer annotated tokens (like %<foo>s) over unannotated tokens (like %s).
      @partition ||= dataset.table("#{table.table_id}$#{@date.strftime('%Y%m%d')}")

[Tagging #4539 for x-ref.]

[envygeeks]

We ran into this with strftime too.

[pocke]

@pegasd Thanks for your bug report! I'll fix it.

@jeremywadsack @envygeeks
The strftime's issue will be fixed by #5230
See #5223 #5230

[davidhrbac]

We ran into this with select like

 "date_format(IF(dateEfficiency>NOW(),dateEfficiency,dateExpiration),'%Y%m%d%H%i%s')" \
                                                                              ^^

[pegasd]

@davidhrbac Yeah, I think that particular case is discussed in #5230

[ctm]

scanf also takes a string that uses formatting characters:

irb(main):001:0> s = '12'
=> "12"
irb(main):002:0> require 'scanf'
=> true
irb(main):003:0> s.scanf('%f').first
=> 12.0
irb(main):004:0> s.scanf('%<ignored>f').first
=> nil

[jeremywadsack]

@pocke Thanks for linking that fix for strftime — sorry I missed that when looking for previous reports.

[envygeeks]

Thanks for the quick fix!

[agross]

I see this as a false positive for "foo/bar.rb".pathmap('%d/baz/%f') (pathmap is defined by Rake).

[agross]

Also in this spec:

context 'something' do
  let(:content) { template % address }
                           ^^
  let(:expected) { template % something(address) }

  let(:address) { '[email protected]' }

  context 'text' do
    let(:template) { 'prefix %s suffix' }

    it 'yields encoded content' do
      expect(subject.run(content)).to eq(expected)
    end
  end
end

[lionel218]

Hi guys, I using the rubocop version rubocop-0.52.1 and rails 5.1.1 ruby version is 2.4.2p198
But I just got rubocop error for strftime.
C: Style/FormatStringToken: Prefer annotated tokens (like %s) over unannotated tokens (like %s).
format('%s %s-%s', starts_at.strftime('%a'), starts_at.strftime('%l%P'), ends_at.strftime('%l%P').strip)

How can I solve this issue?

[mikegee]

@lionel218 the cop wants you to use something like:

format('%<day>s %<start>s-%<end>s', day: starts_at.strftime('%a'), start: starts_at.strftime('%l%P'), end: ends_at.strftime('%l%P').strip)

[lionel218]

Hi @mikegee , Just solved this issue.

Thanks for your kindness.

[lionel218]

Hi @mikegee , How can I solve this issue?
Style/FormatStringToken: Prefer annotated tokens (like %s) over unannotated tokens (like %s).
starts_at.strftime('%A')

Style/FormatStringToken: Prefer annotated tokens (like %s) over unannotated tokens (like %s).
cutoff = 3.years.ago.to_date.strftime('%Y/%m/%d')

[lionel218]

Please help me for this issue.

[mikegee]

@lionel218 I failed to duplicate the issue with the example you provided. Please open a new issue with all the details if you think this bug is still in Rubocop. Thanks!

[lionel218]

@mikegee , Thanks for your kindness.
But I can't create new issue because the submit issue button is disabled now.

Please check for my code and rubocop issue.

def event_date(open_house)
format('%s %s-%s', day: open_house.starts_at.strftime('%a'),
start: open_house.starts_at.strftime('%l'),
end: open_house.ends_at.strftime('%l %p').strip)
end
This is my helper action, but just got this issue on rubocop-0.53.0, ruby-2.4.2, rails-5.1.1

app/helpers/open_houses_helper.rb:3:77: C: Style/FormatStringToken: Prefer annotated tokens (like %s) over unannotated tokens (like %s).
format('%s %s-%s', day: open_house.starts_at.strftime
('%a'),
^^
app/helpers/open_houses_helper.rb:5:78: C: Style/FormatStringToken: Prefer annotated tokens (like %s) over unannotated tokens (like %s).
end: open_house.ends_at.strftime('%l
%p').strip)
^^

[mikegee]

@lionel218 I confirmed with that code and opened another issue for you: #5630

[lionel218]

Thanks so much @mikegee

[Seenox]

Sorry for bumping an old issue, but since it shows up in first page of Google if you search for:
Prefer annotated tokens (like `%<foo>s`) over template tokens (like `%{foo}`)
I thought I'd share what has helped me to work my way around that.

My code:

[...]
    when /RedHat|CentOS|Fedora|SLES/                                                                                  
      installed_packages = Facter::Core::Execution.exec('rpm -qa --qf "%{NAME} %{VERSION}\n"').split('\n')
    when /Debian|Ubuntu/
      ...
[...]

Was throwing an error:
rubocop: Prefer annotated tokens (like `%<foo>s`) over template tokens (like `%{foo}`).

Disabling Style/FormatStringToken right before and re-enabling it right after have helped my pipeline to succeed:

[...]
    when /RedHat|CentOS|Fedora|SLES/                                                                                  
      # rubocop:disable Style/FormatStringToken
      installed_packages = Facter::Core::Execution.exec('rpm -qa --qf "%{NAME} %{VERSION}\n"').split('\n')
      # rubocop:enable Style/FormatStringToken
    when /Debian|Ubuntu/
[...]

It may not be the correct/best way to solve it, but it was the quickest for someone like me who has little knowledge about robocup.