merge in experimental explicitspeed branch

This commit is contained in:
alisdair sullivan 2012-03-26 19:41:45 -07:00
commit d5d7273e1e
7 changed files with 388 additions and 278 deletions

View file

@ -1 +0,0 @@
"﷐"

View file

@ -1,3 +0,0 @@
{name, "noncharacter"}.
{jsx, {error, badjson}}.
{json, "noncharacter.json"}.

View file

@ -1 +0,0 @@
"﷐"

View file

@ -1,4 +0,0 @@
{name, "noncharacter replaced"}.
{jsx, [{string,<<16#fffd/utf8>>},end_json]}.
{json, "noncharacter_replaced.json"}.
{jsx_flags, [loose_unicode]}.

View file

@ -145,11 +145,11 @@ value(<<$f, Rest/binary>>, Handler, Stack, Opts) ->
value(<<$n, Rest/binary>>, Handler, Stack, Opts) ->
nu(Rest, Handler, Stack, Opts);
value(<<?negative, Rest/binary>>, Handler, Stack, Opts) ->
negative(Rest, Handler, [?new_seq($-)|Stack], Opts);
negative(Rest, Handler, [[$-]|Stack], Opts);
value(<<?zero, Rest/binary>>, Handler, Stack, Opts) ->
zero(Rest, Handler, [?new_seq($0)|Stack], Opts);
zero(Rest, Handler, [[$0]|Stack], Opts);
value(<<S, Rest/binary>>, Handler, Stack, Opts) when ?is_nonzero(S) ->
integer(Rest, Handler, [?new_seq(S)|Stack], Opts);
integer(Rest, Handler, [[S]|Stack], Opts);
value(<<?start_object, Rest/binary>>, {Handler, State}, Stack, Opts) ->
object(Rest, {Handler, Handler:handle_event(start_object, State)}, [key|Stack], Opts);
value(<<?start_array, Rest/binary>>, {Handler, State}, Stack, Opts) ->
@ -193,11 +193,11 @@ array(<<$f, Rest/binary>>, Handler, Stack, Opts) ->
array(<<$n, Rest/binary>>, Handler, Stack, Opts) ->
nu(Rest, Handler, Stack, Opts);
array(<<?negative, Rest/binary>>, Handler, Stack, Opts) ->
negative(Rest, Handler, [?new_seq($-)|Stack], Opts);
negative(Rest, Handler, [[$-]|Stack], Opts);
array(<<?zero, Rest/binary>>, Handler, Stack, Opts) ->
zero(Rest, Handler, [?new_seq($0)|Stack], Opts);
zero(Rest, Handler, [[$0]|Stack], Opts);
array(<<S, Rest/binary>>, Handler, Stack, Opts) when ?is_nonzero(S) ->
integer(Rest, Handler, [?new_seq(S)|Stack], Opts);
integer(Rest, Handler, [[S]|Stack], Opts);
array(<<?start_object, Rest/binary>>, {Handler, State}, Stack, Opts) ->
object(Rest, {Handler, Handler:handle_event(start_object, State)}, [key|Stack], Opts);
array(<<?start_array, Rest/binary>>, {Handler, State}, Stack, Opts) ->
@ -260,6 +260,12 @@ partial_utf(<<X, Y, Z>>)
partial_utf(_) -> false.
%% explicitly whitelist ascii set for better efficiency (seriously, it's worth
%% almost a 20% increase)
string(<<32, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 32)|Stack], Opts);
string(<<33, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 33)|Stack], Opts);
string(<<?doublequote, Rest/binary>>, {Handler, State}, S, Opts) ->
case S of
[Acc, key|Stack] ->
@ -269,6 +275,14 @@ string(<<?doublequote, Rest/binary>>, {Handler, State}, S, Opts) ->
[Acc|Stack] ->
maybe_done(Rest, {Handler, Handler:handle_event({string, ?end_seq(Acc)}, State)}, Stack, Opts)
end;
string(<<35, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 35)|Stack], Opts);
string(<<36, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 36)|Stack], Opts);
string(<<37, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 37)|Stack], Opts);
string(<<38, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 38)|Stack], Opts);
string(<<?singlequote, Rest/binary>>, {Handler, State}, S, Opts = #opts{single_quotes=true}) ->
case S of
[Acc, single_quote, key|Stack] ->
@ -278,37 +292,183 @@ string(<<?singlequote, Rest/binary>>, {Handler, State}, S, Opts = #opts{single_q
[Acc|Stack] ->
string(Rest, {Handler, State}, [?acc_seq(Acc, ?singlequote)|Stack], Opts)
end;
string(<<40, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 40)|Stack], Opts);
string(<<41, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 41)|Stack], Opts);
string(<<42, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 42)|Stack], Opts);
string(<<43, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 43)|Stack], Opts);
string(<<44, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 44)|Stack], Opts);
string(<<45, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 45)|Stack], Opts);
string(<<46, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 46)|Stack], Opts);
string(<<47, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 47)|Stack], Opts);
string(<<48, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 48)|Stack], Opts);
string(<<49, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 49)|Stack], Opts);
string(<<50, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 50)|Stack], Opts);
string(<<51, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 51)|Stack], Opts);
string(<<52, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 52)|Stack], Opts);
string(<<53, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 53)|Stack], Opts);
string(<<54, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 54)|Stack], Opts);
string(<<55, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 55)|Stack], Opts);
string(<<56, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 56)|Stack], Opts);
string(<<57, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 57)|Stack], Opts);
string(<<58, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 58)|Stack], Opts);
string(<<59, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 59)|Stack], Opts);
string(<<60, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 60)|Stack], Opts);
string(<<61, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 61)|Stack], Opts);
string(<<62, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 62)|Stack], Opts);
string(<<63, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 63)|Stack], Opts);
string(<<64, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 64)|Stack], Opts);
string(<<65, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 65)|Stack], Opts);
string(<<66, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 66)|Stack], Opts);
string(<<67, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 67)|Stack], Opts);
string(<<68, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 68)|Stack], Opts);
string(<<69, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 69)|Stack], Opts);
string(<<70, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 70)|Stack], Opts);
string(<<71, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 71)|Stack], Opts);
string(<<72, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 72)|Stack], Opts);
string(<<73, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 73)|Stack], Opts);
string(<<74, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 74)|Stack], Opts);
string(<<75, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 75)|Stack], Opts);
string(<<76, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 76)|Stack], Opts);
string(<<77, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 77)|Stack], Opts);
string(<<78, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 78)|Stack], Opts);
string(<<79, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 79)|Stack], Opts);
string(<<80, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 80)|Stack], Opts);
string(<<81, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 81)|Stack], Opts);
string(<<82, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 82)|Stack], Opts);
string(<<83, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 83)|Stack], Opts);
string(<<84, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 84)|Stack], Opts);
string(<<85, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 85)|Stack], Opts);
string(<<86, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 86)|Stack], Opts);
string(<<87, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 87)|Stack], Opts);
string(<<88, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 88)|Stack], Opts);
string(<<89, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 89)|Stack], Opts);
string(<<90, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 90)|Stack], Opts);
string(<<91, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 91)|Stack], Opts);
string(<<?rsolidus/utf8, Rest/binary>>, Handler, Stack, Opts) ->
escape(Rest, Handler, Stack, Opts);
%% things get dumb here. erlang doesn't properly restrict unicode non-characters
%% so you can't trust the codepoints it returns always
%% the range 32..16#fdcf is safe, so allow that
string(<<S/utf8, Rest/binary>>, Handler, [Acc|Stack], Opts)
when ?is_noncontrol(S), S < 16#fdd0 ->
string(Rest, Handler, [?acc_seq(Acc, S)|Stack], Opts);
%% the range 16#fdf0..16#fffd is also safe
string(<<S/utf8, Rest/binary>>, Handler, [Acc|Stack], Opts)
when S > 16#fdef, S < 16#fffe ->
string(Rest, Handler, [?acc_seq(Acc, S)|Stack], Opts);
%% yes, i think it's insane too
string(<<S/utf8, Rest/binary>>, Handler, [Acc|Stack], Opts)
when S > 16#ffff andalso
S =/= 16#1fffe andalso S =/= 16#1ffff andalso
S =/= 16#2fffe andalso S =/= 16#2ffff andalso
S =/= 16#3fffe andalso S =/= 16#3ffff andalso
S =/= 16#4fffe andalso S =/= 16#4ffff andalso
S =/= 16#5fffe andalso S =/= 16#5ffff andalso
S =/= 16#6fffe andalso S =/= 16#6ffff andalso
S =/= 16#7fffe andalso S =/= 16#7ffff andalso
S =/= 16#8fffe andalso S =/= 16#8ffff andalso
S =/= 16#9fffe andalso S =/= 16#9ffff andalso
S =/= 16#afffe andalso S =/= 16#affff andalso
S =/= 16#bfffe andalso S =/= 16#bffff andalso
S =/= 16#cfffe andalso S =/= 16#cffff andalso
S =/= 16#dfffe andalso S =/= 16#dffff andalso
S =/= 16#efffe andalso S =/= 16#effff andalso
S =/= 16#ffffe andalso S =/= 16#fffff andalso
S =/= 16#10fffe andalso S =/= 16#10ffff ->
string(<<93, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 93)|Stack], Opts);
string(<<94, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 94)|Stack], Opts);
string(<<95, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 95)|Stack], Opts);
string(<<96, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 96)|Stack], Opts);
string(<<97, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 97)|Stack], Opts);
string(<<98, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 98)|Stack], Opts);
string(<<99, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 99)|Stack], Opts);
string(<<100, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 100)|Stack], Opts);
string(<<101, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 101)|Stack], Opts);
string(<<102, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 102)|Stack], Opts);
string(<<103, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 103)|Stack], Opts);
string(<<104, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 104)|Stack], Opts);
string(<<105, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 105)|Stack], Opts);
string(<<106, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 106)|Stack], Opts);
string(<<107, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 107)|Stack], Opts);
string(<<108, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 108)|Stack], Opts);
string(<<109, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 109)|Stack], Opts);
string(<<110, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 110)|Stack], Opts);
string(<<111, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 111)|Stack], Opts);
string(<<112, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 112)|Stack], Opts);
string(<<113, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 113)|Stack], Opts);
string(<<114, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 114)|Stack], Opts);
string(<<115, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 115)|Stack], Opts);
string(<<116, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 116)|Stack], Opts);
string(<<117, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 117)|Stack], Opts);
string(<<118, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 118)|Stack], Opts);
string(<<119, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 119)|Stack], Opts);
string(<<120, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 120)|Stack], Opts);
string(<<121, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 121)|Stack], Opts);
string(<<122, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 122)|Stack], Opts);
string(<<123, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 123)|Stack], Opts);
string(<<124, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 124)|Stack], Opts);
string(<<125, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 125)|Stack], Opts);
string(<<126, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 126)|Stack], Opts);
string(<<127, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 127)|Stack], Opts);
string(<<S/utf8, Rest/binary>>, Handler, [Acc|Stack], Opts) when ?is_noncontrol(S) ->
string(Rest, Handler, [?acc_seq(Acc, S)|Stack], Opts);
string(Bin, Handler, Stack, Opts) ->
case partial_utf(Bin) of
@ -321,35 +481,13 @@ string(Bin, Handler, Stack, Opts) ->
end.
%% we don't need to guard against partial utf here, because it's already taken
%% care of in string. theoretically, the last clause of noncharacter/4 is
%% unreachable
%% non-characters erlang doesn't recognize as non-characters
noncharacter(<<S/utf8, Rest/binary>>, Handler, [Acc|Stack], Opts)
when ?is_noncontrol(S) ->
string(Rest, Handler, [?acc_seq(Acc, 16#fffd)|Stack], Opts);
%% u+fffe and u+ffff
noncharacter(<<239, 191, X, Rest/binary>>, Handler, [Acc|Stack], Opts)
when X == 190; X == 191 ->
string(Rest, Handler, [?acc_seq(Acc, 16#fffd)|Stack], Opts);
%% care of in string
%% surrogates
noncharacter(<<237, X, _, Rest/binary>>, Handler, [Acc|Stack], Opts) when X >= 160 ->
string(Rest, Handler, [?acc_seq(Acc, 16#fffd)|Stack], Opts);
noncharacter(<<X, Y, 191, Z, Rest/binary>>, Handler, [Acc|Stack], Opts)
when (
(X == 240 andalso Y == 159) orelse
(X == 240 andalso Y == 175) orelse
(X == 240 andalso Y == 191) orelse
(
(X == 241 orelse X == 242 orelse X == 243) andalso
(Y == 143 orelse Y == 159 orelse Y == 175 orelse Y == 191)
) orelse
(X == 244 andalso Y == 143)
) andalso (Z == 190 orelse Z == 191) ->
string(Rest, Handler, [?acc_seq(Acc, 16#fffd)|Stack], Opts);
%% bad utf8
noncharacter(<<_, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
string(Rest, Handler, [?acc_seq(Acc, 16#fffd)|Stack], Opts);
noncharacter(Bin, Handler, Stack, Opts) ->
?error([Bin, Handler, Stack, Opts]).
string(Rest, Handler, [?acc_seq(Acc, 16#fffd)|Stack], Opts).
escape(<<$b, Rest/binary>>, Handler, [Acc|Stack], Opts) ->
@ -1027,7 +1165,6 @@ comments_test_() ->
)}
].
escape_forward_slash_test_() ->
[
{"escape forward slash test", ?_assertEqual(
@ -1036,27 +1173,6 @@ escape_forward_slash_test_() ->
)}
].
noncharacters_test_() ->
[
{"noncharacters - badjson",
?_assertEqual(check_bad(noncharacters()), [])
},
{"noncharacters - replaced",
?_assertEqual(check_replaced(noncharacters()), [])
}
].
extended_noncharacters_test_() ->
[
{"extended noncharacters - badjson",
?_assertEqual(check_bad(extended_noncharacters()), [])
},
{"extended noncharacters - replaced",
?_assertEqual(check_replaced(extended_noncharacters()), [])
}
].
surrogates_test_() ->
[
{"surrogates - badjson",
@ -1074,16 +1190,6 @@ control_test_() ->
}
].
reserved_test_() ->
[
{"reserved noncharacters - badjson",
?_assertEqual(check_bad(reserved_space()), [])
},
{"reserved noncharacters - replaced",
?_assertEqual(check_replaced(reserved_space()), [])
}
].
good_characters_test_() ->
[
{"acceptable codepoints",
@ -1173,32 +1279,13 @@ decode(JSON, Opts) ->
end.
noncharacters() -> lists:seq(16#fffe, 16#ffff).
extended_noncharacters() ->
[16#1fffe, 16#1ffff, 16#2fffe, 16#2ffff]
++ [16#3fffe, 16#3ffff, 16#4fffe, 16#4ffff]
++ [16#5fffe, 16#5ffff, 16#6fffe, 16#6ffff]
++ [16#7fffe, 16#7ffff, 16#8fffe, 16#8ffff]
++ [16#9fffe, 16#9ffff, 16#afffe, 16#affff]
++ [16#bfffe, 16#bffff, 16#cfffe, 16#cffff]
++ [16#dfffe, 16#dffff, 16#efffe, 16#effff]
++ [16#ffffe, 16#fffff, 16#10fffe, 16#10ffff].
surrogates() -> lists:seq(16#d800, 16#dfff).
control_characters() -> lists:seq(1, 31).
reserved_space() -> lists:seq(16#fdd0, 16#fdef).
good() -> [32, 33] ++ lists:seq(16#23, 16#5b) ++ lists:seq(16#5d, 16#d7ff) ++ lists:seq(16#e000, 16#ffff).
good() -> [32, 33]
++ lists:seq(16#23, 16#5b)
++ lists:seq(16#5d, 16#d7ff)
++ lists:seq(16#e000, 16#fdcf)
++ lists:seq(16#fdf0, 16#fffd).
good_extended() -> lists:seq(16#100000, 16#10fffd).
good_extended() -> lists:seq(16#100000, 16#10ffff).
%% erlang refuses to encode certain codepoints, so fake them all
to_fake_utf(N, utf8) when N < 16#0080 -> <<34/utf8, N:8, 34/utf8>>;

View file

@ -53,7 +53,7 @@ start(Term, {Handler, State}, Opts) ->
value(String, {Handler, State}, Opts) when is_binary(String) ->
Handler:handle_event({string, clean_string(String, <<>>, Opts)}, State);
Handler:handle_event({string, clean_string(String, Opts)}, State);
value(Float, {Handler, State}, _Opts) when is_float(Float) ->
Handler:handle_event({float, Float}, State);
value(Int, {Handler, State}, _Opts) when is_integer(Int) ->
@ -83,7 +83,7 @@ object([{Key, Value}|Rest], {Handler, State}, Opts) ->
Handler,
value(
Value,
{Handler, Handler:handle_event({key, clean_string(fix_key(Key), <<>>, Opts)}, State)},
{Handler, Handler:handle_event({key, clean_string(fix_key(Key), Opts)}, State)},
Opts
)
},
@ -103,96 +103,31 @@ fix_key(Key) when is_atom(Key) -> fix_key(atom_to_binary(Key, utf8));
fix_key(Key) when is_binary(Key) -> Key.
clean_string(<<$\", Rest/binary>>, Acc, Opts=#opts{json_escape=true}) ->
clean_string(Rest, <<Acc/binary, $\\, $\">>, Opts);
clean_string(<<$\\, Rest/binary>>, Acc, Opts=#opts{json_escape=true}) ->
clean_string(Rest, <<Acc/binary, $\\, $\\>>, Opts);
clean_string(<<$\b, Rest/binary>>, Acc, Opts=#opts{json_escape=true}) ->
clean_string(Rest, <<Acc/binary, $\\, $b>>, Opts);
clean_string(<<$\f, Rest/binary>>, Acc, Opts=#opts{json_escape=true}) ->
clean_string(Rest, <<Acc/binary, $\\, $f>>, Opts);
clean_string(<<$\n, Rest/binary>>, Acc, Opts=#opts{json_escape=true}) ->
clean_string(Rest, <<Acc/binary, $\\, $n>>, Opts);
clean_string(<<$\r, Rest/binary>>, Acc, Opts=#opts{json_escape=true}) ->
clean_string(Rest, <<Acc/binary, $\\, $r>>, Opts);
clean_string(<<$\t, Rest/binary>>, Acc, Opts=#opts{json_escape=true}) ->
clean_string(Rest, <<Acc/binary, $\\, $t>>, Opts);
clean_string(<<$/, Rest/binary>>, Acc, Opts=#opts{json_escape=true, escape_forward_slash=true}) ->
clean_string(Rest, <<Acc/binary, $\\, $/>>, Opts);
clean_string(<<16#2028/utf8, Rest/binary>>, Acc, Opts=#opts{json_escape=true, no_jsonp_escapes=true}) ->
clean_string(Rest, <<Acc/binary, 16#2028/utf8>>, Opts);
clean_string(<<16#2029/utf8, Rest/binary>>, Acc, Opts=#opts{json_escape=true, no_jsonp_escapes=true}) ->
clean_string(Rest, <<Acc/binary, 16#2029/utf8>>, Opts);
clean_string(<<16#2028/utf8, Rest/binary>>, Acc, Opts=#opts{json_escape=true}) ->
clean_string(Rest, <<Acc/binary, (json_escape_sequence(16#2028))/binary>>, Opts);
clean_string(<<16#2029/utf8, Rest/binary>>, Acc, Opts=#opts{json_escape=true}) ->
clean_string(Rest, <<Acc/binary, (json_escape_sequence(16#2029))/binary>>, Opts);
clean_string(<<C/utf8, Rest/binary>>, Acc, Opts=#opts{json_escape=true}) when C < 32 ->
clean_string(Rest, <<Acc/binary, (json_escape_sequence(C))/binary>>, Opts);
clean_string(<<C/utf8, Rest/binary>>, Acc, Opts) when C < 16#fdd0 ->
clean_string(Rest, <<Acc/binary, C/utf8>>, Opts);
clean_string(<<C/utf8, Rest/binary>>, Acc, Opts) when C > 16#fdef, C < 16#fffe ->
clean_string(Rest, <<Acc/binary, C/utf8>>, Opts);
clean_string(<<C/utf8, Rest/binary>>, Acc, Opts)
when C > 16#ffff andalso
C =/= 16#1fffe andalso C =/= 16#1ffff andalso
C =/= 16#2fffe andalso C =/= 16#2ffff andalso
C =/= 16#3fffe andalso C =/= 16#3ffff andalso
C =/= 16#4fffe andalso C =/= 16#4ffff andalso
C =/= 16#5fffe andalso C =/= 16#5ffff andalso
C =/= 16#6fffe andalso C =/= 16#6ffff andalso
C =/= 16#7fffe andalso C =/= 16#7ffff andalso
C =/= 16#8fffe andalso C =/= 16#8ffff andalso
C =/= 16#9fffe andalso C =/= 16#9ffff andalso
C =/= 16#afffe andalso C =/= 16#affff andalso
C =/= 16#bfffe andalso C =/= 16#bffff andalso
C =/= 16#cfffe andalso C =/= 16#cffff andalso
C =/= 16#dfffe andalso C =/= 16#dffff andalso
C =/= 16#efffe andalso C =/= 16#effff andalso
C =/= 16#ffffe andalso C =/= 16#fffff andalso
C =/= 16#10fffe andalso C =/= 16#10ffff ->
clean_string(Rest, <<Acc/binary, C/utf8>>, Opts);
clean_string(Bin, Opts) ->
case Opts#opts.json_escape of
true -> jsx_utils:json_escape(Bin, Opts);
false ->
case is_clean(Bin) of
true -> Bin;
false -> clean_string(Bin, [], Opts)
end
end.
is_clean(<<>>) -> true;
is_clean(<<_/utf8, Rest/binary>>) -> is_clean(Rest);
is_clean(_) -> false.
clean_string(Bin, _Acc, Opts=#opts{loose_unicode=false}) -> ?error([Bin, Opts]);
clean_string(<<>>, Acc, _Opts) -> unicode:characters_to_binary(lists:reverse(Acc));
clean_string(<<X/utf8, Rest/binary>>, Acc, Opts) -> clean_string(Rest, [X] ++ Acc, Opts);
%% surrogates
clean_string(<<237, X, _, Rest/binary>>, Acc, Opts=#opts{loose_unicode=true}) when X >= 160 ->
clean_string(Rest, <<Acc/binary, 16#fffd/utf8>>, Opts);
%% private use noncharacters
clean_string(<<239, 183, X, Rest/binary>>, Acc, Opts=#opts{loose_unicode=true}) when X >= 143, X =< 175 ->
clean_string(Rest, <<Acc/binary, 16#fffd/utf8>>, Opts);
%% u+fffe and u+ffff
clean_string(<<239, 191, X, Rest/binary>>, Acc, Opts=#opts{loose_unicode=true}) when X == 190; X == 191 ->
clean_string(Rest, <<Acc/binary, 16#fffd/utf8>>, Opts);
%% the u+Xfffe and u+Xffff noncharacters
clean_string(<<X, Y, 191, Z, Rest/binary>>, Acc, Opts=#opts{loose_unicode=true}) when (
(X == 240 andalso Y == 159) orelse
(X == 240 andalso Y == 175) orelse
(X == 240 andalso Y == 191) orelse
(
(X == 241 orelse X == 242 orelse X == 243) andalso
(Y == 143 orelse Y == 159 orelse Y == 175 orelse Y == 191)
) orelse
(X == 244 andalso Y == 143)
) andalso (Z == 190 orelse Z == 191) ->
clean_string(Rest, <<Acc/binary, 16#fffd/utf8>>, Opts);
clean_string(<<_, Rest/binary>>, Acc, Opts=#opts{loose_unicode=true}) ->
clean_string(Rest, <<Acc/binary, 16#fffd/utf8>>, Opts);
clean_string(<<>>, Acc, _) -> Acc;
clean_string(Bin, _Acc, Opts) -> erlang:error(badarg, [Bin, Opts]).
clean_string(<<237, X, _, Rest/binary>>, Acc, Opts) when X >= 160 -> clean_string(Rest, [16#fffd] ++ Acc, Opts);
%% bad codepoints
clean_string(<<_, Rest/binary>>, Acc, Opts) -> clean_string(Rest, [16#fffd] ++ Acc, Opts).
%% convert a codepoint to it's \uXXXX equiv.
json_escape_sequence(X) ->
<<A:4, B:4, C:4, D:4>> = <<X:16>>,
unicode:characters_to_binary([$\\, $u, (to_hex(A)), (to_hex(B)), (to_hex(C)), (to_hex(D))]).
to_hex(10) -> $a;
to_hex(11) -> $b;
to_hex(12) -> $c;
to_hex(13) -> $d;
to_hex(14) -> $e;
to_hex(15) -> $f;
to_hex(X) -> X + 48. %% ascii "1" is [49], "2" is [50], etc...
-ifdef(TEST).
-include_lib("eunit/include/eunit.hrl").
@ -275,26 +210,6 @@ encode_test_() ->
}
].
noncharacters_test_() ->
[
{"noncharacters - badjson",
?_assertEqual(check_bad(noncharacters()), [])
},
{"noncharacters - replaced",
?_assertEqual(check_replaced(noncharacters()), [])
}
].
extended_noncharacters_test_() ->
[
{"extended noncharacters - badjson",
?_assertEqual(check_bad(extended_noncharacters()), [])
},
{"extended noncharacters - replaced",
?_assertEqual(check_replaced(extended_noncharacters()), [])
}
].
surrogates_test_() ->
[
{"surrogates - badjson",
@ -305,16 +220,6 @@ surrogates_test_() ->
}
].
reserved_test_() ->
[
{"reserved noncharacters - badjson",
?_assertEqual(check_bad(reserved_space()), [])
},
{"reserved noncharacters - replaced",
?_assertEqual(check_replaced(reserved_space()), [])
}
].
good_characters_test_() ->
[
{"acceptable codepoints",
@ -385,26 +290,11 @@ check([H|T], Opts, Acc) ->
check(T, Opts, [{H, R}] ++ Acc).
noncharacters() -> lists:seq(16#fffe, 16#ffff).
extended_noncharacters() ->
[16#1fffe, 16#1ffff, 16#2fffe, 16#2ffff]
++ [16#3fffe, 16#3ffff, 16#4fffe, 16#4ffff]
++ [16#5fffe, 16#5ffff, 16#6fffe, 16#6ffff]
++ [16#7fffe, 16#7ffff, 16#8fffe, 16#8ffff]
++ [16#9fffe, 16#9ffff, 16#afffe, 16#affff]
++ [16#bfffe, 16#bffff, 16#cfffe, 16#cffff]
++ [16#dfffe, 16#dffff, 16#efffe, 16#effff]
++ [16#ffffe, 16#fffff, 16#10fffe, 16#10ffff].
surrogates() -> lists:seq(16#d800, 16#dfff).
reserved_space() -> lists:seq(16#fdd0, 16#fdef).
good() -> lists:seq(1, 16#d7ff) ++ lists:seq(16#e000, 16#ffff).
good() -> lists:seq(1, 16#d7ff) ++ lists:seq(16#e000, 16#fdcf) ++ lists:seq(16#fdf0, 16#fffd).
good_extended() -> lists:seq(16#100000, 16#10fffd).
good_extended() -> lists:seq(16#100000, 16#10ffff).
%% erlang refuses to encode certain codepoints, so fake them all
to_fake_utf(N, utf8) when N < 16#0080 -> <<N:8>>;

View file

@ -90,22 +90,66 @@ extract_parser_opts([K|Rest], Acc) ->
json_escape(String, Opts) when is_binary(String) ->
json_escape(String, Opts, 0, size(String)).
-define(control_character(X),
<<H:L/binary, X, T/binary>> ->
json_escape(
<<H/binary, (unicode:characters_to_binary(json_escape_sequence(X)))/binary, T/binary>>,
Opts,
L + 6,
Len + 5
)
).
json_escape(Str, Opts, L, Len) when L < Len ->
case Str of
<<H:L/binary, $\", T/binary>> -> %"
json_escape(<<H/binary, $\\, $", T/binary>>, Opts, L + 2, Len + 1);
<<H:L/binary, $\\, T/binary>> ->
json_escape(<<H/binary, $\\, $\\, T/binary>>, Opts, L + 2, Len + 1);
<<H:L/binary, $\b, T/binary>> ->
json_escape(<<H/binary, $\\, $b, T/binary>>, Opts, L + 2, Len + 1);
<<H:L/binary, $\f, T/binary>> ->
json_escape(<<H/binary, $\\, $f, T/binary>>, Opts, L + 2, Len + 1);
<<H:L/binary, $\n, T/binary>> ->
json_escape(<<H/binary, $\\, $n, T/binary>>, Opts, L + 2, Len + 1);
<<H:L/binary, $\r, T/binary>> ->
json_escape(<<H/binary, $\\, $r, T/binary>>, Opts, L + 2, Len + 1);
<<H:L/binary, $\t, T/binary>> ->
json_escape(<<H/binary, $\\, $t, T/binary>>, Opts, L + 2, Len + 1);
?control_character(0);
?control_character(1);
?control_character(2);
?control_character(3);
?control_character(4);
?control_character(5);
?control_character(6);
?control_character(7);
<<H:L/binary, $\b, T/binary>> -> json_escape(<<H/binary, $\\, $b, T/binary>>, Opts, L + 2, Len + 1);
<<H:L/binary, $\t, T/binary>> -> json_escape(<<H/binary, $\\, $t, T/binary>>, Opts, L + 2, Len + 1);
<<H:L/binary, $\n, T/binary>> -> json_escape(<<H/binary, $\\, $n, T/binary>>, Opts, L + 2, Len + 1);
?control_character(11);
<<H:L/binary, $\f, T/binary>> -> json_escape(<<H/binary, $\\, $f, T/binary>>, Opts, L + 2, Len + 1);
<<H:L/binary, $\r, T/binary>> -> json_escape(<<H/binary, $\\, $r, T/binary>>, Opts, L + 2, Len + 1);
?control_character(14);
?control_character(15);
?control_character(16);
?control_character(17);
?control_character(18);
?control_character(19);
?control_character(20);
?control_character(21);
?control_character(22);
?control_character(23);
?control_character(24);
?control_character(25);
?control_character(26);
?control_character(27);
?control_character(28);
?control_character(29);
?control_character(30);
?control_character(31);
<<_:L/binary, 32, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 33, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<H:L/binary, $\", T/binary>> -> json_escape(<<H/binary, $\\, $", T/binary>>, Opts, L + 2, Len + 1);
<<_:L/binary, 35, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 36, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 37, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 38, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 39, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 40, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 41, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 42, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 43, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 44, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 45, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 46, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<H:L/binary, $/, T/binary>> ->
case Opts#opts.escape_forward_slash of
true ->
@ -113,6 +157,86 @@ json_escape(Str, Opts, L, Len) when L < Len ->
false ->
json_escape(<<H/binary, $/, T/binary>>, Opts, L + 1, Len)
end;
<<_:L/binary, 48, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 49, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 50, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 51, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 52, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 53, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 54, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 55, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 56, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 57, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 58, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 59, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 60, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 61, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 62, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 63, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 64, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 65, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 66, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 67, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 68, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 69, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 70, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 71, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 72, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 73, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 74, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 75, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 76, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 77, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 78, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 79, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 80, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 81, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 82, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 83, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 84, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 85, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 86, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 87, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 88, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 89, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 90, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 91, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<H:L/binary, $\\, T/binary>> -> json_escape(<<H/binary, $\\, $\\, T/binary>>, Opts, L + 2, Len + 1);
<<_:L/binary, 93, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 94, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 95, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 96, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 97, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 98, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 99, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 100, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 101, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 102, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 103, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 104, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 105, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 106, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 107, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 108, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 109, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 110, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 111, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 112, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 113, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 114, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 115, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 116, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 117, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 118, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 119, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 120, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 121, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 122, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 123, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 124, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 125, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 126, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, 127, _/binary>> -> json_escape(Str, Opts, L + 1, Len);
<<H:L/binary, 16#2028/utf8, T/binary>> ->
case Opts#opts.no_jsonp_escapes of
true ->
@ -129,9 +253,6 @@ json_escape(Str, Opts, L, Len) when L < Len ->
B = unicode:characters_to_binary(json_escape_sequence(16#2029)),
json_escape(<<H/binary, B/binary, T/binary>>, Opts, L + size(B), Len + size(B) - size(<<16#2029/utf8>>))
end;
<<H:L/binary, X/utf8, T/binary>> when X < 32 ->
B = unicode:characters_to_binary(json_escape_sequence(X)),
json_escape(<<H/binary, B/binary, T/binary>>, Opts, L + size(B), Len + size(B) - size(<<X/utf8>>));
<<_:L/binary, X/utf8, _/binary>> when X < 16#0080 ->
json_escape(Str, Opts, L + 1, Len);
<<_:L/binary, X/utf8, _/binary>> when X < 16#0800 ->
@ -140,8 +261,16 @@ json_escape(Str, Opts, L, Len) when L < Len ->
json_escape(Str, Opts, L + 3, Len);
<<_:L/binary, _/utf8, _/binary>> ->
json_escape(Str, Opts, L + 4, Len);
<<H:L/binary, X, T/binary>> ->
erlang:error(badarg, [[<<H:L/binary, X, T/binary>>, Opts]])
<<H:L/binary, 237, X, _, T/binary>> when X >= 160 ->
case Opts#opts.loose_unicode of
true -> json_escape(<<H/binary, 16#fffd/utf8, T/binary>>, Opts, L + 3, Len);
false -> erlang:error(badarg, [Str, Opts])
end;
<<H:L/binary, _, T/binary>> ->
case Opts#opts.loose_unicode of
true -> json_escape(<<H/binary, 16#fffd/utf8, T/binary>>, Opts, L + 3, Len + 2);
false -> erlang:error(badarg, [Str, Opts])
end
end;
json_escape(Str, _, L, Len) when L =:= Len ->
Str.
@ -201,7 +330,20 @@ binary_escape_test_() ->
)
},
{"bad utf8",
?_assertError(badarg, json_escape(<<32, 64, 128, 256>>, #opts{}))
?_assertError(badarg, json_escape(<<32, 64, 128, 255>>, #opts{}))
},
{"bad utf8 ok",
?_assertEqual(
json_escape(<<32, 64, 128, 255>>, #opts{loose_unicode=true}),
<<32, 64, 16#fffd/utf8, 16#fffd/utf8>>
)
},
{"bad surrogate", ?_assertError(badarg, json_escape(<<237, 160, 127>>, #opts{}))},
{"bad surrogate ok",
?_assertEqual(
json_escape(<<237, 160, 127>>, #opts{loose_unicode=true}),
<<16#fffd/utf8>>
)
},
{"all sizes of codepoints",
?_assertEqual(