umgeher/jsx
0
Fork 0
Code Issues Pull requests Activity

converted jsx_test to etap, fixed resulting bugs with autodetection

Browse source
This commit is contained in:
alisdair sullivan 2010-06-01 23:33:56 -07:00
parent b82d4c5337
commit ef952ceee8
6 changed files with 78 additions and 43 deletions
Download patch file Download diff file Expand all files Collapse all files

40
src/jsx.erl
View file

@ -99,9 +99,6 @@ detect_encoding(<<16#ff, 16#fe, Rest/binary>>, Stack, Callbacks, Opts) ->
detect_encoding(<<0, 0, 16#fe, 16#ff, Rest/binary>>, Stack, Callbacks, Opts) ->
jsx_utf32:start(Rest, Stack, Callbacks, Opts);
%% utf8 null order detection
detect_encoding(<<X, Y, _Rest/binary>> = JSON, Stack, Callbacks, Opts) when X =/= 0, Y =/= 0 ->
jsx_utf8:start(JSON, Stack, Callbacks, Opts);
%% utf32-little null order detection
detect_encoding(<<X, 0, 0, 0, _Rest/binary>> = JSON, Stack, Callbacks, Opts) when X =/= 0 ->
@ -119,17 +116,38 @@ detect_encoding(<<X, 0, Y, 0, _Rest/binary>> = JSON, Stack, Callbacks, Opts) whe
detect_encoding(<<0, 0, 0, X, _Rest/binary>> = JSON, Stack, Callbacks, Opts) when X =/= 0 ->
jsx_utf32:start(JSON, Stack, Callbacks, Opts);
%% trying to parse a json string of a single character encoded in utf8 will fail
%% unless special cased
detect_encoding(<<X>> = JSON, Stack, Callbacks, Opts) when X =/= 0 ->
try jsx_utf8:start(JSON, Stack, Callbacks, Opts)
catch error:function_clause ->
{incomplete,
%% utf8 null order detection
detect_encoding(<<X, Y, _Rest/binary>> = JSON, Stack, Callbacks, Opts) when X =/= 0, Y =/= 0 ->
jsx_utf8:start(JSON, Stack, Callbacks, Opts);
%% a problem, to autodetect naked single digits' encoding, there is not enough data
%% to conclusively determine the encoding correctly. below is an attempt to solve
%% the problem
detect_encoding(<<X>>, Stack, Callbacks, Opts) when X =/= 0 ->
{
try {Result, _} = jsx_utf8:start(<<X>>, [], Callbacks, Opts), Result
catch error:function_clause -> incomplete end,
fun(Stream) ->
detect_encoding(<<X, Stream/binary>>, Stack, Callbacks, Opts)
end
}
end;
};
detect_encoding(<<0, X>>, Stack, Callbacks, Opts) when X =/= 0 ->
{
try {Result, _} = jsx_utf16:start(<<0, X>>, [], Callbacks, Opts), Result
catch error:function_clause -> incomplete end,
fun(Stream) ->
detect_encoding(<<0, X, Stream/binary>>, Stack, Callbacks, Opts)
end
};
detect_encoding(<<X, 0>>, Stack, Callbacks, Opts) when X =/= 0 ->
{
try {Result, _} = jsx_utf16le:start(<<X, 0>>, [], Callbacks, Opts), Result
catch error:function_clause -> incomplete end,
fun(Stream) ->
detect_encoding(<<X, 0, Stream/binary>>, Stack, Callbacks, Opts)
end
};
%% not enough input, request more
detect_encoding(Bin, Stack, Callbacks, Opts) ->

2
src/jsx_utf16.erl
View file

@ -371,7 +371,7 @@ initial_decimal(<<S/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) when
initial_decimal(<<?zero/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
decimal(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
initial_decimal(Bin, Stack, Callbacks, Opts, Acc) when byte_size(Bin) < 2 ->
{incomplete, fun(Stream) -> initial_decimal(Stream, Stack, Callbacks, Opts, Acc) end}.
{incomplete, fun(Stream) -> initial_decimal(<<Bin/binary, Stream/binary>>, Stack, Callbacks, Opts, Acc) end}.
decimal(<<S/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->

2
src/jsx_utf16le.erl
View file

@ -371,7 +371,7 @@ initial_decimal(<<S/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) when
initial_decimal(<<?zero/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
decimal(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
initial_decimal(Bin, Stack, Callbacks, Opts, Acc) when byte_size(Bin) < 2 ->
{incomplete, fun(Stream) -> initial_decimal(Stream, Stack, Callbacks, Opts, Acc) end}.
{incomplete, fun(Stream) -> initial_decimal(<<Bin/binary, Stream/binary>>, Stack, Callbacks, Opts, Acc) end}.
decimal(<<S/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->

2
src/jsx_utf32.erl
View file

@ -358,7 +358,7 @@ initial_decimal(<<S/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) when
initial_decimal(<<?zero/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
decimal(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
initial_decimal(Bin, Stack, Callbacks, Opts, Acc) when byte_size(Bin) < 4 ->
{incomplete, fun(Stream) -> initial_decimal(Stream, Stack, Callbacks, Opts, Acc) end}.
{incomplete, fun(Stream) -> initial_decimal(<<Bin/binary, Stream/binary>>, Stack, Callbacks, Opts, Acc) end}.
decimal(<<S/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->

2
src/jsx_utf32le.erl
View file

@ -358,7 +358,7 @@ initial_decimal(<<S/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) when
initial_decimal(<<?zero/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
decimal(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
initial_decimal(Bin, Stack, Callbacks, Opts, Acc) when byte_size(Bin) < 4 ->
{incomplete, fun(Stream) -> initial_decimal(Stream, Stack, Callbacks, Opts, Acc) end}.
{incomplete, fun(Stream) -> initial_decimal(<<Bin/binary, Stream/binary>>, Stack, Callbacks, Opts, Acc) end}.
decimal(<<S/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->

65
test/jsx_test.erl
View file

@ -24,14 +24,18 @@
-module(jsx_test).
-author("alisdairsullivan@yahoo.ca").
-export([test/1, test_event/2, incremental_decode/2, decode/2]).
-include_lib("eunit/include/eunit.hrl").
-export([test/0, test/1, test_event/2]).
test() ->
F = decoder([]),
incremental_decode(F, unicode:characters_to_binary(<<"0">>, utf8, utf16)).
test(Dir) ->
Tests = gen_tests(Dir),
eunit:test(Tests, [verbose]).
ValidJSONTests = load_tests(Dir),
etap:plan(length(ValidJSONTests) * 10),
run_tests(ValidJSONTests),
etap:end_tests().
decoder(Flags) ->
@ -43,33 +47,40 @@ test_event(Event, Acc) ->
[Event] ++ Acc.
gen_tests(Dir) ->
load_tests(Dir) ->
TestSpecs = filelib:wildcard("*.test", Dir),
gen_tests(TestSpecs, Dir, []).
load_tests(TestSpecs, Dir, []).
gen_tests([], _, Acc) ->
load_tests([], _Dir, Acc) ->
lists:reverse(Acc);
gen_tests([Test|Rest], Dir, Acc) ->
gen_tests(Rest, Dir, test_body(Test, Dir) ++ Acc).
test_body(TestSpec, Dir) ->
load_tests([Test|Rest], Dir, Acc) ->
try
TestName = filename:basename(TestSpec, ".test"),
TestName = filename:basename(Test, ".test"),
{ok, JSON} = file:read_file(Dir ++ "/" ++ TestName ++ ".json"),
case file:consult(Dir ++ "/" ++ TestSpec) of
case file:consult(Dir ++ "/" ++ Test) of
{ok, [Events]} ->
Decoder = jsx:decoder(),
[{TestName ++ "_incremental", ?_assertEqual(incremental_decode(Decoder, JSON), Events)}] ++
[{TestName, ?_assertEqual(decode(Decoder, JSON), Events)}]
load_tests(Rest, Dir, [{TestName, JSON, Events, []}] ++ Acc)
; {ok, [Events, Flags]} ->
Decoder = jsx:decoder(Flags),
[{TestName ++ "_incremental", ?_assertEqual(incremental_decode(Decoder, JSON), Events)}] ++
[{TestName, ?_assertEqual(decode(Decoder, JSON), Events)}]
load_tests(Rest, Dir, [{TestName, JSON, Events, Flags}] ++ Acc)
end
catch _:_ -> []
end.
catch _:_ -> load_tests(Rest, Dir, Acc) end.
run_tests([]) ->
ok;
run_tests([{TestName, JSON, Events, Flags}|Rest]) ->
F = decoder(Flags),
etap:is(decode(F, JSON), Events, TestName ++ ": utf8"),
etap:is(incremental_decode(F, JSON), Events, TestName ++ ": incremental utf8"),
etap:is(decode(F, to_utf16(JSON)), Events, TestName ++ ": utf16"),
etap:is(incremental_decode(F, to_utf16(JSON)), Events, TestName ++ ": incremental utf16"),
etap:is(decode(F, to_utf16le(JSON)), Events, TestName ++ ": utf16le"),
etap:is(incremental_decode(F, to_utf16le(JSON)), Events, TestName ++ ": incremental utf16le"),
etap:is(decode(F, to_utf32(JSON)), Events, TestName ++ ": utf32"),
etap:is(incremental_decode(F, to_utf32(JSON)), Events, TestName ++ ": incremental utf32"),
etap:is(decode(F, to_utf32le(JSON)), Events, TestName ++ ": utf32le"),
etap:is(incremental_decode(F, to_utf32le(JSON)), Events, TestName ++ ": incremental utf32le"),
run_tests(Rest).
incremental_decode(F, <<>>) ->
case F(<<>>) of
@ -88,3 +99,9 @@ decode(F, JSON) ->
Result
end.
to_utf16(Bin) -> unicode:characters_to_binary(Bin, utf8, utf16).
to_utf16le(Bin) -> unicode:characters_to_binary(Bin, utf8, {utf16,little}).
to_utf32(Bin) -> unicode:characters_to_binary(Bin, utf8, utf32).
to_utf32le(Bin) -> unicode:characters_to_binary(Bin, utf8, {utf32,little}).