Introduce PhraseIdentifier::PATTERN

The `PhraseIdentifier.generate` method is used to generate an identifier
for projects created by teachers and students. However, public projects,
e.g. those created via the `GithubWebhooksController#github_push` [1]
action and those created via the (possibly defunct)
`projects:create_all` rake task [2], have identifiers specified in their
project config via `ProjectImporter` [3].

While there is no *explicit* validation to enforce it, the identifiers
for these public projects seems to follow the pattern described by the
regular expression I have defined in `PhraseIdentifier::PATTERN`, i.e.
kebab-case with digits allowed, at least in the 2nd word [4]. The
pattern could probably be stricter, but since I don't have access to the
production data, it seems safer for it to be more relaxed.

We want admin users in experience-cs to be able to create public scratch
projects via the editor-api API and I think it makes sense to restrict
them to using identifiers following a similar pattern. I plan to add
some validation to enforce that and so we can provide feedback to the
admin user if they choose an identifier with the wrong format. I'll be
able to use this `PhraseIdentifier::PATTERN` in that validation.

[1]: https://github.com/RaspberryPiFoundation/editor-api/blob/f348dfc92fdd8b19fbfd5434e7751340f33b4093/app/controllers/github_webhooks_controller.rb#L6-L8
[2]: https://github.com/RaspberryPiFoundation/editor-api/blob/f348dfc92fdd8b19fbfd5434e7751340f33b4093/lib/tasks/projects.rake#L4-L7
[3]: https://github.com/RaspberryPiFoundation/editor-api/blob/f348dfc92fdd8b19fbfd5434e7751340f33b4093/lib/project_importer.rb
[4]: https://github.com/RaspberryPiFoundation/editor-api/blob/f348dfc92fdd8b19fbfd5434e7751340f33b4093/lib/tasks/project_components/top_5_emoji_list/project_config.yml#L2

Author: floehopper

lib/phrase_identifier.rb

@@ -1,6 +1,8 @@
 # frozen_string_literal: true
 
 class PhraseIdentifier
+  PATTERN = /\A[a-z0-9]+(-[a-z0-9]+)*\z/
+
   class Error < RuntimeError
   end
 

spec/lib/phrase_identifier_spec.rb

@@ -4,6 +4,24 @@
 require 'phrase_identifier'
 
 RSpec.describe PhraseIdentifier do
+  describe 'PATTERN' do
+    subject(:pattern) { described_class::PATTERN }
+
+    it { is_expected.to match('abc-def-ghi') }
+    it { is_expected.to match('123-456-789') }
+    it { is_expected.to match('a2c-d5f-g8i') }
+
+    it { is_expected.not_to match('Abc-def-ghi') }
+    it { is_expected.not_to match('Abc-def-GHI') }
+    it { is_expected.not_to match('abc--def-ghi') }
+    it { is_expected.not_to match('-abc-def-ghi') }
+    it { is_expected.not_to match('abc-def-ghi-') }
+    it { is_expected.not_to match('abc def-ghi') }
+    it { is_expected.not_to match(' abc-def-ghi') }
+    it { is_expected.not_to match('abc-def-ghi ') }
+    it { is_expected.not_to match('abc_def_ghi') }
+  end
+
   describe '#generate' do
     subject(:generate) { described_class.generate }
 
@@ -18,6 +36,14 @@
       it { is_expected.to match phrase_regex }
     end
 
+    context 'when using the default words.txt file' do
+      it 'returns identifiers conforming to the expected pattern' do
+        10.times do
+          expect(generate).to match(described_class::PATTERN)
+        end
+      end
+    end
+
     context 'when there are no available combinations' do
       let(:identifier) { Faker::Verb.base }