mini JSON

minijson.rb

class ParseError < StandardError; end

def object(s, i)
  o = {}
  state = :key
  key = nil
  j = i + 1

  while (c = s[j])
    case state
    when :key
      case c
      when ?}
        return [o, j + 1] if o.empty?
        # This came after a comma, so it's not allowed.
        raise ParseError, "expected string key (after key #{key} for object started at #{i}), instead got #{c} at #{j}"
      when ?"
        closing_quote = s.index(?", j + 1)
        raise ParseError, "missing closing quote (for key started at #{j + 1} for object started at #{i})" unless closing_quote
        key = s[(j + 1)...closing_quote]
        state = :colon
        j = closing_quote + 1
      else raise ParseError, "expected string key (for object started at #{i}), instead got #{c} at #{j}"
      end
    when :colon
      if c == ?:
        state = :value
        j += 1
      else raise ParseError, "expected colon (for key #{key} for object started at #{i}), instead got #{c} at #{j}"
      end
    when :value
      case c
      when ?{
        subobj, j = object(s, j)
        o[key] = subobj
        state = :comma
      when ?"
        closing_quote = s.index(?", j + 1)
        raise ParseError, "missing closing quote (for value started at #{j + 1} for key #{key} for object started at #{i})" unless closing_quote
        o[key] = s[(j + 1)...closing_quote]
        state = :comma
        j = closing_quote + 1
      else raise ParseError, "expected value (for key #{key} for object started at #{i}), instead got #{c} at #{j}"
      end
    when :comma
      case c
      when ?,
        j += 1
        state = :key
      when ?}; return [o, j + 1]
      else raise ParseError, "expected comma (after key #{key} for object started at #{i}), instead got #{c} at #{j}"
      end
    else
      raise ParseError, "Unknown state #{state}"
    end
  end

  # TODO explain what I wanted, based on my state
  raise ParseError, "incomplete object started at #{i}"
end

def json(s)
  i = s.each_char.find_index { |c| !c.strip.empty? }
  raise ParseError, "whitespace only '#{s}' invalid" unless i
  if s[i] == ?{
    o, i = object(s, 0)
    raise ParseError, "extra characters after #{o} at #{i}" if i != s.size
    o
  else
    raise ParseError, "unexpected #{s[i]} at #{i}"
  end
end

kv = '{"k":"v"}'.freeze

tests = [
  ['{}', {}],
  ['{"k":"v"}', {?k => ?v}],
  # trailing comma is sadly not allowed
  ['{"k":"v",}', :raise],
  ['{"k1":"v1","k2":"v2"}', {'k1' => 'v1', 'k2' => 'v2'}],
  ['{}{}', :raise],
  *(0...kv.size).map { |l| [kv[0...l], :raise] },
  ['{k:"v"}', :raise],
  ['{"k":v}', :raise],
  # TIMING: Done with all string cases at 0:40
  # (this included the time taken to actually write the tests)
  ['{"a":{"b":"c"}}', {?a => {?b => ?c}}],
  # TIMING: Nested object took only four extra lines and < 1 minute.
]

passed = 0
failed = 0

tests.each { |i, exp|
  puts i
  begin
    p = json(i)
  rescue ParseError => e
    if exp == :raise
      passed += 1
    else
      puts "  raised #{e}"
      failed += 1
    end
    next
  else
    if exp == :raise
      failed += 1
      puts "  didn't raise (=> #{p}), but should have"
      next
    end
  end

  if p == exp
    passed += 1
  else
    failed += 1
    puts "  => #{p}, should be #{exp}"
  end
}

puts "#{passed} passed, #{failed} failed"
raise "#{failed} failed" if failed > 0