Merge pull request #2703 from dduugg/dedupe-duplicate-file-comments

De-duplicate comments shared by multiple definitions

Author: vinistock

lib/tapioca/gem/listeners/documentation.rb

@@ -69,20 +69,25 @@ def documentation_comments(name, sigs: [])
           end
           return [] unless declaration
 
-          comments = declaration.definitions.flat_map(&:comments)
-          comments.uniq!
-          return [] if comments.empty?
-
-          lines = comments
-            .map { |comment| comment.string.gsub(/^#+ ?/, "") }
-            .reject { |line| IGNORED_COMMENTS.any? { |comment| line.include?(comment) } || rbs_comment?(line) }
+          # Definitions often share a comment block, such as a license header, so de-duplicate them
+          lines = declaration.definitions
+            .map { |definition| comment_lines(definition) }
+            .uniq
+            .flatten
 
           # Strip leading and trailing blank lines, matching YARD's behavior
           lines = lines.reverse_each.drop_while(&:empty?).reverse_each.drop_while(&:empty?)
 
           lines.map! { |line| RBI::Comment.new(line) }
         end
 
+        #: (Rubydex::Definition definition) -> Array[String]
+        def comment_lines(definition)
+          lines = definition.comments.map { |comment| comment.string.gsub(/^#+ ?/, "") }
+          lines.reject! { |line| IGNORED_COMMENTS.any? { |comment| line.include?(comment) } || rbs_comment?(line) }
+          lines
+        end
+
         # @override
         #: (NodeAdded event) -> bool
         def ignore?(event)

spec/tapioca/gem/pipeline_spec.rb

@@ -4468,6 +4468,78 @@ def something; end
       assert_equal(output, compile(include_doc: false))
     end
 
+    it "does not repeat comments shared by multiple definitions" do
+      add_ruby_file("bar.rb", <<~RUBY)
+        # rubocop:disable Style/Documentation
+        # Licensed under the Foo License, Version 2.0
+        module Namespace
+          # The answer
+          ANSWER = 42
+
+          # Does the thing
+          def self.thing; end
+        end
+      RUBY
+
+      add_ruby_file("baz.rb", <<~RUBY)
+        # Namespace also holds Baz
+        module Namespace
+          class Baz; end
+        end
+      RUBY
+
+      add_ruby_file("foo.rb", <<~RUBY)
+        # Licensed under the Foo License, Version 2.0
+        module Namespace
+          # The answer
+          ANSWER = 42
+
+          # Does the thing
+          def self.thing; end
+        end
+      RUBY
+
+      output = template(<<~RBI)
+        # Licensed under the Foo License, Version 2.0
+        # Namespace also holds Baz
+        module Namespace
+          class << self
+            # Does the thing
+            def thing; end
+          end
+        end
+
+        # The answer
+        Namespace::ANSWER = T.let(T.unsafe(nil), Integer)
+
+        class Namespace::Baz; end
+      RBI
+
+      assert_equal(output, compile(include_doc: true))
+    end
+
+    it "does not de-duplicate comment blocks that partially overlap" do
+      add_ruby_file("bar.rb", <<~RUBY)
+        # Licensed under the Foo License, Version 2.0
+        module Namespace; end
+      RUBY
+
+      add_ruby_file("foo.rb", <<~RUBY)
+        # Licensed under the Foo License, Version 2.0
+        # Namespace is defined here because of whatever
+        module Namespace; end
+      RUBY
+
+      output = template(<<~RBI)
+        # Licensed under the Foo License, Version 2.0
+        # Licensed under the Foo License, Version 2.0
+        # Namespace is defined here because of whatever
+        module Namespace; end
+      RBI
+
+      assert_equal(output, compile(include_doc: true))
+    end
+
     it "properly processes void in type aliases" do
       add_ruby_file("foo.rb", <<~RUBY)
         module Foo