Initial implementation

lib/yard_sorbet.rb

@@ -0,0 +1,10 @@
+# frozen_string_literal: true
+# typed: strong
+require 'sorbet-runtime'
+require 'yard'
+
+# top-level namespace
+module YARDSorbet; end
+
+require_relative 'yard_sorbet/sig_handler'
+require_relative 'yard_sorbet/sig_to_yard'

lib/yard_sorbet/sig_handler.rb

@@ -0,0 +1,182 @@
+# frozen_string_literal: true
+module YARDSorbet
+  class SigHandler < YARD::Handlers::Ruby::Base
+    extend T::Sig
+    handles :class, :module, :singleton_class?
+
+    sig { returns(String).checked(:never) }
+    def process
+      # Find the list of declarations inside the class
+      class_def = statement.children.find { |c| c.type == :list }
+      class_contents = class_def.children
+
+      class_contents.each_with_index do |child, i|
+        next unless type_signature?(child)
+
+        next_statement = class_contents[i + 1]
+        if %i[def defs command].include?(next_statement&.type) && !next_statement.docstring
+          # Swap the method definition docstring and the sig docstring.
+          # Parse relevant parts of the `sig` and include them as well.
+          parser = YARD::DocstringParser.new.parse(child.docstring)
+          # Directives are already parsed at this point, and there doesn't
+          # seem to be an API to tweeze them from one node to another without
+          # managing YARD internal state. Instead, we just extract them from
+          # the raw text and re-attach them.
+          directives = parser.raw_text&.split("\n")&.select do |line|
+            line.start_with?('@!')
+          end || []
+          docstring = parser.to_docstring
+          parsed_sig = parse_sig(child)
+          enhance_tag(docstring, :abstract, parsed_sig)
+          enhance_tag(docstring, :return, parsed_sig)
+          if next_statement.type != :command
+            parsed_sig[:params]&.each do |name, types|
+              enhance_param(docstring, name, types)
+            end
+          end
+          next_statement.docstring = docstring.to_raw
+          directives.each do |directive|
+            next_statement.docstring.concat("\n#{directive}")
+          end
+          child.docstring = nil
+        end
+      end
+    end
+
+    # sig do
+    #  params(
+    #    docstring: YARD::Docstring,
+    #    name: String,
+    #    types: T::Array[String]
+    #  )
+    #    .void
+    #    .checked(:tests)
+    # end
+    private def enhance_param(docstring, name, types)
+      tag = docstring.tags.find { |t| t.tag_name == 'param' && t.name == name }
+      if tag
+        docstring.delete_tag_if { |t| t == tag }
+        tag.types = types
+      else
+        tag = YARD::Tags::Tag.new(:param, '', types, name)
+      end
+      docstring.add_tag(tag)
+    end
+
+    # sig do
+    #  params(
+    #    docstring: YARD::Docstring,
+    #    type: Symbol,
+    #    parsed_sig: T::Hash[T.untyped, T.untyped]
+    #  )
+    #    .void
+    #    .checked(:tests)
+    # end
+    private def enhance_tag(docstring, type, parsed_sig)
+      return if !parsed_sig[type]
+
+      tag = docstring.tags.find { |t| t.tag_name == type.to_s }
+      if tag
+        docstring.delete_tags(type)
+      else
+        tag = YARD::Tags::Tag.new(type, '')
+      end
+      if parsed_sig[type].is_a?(Array)
+        tag.types = parsed_sig[type]
+      end
+      docstring.add_tag(tag)
+    end
+
+    # sig do
+    #  params(sig_node: YARD::Parser::Ruby::MethodCallNode)
+    #    .returns(T::Hash[T.untyped, T.untyped])
+    #    .checked(:tests)
+    # end
+    private def parse_sig(sig_node)
+      parsed = {}
+      parsed[:abstract] = false
+      parsed[:params] = {}
+      found_params = T.let(false, T::Boolean)
+      found_return = T.let(false, T::Boolean)
+      bfs_traverse(sig_node, exclude: %i[array hash]) do |n|
+        if n.source == 'abstract'
+          parsed[:abstract] = true
+        elsif n.source == 'params' && !found_params
+          found_params = true
+          sibling = T.must(sibling_node(n))
+          bfs_traverse(sibling, exclude: %i[array call hash]) do |p|
+            if p.type == :assoc
+              param_name = p.children.first.source[0...-1]
+              types = YARDSorbet::SigToYARD.convert(p.children.last)
+              parsed[:params][param_name] = types
+            end
+          end
+        elsif n.source == 'returns' && !found_return
+          found_return = true
+          parsed[:return] = YARDSorbet::SigToYARD.convert(T.must(sibling_node(n)))
+        elsif n.source == 'void'
+          parsed[:return] ||= ['void']
+        end
+      end
+      parsed
+    end
+
+    # Returns true if the given node is part of a type signature.
+    # sig do
+    # node is T.nilable(YARD::Parser::Ruby::AstNode) but that has issues
+    # with loop code
+    # params(node: T.untyped).returns(T.nilable(Boolean)).checked(:tests)
+    # end
+    private def type_signature?(node)
+      loop do
+        return false if node.nil?
+        return false unless %i[call vcall fcall].include?(node.type)
+        return true if T.unsafe(node).method_name(true) == :sig
+
+        node = node.children.first
+      end
+    end
+
+    # sig do
+    #  params(node: YARD::Parser::Ruby::AstNode)
+    #    .returns(T.nilable(YARD::Parser::Ruby::AstNode))
+    #    .checked(:tests)
+    # end
+    private def sibling_node(node)
+      found_sibling = T.let(false, T::Boolean)
+      node.parent.children.each do |n|
+        if found_sibling
+          return n
+        end
+
+        if n == node
+          found_sibling = true
+        end
+      end
+      nil
+    end
+
+    # @yield [YARD::Parser::Ruby::AstNode]
+    # sig do
+    #  params(
+    #    node: YARD::Parser::Ruby::AstNode,
+    #    exclude: T::Array[Symbol],
+    #    blk: T.proc.params(arg0: T.untyped).returns(T.untyped)
+    #  )
+    #    .void
+    #    .checked(:tests)
+    # end
+    private def bfs_traverse(node, exclude: [])
+      queue = [node]
+      while !queue.empty?
+        n = T.must(queue.shift)
+        yield n
+        n.children.each do |c|
+          if !exclude.include?(c.type)
+            queue.push(c)
+          end
+        end
+      end
+    end
+  end
+end

lib/yard_sorbet/sig_to_yard.rb

@@ -0,0 +1,85 @@
+# frozen_string_literal: true
+# typed: strict
+
+module YARDSorbet::SigToYARD
+  # Translate sig type syntax to YARD type syntax.
+  #
+  # @see https://yardoc.org/types.html
+  #   sig {params(node: YARD::Parser::Ruby::AstNode).returns(T::Array[String]).checked(:tests)}
+  def self.convert(node)
+    types = convert_type(node)
+    # scrub newlines, as they break the YARD parser
+    types.map { |type| type.gsub(/\n\s*/, ' ') }
+  end
+
+  def self.convert_type(node)
+    children = node.children
+    case node.type
+    when :aref
+      # https://www.rubydoc.info/gems/yard/file/docs/Tags.md#Parametrized_Types
+      case children.first.source
+      when 'T::Array', 'T::Enumerable', 'T::Range', 'T::Set'
+        collection_type = children.first.source.split('::').last
+        member_type = convert(children.last.children.first).join(', ')
+        ["#{collection_type}<#{member_type}>"]
+      when 'T::Hash'
+        key_type = convert(children.last.children.first).join(', ')
+        value_type = convert(children.last.children.last).join(', ')
+        ["Hash{#{key_type} => #{value_type}}"]
+      else
+        log.warn("Unsupported sig aref node #{node.source}")
+        [node.source]
+      end
+    when :arg_paren
+      convert(children.first.children.first)
+    when :array
+      # https://www.rubydoc.info/gems/yard/file/docs/Tags.md#Order-Dependent_Lists
+      member_types = children.first.children.map { |n| convert(n) }.join(', ')
+      ["Array(#{member_types})"]
+    when :call
+      if children[0].source == 'T'
+        case children[2].source
+        when 'all', 'class_of', 'enum', 'noreturn', 'self_type', 'type_parameter', 'untyped'
+          # YARD doesn't have equivalent notions, so we just use the raw source
+          [node.source]
+        when 'any'
+          children.last.children.first.children.map { |n| convert(n) }.flatten
+        when 'nilable'
+          # Order matters here, putting `nil` last results in a more concise
+          # return syntax in the UI (superscripted `?`)
+          convert(children.last) + ['nil']
+        else
+          log.warn("Unsupported T method #{node.source}")
+          [node.source]
+        end
+      else
+        [node.source]
+      end
+    when :const_path_ref
+      [node.source]
+    when :hash, :list
+      # Fixed hashes as return values are unsupported:
+      # https://github.com/lsegal/yard/issues/425
+      #
+      # Hash key params can be individually documented with `@option`, but
+      # sig translation is unsupported.
+      ['Hash']
+    when :var_ref
+      # YARD convention is use singleton objects when applicable:
+      # https://www.rubydoc.info/gems/yard/file/docs/Tags.md#Literals
+      case node.source
+      when 'FalseClass'
+        ['false']
+      when 'NilClass'
+        ['nil']
+      when 'TrueClass'
+        ['true']
+      else
+        [node.source]
+      end
+    else
+      log.warn("Unsupported sig #{node.type} node #{node.source}")
+      [node.source]
+    end
+  end
+end