ripper_test.rb

# frozen_string_literal: true

return if RUBY_VERSION < "3.3" || RUBY_ENGINE != "ruby"

require_relative "../test_helper"
require "ripper"

module Prism
  class RipperTest < TestCase
    # Skip these tests that Ripper is reporting the wrong results for.
    incorrect = [
      # Ripper incorrectly attributes the block to the keyword.
      "seattlerb/block_return.txt",
      "whitequark/return_block.txt",

      # Ripper cannot handle named capture groups in regular expressions.
      "regex.txt",

      # Ripper fails to understand some structures that span across heredocs.
      "spanning_heredoc.txt",

      # Ripper interprets circular keyword arguments as method calls.
      "3.4/circular_parameters.txt",

      # Ripper doesn't emit `args_add_block` when endless method is prefixed by modifier.
      "4.0/endless_methods_command_call.txt",

      # https://bugs.ruby-lang.org/issues/21168#note-5
      "command_method_call_2.txt",
    ]

    if RUBY_VERSION.start_with?("3.3.")
      incorrect += [
        "whitequark/lvar_injecting_match.txt",
        "seattlerb/parse_pattern_058.txt",
        "regex_char_width.txt",
      ]
    end

    # Skip these tests that we haven't implemented yet.
    omitted_sexp_raw = [
      "dos_endings.txt",
      "heredocs_with_fake_newlines.txt",
      "heredocs_with_ignored_newlines.txt",
      "seattlerb/block_call_dot_op2_brace_block.txt",
      "seattlerb/block_command_operation_colon.txt",
      "seattlerb/block_command_operation_dot.txt",
      "seattlerb/heredoc__backslash_dos_format.txt",
      "seattlerb/heredoc_backslash_nl.txt",
      "seattlerb/heredoc_nested.txt",
      "seattlerb/heredoc_squiggly_blank_line_plus_interpolation.txt",
      "tilde_heredocs.txt",
      "unparser/corpus/semantic/dstr.txt",
      "whitequark/dedenting_heredoc.txt",
      "whitequark/parser_drops_truncated_parts_of_squiggly_heredoc.txt",
      "whitequark/parser_slash_slash_n_escaping_in_literals.txt",
      "whitequark/ruby_bug_18878.txt",
      "whitequark/send_block_chain_cmd.txt",
      "whitequark/slash_newline_in_heredocs.txt"
    ]

    omitted_lexer_parse = [
      "comments.txt",
      "heredoc_percent_q_newline_delimiter.txt",
      "heredoc_with_escaped_newline_at_start.txt",
      "heredocs_with_fake_newlines.txt",
      "indented_file_end.txt",
      "seattlerb/TestRubyParserShared.txt",
      "seattlerb/class_comments.txt",
      "seattlerb/module_comments.txt",
      "seattlerb/parse_line_block_inline_comment_leading_newlines.txt",
      "seattlerb/parse_line_block_inline_multiline_comment.txt",
      "spanning_heredoc_newlines.txt",
      "strings.txt",
      "whitequark/dedenting_heredoc.txt",
      "whitequark/procarg0.txt",
    ]

    Fixture.each_for_current_ruby(except: incorrect | omitted_sexp_raw) do |fixture|
      define_method("#{fixture.test_name}_sexp_raw") { assert_ripper_sexp_raw(fixture.read) }
    end

    Fixture.each_for_current_ruby(except: incorrect | omitted_lexer_parse) do |fixture|
      define_method("#{fixture.test_name}_lexer_parse") { assert_ripper_lexer_parse(fixture.read) }
    end

    # Check that the hardcoded values don't change without us noticing.
    def test_internals
      actual = Translation::Ripper.constants.select { |name| name.start_with?("EXPR_") }.sort
      expected = Ripper.constants.select { |name| name.start_with?("EXPR_") }.sort

      assert_equal(expected, actual)
      expected.zip(actual).each do |ripper, prism|
        assert_equal(Ripper.const_get(ripper), Translation::Ripper.const_get(prism))
      end
    end

    private

    def assert_ripper_sexp_raw(source)
      assert_equal Ripper.sexp_raw(source), Prism::Translation::Ripper.sexp_raw(source)
    end

    def assert_ripper_lexer_parse(source)
      prism = Translation::Ripper::Lexer.new(source).parse
      ripper = Ripper::Lexer.new(source).parse
      ripper.reject! { |elem| elem.event == :on_sp } # Prism doesn't emit on_sp
      ripper.sort_by!(&:pos) # Prism emits tokens by their order in the code, not in parse order

      [prism.size, ripper.size].max.times do |i|
        expected = ripper[i].to_a
        actual = prism[i].to_a
        # Since tokens related to heredocs are not emitted in the same order,
        # the state also doesn't line up.
        if expected[1] == :on_heredoc_end && actual[1] == :on_heredoc_end
          expected[3] = actual[3] = nil
        end

        assert_equal(expected, actual)
      end
    end
  end
end