rbi 0.0.2 → 0.0.6

data/lib/rbi/parser.rb

@@ -1,13 +1,24 @@
 # typed: strict
 # frozen_string_literal: true
 
-require "unparser"
+require "parser"
 
 module RBI
-  class Parser
+  class ParseError < StandardError
     extend T::Sig
 
-    class Error < StandardError; end
+    sig { returns(Loc) }
+    attr_reader :location
+
+    sig { params(message: String, location: Loc).void }
+    def initialize(message, location)
+      super(message)
+      @location = location
+    end
+  end
+
+  class Parser
+    extend T::Sig
 
     # opt-in to most recent AST format
     ::Parser::Builders::Default.emit_lambda               = true
@@ -16,6 +27,12 @@ module RBI
     ::Parser::Builders::Default.emit_index                = true
     ::Parser::Builders::Default.emit_arg_inside_procarg0  = true
 
+    sig { void }
+    def initialize
+      # Delay load unparser and only if it has not been loaded already.
+      require "unparser" unless defined?(::Unparser)
+    end
+
     sig { params(string: String).returns(Tree) }
     def self.parse_string(string)
       Parser.new.parse_string(string)
@@ -29,15 +46,11 @@ module RBI
     sig { params(string: String).returns(Tree) }
     def parse_string(string)
       parse(string, file: "-")
-    rescue ::Parser::SyntaxError => e
-      raise Error, e.message
     end
 
     sig { params(path: String).returns(Tree) }
     def parse_file(path)
       parse(::File.read(path), file: path)
-    rescue ::Parser::SyntaxError => e
-      raise Error, e.message
     end
 
     private
@@ -47,9 +60,12 @@ module RBI
       node, comments = Unparser.parse_with_comments(content)
       assoc = ::Parser::Source::Comment.associate_locations(node, comments)
       builder = TreeBuilder.new(file: file, comments: assoc)
+      builder.separate_header_comments
       builder.visit(node)
       builder.assoc_dangling_comments(comments)
       builder.tree
+    rescue ::Parser::SyntaxError => e
+      raise ParseError.new(e.message, Loc.from_ast_loc(file, e.diagnostic.location))
     end
   end
 
@@ -70,12 +86,12 @@ module RBI
     private
 
     sig { params(node: AST::Node).returns(String) }
-    def visit_name(node)
+    def parse_name(node)
       T.must(ConstBuilder.visit(node))
     end
 
     sig { params(node: AST::Node).returns(String) }
-    def visit_expr(node)
+    def parse_expr(node)
       Unparser.unparse(node)
     end
   end
@@ -89,10 +105,10 @@ module RBI
     sig do
       params(
         file: String,
-        comments: T.nilable(T::Hash[::Parser::Source::Map, T::Array[::Parser::Source::Comment]])
+        comments: T::Hash[::Parser::Source::Map, T::Array[::Parser::Source::Comment]]
       ).void
     end
-    def initialize(file:, comments: nil)
+    def initialize(file:, comments: {})
       super()
       @file = file
       @comments = comments
@@ -106,130 +122,171 @@ module RBI
       return unless node.is_a?(AST::Node)
       case node.type
       when :module, :class, :sclass
-        visit_scope(node)
+        scope = parse_scope(node)
+        current_scope << scope
+        @scopes_stack << scope
+        visit_all(node.children)
+        @scopes_stack.pop
       when :casgn
-        visit_const_assign(node)
+        current_scope << parse_const_assign(node)
       when :def, :defs
-        visit_def(node)
+        current_scope << parse_def(node)
       when :send
-        visit_send(node)
+        node = parse_send(node)
+        current_scope << node if node
       when :block
-        visit_block(node)
+        node = parse_block(node)
+        if node.is_a?(Sig)
+          @last_sigs << node
+        elsif node
+          current_scope << node
+        end
       else
         visit_all(node.children)
       end
     end
 
+    sig { void }
+    def separate_header_comments
+      return if @comments.empty?
+
+      keep = []
+      node = T.must(@comments.keys.first)
+      comments = T.must(@comments.values.first)
+
+      last_line = T.let(nil, T.nilable(Integer))
+      comments.reverse.each do |comment|
+        comment_line = comment.location.last_line
+
+        break if last_line && comment_line < last_line - 1 ||
+          !last_line && comment_line < node.first_line - 1
+
+        keep << comment
+        last_line = comment_line
+      end
+
+      @comments[node] = keep.reverse
+    end
+
     sig { params(comments: T::Array[::Parser::Source::Comment]).void }
     def assoc_dangling_comments(comments)
-      return unless tree.empty?
-      comments.each do |comment|
+      last_line = T.let(nil, T.nilable(Integer))
+      (comments - @comments.values.flatten).each do |comment|
+        comment_line = comment.location.last_line
         text = comment.text[1..-1].strip
-        loc = ast_to_rbi_loc(comment.location)
+        loc = Loc.from_ast_loc(@file, comment.location)
+
+        if last_line && comment_line > last_line + 1
+          # Preserve empty lines in file headers
+          tree.comments << EmptyComment.new(loc: loc)
+        end
+
         tree.comments << Comment.new(text, loc: loc)
+        last_line = comment_line
       end
     end
 
     private
 
-    sig { params(node: AST::Node).void }
-    def visit_scope(node)
+    sig { params(node: AST::Node).returns(Scope) }
+    def parse_scope(node)
       loc = node_loc(node)
       comments = node_comments(node)
 
-      scope = case node.type
+      case node.type
       when :module
-        name = visit_name(node.children[0])
+        name = parse_name(node.children[0])
         Module.new(name, loc: loc, comments: comments)
       when :class
-        name = visit_name(node.children[0])
+        name = parse_name(node.children[0])
         superclass_name = ConstBuilder.visit(node.children[1])
         Class.new(name, superclass_name: superclass_name, loc: loc, comments: comments)
       when :sclass
         SingletonClass.new(loc: loc, comments: comments)
       else
-        raise "Unsupported node #{node.type}"
+        raise ParseError.new("Unsupported scope node type `#{node.type}`", loc)
       end
-      current_scope << scope
+    end
 
-      @scopes_stack << scope
-      visit_all(node.children)
-      @scopes_stack.pop
+    sig { params(node: AST::Node).returns(RBI::Node) }
+    def parse_const_assign(node)
+      node_value = node.children[2]
+      if struct_definition?(node_value)
+        parse_struct(node)
+      else
+        name = parse_name(node)
+        value = parse_expr(node_value)
+        loc = node_loc(node)
+        comments = node_comments(node)
+        Const.new(name, value, loc: loc, comments: comments)
+      end
     end
 
-    sig { params(node: AST::Node).void }
-    def visit_const_assign(node)
-      name = visit_name(node)
-      value = visit_expr(node.children[2])
+    sig { params(node: AST::Node).returns(Method) }
+    def parse_def(node)
       loc = node_loc(node)
-      comments = node_comments(node)
-
-      current_scope << Const.new(name, value, loc: loc, comments: comments)
-    end
 
-    sig { params(node: AST::Node).void }
-    def visit_def(node)
-      current_scope << case node.type
+      case node.type
       when :def
         Method.new(
           node.children[0].to_s,
-          params: node.children[1].children.map { |child| visit_param(child) },
+          params: node.children[1].children.map { |child| parse_param(child) },
           sigs: current_sigs,
-          loc: node_loc(node),
+          loc: loc,
           comments: node_comments(node)
         )
       when :defs
         Method.new(
           node.children[1].to_s,
-          params: node.children[2].children.map { |child| visit_param(child) },
+          params: node.children[2].children.map { |child| parse_param(child) },
           is_singleton: true,
           sigs: current_sigs,
-          loc: node_loc(node),
+          loc: loc,
           comments: node_comments(node)
         )
       else
-        raise "Unsupported node #{node.type}"
+        raise ParseError.new("Unsupported def node type `#{node.type}`", loc)
       end
     end
 
     sig { params(node: AST::Node).returns(Param) }
-    def visit_param(node)
+    def parse_param(node)
       name = node.children[0].to_s
       loc = node_loc(node)
       comments = node_comments(node)
 
       case node.type
       when :arg
-        Param.new(name, loc: loc, comments: comments)
+        ReqParam.new(name, loc: loc, comments: comments)
       when :optarg
-        value = visit_expr(node.children[1])
+        value = parse_expr(node.children[1])
         OptParam.new(name, value, loc: loc, comments: comments)
       when :restarg
         RestParam.new(name, loc: loc, comments: comments)
       when :kwarg
         KwParam.new(name, loc: loc, comments: comments)
       when :kwoptarg
-        value = visit_expr(node.children[1])
+        value = parse_expr(node.children[1])
         KwOptParam.new(name, value, loc: loc, comments: comments)
       when :kwrestarg
         KwRestParam.new(name, loc: loc, comments: comments)
       when :blockarg
         BlockParam.new(name, loc: loc, comments: comments)
       else
-        raise "Unsupported node #{node.type}"
+        raise ParseError.new("Unsupported param node type `#{node.type}`", loc)
       end
     end
 
-    sig { params(node: AST::Node).void }
-    def visit_send(node)
+    sig { params(node: AST::Node).returns(T.nilable(RBI::Node)) }
+    def parse_send(node)
       recv = node.children[0]
-      return if recv && recv != :self
+      return nil if recv && recv != :self
 
       method_name = node.children[1]
       loc = node_loc(node)
       comments = node_comments(node)
 
-      current_scope << case method_name
+      case method_name
       when :attr_reader
         symbols = node.children[2..-1].map { |child| child.children[0] }
         AttrReader.new(*symbols, sigs: current_sigs, loc: loc, comments: comments)
@@ -240,53 +297,114 @@ module RBI
         symbols = node.children[2..-1].map { |child| child.children[0] }
         AttrAccessor.new(*symbols, sigs: current_sigs, loc: loc, comments: comments)
       when :include
-        names = node.children[2..-1].map { |child| visit_name(child) }
+        names = node.children[2..-1].map { |child| parse_name(child) }
         Include.new(*names, loc: loc, comments: comments)
       when :extend
-        names = node.children[2..-1].map { |child| visit_name(child) }
+        names = node.children[2..-1].map { |child| parse_name(child) }
         Extend.new(*names, loc: loc, comments: comments)
       when :abstract!, :sealed!, :interface!
         Helper.new(method_name.to_s.delete_suffix("!"), loc: loc, comments: comments)
       when :mixes_in_class_methods
-        names = node.children[2..-1].map { |child| visit_name(child) }
+        names = node.children[2..-1].map { |child| parse_name(child) }
         MixesInClassMethods.new(*names, loc: loc, comments: comments)
       when :public, :protected, :private
-        Visibility.new(method_name, loc: loc)
+        visibility = Visibility.new(method_name, loc: loc)
+        nested_node = node.children[2]
+        case nested_node&.type
+        when :def, :defs
+          method = parse_def(nested_node)
+          method.visibility = visibility
+          method
+        when :send
+          snode = parse_send(nested_node)
+          raise ParseError.new("Unexpected token `private` before `#{nested_node.type}`", loc) unless snode.is_a?(Attr)
+          snode.visibility = visibility
+          snode
+        when nil
+          visibility
+        else
+          raise ParseError.new("Unexpected token `private` before `#{nested_node.type}`", loc)
+        end
       when :prop
-        name, type, default_value = visit_struct_prop(node)
+        name, type, default_value = parse_tstruct_prop(node)
         TStructProp.new(name, type, default: default_value, loc: loc, comments: comments)
       when :const
-        name, type, default_value = visit_struct_prop(node)
+        name, type, default_value = parse_tstruct_prop(node)
         TStructConst.new(name, type, default: default_value, loc: loc, comments: comments)
       else
-        raise "Unsupported node #{node.type} with name #{method_name}"
+        raise ParseError.new("Unsupported send node with name `#{method_name}`", loc)
       end
     end
 
-    sig { params(node: AST::Node).void }
-    def visit_block(node)
+    sig { params(node: AST::Node).returns(T.nilable(RBI::Node)) }
+    def parse_block(node)
       name = node.children[0].children[1]
 
       case name
       when :sig
-        @last_sigs << visit_sig(node)
+        parse_sig(node)
       when :enums
-        current_scope << visit_enum(node)
+        parse_enum(node)
       else
-        raise "Unsupported node #{node.type} with name #{name}"
+        raise ParseError.new("Unsupported block node type `#{name}`", node_loc(node))
       end
     end
 
+    sig { params(node: AST::Node).returns(T::Boolean) }
+    def struct_definition?(node)
+      (node.type == :send && node.children[0]&.type == :const && node.children[0].children[1] == :Struct) ||
+        (node.type == :block && struct_definition?(node.children[0]))
+    end
+
+    sig { params(node: AST::Node).returns(RBI::Struct) }
+    def parse_struct(node)
+      name = parse_name(node)
+      loc = node_loc(node)
+      comments = node_comments(node)
+
+      send = node.children[2]
+      body = []
+
+      if send.type == :block
+        if send.children[2].type == :begin
+          body = send.children[2].children
+        else
+          body << send.children[2]
+        end
+        send = send.children[0]
+      end
+
+      members = []
+      keyword_init = T.let(false, T::Boolean)
+      send.children[2..].each do |child|
+        if child.type == :sym
+          members << child.children[0]
+        elsif child.type == :kwargs
+          pair = child.children[0]
+          if pair.children[0].children[0] == :keyword_init
+            keyword_init = true if pair.children[1].type == :true
+          end
+        end
+      end
+
+      struct = Struct.new(name, members: members, keyword_init: keyword_init, loc: loc, comments: comments)
+      @scopes_stack << struct
+      visit_all(body)
+      @scopes_stack.pop
+
+      struct
+    end
+
     sig { params(node: AST::Node).returns([String, String, T.nilable(String)]) }
-    def visit_struct_prop(node)
+    def parse_tstruct_prop(node)
       name = node.children[2].children[0].to_s
-      type = visit_expr(node.children[3])
+      type = parse_expr(node.children[3])
       has_default = node.children[4]
         &.children&.fetch(0, nil)
         &.children&.fetch(0, nil)
         &.children&.fetch(0, nil) == :default
       default_value = if has_default
-        visit_expr(node.children.fetch(4, nil)
+        parse_expr(node.children.fetch(4, nil)
           &.children&.fetch(0, nil)
           &.children&.fetch(1, nil))
       end
@@ -294,17 +412,17 @@ module RBI
     end
 
     sig { params(node: AST::Node).returns(Sig) }
-    def visit_sig(node)
+    def parse_sig(node)
       sig = SigBuilder.build(node)
       sig.loc = node_loc(node)
       sig
     end
 
     sig { params(node: AST::Node).returns(TEnumBlock) }
-    def visit_enum(node)
+    def parse_enum(node)
       enum = TEnumBlock.new
       node.children[2].children.each do |child|
-        enum << visit_name(child)
+        enum << parse_name(child)
       end
       enum.loc = node_loc(node)
       enum
@@ -312,28 +430,16 @@ module RBI
 
     sig { params(node: AST::Node).returns(Loc) }
     def node_loc(node)
-      ast_to_rbi_loc(node.location)
-    end
-
-    sig { params(ast_loc: ::Parser::Source::Map).returns(Loc) }
-    def ast_to_rbi_loc(ast_loc)
-      Loc.new(
-        file: @file,
-        begin_line: ast_loc.line,
-        begin_column: ast_loc.column,
-        end_line: ast_loc.last_line,
-        end_column: ast_loc.last_column
-      )
+      Loc.from_ast_loc(@file, node.location)
     end
 
     sig { params(node: AST::Node).returns(T::Array[Comment]) }
     def node_comments(node)
-      return [] unless @comments
       comments = @comments[node.location]
       return [] unless comments
       comments.map do |comment|
         text = comment.text[1..-1].strip
-        loc = ast_to_rbi_loc(comment.location)
+        loc = Loc.from_ast_loc(@file, comment.location)
         Comment.new(text, loc: loc)
       end
     end
@@ -434,11 +540,11 @@ module RBI
       when :params
         node.children[2].children.each do |child|
           name = child.children[0].children[0].to_s
-          type = visit_expr(child.children[1])
+          type = parse_expr(child.children[1])
           @current << SigParam.new(name, type)
         end
       when :returns
-        @current.return_type = visit_expr(node.children[2])
+        @current.return_type = parse_expr(node.children[2])
       when :void
         @current.return_type = nil
       else
@@ -446,4 +552,17 @@ module RBI
       end
     end
   end
+
+  class Loc
+    sig { params(file: String, ast_loc: T.any(::Parser::Source::Map, ::Parser::Source::Range)).returns(Loc) }
+    def self.from_ast_loc(file, ast_loc)
+      Loc.new(
+        file: file,
+        begin_line: ast_loc.line,
+        begin_column: ast_loc.column,
+        end_line: ast_loc.last_line,
+        end_column: ast_loc.last_column
+      )
+    end
+  end
 end