path: root/spec/practical/json_spec.cr

require "../spec_helper"

include Parcom

alias JSONType = Nil \
               | Bool \
               | Int64 \
               | Float64 \
               | String \
               | Array(JSONValue) \
               | Hash(String, JSONValue)

struct JSONValue
  property data : JSONType

  def initialize(@data)
  end
end

def assoc_to_hash(arr : Array({K, V})) : Hash(K, V) forall K, V
  h = {} of K => V
  arr.each do |k, v|
    h[k] = v
  end
  h
end

describe "example: JSON parsing" do
  json_null = Parser.token_sequence("null".chars).map_const(JSONValue.new(nil))

  t = Parser.token_sequence("true".chars).map_const(true)
  f = Parser.token_sequence("false".chars).map_const(false)
  json_bool = (t | f).map { |b| JSONValue.new(b) }

  digits = Parser(Char, Char).satisfy(&.ascii_number?).many
  sign = Parser.token('-').map_const(-1).recover(1)
  point = Parser.token('.').optional
  json_number = sign.and_then do |s|
    digits.and_then do |front|
      point.and_then do |p|
        digits.and_then do |back|
          if front.empty? && back.empty?
            Parser(Char, JSONValue).flunk
          else
            x = 
              case
                when front.empty? && !p.nil? # {[], '.', _} # 0 <= x < 1
                  "0.#{back.join}".to_f64
                when back.empty? #{_, _, []} # whole number
                  front.join.to_i64
                else #when !front.empty? && !p.nil? && !back.empty? # some other float
                  "#{front.join}.#{back.join}".to_f64
                #else
                #  raise "json_number, Should be unreachable: #{front} #{p} #{back}"
              end
            x = x.to_i64 if x.is_a?(Float64) && (x % 1).zero?
            Parser(Char, JSONValue).pure(JSONValue.new(x * s))
          end
        end
      end
    end
  end

  q = Parser.token('"')
  s_char = Parser(Char, Char).first_of([
    Parser.token_sequence(['\\', '"']).map_const('"'),
    Parser(Char, Char).satisfy(&.!=('"')),
  ])
  string_literal = (q >> s_char.many.map(&.join) << q)
  json_string = string_literal.map { |s| JSONValue.new(s) }

  # Forward-declared here to allow for mutual recursion
  json_value   = uninitialized Parser(Char, JSONValue)
  p_json_value = pointerof(json_value)

  ws = Parser(Char, Char).satisfy(&.whitespace?).many
  a_sep = ws >> Parser.token(',') >> ws
  a_front = Parser.token('[') >> ws
  a_back = ws >> Parser.token(']')
  json_array = Parser(Char, JSONValue).new("json_array") do |tokens|
    # Has to be created and re-created here to allow mutual recursion
    # `a_sep.optional` matches a trailing comma in this expression
    non_empty_elements = p_json_value.value.sep_by(a_sep) << a_sep.optional
    elements = non_empty_elements.recover([] of JSONValue)
    p = (a_front >> elements << a_back).map { |es| JSONValue.new(es) }
    p.parse(tokens)
  end

  kv_sep = ws >> Parser.token(':') >> ws
  o_sep = ws >> Parser.token(',') >> ws
  o_front = Parser.token('{') >> ws
  o_back  = ws >> Parser.token('}')
  json_object = Parser(Char, JSONValue).new("json_object") do |tokens|
    # Has to be created and re-created here to allow mutual recursion
    kv_pair = string_literal + (kv_sep >> p_json_value.value)
    # `a_sep.optional` matches a trailing comma in this expression
    attributes = kv_pair.sep_by(o_sep) << o_sep.optional
    p = o_front >> attributes.recover([] of {String, JSONValue}) << o_back
    p.map { |pairs| JSONValue.new(assoc_to_hash(pairs)) }.parse(tokens)
  end

  json_value = Parser(Char, JSONValue).new("json_value") do |tokens|
    Parser(Char, JSONValue).first_of([
      json_null,
      json_bool,
      json_number,
      json_string,
      json_array,
      json_object,
    ]).parse(tokens)
  end

  describe "json_null" do
    it "parses the string 'null' and returns nil when successful" do
      result = json_null.parse(Tokens.from_string("null"))
      result.value.data.should be_nil
      result.tokens.empty?.should be_true

      expect_raises(ParserFail) { json_null.parse(Tokens.from_string("")) }
    end
  end

  describe "json_bool" do
    it "parses the strings 'true' or 'false' and returns bool" do
      result = json_bool.parse(Tokens.from_string("true"))
      result.value.data.should be_true
      result.tokens.empty?.should be_true

      result = json_bool.parse(Tokens.from_string("false"))
      result.value.data.should be_false
      result.tokens.empty?.should be_true

      expect_raises(ParserFail) { json_bool.parse(Tokens.from_string("")) }
    end
  end

  describe "json_number" do
    it "parses positive and negative integers" do
      {"42", "42.", "42.000000"}.each do |s|
        result = json_number.parse(Tokens.from_string(s))
        result.value.data.should eq(42)
        result.value.data.should be_a(Int64)
        result.tokens.empty?.should be_true

        result = json_number.parse(Tokens.from_string("-#{s}"))
        result.value.data.should eq(-42)
        result.value.data.should be_a(Int64)
        result.tokens.empty?.should be_true
      end
    end

    it "parses positive and negative floats between -1 and 1" do
      {"0.1234", ".1234", ".12340000"}.each do |s|
        result = json_number.parse(Tokens.from_string(s))
        result.value.data.should eq(0.1234)
        result.value.data.should be_a(Float64)
        result.tokens.empty?.should be_true

        result = json_number.parse(Tokens.from_string("-#{s}"))
        result.value.data.should eq(-0.1234)
        result.value.data.should be_a(Float64)
        result.tokens.empty?.should be_true
      end
    end

    it "parses positive and negative floats with a whole-number component" do
      {"12.34", "0012.3400"}.each do |s|
        result = json_number.parse(Tokens.from_string(s))
        result.value.data.should eq(12.34)
        result.value.data.should be_a(Float64)
        result.tokens.empty?.should be_true

        result = json_number.parse(Tokens.from_string("-#{s}"))
        result.value.data.should eq(-12.34)
        result.value.data.should be_a(Float64)
        result.tokens.empty?.should be_true
      end
    end

    it "parses 0 as an int in various forms" do
      {"0", "0.0", "0000", ".000", "000.", "-0", "-0.", "-.0", "-0.0"}.each do |s|
        result = json_number.parse(Tokens.from_string(s))
        result.value.data.should eq(0)
        result.value.data.should be_a(Int64)
        result.tokens.empty?.should be_true
      end
    end

    it "does not parse '.' or '-.' as 0" do
      expect_raises(ParserFail) { json_number.parse(Tokens.from_string(".")) }
    end

    it "fails when input if not a number" do
      expect_raises(ParserFail) { json_number.parse(Tokens.from_string("")) }
      expect_raises(ParserFail) { json_number.parse(Tokens.from_string("foo")) }
    end
  end

  describe "json_string" do
    it "can parse non-empty strings" do
      result = json_string.parse(Tokens.from_string("\"foo\""))
      result.value.data.should eq("foo")
      result.tokens.empty?.should be_true
    end

    it "can parse empty strings" do
      result = json_string.parse(Tokens.from_string("\"\""))
      result.value.data.should eq("")
      result.tokens.empty?.should be_true
    end

    it "can parse escaped quotes" do
      result = json_string.parse(Tokens.from_string("\"escaped \\\"\\\" quotes\""))
      result.value.data.should eq("escaped \"\" quotes")
      result.tokens.empty?.should be_true
    end
  end

  describe "json_array" do
    it "can parse empty arrays" do
      empty = JSONValue.new([] of JSONValue)
      {"[]", "[ \t\n\n ]"}.each do |s|
        tokens = Tokens.from_string(s)
        result = json_array.parse(tokens)
        result.value.should eq(empty)
        result.tokens.empty?.should be_true
      end
    end

    it "can parse arrays with one element" do
      {
        "[null]",
        "[ \n1]",
        "[0.123 \n]",
        "[true,]",
        "[false , ]",
        "[\"a string\"]",
        "[{\"hsrths\": 78.9}]"
      }.each do |s|
        tokens = Tokens.from_string(s)
        result = json_array.parse(tokens)
        result.value.data.should be_a(Array(JSONValue))
        result.value.data.as(Array(JSONValue)).size.should eq(1)
        result.tokens.empty?.should be_true
      end
    end

    it "can parse arrays with many elements" do
      {"[\"one\" , \"two\"]", "[1,2,3,4,]", "[ null\n, false\n, true\n]"}.each do |s|
        tokens = Tokens.from_string(s)
        # TODO: come up with better test format
        result = json_array.parse(tokens)
        result.value.data.as(Array(JSONValue)).size.>(1).should be_true
        result.tokens.empty?.should be_true
      end
    end

    it "can parse nested arrays" do
      expected = JSONValue.new([
        JSONValue.new([] of JSONValue),
        JSONValue.new([
          JSONValue.new(nil),
          JSONValue.new(false),
          JSONValue.new("two"),
          JSONValue.new(3_i64),
          JSONValue.new(4.1_f64),
          JSONValue.new([] of JSONValue),
        ]),
        JSONValue.new([JSONValue.new(5_i64)]),
      ])
      s = "[ [], [null, false, \"two\", 3, 4.1, [],] ,[5]]"
      tokens = Tokens.from_string(s)
      result = json_array.parse(tokens)
      result.value.should eq(expected)
      result.tokens.empty?.should be_true
    end

    it "does not allow trailing commas in empty arrays" do
      {"[,]", "[ , ]"}.each do |s|
        expect_raises(ParserFail) { json_array.parse(Tokens.from_string(s)) }
      end
    end
  end

  describe "json_object" do
    it "can parse empty objects" do
      empty = JSONValue.new({} of String => JSONValue)
      {"{}", "{  }", "{ \n\t\r}"}.each do |s|
        tokens = Tokens.from_string(s)
        result = json_object.parse(tokens)
        result.value.should eq(empty)
        result.tokens.empty?.should be_true
      end
    end

    it "can parse objects with one attribute" do
      {
        "{\"foo\":null}",
        "{\"foo\": true,}",
        "{\"foo\" :1}",
        "{\"foo\" : 2.3 , }",
        "{\"foo\"  :  \"bar\"}",
        "{\"foo\"\t:\n[1],    }",
      }.each do |s|
        tokens = Tokens.from_string(s)
        result = json_object.parse(tokens)
        result.value.data.should be_a(Hash(String, JSONValue))
        result.value.data.as(Hash(String, JSONValue)).size.should eq(1)
        result.tokens.empty?.should be_true
      end
    end

    it "can parse objects with more than one attribute" do
      {
        "{\"foo\":null,\"bar\":false}",
        "{ \"foo\" : 3.4 , \"bar\" : 1 , }",
        "{\"foo\": \"some string\", \"bar\": [1, 2, 3] }",
      }.each do |s|
        tokens = Tokens.from_string(s)
        result = json_object.parse(tokens)
        result.value.data.should be_a(Hash(String, JSONValue))
        result.value.data.as(Hash(String, JSONValue)).size.>(1).should be_true
        result.tokens.empty?.should be_true
      end
    end

    it "can parse nested objects" do
      expected = JSONValue.new({
        "one"   => JSONValue.new({"foo" => JSONValue.new(8_i64)}),
        "two"   => JSONValue.new(nil),
        "three" => JSONValue.new({
          "foo" => JSONValue.new(7_i64),
          "bar" => JSONValue.new([
            JSONValue.new(9_i64),
            JSONValue.new(8_i64),
            JSONValue.new(7_i64),
            JSONValue.new(nil),
          ]),
        }),
      })
      s = "{
    \"one\": {\"foo\": 8,},
    \"two\": null,
    \"three\": {\"foo\": 7, \"bar\" : [9,8,7,null]},
}"
      tokens = Tokens.from_string(s)
      result = json_object.parse(tokens)
      result.value.should eq(expected)
      result.tokens.empty?.should be_true
    end

    it "does not allow trailing commas in empty objects" do
      {"{,}", "{  ,   }"}.each do |s|
        expect_raises(ParserFail) { json_object.parse(Tokens.from_string(s)) }
      end
    end
  end

  describe "json_value" do
    it "parses null" do
      result = json_value.parse(Tokens.from_string("null"))
      result.value.data.should be_nil
      result.tokens.empty?.should be_true
    end

    it "parses bools" do
      result = json_value.parse(Tokens.from_string("true"))
      result.value.data.should be_true
      result.tokens.empty?.should be_true

      result = json_value.parse(Tokens.from_string("false"))
      result.value.data.should be_false
      result.tokens.empty?.should be_true
    end

    it "parses ints" do
      result = json_value.parse(Tokens.from_string("42"))
      result.value.data.should eq(42)
      result.value.data.should be_a(Int64)
      result.tokens.empty?.should be_true

      result = json_value.parse(Tokens.from_string("-42"))
      result.value.data.should eq(-42)
      result.value.data.should be_a(Int64)
      result.tokens.empty?.should be_true
    end

    it "parses floats" do
      result = json_value.parse(Tokens.from_string("12.34"))
      result.value.data.should eq(12.34)
      result.value.data.should be_a(Float64)
      result.tokens.empty?.should be_true

      result = json_value.parse(Tokens.from_string("-12.34"))
      result.value.data.should eq(-12.34)
      result.value.data.should be_a(Float64)
      result.tokens.empty?.should be_true
    end

    it "parses strings" do
      result = json_value.parse(Tokens.from_string("\"foo\""))
      result.value.data.should eq("foo")
      result.tokens.empty?.should be_true
    end

    it "parses arrays" do
      {"[]", "[ null, ]", "[1, 2, 3, 4.1, 5]"}.each do |s|
        result = json_value.parse(Tokens.from_string(s))
        result.value.data.should be_a(Array(JSONValue))
        result.tokens.empty?.should be_true
      end
    end

    it "parses objects" do
      {"{}", "{\"foo\": 1}"}.each do |s|
        result = json_value.parse(Tokens.from_string(s))
        result.value.data.should be_a(Hash(String, JSONValue))
        result.tokens.empty?.should be_true
      end
    end
  end
end