jsone_decode_tests.erl 14 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295
  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. -endif.
  18. decode_test_() ->
  19. [
  20. %% Symbols
  21. {"false",
  22. fun () ->
  23. ?assertEqual({ok, false, <<"">>}, jsone_decode:decode(<<"false">>))
  24. end},
  25. {"true",
  26. fun () ->
  27. ?assertEqual({ok, true, <<"">>}, jsone_decode:decode(<<"true">>))
  28. end},
  29. {"null",
  30. fun () ->
  31. ?assertEqual({ok, null, <<"">>}, jsone_decode:decode(<<"null">>))
  32. end},
  33. %% Numbers: Integer
  34. {"positive integer",
  35. fun () ->
  36. ?assertEqual({ok, 1, <<"">>}, jsone_decode:decode(<<"1">>))
  37. end},
  38. {"zero",
  39. fun () ->
  40. ?assertEqual({ok, 0, <<"">>}, jsone_decode:decode(<<"0">>))
  41. end},
  42. {"negative integer",
  43. fun () ->
  44. ?assertEqual({ok, -1, <<"">>}, jsone_decode:decode(<<"-1">>))
  45. end},
  46. {"large integer (no limit on size)",
  47. fun () ->
  48. ?assertEqual({ok, 111111111111111111111111111111111111111111111111111111111111111111111111111111, <<"">>},
  49. jsone_decode:decode(<<"111111111111111111111111111111111111111111111111111111111111111111111111111111">>))
  50. end},
  51. {"integer with leading zero (interpreted as zero and remaining binary)",
  52. fun () ->
  53. ?assertEqual({ok, 0, <<"0">>}, jsone_decode:decode(<<"00">>)),
  54. ?assertEqual({ok, 0, <<"1">>}, jsone_decode:decode(<<"01">>)),
  55. ?assertEqual({ok, 0, <<"1">>}, jsone_decode:decode(<<"-01">>))
  56. end},
  57. {"integer can't begin with an explicit plus sign",
  58. fun () ->
  59. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"+1">>))
  60. end},
  61. %% Numbers: Floats
  62. {"float: decimal notation",
  63. fun () ->
  64. ?assertEqual({ok, 1.23, <<"">>}, jsone_decode:decode(<<"1.23">>))
  65. end},
  66. {"float: exponential notation",
  67. fun () ->
  68. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"12345e-3">>)), % lower case 'e'
  69. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"12345E-3">>)), % upper case 'E'
  70. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"12345.0e-3">>)),
  71. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"0.12345E2">>)),
  72. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"0.12345e+2">>)), % exponent part can begin with plus sign
  73. ?assertEqual({ok, 12.345, <<"">>}, jsone_decode:decode(<<"0.12345E+2">>)),
  74. ?assertEqual({ok, -12.345, <<"">>}, jsone_decode:decode(<<"-0.012345e3">>))
  75. end},
  76. {"float: invalid format",
  77. fun () ->
  78. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<".123">>)), % omitted integer part
  79. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.">>)), % omitted fraction part: EOS
  80. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.e+3">>)), % omitted fraction part: with exponent part
  81. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.1e">>)), % imcomplete fraction part
  82. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.1e-">>)), % imcomplete fraction part
  83. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.1ee-1">>)), % duplicated 'e'
  84. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.1e--1">>)), % duplicated sign
  85. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"1e999">>)), % exponent out of range
  86. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"0.1e999">>)), % exponent out of range
  87. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"100000000000000000000000000000000000e300">>)), % product out of range
  88. ?assertEqual({ok, 0.1, <<".2">>}, jsone_decode:decode(<<"0.1.2">>)) % duplicated '.': interpreted as individual tokens
  89. end},
  90. %% Strings
  91. {"simple string",
  92. fun () ->
  93. ?assertEqual({ok, <<"abc">>, <<"">>}, jsone_decode:decode(<<"\"abc\"">>))
  94. end},
  95. {"string: escaped characters",
  96. fun () ->
  97. Input = list_to_binary([$", [[$\\, C] || C <- [$", $/, $\\, $b, $f, $n, $r, $t]], $"]),
  98. Expected = <<"\"\/\\\b\f\n\r\t">>,
  99. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input))
  100. end},
  101. {"string: escaped Unicode characters",
  102. fun () ->
  103. %% japanese
  104. Input1 = <<"\"\\u3042\\u3044\\u3046\\u3048\\u304A\"">>,
  105. Expected1 = <<"あいうえお">>, % assumed that the encoding of this file is UTF-8
  106. ?assertEqual({ok, Expected1, <<"">>}, jsone_decode:decode(Input1)),
  107. %% ascii
  108. Input2 = <<"\"\\u0061\\u0062\\u0063\"">>,
  109. Expected2 = <<"abc">>,
  110. ?assertEqual({ok, Expected2, <<"">>}, jsone_decode:decode(Input2)),
  111. %% other multi-byte characters
  112. Input3 = <<"\"\\u06DD\\u06DE\\u10AE\\u10AF\"">>,
  113. Expected3 = <<"۝۞ႮႯ">>,
  114. ?assertEqual({ok, Expected3, <<"">>}, jsone_decode:decode(Input3)),
  115. %% mixture of ascii and japanese characters
  116. Input4 = <<"\"a\\u30421\\u3044bb\\u304622\\u3048ccc\\u304A333\"">>,
  117. Expected4 = <<"aあ1いbbう22えcccお333">>, % assumed that the encoding of this file is UTF-8
  118. ?assertEqual({ok, Expected4, <<"">>}, jsone_decode:decode(Input4))
  119. end},
  120. {"string: surrogate pairs",
  121. fun () ->
  122. Input = <<"\"\\ud848\\udc49\\ud848\\udc9a\\ud848\\udcfc\"">>,
  123. Expected = <<"𢁉𢂚𢃼">>,
  124. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input))
  125. end},
  126. {"string: control characters",
  127. fun () ->
  128. Ctrls = lists:seq(0, 16#1f),
  129. lists:foreach(
  130. fun (C) ->
  131. %% Control characters are unacceptable
  132. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<$", C, $">>))
  133. end,
  134. Ctrls),
  135. lists:foreach(
  136. fun (C) ->
  137. %% `allow_ctrl_chars' option allows strings which contain unescaped control characters
  138. ?assertEqual({ok, <<C>>, <<"">>}, jsone_decode:decode(<<$", C, $">>, [{allow_ctrl_chars, true}]))
  139. end,
  140. Ctrls)
  141. end},
  142. {"string: invalid escape characters",
  143. fun () ->
  144. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\z\"">>)), % '\z' is undefined
  145. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\uab\"">>)), % too few hex characters
  146. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\ud848\"">>)), % high(first) surrogate only
  147. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\udc49\"">>)), % low(second) surrogate only
  148. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\ud848\\u0061\"">>)), % missing low(second) surrogate
  149. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\udf0u\"">>)),
  150. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\ud848\\udf0u\"">>)),
  151. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(<<"\"\\u-3351\"">>))
  152. end},
  153. %% Arrays
  154. {"simple array",
  155. fun () ->
  156. Input = <<"[1,2,\"abc\",null]">>,
  157. Expected = [1, 2, <<"abc">>, null],
  158. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input))
  159. end},
  160. {"array: contains whitespaces",
  161. fun () ->
  162. Input = <<"[ 1,\t2, \n \"abc\",\r null]">>,
  163. Expected = [1, 2, <<"abc">>, null],
  164. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input))
  165. end},
  166. {"empty array",
  167. fun () ->
  168. ?assertEqual({ok, [], <<"">>}, jsone_decode:decode(<<"[]">>)),
  169. ?assertEqual({ok, [], <<"">>}, jsone_decode:decode(<<"[ \t\r\n]">>))
  170. end},
  171. {"array: trailing comma is disallowed",
  172. fun () ->
  173. Input = <<"[1, 2, \"abc\", null, ]">>,
  174. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  175. end},
  176. {"array: missing comma",
  177. fun () ->
  178. Input = <<"[1 2, \"abc\", null]">>, % a missing comma between '1' and '2'
  179. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  180. end},
  181. {"array: missing closing bracket",
  182. fun () ->
  183. Input = <<"[1, 2, \"abc\", null">>,
  184. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  185. end},
  186. %% Objects
  187. {"simple object",
  188. fun () ->
  189. Input = <<"{\"1\":2,\"key\":\"value\"}">>,
  190. Expected = ?OBJ2(<<"1">>, 2, <<"key">>, <<"value">>),
  191. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input)), % `map' is the default format
  192. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input, [{object_format, ?MAP_OBJECT_TYPE}]))
  193. end},
  194. {"simple object: tuple or proplist",
  195. fun () ->
  196. Input = <<"{\"1\":2,\"key\":\"value\"}">>,
  197. Expected = {[{<<"1">>, 2},{<<"key">>, <<"value">>}]},
  198. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input, [{object_format, tuple}])),
  199. ?assertEqual({ok, element(1, Expected), <<"">>}, jsone_decode:decode(Input, [{object_format, proplist}]))
  200. end},
  201. {"object: contains whitespaces",
  202. fun () ->
  203. Input = <<"{ \"1\" :\t 2,\n\r\"key\" : \n \"value\"}">>,
  204. Expected = ?OBJ2(<<"1">>, 2, <<"key">>, <<"value">>),
  205. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input))
  206. end},
  207. {"empty object",
  208. fun () ->
  209. ?assertEqual({ok, ?OBJ0, <<"">>}, jsone_decode:decode(<<"{}">>)),
  210. ?assertEqual({ok, ?OBJ0, <<"">>}, jsone_decode:decode(<<"{ \t\r\n}">>)),
  211. ?assertEqual({ok, {[]}, <<"">>}, jsone_decode:decode(<<"{}">>, [{object_format, tuple}])),
  212. ?assertEqual({ok, [{}], <<"">>}, jsone_decode:decode(<<"{}">>, [{object_format, proplist}]))
  213. end},
  214. {"empty object: map",
  215. fun () ->
  216. ?assertEqual({ok, ?OBJ0, <<"">>}, jsone_decode:decode(<<"{}">>, [{object_format, ?MAP_OBJECT_TYPE}]))
  217. end},
  218. {"duplicated members: map",
  219. fun () ->
  220. Input = <<"{\"1\":\"first\",\"1\":\"second\"}">>,
  221. Expected = ?OBJ2_DUP_KEY(<<"1">>, <<"first">>, <<"1">>, <<"second">>),
  222. ?assertEqual({ok, Expected, <<"">>}, jsone_decode:decode(Input, [{object_format, ?MAP_OBJECT_TYPE}]))
  223. end},
  224. {"object: trailing comma is disallowed",
  225. fun () ->
  226. Input = <<"{\"1\":2, \"key\":\"value\", }">>,
  227. io:format("~p\n", [catch jsone_decode:decode(Input)]),
  228. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input, [{object_format, tuple}]))
  229. end},
  230. {"object: missing comma",
  231. fun () ->
  232. Input = <<"{\"1\":2 \"key\":\"value\"}">>,
  233. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  234. end},
  235. {"object: missing field key",
  236. fun () ->
  237. Input = <<"{:2, \"key\":\"value\"}">>,
  238. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  239. end},
  240. {"object: non string key",
  241. fun () ->
  242. Input = <<"{1:2, \"key\":\"value\"}">>,
  243. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  244. end},
  245. {"object: missing field value",
  246. fun () ->
  247. Input = <<"{\"1\", \"key\":\"value\"}">>,
  248. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  249. end},
  250. {"object: missing closing brace",
  251. fun () ->
  252. Input = <<"{\"1\":2 \"key\":\"value\"">>,
  253. ?assertMatch({error, {badarg, _}}, jsone_decode:decode(Input))
  254. end},
  255. {"atom keys",
  256. fun () ->
  257. KeyOpt = fun(Keys) -> [{keys, Keys}, {object_format, proplist}]
  258. end,
  259. Input = <<"{\"foo\":\"ok\"}">>,
  260. ?assertEqual([{<<"foo">>, <<"ok">>}], jsone:decode(Input, KeyOpt(binary))),
  261. ?assertEqual([{foo, <<"ok">>}], jsone:decode(Input, KeyOpt(atom))),
  262. ?assertEqual([{foo, <<"ok">>}], jsone:decode(Input, KeyOpt(existing_atom))),
  263. ?assertError(badarg, jsone:decode(<<"{\"@#$%^!\":\"ok\"}">>, KeyOpt(existing_atom))),
  264. ?assertEqual([{foo, <<"ok">>}], jsone:decode(Input, KeyOpt(attempt_atom))),
  265. ?assertEqual([{<<"@#$%^!">>, <<"ok">>}], jsone:decode(<<"{\"@#$%^!\":\"ok\"}">>, KeyOpt(attempt_atom))),
  266. Value = integer_to_binary(1234),
  267. % do not make atom in test code
  268. [{Atom, <<"ok">>}] = jsone:decode(<<"{\"", Value/binary, "\":\"ok\"}">>, KeyOpt(atom)),
  269. ?assertEqual(Value, atom_to_binary(Atom, latin1))
  270. end},
  271. %% Others
  272. {"compound data",
  273. fun () ->
  274. Input = <<" [true, {\"1\" : 2, \"array\":[[[[1]]], {\"ab\":\"cd\"}, false]}, null] ">>,
  275. Expected = [true, ?OBJ2(<<"1">>, 2, <<"array">>, [[[[1]]], ?OBJ1(<<"ab">>, <<"cd">>), false]), null],
  276. ?assertEqual({ok, Expected, <<" ">>}, jsone_decode:decode(Input))
  277. end},
  278. {"undefined_as_null option",
  279. fun() ->
  280. ?assertEqual({ok, undefined, <<>>}, jsone_decode:decode(<<"null">>,[undefined_as_null])), % OK
  281. ?assertEqual({ok, null, <<>>}, jsone_decode:decode(<<"null">>,[])) % OK
  282. end}
  283. ].