Merge pull request #3 from euglena1215/refact-parser-2

Change parsing from regular expression to Prism AST

.rubocop.yml

@@ -1,5 +1,6 @@
 AllCops:
   TargetRubyVersion: 3.3
+  NewCops: enable
 
 Style/StringLiterals:
   EnforcedStyle: double_quotes
@@ -17,3 +18,8 @@ Metrics/ClassLength:
 Metrics/MethodLength:
   Exclude:
     - 'test/**/*'
+  Max: 20
+
+Style/Style/Documentation:
+  Enabled: false
+

lib/rbs_inline_data/cli.rb

@@ -19,22 +19,29 @@ def run(args)
         end
       end.parse!(args)
 
-      targets = Pathname.glob(args[0]).flat_map do |path|
-        if path.directory?
-          Pathname.glob(path.join("**/*.rb").to_s)
+      get_targets(args[0]).each do |target|
+        result = Prism.parse_file(target.to_s)
+        definitions = Parser.parse(result)
+        Writer.write(definitions, output_path ? (output_path + target).sub_ext(".rbs") : nil)
+      end
+    end
+
+    private
+
+    #:: (String) -> Array[Pathname]
+    def get_targets(path)
+      targets = Pathname.glob(path).flat_map do |pathname|
+        if pathname.directory?
+          Pathname.glob(pathname.join("**/*.rb").to_s)
         else
-          path
+          pathname
         end
       end
 
       targets.sort!
       targets.uniq!
 
-      targets.each do |target|
-        result = Prism.parse_file(target.to_s)
-        definitions = Parser.parse(result)
-        Writer.write(definitions, output_path ? (output_path + target).sub_ext(".rbs") : nil)
-      end
+      targets
     end
   end
 end

lib/rbs_inline_data/parser.rb

@@ -14,14 +14,37 @@ class Parser < Prism::Visitor
       :field_name, #:: String
       :type #:: String
     )
+    # @rbs skip
+    Comments = Data.define(
+      :comment_lines #:: Hash[Integer, String]
+    )
+    class Comments
+      MARKER = "#::"
+
+      #:: (Array[Prism::Comment]) -> RbsInlineData::Parser::Comments
+      def self.from_prism_comments(comments)
+        # @type var comment_lines: Hash[Integer, String]
+        comment_lines = {}
+        comments.each do |comment|
+          sliced = comment.slice
+          next unless sliced.start_with?(MARKER)
+
+          comment_lines[comment.location.start_line] = sliced.sub(MARKER, "").strip
+        end
+
+        new(comment_lines:)
+      end
+    end
 
     # @rbs @definitions: Array[RbsInlineData::Parser::TypedDefinition]
     # @rbs @surronding_class_or_module: Array[Symbol]
+    # @rbs @comments: RbsInlineData::Parser::Comments
 
     # rubocop:disable Lint/MissingSuper
-    #:: (Array[RbsInlineData::Parser::TypedDefinition]) -> void
-    def initialize(definitions)
+    #:: (Array[RbsInlineData::Parser::TypedDefinition], RbsInlineData::Parser::Comments) -> void
+    def initialize(definitions, comments)
       @definitions = definitions
+      @comments = comments
       @surronding_class_or_module = []
     end
     # rubocop:enable Lint/MissingSuper
@@ -30,7 +53,8 @@ def initialize(definitions)
     def self.parse(result)
       # @type var definitions: Array[RbsInlineData::Parser::TypedDefinition]
       definitions = []
-      instance = new(definitions)
+      comments = Comments.from_prism_comments(result.comments)
+      instance = new(definitions, comments)
       result.value.accept(instance)
       definitions
     end
@@ -80,30 +104,27 @@ def define_data?(node)
 
     #:: (Prism::ConstantWriteNode) -> RbsInlineData::Parser::TypedDefinition?
     def extract_definition(node)
-      source = node.slice
-      _, class_name, field_text = source.match(/\A([a-zA-Z0-9]+) ?= ?Data\.define\(([\n\s\w\W]+)\)\z/).to_a
-      return nil if field_text.nil? || class_name.nil?
-
-      class_name = "#{@surronding_class_or_module.join("::")}::#{class_name}"
-
-      fields = field_text.split("\n").map(&:strip).reject(&:empty?).map do |str|
-        case str
-        when /:(\w+),? #:: ([\w:\[\], ]+)/
-          [::Regexp.last_match(1), ::Regexp.last_match(2)]
-        when /:(\w+),?/
-          [::Regexp.last_match(1), "untyped"]
-        end
-      end.compact.map do |field_name, type|
-        TypedField.new(
-          field_name:,
-          type:
-        )
+      arguments_node = node.value.arguments
+      if arguments_node
+        typed_fields = arguments_node.arguments.map do |sym_node|
+          return nil unless sym_node.is_a?(Prism::SymbolNode)
+
+          TypedField.new(
+            field_name: sym_node.unescaped,
+            type: type_of(sym_node)
+          )
+        end.compact
       end
 
       TypedDefinition.new(
-        class_name:,
-        fields:
+        class_name: "#{@surronding_class_or_module.join("::")}::#{node.name}",
+        fields: typed_fields || []
       )
     end
+
+    #:: (Prism::SymbolNode) -> String
+    def type_of(node)
+      @comments.comment_lines[node.location.start_line] || "untyped"
+    end
   end
 end

rbs_inline_data.gemspec

@@ -30,4 +30,5 @@ Gem::Specification.new do |spec|
 
   spec.add_dependency "prism"
   spec.add_dependency "rbs"
+  spec.metadata["rubygems_mfa_required"] = "true"
 end

sig/generated/data/lib/rbs_inline_data/parser.rbs

@@ -15,3 +15,11 @@ class RbsInlineData::Parser::TypedField
               | (**untyped) -> ::RbsInlineData::Parser::TypedField
               | ...
 end
+
+class RbsInlineData::Parser::Comments
+  extend Data::_DataClass
+  attr_reader comment_lines: Hash[Integer, String]
+  def self.new: (*untyped) -> ::RbsInlineData::Parser::Comments
+              | (**untyped) -> ::RbsInlineData::Parser::Comments
+              | ...
+end

sig/generated/rbs_inline_data/cli.rbs

@@ -1,8 +1,14 @@
 # Generated from lib/rbs_inline_data/cli.rb with RBS::Inline
 
 module RbsInlineData
+  # Process executed when running the rbs-inline-data command.
   class CLI
     # :: (Array[String]) -> void
     def run: (Array[String]) -> void
+
+    private
+
+    # :: (String) -> Array[Pathname]
+    def get_targets: (String) -> Array[Pathname]
   end
 end

sig/generated/rbs_inline_data/parser.rbs

@@ -2,12 +2,22 @@
 
 module RbsInlineData
   class Parser < Prism::Visitor
+    class Comments
+      MARKER: ::String
+
+      # :: (Array[Prism::Comment]) -> RbsInlineData::Parser::Comments
+      def self.from_prism_comments: (Array[Prism::Comment]) -> RbsInlineData::Parser::Comments
+    end
+
     @definitions: Array[RbsInlineData::Parser::TypedDefinition]
 
     @surronding_class_or_module: Array[Symbol]
 
-    # :: (Array[RbsInlineData::Parser::TypedDefinition]) -> void
-    def initialize: (Array[RbsInlineData::Parser::TypedDefinition]) -> void
+    @comments: RbsInlineData::Parser::Comments
+
+    # rubocop:disable Lint/MissingSuper
+    # :: (Array[RbsInlineData::Parser::TypedDefinition], RbsInlineData::Parser::Comments) -> void
+    def initialize: (Array[RbsInlineData::Parser::TypedDefinition], RbsInlineData::Parser::Comments) -> void
 
     # :: (Prism::ParseResult) -> Array[RbsInlineData::Parser::TypedDefinition]
     def self.parse: (Prism::ParseResult) -> Array[RbsInlineData::Parser::TypedDefinition]
@@ -31,5 +41,8 @@ module RbsInlineData
 
     # :: (Prism::ConstantWriteNode) -> RbsInlineData::Parser::TypedDefinition?
     def extract_definition: (Prism::ConstantWriteNode) -> RbsInlineData::Parser::TypedDefinition?
+
+    # :: (Prism::SymbolNode) -> String
+    def type_of: (Prism::SymbolNode) -> String
   end
 end

sig/patch.rbs

@@ -8,4 +8,13 @@ module Prism
     def constant_path: () -> Prism::ConstantReadNode
                      | ...
   end
+
+  class Comment
+    def slice: () -> String
+  end
+
+  class ConstantWriteNode
+    def value: () -> CallNode
+             | ...
+  end
 end

test/rbs_inline_data/parser_test.rb

@@ -132,5 +132,18 @@ class A
         ]
       )
     end
+
+    def test_no_args
+      definitions = Parser.parse(parse_ruby(<<~RUBY))
+        class A
+          B = Data.define
+        end
+      RUBY
+
+      assert_equal definitions[0], Parser::TypedDefinition.new(
+        class_name: "A::B",
+        fields: []
+      )
+    end
   end
 end