ahem/spec/lexer_spec.rb

RSpec.describe Lexer do
  it 'lexes null' do
    expect(Lexer.new('null').scan_all).to eq(
      [Token.new(TokenKinds::NULL), Token.new(TokenKinds::EOF)]
    )
  end

  it 'lexes booleans' do
    expect(Lexer.new('true false').scan_all).to eq(
      [
        Token.new(TokenKinds::BOOLEAN, true),
        Token.new(TokenKinds::BOOLEAN, false),
        Token.new(TokenKinds::EOF)
      ]
    )
  end

  it 'lexes numbers' do
    expect(Lexer.new('1 2.3').scan_all).to eq(
      [
        Token.new(TokenKinds::NUMBER, 1.0),
        Token.new(TokenKinds::NUMBER, 2.3),
        Token.new(TokenKinds::EOF)
      ]
    )
  end

  it 'lexes strings' do
    expect(Lexer.new('"hello world"').scan_all).to eq(
      [Token.new(TokenKinds::STRING, 'hello world'), Token.new(TokenKinds::EOF)]
    )
  end

  it 'lexes operators' do
    expect(Lexer.new('+-*/<<=>>===# and or not').scan_all).to eq(
      [
        Token.new(TokenKinds::OPERATOR, :+),
        Token.new(TokenKinds::OPERATOR, :-),
        Token.new(TokenKinds::OPERATOR, :*),
        Token.new(TokenKinds::OPERATOR, :/),
        Token.new(TokenKinds::OPERATOR, :<),
        Token.new(TokenKinds::OPERATOR, :<=),
        Token.new(TokenKinds::OPERATOR, :>),
        Token.new(TokenKinds::OPERATOR, :>=),
        Token.new(TokenKinds::OPERATOR, :==),
        Token.new(TokenKinds::OPERATOR, :'#'),
        Token.new(TokenKinds::OPERATOR, :and),
        Token.new(TokenKinds::OPERATOR, :or),
        Token.new(TokenKinds::OPERATOR, :not),
        Token.new(TokenKinds::EOF)
      ]
    )
  end

  it 'lexes punctuation' do
    expect(Lexer.new('{}()[];,.').scan_all).to eq(
      [
        Token.new(TokenKinds::LBRACE),
        Token.new(TokenKinds::RBRACE),
        Token.new(TokenKinds::LPAREN),
        Token.new(TokenKinds::RPAREN),
        Token.new(TokenKinds::LBRACKET),
        Token.new(TokenKinds::RBRACKET),
        Token.new(TokenKinds::SEMICOLON),
        Token.new(TokenKinds::COMMA),
        Token.new(TokenKinds::DOT),
        Token.new(TokenKinds::EOF)
      ]
    )
  end

  it 'lexes if elseif else' do
    expect(Lexer.new('if elseif else').scan_all).to eq(
      [
        Token.new(TokenKinds::IF),
        Token.new(TokenKinds::ELSEIF),
        Token.new(TokenKinds::ELSE),
        Token.new(TokenKinds::EOF)
      ]
    )
  end

  it 'lexes =' do
    expect(Lexer.new('=').scan_all).to eq(
      [Token.new(TokenKinds::EQUALS), Token.new(TokenKinds::EOF)]
    )
  end

  it 'lexes let' do
    expect(Lexer.new('let').scan_all).to eq(
      [Token.new(TokenKinds::LET), Token.new(TokenKinds::EOF)]
    )
  end

  it 'lexes identifiers' do
    expect(Lexer.new('x').scan_all).to eq(
      [Token.new(TokenKinds::IDENTIFIER, 'x'), Token.new(TokenKinds::EOF)]
    )
  end

  it 'lexes function' do
    expect(Lexer.new('function').scan_all).to eq(
      [Token.new(TokenKinds::FUNCTION), Token.new(TokenKinds::EOF)]
    )
  end

  it 'lexes class' do
    expect(Lexer.new('class').scan_all).to eq(
      [Token.new(TokenKinds::CLASS), Token.new(TokenKinds::EOF)]
    )
  end

  it 'lexes class names' do
    expect(Lexer.new('Class').scan_all).to eq(
      [Token.new(TokenKinds::CLASS_NAME, 'Class'), Token.new(TokenKinds::EOF)]
    )
  end

  it 'lexes public and private' do
    expect(Lexer.new('public private').scan_all).to eq(
      [
        Token.new(TokenKinds::PUBLIC),
        Token.new(TokenKinds::PRIVATE),
        Token.new(TokenKinds::EOF)
      ]
    )
  end

  it 'lexes hash' do
    expect(Lexer.new('{ :a => 1 }').scan_all).to eq(
      [
        Token.new(TokenKinds::LBRACE),
        Token.new(TokenKinds::ATOM, :a),
        Token.new(TokenKinds::ROCKET),
        Token.new(TokenKinds::NUMBER, 1.0),
        Token.new(TokenKinds::RBRACE),
        Token.new(TokenKinds::EOF)
      ]
    )
  end

  it 'lexes for' do
    expect(Lexer.new('for in').scan_all).to eq(
      [
        Token.new(TokenKinds::FOR),
        Token.new(TokenKinds::IN),
        Token.new(TokenKinds::EOF)
      ]
    )
  end

  it 'allows escaping quotes in strings' do
    expect(Lexer.new('"hello \"world\""').scan_all).to eq(
      [
        Token.new(TokenKinds::STRING, 'hello "world"'),
        Token.new(TokenKinds::EOF)
      ]
    )
  end

  it 'supports newlines and tabs in strings' do
    expect(Lexer.new('"hello\nworld" "hello\tworld"').scan_all).to eq(
      [
        Token.new(TokenKinds::STRING, "hello\nworld"),
        Token.new(TokenKinds::STRING, "hello\tworld"),
        Token.new(TokenKinds::EOF)
      ]
    )
  end

  it 'ignores comments' do
    expect(Lexer.new("// prints 5\nprint(5)").scan_all).to eq(
      [
        Token.new(TokenKinds::IDENTIFIER, 'print'),
        Token.new(TokenKinds::LPAREN),
        Token.new(TokenKinds::NUMBER, 5),
        Token.new(TokenKinds::RPAREN),
        Token.new(TokenKinds::EOF)
      ]
    )
  end
end