jsone_decode_tests.erl 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288
  1. %% Copyright (c) 2013-2015, Takeru Ohta <phjgt308@gmail.com>
  2. %% coding: latin-1
  3. -module(jsone_decode_tests).
  4. -include_lib("eunit/include/eunit.hrl").
  5. -ifdef('NO_MAP_TYPE').
  6. -define(MAP_OBJECT_TYPE, tuple).
  7. -define(OBJ0, {[]}).
  8. -define(OBJ1(K, V), {[{K, V}]}).
  9. -define(OBJ2(K1, V1, K2, V2), {[{K1, V1}, {K2, V2}]}).
  10. -define(OBJ2_DUP_KEY(K1, V1, K2, V2), ?OBJ2(K1, V1, K2, V2)).
  11. -else.
  12. -define(MAP_OBJECT_TYPE, map).
  13. -define(OBJ0, #{}).
  14. -define(OBJ1(K, V), #{K => V}).
  15. -define(OBJ2(K1, V1, K2, V2), #{K1 => V1, K2 => V2}).
  16. -define(OBJ2_DUP_KEY(K1, V1, _K2, _V2), #{K1 => V1}). % the first (leftmost) value is used
  17. -define(OBJ2_DUP_KEY_LAST(_K1, _V1, K2, V2), #{K2 => V2}). % the last value is used
  18. -endif.
  19. decode_test_() ->
  20. [
  21. %% Symbols
  22. {"false", fun () -> ?assertEqual({ok, false, <<"">>}, jsone_decode:decode(<<"false">>))end},
  23. {"true", fun () -> ?assertEqual({ok, true, <<"">>}, jsone_decode:decode(<<"true">>))end},
  24. {"null", fun () -> ?assertEqual({ok, null, <<"">>}, jsone_decode:decode(<<"null">>))end},
  25. %% Numbers: Integer
  26. {"positive integer", fun () -> ?assertEqual({ok, 1, <<"">>}, jsone_decode:decode(<<"1">>))end},
  27. {"zero", fun () -> ?assertEqual({ok, 0, <<"">>}, jsone_decode:decode(<<"0">>))end},
  28. {"negative integer", fun () -> ?assertEqual({ok, -1, <<"">>}, jsone_decode:decode(<<"-1">>))end},
  29. {"large integer (no limit on size)",
  30. fun () ->
  31. ?assertEqual({ok, 111111111111111111111111111111111111111111111111111111111111111111111111111111, <<"">>},
  32. jsone_decode:decode(<<"111111111111111111111111111111111111111111111111111111111111111111111111111111">>))
  33. end},
  34. {"integer with leading zero (interpreted as zero and remaining binary)",
  35. fun () ->
  36. ?assertEqual({ok, 0, <<"0">>}, jsone_decode:decode(<<"00">>)),
  37. ?assertEqual({ok, 0, <<"1">>}, jsone_decode:decode(<<"01">>)),
  38. ?assertEqual({ok, 0, <<"1">>}, jsone_decode:decode(<<"-01">>))
  39. end},
  40. {"integer can't begin with an explicit plus sign",
  41. fun () -> ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"+1">>))end},
  42. %% Numbers: Floats
  43. {"float: decimal notation", fun () -> ?assertEqual({ok, 1.23, <<"">>}, jsone_decode:decode(<<"1.23">>))end},
  44. {"float: exponential notation",
  45. fun () ->
  46. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"12345e-3">>)), % lower case 'e'
  47. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"12345E-3">>)), % upper case 'E'
  48. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"12345.0e-3">>)),
  49. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"0.12345E2">>)),
  50. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"0.12345e+2">>)), % exponent part can begin with plus sign
  51. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"0.12345E+2">>)),
  52. ?assertEqual({ok, -12.345, <<"">>}, jsone_decode:decode(<<"-0.012345e3">>)),
  53. ?assertEqual({ok, 123.0, <<"">>}, jsone_decode:decode(<<"1.23000000000000000000e+02">>))
  54. end},
  55. {"float: invalid format",
  56. fun () ->
  57. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<".123">>)), % omitted integer part
  58. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.">>)), % omitted fraction part: EOS
  59. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.e+3">>)), % omitted fraction part: with exponent part
  60. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.1e">>)), % incomplete fraction part
  61. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.1e-">>)), % incomplete fraction part
  62. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.1ee-1">>)), % duplicated 'e'
  63. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.1e--1">>)), % duplicated sign
  64. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"1e999">>)), % exponent out of range
  65. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.1e999">>)), % exponent out of range
  66. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"100000000000000000000000000000000000e300">>)), % product out of range
  67. ?assertEqual({ok, 0.1, <<".2">>}, jsone_decode:decode(<<"0.1.2">>)) % duplicated '.': interpreted as individual tokens
  68. end},
  69. %% Strings
  70. {"simple string", fun () -> ?assertEqual({ok, <<"abc">>, <<"">>}, jsone_decode:decode(<<"\"abc\"">>))end},
  71. {"string: escaped characters",
  72. fun () ->
  73. Input = list_to_binary([$", [[$\\, C] || C <- [$", $/, $\\, $b, $f, $n, $r, $t]], $"]),
  74. Expected = <<"\"\/\\\b\f\n\r\t">>,
  75. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input))
  76. end},
  77. {"string: escaped Unicode characters",
  78. fun () ->
  79. %% japanese
  80. Input1 = <<"\"\\u3042\\u3044\\u3046\\u3048\\u304A\"">>,
  81. Expected1 = <<"あいうえお">>, % assumed that the encoding of this file is UTF-8
  82. ?assertEqual({ok, Expected1, <<"">>}, jsone_decode:decode(Input1)),
  83. %% ascii
  84. Input2 = <<"\"\\u0061\\u0062\\u0063\"">>,
  85. Expected2 = <<"abc">>,
  86. ?assertEqual({ok, Expected2, <<"">>}, jsone_decode:decode(Input2)),
  87. %% other multi-byte characters
  88. Input3 = <<"\"\\u06DD\\u06DE\\u10AE\\u10AF\"">>,
  89. Expected3 = <<"۝۞ႮႯ">>,
  90. ?assertEqual({ok, Expected3, <<"">>}, jsone_decode:decode(Input3)),
  91. %% mixture of ascii and japanese characters
  92. Input4 = <<"\"a\\u30421\\u3044bb\\u304622\\u3048ccc\\u304A333\"">>,
  93. Expected4 = <<"aあ1いbbう22えcccお333">>, % assumed that the encoding of this file is UTF-8
  94. ?assertEqual({ok, Expected4, <<"">>}, jsone_decode:decode(Input4))
  95. end},
  96. {"string: surrogate pairs",
  97. fun () ->
  98. Input = <<"\"\\ud848\\udc49\\ud848\\udc9a\\ud848\\udcfc\"">>,
  99. Expected = <<"𢁉𢂚𢃼">>,
  100. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input))
  101. end},
  102. {"string: control characters",
  103. fun () ->
  104. Ctrls = lists:seq(0, 16#1f),
  105. lists:foreach(fun (C) ->
  106. %% Control characters are unacceptable
  107. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<$", C, $">>))
  108. end,
  109. Ctrls),
  110. lists:foreach(fun (C) ->
  111. %% `allow_ctrl_chars' option allows strings which contain unescaped control characters
  112. ?assertEqual({ok, <<C>>, <<"">>},
  113. jsone_decode:decode(<<$", C, $">>, [{allow_ctrl_chars, true}]))
  114. end,
  115. Ctrls)
  116. end},
  117. {"string: invalid escape characters",
  118. fun () ->
  119. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\z\"">>)), % '\z' is undefined
  120. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\uab\"">>)), % too few hex characters
  121. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\ud848\"">>)), % high(first) surrogate only
  122. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\udc49\"">>)), % low(second) surrogate only
  123. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\ud848\\u0061\"">>)), % missing low(second) surrogate
  124. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\udf0u\"">>)),
  125. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\ud848\\udf0u\"">>)),
  126. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\u-3351\"">>))
  127. end},
  128. %% Arrays
  129. {"simple array",
  130. fun () ->
  131. Input = <<"[1,2,\"abc\",null]">>,
  132. Expected = [1, 2, <<"abc">>, null],
  133. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input))
  134. end},
  135. {"array: contains whitespaces",
  136. fun () ->
  137. Input = <<"[ 1,\t2, \n \"abc\",\r null]">>,
  138. Expected = [1, 2, <<"abc">>, null],
  139. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input))
  140. end},
  141. {"empty array",
  142. fun () ->
  143. ?assertEqual({ok, [], <<"">>}, jsone_decode:decode(<<"[]">>)),
  144. ?assertEqual({ok, [], <<"">>}, jsone_decode:decode(<<"[ \t\r\n]">>))
  145. end},
  146. {"array: trailing comma is disallowed",
  147. fun () ->
  148. Input = <<"[1, 2, \"abc\", null, ]">>,
  149. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  150. end},
  151. {"array: missing comma",
  152. fun () ->
  153. Input = <<"[1 2, \"abc\", null]">>, % a missing comma between '1' and '2'
  154. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  155. end},
  156. {"array: missing closing bracket",
  157. fun () ->
  158. Input = <<"[1, 2, \"abc\", null">>,
  159. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  160. end},
  161. %% Objects
  162. {"simple object",
  163. fun () ->
  164. Input = <<"{\"1\":2,\"key\":\"value\"}">>,
  165. Expected = ?OBJ2(<<"1">>, 2, <<"key">>, <<"value">>),
  166. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input)), % `map' is the default format
  167. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input, [{object_format, ?MAP_OBJECT_TYPE}]))
  168. end},
  169. {"simple object: tuple or proplist",
  170. fun () ->
  171. Input = <<"{\"1\":2,\"key\":\"value\"}">>,
  172. Expected = {[{<<"1">>, 2}, {<<"key">>, <<"value">>}]},
  173. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input, [{object_format, tuple}])),
  174. ?assertEqual({ok, element(1, Expected), <<"">>}, jsone_decode:decode(Input, [{object_format, proplist}]))
  175. end},
  176. {"object: contains whitespaces",
  177. fun () ->
  178. Input = <<"{ \"1\" :\t 2,\n\r\"key\" : \n \"value\"}">>,
  179. Expected = ?OBJ2(<<"1">>, 2, <<"key">>, <<"value">>),
  180. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input))
  181. end},
  182. {"empty object",
  183. fun () ->
  184. ?assertEqual({ok, ?OBJ0, <<"">>}, jsone_decode:decode(<<"{}">>)),
  185. ?assertEqual({ok, ?OBJ0, <<"">>}, jsone_decode:decode(<<"{ \t\r\n}">>)),
  186. ?assertEqual({ok, {[]}, <<"">>}, jsone_decode:decode(<<"{}">>, [{object_format, tuple}])),
  187. ?assertEqual({ok, [{}], <<"">>}, jsone_decode:decode(<<"{}">>, [{object_format, proplist}]))
  188. end},
  189. {"empty object: map",
  190. fun () ->
  191. ?assertEqual({ok, ?OBJ0, <<"">>}, jsone_decode:decode(<<"{}">>, [{object_format, ?MAP_OBJECT_TYPE}]))
  192. end},
  193. {"duplicated members: map",
  194. fun () ->
  195. Input = <<"{\"1\":\"first\",\"1\":\"second\"}">>,
  196. Expected = ?OBJ2_DUP_KEY(<<"1">>, <<"first">>, <<"1">>, <<"second">>),
  197. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input, [{object_format, ?MAP_OBJECT_TYPE}]))
  198. end},
  199. {"duplicated members last: map",
  200. fun () ->
  201. Input = <<"{\"1\":\"first\",\"1\":\"second\"}">>,
  202. Expected = ?OBJ2_DUP_KEY_LAST(<<"1">>, <<"first">>, <<"1">>, <<"second">>),
  203. ?assertEqual({ok, Expected, <<"">>},
  204. jsone_decode:decode(Input, [{object_format, ?MAP_OBJECT_TYPE}, {duplicate_map_keys, last}]))
  205. end},
  206. {"object: trailing comma is disallowed",
  207. fun () ->
  208. Input = <<"{\"1\":2, \"key\":\"value\", }">>,
  209. io:format("~p\n", [catch jsone_decode:decode(Input)]),
  210. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input, [{object_format, tuple}]))
  211. end},
  212. {"object: missing comma",
  213. fun () ->
  214. Input = <<"{\"1\":2 \"key\":\"value\"}">>,
  215. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  216. end},
  217. {"object: missing field key",
  218. fun () ->
  219. Input = <<"{:2, \"key\":\"value\"}">>,
  220. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  221. end},
  222. {"object: non string key",
  223. fun () ->
  224. Input = <<"{1:2, \"key\":\"value\"}">>,
  225. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  226. end},
  227. {"object: missing field value",
  228. fun () ->
  229. Input = <<"{\"1\", \"key\":\"value\"}">>,
  230. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  231. end},
  232. {"object: missing closing brace",
  233. fun () ->
  234. Input = <<"{\"1\":2 \"key\":\"value\"">>,
  235. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  236. end},
  237. {"atom keys",
  238. fun () ->
  239. KeyOpt = fun (Keys) -> [{keys, Keys}, {object_format, proplist}]end,
  240. Input = <<"{\"foo\":\"ok\"}">>,
  241. ?assertEqual([{<<"foo">>, <<"ok">>}], jsone:decode(Input, KeyOpt(binary))),
  242. ?assertEqual([{foo, <<"ok">>}], jsone:decode(Input, KeyOpt(atom))),
  243. ?assertEqual([{foo, <<"ok">>}], jsone:decode(Input, KeyOpt(existing_atom))),
  244. ?assertError(badarg, jsone:decode(<<"{\"@#$%^!\":\"ok\"}">>, KeyOpt(existing_atom))),
  245. ?assertEqual([{foo, <<"ok">>}], jsone:decode(Input, KeyOpt(attempt_atom))),
  246. ?assertEqual([{<<"@#$%^!">>, <<"ok">>}], jsone:decode(<<"{\"@#$%^!\":\"ok\"}">>, KeyOpt(attempt_atom))),
  247. Value = integer_to_binary(1234),
  248. % do not make atom in test code
  249. [{Atom, <<"ok">>}] = jsone:decode(<<"{\"", Value/binary, "\":\"ok\"}">>, KeyOpt(atom)),
  250. ?assertEqual(Value, atom_to_binary(Atom, latin1))
  251. end},
  252. {"garbage remainings chars",
  253. fun () ->
  254. ?assertError(badarg, jsone:decode(<<"1@">>)),
  255. ?assertEqual(1, jsone:decode(<<"1 \n\t\r ">>)) % Whitespaces are OK
  256. end},
  257. %% Others
  258. {"compound data",
  259. fun () ->
  260. Input = <<" [true, {\"1\" : 2, \"array\":[[[[1]]], {\"ab\":\"cd\"}, false]}, null] ">>,
  261. Expected = [true, ?OBJ2(<<"1">>, 2, <<"array">>, [[[[1]]], ?OBJ1(<<"ab">>, <<"cd">>), false]), null],
  262. ?assertEqual({ok, Expected, <<" ">>}, jsone_decode:decode(Input))
  263. end},
  264. {"undefined_as_null option",
  265. fun () ->
  266. ?assertEqual({ok, undefined, <<>>}, jsone_decode:decode(<<"null">>, [undefined_as_null])), % OK
  267. ?assertEqual({ok, null, <<>>}, jsone_decode:decode(<<"null">>, [])) % OK
  268. end},
  269. {"Invalid UTF-8 characters",
  270. fun () ->
  271. Input =
  272. <<123, 34, 105, 100, 34, 58, 34, 190, 72, 94, 90, 253, 121, 94, 71, 73, 68, 91, 122, 211, 253, 32, 94,
  273. 86, 67, 163, 253, 230, 34, 125>>,
  274. ?assertMatch({ok, _, _}, jsone:try_decode(Input)),
  275. ?assertMatch({error, {badarg, _}}, jsone:try_decode(Input, [reject_invalid_utf8]))
  276. end}].