Port prototype rbi generation to prism

Apart from porting to prism, this also does the following:

1. Remove `variable` tracking for `type_of0`. It contains `AST::TypeParam` but checked for inclusion of a Symbol.
There's no difference in the output even when this is fixed, so I just removed it entirely
2. Have the parse method return declarations, make it a class method.
Just more convenient with the new structure. Eventually `rb prototype` will do this as well
3. Allow to use it on jruby/truffleruby.
4. Split comment parsing from comment processing in the helper.
In rbi the comments now come from a plain parse. When `prototype rb` uses prism as well, `parse_comments` can be
removed (as well as most of the other helper methods there)

I tested this against code samples from https://github.com/Shopify/tapioca/blob/d029cc9c3f76865f61fdbaff68d75e18a0014764/spec/tapioca/gem/pipeline_spec.rb
The output is largely the same, and improved in some areas. For example, `type_member` with no paren
is no longer considered an untyped constant. Previously that was only the case for `type_member()` or when
an argument was passed like `type_member(:out)`.

Author: Earlopain

lib/rbs/cli.rb

@@ -108,7 +108,8 @@ def parse_logging_options(opts)
       opts
     end
 
-    def has_parser?
+    def has_parser?(format)
+      return true if format == "rbi"
       defined?(RubyVM::AbstractSyntaxTree) ? true : false
     end
 
@@ -683,7 +684,7 @@ def autoload(name, path)
     end
 
     def run_prototype_file(format, args)
-      availability = unless has_parser?
+      availability = unless has_parser?(format)
                        "\n** This command does not work on this interpreter (#{RUBY_ENGINE}) **\n"
                      end
 
@@ -728,7 +729,7 @@ def run_prototype_file(format, args)
 
       opts.parse!(args)
 
-      unless has_parser?
+      unless has_parser?(format)
         stdout.puts "Not supported on this interpreter (#{RUBY_ENGINE})."
         return 1
       end
@@ -741,7 +742,7 @@ def run_prototype_file(format, args)
       new_parser = -> do
         case format
         when "rbi"
-          Prototype::RBI.new()
+          Prototype::RBI
         when "rb"
           Prototype::RB.new()
         else
@@ -796,7 +797,7 @@ def run_prototype_file(format, args)
 
             parser = new_parser[]
             begin
-              parser.parse file_path.read()
+              decls = parser.parse file_path.read()
             rescue SyntaxError
               stdout.puts "  ⚠️  Unable to parse due to SyntaxError: `#{file_path}`"
               next
@@ -817,7 +818,7 @@ def run_prototype_file(format, args)
             (output_path.parent).mkpath
             output_path.open("w") do |io|
               writer = Writer.new(out: io)
-              writer.write(parser.decls)
+              writer.write(decls)
             end
           end
         end
@@ -837,13 +838,12 @@ def run_prototype_file(format, args)
       else
         # file mode
         parser = new_parser[]
+        writer = Writer.new(out: stdout)
 
         input_paths.each do |file|
-          parser.parse file.read()
+          writer.write parser.parse(file.read())
         end
 
-        writer = Writer.new(out: stdout)
-        writer.write parser.decls
       end
 
       0

lib/rbs/prototype/helpers.rb

@@ -7,24 +7,28 @@ module Helpers
 
       def parse_comments(string, include_trailing:)
         Prism.parse_comments(string, version: "current").yield_self do |prism_comments| # steep:ignore UnexpectedKeywordArgument
-          prism_comments.each_with_object({}) do |comment, hash| #$ Hash[Integer, AST::Comment]
-            # Skip EmbDoc comments
-            next unless comment.is_a?(Prism::InlineComment)
-            # skip like `module Foo # :nodoc:`
-            next if comment.trailing? && !include_trailing
-
-            line = comment.location.start_line
-            body = "#{comment.location.slice}\n"
-            body = body[2..-1] or raise
-            body = "\n" if body.empty?
-
-            comment = AST::Comment.new(string: body, location: nil)
-            if prev_comment = hash.delete(line - 1)
-              hash[line] = AST::Comment.new(string: prev_comment.string + comment.string,
-                                            location: nil)
-            else
-              hash[line] = comment
-            end
+          process_comments(prism_comments, include_trailing: include_trailing)
+        end
+      end
+
+      def process_comments(comments, include_trailing:)
+        comments.each_with_object({}) do |comment, hash| #$ Hash[Integer, AST::Comment]
+          # Skip EmbDoc comments
+          next unless comment.is_a?(Prism::InlineComment)
+          # skip like `module Foo # :nodoc:`
+          next if comment.trailing? && !include_trailing
+
+          line = comment.location.start_line
+          body = "#{comment.location.slice}\n"
+          body = body[2..-1] or raise
+          body = "\n" if body.empty?
+
+          comment = AST::Comment.new(string: body, location: nil)
+          if prev_comment = hash.delete(line - 1)
+            hash[line] = AST::Comment.new(string: prev_comment.string + comment.string,
+                                          location: nil)
+          else
+            hash[line] = comment
           end
         end
       end

lib/rbs/prototype/rb.rb

@@ -77,6 +77,7 @@ def parse(string)
         comments = parse_comments(string, include_trailing: false)
 
         process RubyVM::AbstractSyntaxTree.parse(string), decls: source_decls, comments: comments, context: Context.initial
+        decls
       end
 
       def process(node, decls:, comments:, context:)