Merge pull request #134 from sue445/feature/move_to_data_yaml

Move to `RubyHeaderParser::Data`

Author: sue445

.rubocop_todo.yml

@@ -1,6 +1,6 @@
 # This configuration was generated by
 # `rubocop --auto-gen-config`
-# on 2024-09-17 16:12:35 UTC using RuboCop version 1.66.1.
+# on 2024-09-18 15:36:12 UTC using RuboCop version 1.66.1.
 # The point is for the user to remove these configuration records
 # one by one as the offenses are removed from the code base.
 # Note that changes in the inspected code, or installation of new
@@ -25,7 +25,7 @@ Metrics/BlockNesting:
 # Offense count: 3
 # Configuration parameters: CountComments, CountAsOne.
 Metrics/ClassLength:
-  Max: 175
+  Max: 131
 
 # Offense count: 4
 # Configuration parameters: AllowedMethods, AllowedPatterns.

_tools/ruby_h_to_go/lib/ruby_header_parser.rb

@@ -3,6 +3,7 @@
 require "yaml"
 
 require_relative "ruby_header_parser/argument_definition"
+require_relative "ruby_header_parser/data"
 require_relative "ruby_header_parser/function_definition"
 require_relative "ruby_header_parser/parser"
 require_relative "ruby_header_parser/struct_definition"

_tools/ruby_h_to_go/lib/ruby_header_parser/data.rb

@@ -0,0 +1,54 @@
+# frozen_string_literal: true
+
+module RubyHeaderParser
+  # Manager for `data.yml`
+  class Data
+    # @!attribute [r] data
+    #   @return [Hash]
+    attr_reader :data
+
+    def initialize
+      yaml = File.read(File.join(__dir__, "data.yml"))
+      @data = YAML.safe_load(yaml, permitted_classes: [Regexp])
+    end
+
+    # @param function_name [String]
+    # @param index [Integer] arg position (0 origin)
+    # @return [Symbol] :ref, :array
+    def function_arg_pointer_hint(function_name:, index:)
+      pointer_hint = data["function"]["pointer_hint"].dig(function_name, index)
+      return pointer_hint.to_sym if pointer_hint
+
+      :ref
+    end
+
+    # Whether generate C function to go
+    # @param function_name [String]
+    # @return [Boolean]
+    def should_generate_function?(function_name)
+      function_name = function_name.downcase
+
+      return false if data["function"]["deny_name"].any? { |format| format === function_name } # rubocop:disable Style/CaseEquality
+
+      data["function"]["allow_name"].any? { |format| format === function_name } # rubocop:disable Style/CaseEquality
+    end
+
+    # Whether generate C struct to go
+    # @param struct_name [String]
+    # @return [Boolean]
+    def should_generate_struct?(struct_name)
+      struct_name = struct_name.downcase
+
+      data["struct"]["allow_name"].any? { |format| format === struct_name } # rubocop:disable Style/CaseEquality
+    end
+
+    # Whether generate C type to go
+    # @param type_name [String]
+    # @return [Boolean]
+    def should_generate_type?(type_name)
+      type_name = type_name.downcase
+
+      data["type"]["allow_name"].any? { |format| format === type_name } # rubocop:disable Style/CaseEquality
+    end
+  end
+end

_tools/ruby_h_to_go/lib/ruby_header_parser/data.yml

@@ -0,0 +1,47 @@
+function:
+  allow_name:
+    - !ruby/regexp /^rb_/
+    - !ruby/regexp /^rstring_/
+
+  deny_name:
+    # deprecated functions
+    - rb_check_safe_str
+    - rb_clear_constant_cache
+    - rb_clone_setup
+    - rb_complex_polar
+    - rb_data_object_alloc
+    - rb_data_object_get_warning
+    - rb_data_object_wrap_warning
+    - rb_data_typed_object_alloc
+    - rb_dup_setup
+    - rb_gc_force_recycle
+    - rb_iterate
+    - rb_obj_infect
+    - rb_obj_infect_raw
+    - rb_obj_taint
+    - rb_obj_taint_raw
+    - rb_obj_taintable
+    - rb_obj_tainted
+    - rb_obj_tainted_raw
+    - rb_scan_args_length_mismatch
+    - rb_varargs_bad_length
+
+    # internal functions in ruby.h
+    - rb_scan_args_bad_format
+
+  pointer_hint:
+    rb_funcallv:
+      3: array
+    rb_funcallv_public:
+      3: array
+
+struct:
+  allow_name:
+    - !ruby/regexp /^rb_/
+
+type:
+  allow_name:
+    - !ruby/regexp /^rb_/
+    - !ruby/regexp /^st_/
+    - id
+    - value

_tools/ruby_h_to_go/lib/ruby_header_parser/parser.rb

@@ -8,13 +8,13 @@ class Parser
     attr_reader :header_dir
 
     # @!attribute [r] data
-    #   @return [Hash]
+    #   @return [RubyHeaderParser::Data]
     attr_reader :data
 
     # @param header_dir [String]
     def initialize(header_dir)
       @header_dir = header_dir
-      @data = YAML.load_file(File.join(__dir__, "ruby_header_parser.yml"))
+      @data = Data.new
     end
 
     # @return [Array<RubyHeaderParser::FunctionDefinition>]
@@ -26,7 +26,7 @@ def extract_function_definitions
 
         function_name = parts[0]
 
-        next unless should_generate_function?(function_name)
+        next unless data.should_generate_function?(function_name)
 
         definition =
           if parts[2].end_with?(";$/;\"")
@@ -62,7 +62,7 @@ def extract_struct_definitions
 
         struct_name = parts[0]
 
-        next unless should_generate_struct?(struct_name)
+        next unless data.should_generate_struct?(struct_name)
 
         definitions << StructDefinition.new(
           name:     struct_name,
@@ -80,7 +80,7 @@ def extract_type_definitions
 
         type_name = parts[0]
 
-        next unless should_generate_type?(type_name)
+        next unless data.should_generate_type?(type_name)
 
         definitions << TypeDefinition.new(
           name:     type_name,
@@ -91,48 +91,6 @@ def extract_type_definitions
 
     private
 
-    ALLOW_FUNCTION_NAME_PREFIXES = %w[rb_ rstring_].freeze
-
-    DENY_FUNCTION_NAMES = [
-      # deprecated functions
-      "rb_check_safe_str",
-      "rb_clear_constant_cache",
-      "rb_clone_setup",
-      "rb_complex_polar",
-      "rb_data_object_alloc",
-      "rb_data_object_get_warning",
-      "rb_data_object_wrap_warning",
-      "rb_data_typed_object_alloc",
-      "rb_dup_setup",
-      "rb_gc_force_recycle",
-      "rb_iterate",
-      "rb_obj_infect",
-      "rb_obj_infect_raw",
-      "rb_obj_taint",
-      "rb_obj_taint_raw",
-      "rb_obj_taintable",
-      "rb_obj_tainted",
-      "rb_obj_tainted_raw",
-      "rb_scan_args_length_mismatch",
-      "rb_varargs_bad_length",
-
-      # internal functions in ruby.h
-      "rb_scan_args_bad_format",
-    ].freeze
-
-    # Whether generate C function to go
-    # @param function_name [String]
-    # @return [Boolean]
-    def should_generate_function?(function_name)
-      function_name = function_name.downcase
-
-      return false if DENY_FUNCTION_NAMES.include?(function_name)
-
-      return true if ALLOW_FUNCTION_NAME_PREFIXES.any? { |prefix| function_name.start_with?(prefix) }
-
-      false
-    end
-
     # @param file [String]
     # @param line_num [Integer]
     def read_definition_from_header_file(file, line_num)
@@ -149,29 +107,6 @@ def read_definition_from_header_file(file, line_num)
       ""
     end
 
-    # Whether generate C struct to go
-    # @param struct_name [String]
-    # @return [Boolean]
-    def should_generate_struct?(struct_name)
-      struct_name = struct_name.downcase
-
-      struct_name.start_with?("rb_")
-    end
-
-    ALLOW_TYPE_NAME_PREFIXES = %w[rb_ st_].freeze
-    ALLOW_TYPE_NAMES = %w[id value].freeze
-
-    # Whether generate C type to go
-    # @param type_name [String]
-    # @return [Boolean]
-    def should_generate_type?(type_name)
-      type_name = type_name.downcase
-
-      return true if ALLOW_TYPE_NAME_PREFIXES.any? { |prefix| type_name.start_with?(prefix) }
-
-      ALLOW_TYPE_NAMES.include?(type_name)
-    end
-
     # @param function_name [String]
     # @param signature [String,nil]
     # @return [Array<RubyHeaderParser::ArgumentDefinition>]
@@ -222,7 +157,7 @@ def parse_definition_args(function_name, signature)
 
           if type.match?(/\*+$/)
             type = type.gsub(/\*+$/, "").strip
-            pointer ||= function_arg_pointer_hint(function_name, arg_pos - 1)
+            pointer ||= data.function_arg_pointer_hint(function_name:, index: arg_pos - 1)
           elsif /^void\s*\s/.match?(type) || /\(.*\)/.match?(type)
             # function pointer (e.g. void *(*func)(void *)) is treated as `void*`
             type = "void"
@@ -239,16 +174,6 @@ def parse_definition_args(function_name, signature)
       end.compact
     end
 
-    # @param function_name [String]
-    # @param index [Integer] arg position
-    # @return [Symbol] :ref, :array
-    def function_arg_pointer_hint(function_name, index)
-      pointer_hint = data["pointer_hint"]["function"].dig(function_name, index)
-      return pointer_hint.to_sym if pointer_hint
-
-      :ref
-    end
-
     # @param definition [String]
     # @param function_name [String]
     # @return [RubyHeaderParser::TyperefDefinition]

_tools/ruby_h_to_go/lib/ruby_header_parser/ruby_header_parser.yml

@@ -0,0 +1,83 @@
+# frozen_string_literal: true
+
+RSpec.describe RubyHeaderParser::Data do
+  let(:data) { RubyHeaderParser::Data.new }
+
+  describe "#function_arg_pointer_hint" do
+    subject { data.function_arg_pointer_hint(function_name:, index:) }
+
+    context "found in data.yml" do
+      let(:function_name) { "rb_funcallv" }
+      let(:index) { 3 }
+
+      it { should eq :array }
+    end
+
+    context "not found in data.yml" do
+      let(:function_name) { "rb_funcallv" }
+      let(:index) { 4 }
+
+      it { should eq :ref }
+    end
+  end
+
+  describe "#should_generate_function?" do
+    subject { data.should_generate_function?(function_name) }
+
+    context "rb function (denied)" do
+      let(:function_name) { "rb_check_safe_str" }
+
+      it { should eq false }
+    end
+
+    context "rb function (allowed)" do
+      let(:function_name) { "rb_define_class" }
+
+      it { should eq true }
+    end
+
+    context "no rb function" do
+      let(:function_name) { "sprintf" }
+
+      it { should eq false }
+    end
+  end
+
+  describe "#should_generate_struct?" do
+    subject { data.should_generate_struct?(struct_name) }
+
+    context "rb struct" do
+      let(:struct_name) { "rb_random_struct" }
+
+      it { should eq true }
+    end
+
+    context "non rb struct" do
+      let(:struct_name) { "fuga" }
+
+      it { should eq false }
+    end
+  end
+
+  describe "#should_generate_type?" do
+    subject { data.should_generate_type?(type_name) }
+
+    context "rb type" do
+      let(:type_name) { "rb_data_type_t" }
+
+      it { should eq true }
+    end
+
+    context "VALUE" do
+      let(:type_name) { "VALUE" }
+
+      it { should eq true }
+    end
+
+    context "unknown" do
+      let(:type_name) { "unknown" }
+
+      it { should eq false }
+    end
+  end
+end