added option for explicit termination requirement when parsing streams

This commit is contained in:
alisdair sullivan 2010-05-24 15:47:29 -07:00
parent 3b3e33c6e3
commit 6515f9981d
11 changed files with 59 additions and 2232 deletions

View file

@ -13,17 +13,7 @@ decoder(Callbacks, OptsList) when is_list(OptsList) ->
decoder(Callbacks, Opts) ->
case Opts#opts.encoding of
utf8 ->
fun(Stream) -> jsx_utf8:start(Stream, [], init_callbacks(Callbacks), Opts) end
; utf16b ->
fun(Stream) -> jsx_utf16b:start(Stream, [], init_callbacks(Callbacks), Opts) end
; utf16l ->
fun(Stream) -> jsx_utf16l:start(Stream, [], init_callbacks(Callbacks), Opts) end
; utf32b ->
fun(Stream) -> jsx_utf32b:start(Stream, [], init_callbacks(Callbacks), Opts) end
; utf32l ->
fun(Stream) -> jsx_utf32l:start(Stream, [], init_callbacks(Callbacks), Opts) end
; auto ->
fun(Stream) -> detect_encoding(Stream, Callbacks, Opts) end
fun(Stream) -> jsx_decoder:start(Stream, [], init_callbacks(Callbacks), Opts) end
end.
@ -40,10 +30,13 @@ parse_opts([{escaped_unicode, Value}|Rest], Opts) ->
parse_opts(Rest, Opts#opts{escaped_unicode = Value});
parse_opts([{naked_values, Value}|Rest], Opts) ->
true = lists:member(Value, [true, false]),
parse_opts(Rest, Opts#opts{naked_values = Value});
parse_opts(Rest, Opts#opts{naked_values = Value, explicit_termination = true});
parse_opts([{encoding, Value}|Rest], Opts) ->
true = lists:member(Value, [auto, utf8, utf16b, utf16l, utf32b, utf32l]),
parse_opts(Rest, Opts#opts{encoding = Value}).
true = lists:member(Value, [utf8]),
parse_opts(Rest, Opts#opts{encoding = Value});
parse_opts([{explicit_termination, Value}|Rest], Opts) ->
true = lists:member(Value, [true, false]),
parse_opts(Rest, Opts#opts{explicit_termination = Value}).
init_callbacks(none) ->
{none, []};
@ -52,15 +45,3 @@ init_callbacks({M, S}) when is_atom(M) ->
init_callbacks({F, S}) when is_function(F) ->
{F, S}.
detect_encoding(<<A:8, B:8, C:8, D:8, _/binary>> = Stream, Callbacks, Opts) ->
Encoding = case [A, B, C, D] of
[0, 0, 0, _] -> utf32b
; [0, _, 0, _] -> utf16b
; [_, 0, 0, 0] -> utf32l
; [_, 0, _, 0] -> utf16l
; _ -> utf8
end,
(decoder(Callbacks, Opts#opts{encoding = Encoding}))(Stream);
detect_encoding(Else, Callbacks, Opts) ->
fun(Stream) -> detect_encoding(<<Else/binary, Stream/binary>>, Callbacks, Opts) end.

View file

@ -2,7 +2,8 @@
comments = false,
escaped_unicode = ascii,
naked_values = false,
encoding = auto
encoding = utf8,
explicit_termination = false
}).
%% whitespace

View file

@ -38,7 +38,10 @@ start(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
start(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> start(Stream, Stack, Callbacks, Opts) end.
maybe_done(eof, [], Callbacks, Opts) when Opts#opts.explicit_termination == true ->
{callback(eof, Callbacks), <<>>};
maybe_done(Rest, [], Callbacks, Opts) when Opts#opts.explicit_termination == false ->
{callback(eof, Callbacks), Rest};
maybe_done(<<?end_object/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
maybe_done(<<?end_array/utf8, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
@ -51,8 +54,6 @@ maybe_done(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opt
maybe_comment(Rest, fun(Resume) -> maybe_done(Resume, Stack, Callbacks, Opts) end);
maybe_done(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, Callbacks, Opts);
maybe_done(<<>>, [], Callbacks, _Opts) ->
callback(eof, Callbacks);
maybe_done(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> maybe_done(Stream, Stack, Callbacks, Opts) end.
@ -236,10 +237,14 @@ zero(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> zero(Resume, Stack, Callbacks, Opts, Acc) end);
zero(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
zero(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> zero(Stream, Stack, Callbacks, Opts, Acc) end.
zero(<<>>, [], Callbacks, Opts, Acc) when Opts#opts.explicit_termination == false ->
{callback(eof, callback({number, lists:reverse(Acc)}, Callbacks)), <<>>};
zero(<<>>, Stack, Callbacks, Opts, Acc) when Opts#opts.explicit_termination == false ->
fun(Stream) -> zero(Stream, Stack, Callbacks, Opts, Acc) end;
zero(<<>>, Stack, Callbacks, Opts, Acc) when Opts#opts.explicit_termination == true ->
fun(eof) -> {callback(eof, callback({number, lists:reverse(Acc)}, Callbacks)), <<>>}
;(Stream) -> zero(Stream, Stack, Callbacks, Opts, Acc)
end.
integer(<<?end_object/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
@ -264,10 +269,14 @@ integer(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespac
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> integer(Resume, Stack, Callbacks, Opts, Acc) end);
integer(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
integer(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> integer(Stream, Stack, Callbacks, Opts, Acc) end.
integer(<<>>, [], Callbacks, Opts, Acc) when Opts#opts.explicit_termination == false ->
{callback(eof, callback({number, lists:reverse(Acc)}, Callbacks)), <<>>};
integer(<<>>, Stack, Callbacks, Opts, Acc) when Opts#opts.explicit_termination == false ->
fun(Stream) -> integer(Stream, Stack, Callbacks, Opts, Acc) end;
integer(<<>>, Stack, Callbacks, Opts, Acc) when Opts#opts.explicit_termination == true ->
fun(eof) -> {callback(eof, callback({number, lists:reverse(Acc)}, Callbacks)), <<>>}
;(Stream) -> integer(Stream, Stack, Callbacks, Opts, Acc)
end.
fraction(<<?end_object/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
@ -290,10 +299,14 @@ fraction(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespa
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> fraction(Resume, Stack, Callbacks, Opts, Acc) end);
fraction(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
fraction(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> fraction(Stream, Stack, Callbacks, Opts, Acc) end.
fraction(<<>>, [], Callbacks, Opts, Acc) when Opts#opts.explicit_termination == false ->
{callback(eof, callback({number, lists:reverse(Acc)}, Callbacks)), <<>>};
fraction(<<>>, Stack, Callbacks, Opts, Acc) when Opts#opts.explicit_termination == false ->
fun(Stream) -> fraction(Stream, Stack, Callbacks, Opts, Acc) end;
fraction(<<>>, Stack, Callbacks, Opts, Acc) when Opts#opts.explicit_termination == true ->
fun(eof) -> {callback(eof, callback({number, lists:reverse(Acc)}, Callbacks)), <<>>}
;(Stream) -> fraction(Stream, Stack, Callbacks, Opts, Acc)
end.
e(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?positive; S =:= ?negative ->
@ -326,10 +339,14 @@ exp(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S)
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?rsolidus/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> exp(Resume, Stack, Callbacks, Opts, Acc) end);
exp(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
exp(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> exp(Stream, Stack, Callbacks, Opts, Acc) end.
exp(<<>>, [], Callbacks, Opts, Acc) when Opts#opts.explicit_termination == false ->
{callback(eof, callback({number, lists:reverse(Acc)}, Callbacks)), <<>>};
exp(<<>>, Stack, Callbacks, Opts, Acc) when Opts#opts.explicit_termination == false ->
fun(Stream) -> exp(Stream, Stack, Callbacks, Opts, Acc) end;
exp(<<>>, Stack, Callbacks, Opts, Acc) when Opts#opts.explicit_termination == true ->
fun(eof) -> {callback(eof, callback({number, lists:reverse(Acc)}, Callbacks)), <<>>}
;(Stream) -> exp(Stream, Stack, Callbacks, Opts, Acc)
end.
tr(<<"r"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->

View file

@ -1,436 +0,0 @@
-module(jsx_utf16b).
-export([start/4]).
-include("jsx_common.hrl").
%% this code is mostly autogenerated and mostly ugly. apologies. for more insight on
%% Callbacks or Opts, see the comments accompanying decoder/2 (in jsx.erl). Stack
%% is a stack of flags used to track depth and to keep track of whether we are
%% returning from a value or a key inside objects. all pops, peeks and pushes are
%% inlined. the code that handles naked values and comments is not optimized by the
%% compiler for efficient matching, but you shouldn't be using naked values or comments
%% anyways, they are horrible and contrary to the spec.
start(<<?start_object/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
start(<<?start_array/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
start(<<?quote/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
string(Rest, Stack, Callbacks, Opts, []);
start(<<$t/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
tr(Rest, Stack, Callbacks, Opts);
start(<<$f/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
fa(Rest, Stack, Callbacks, Opts);
start(<<$n/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
nu(Rest, Stack, Callbacks, Opts);
start(<<?negative/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
negative(Rest, Stack, Callbacks, Opts, "-");
start(<<?zero/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
zero(Rest, Stack, Callbacks, Opts, "0");
start(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S), Opts#opts.naked_values == true ->
integer(Rest, Stack, Callbacks, Opts, [S]);
start(<<?solidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> start(Resume, Stack, Callbacks, Opts) end);
start(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
start(Rest, Stack, Callbacks, Opts);
start(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> start(Stream, Stack, Callbacks, Opts) end.
maybe_done(<<?end_object/utf16-big, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
maybe_done(<<?end_array/utf16-big, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_array, Callbacks), Opts);
maybe_done(<<?comma/utf16-big, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
key(Rest, [key|Stack], Callbacks, Opts);
maybe_done(<<?comma/utf16-big, Rest/binary>>, [array|_] = Stack, Callbacks, Opts) ->
value(Rest, Stack, Callbacks, Opts);
maybe_done(<<?solidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> maybe_done(Resume, Stack, Callbacks, Opts) end);
maybe_done(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, Callbacks, Opts);
maybe_done(<<>>, [], Callbacks, _Opts) ->
callback(eof, Callbacks);
maybe_done(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> maybe_done(Stream, Stack, Callbacks, Opts) end.
object(<<?end_object/utf16-big, Rest/binary>>, [key|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
object(<<?quote/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
object(<<?solidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> object(Resume, Stack, Callbacks, Opts) end);
object(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
object(Rest, Stack, Callbacks, Opts);
object(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> object(Stream, Stack, Callbacks, Opts) end.
array(<<?quote/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
array(<<?start_object/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
array(<<?start_array/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
array(<<?end_array/utf16-big, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_array, Callbacks), Opts);
array(<<$t/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
tr(Rest, Stack, Callbacks, Opts);
array(<<$f/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
fa(Rest, Stack, Callbacks, Opts);
array(<<$n/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
nu(Rest, Stack, Callbacks, Opts);
array(<<?negative/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
negative(Rest, Stack, Callbacks, Opts, "-");
array(<<?zero/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
zero(Rest, Stack, Callbacks, Opts, "0");
array(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S]);
array(<<?solidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> array(Resume, Stack, Callbacks, Opts) end);
array(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
array(Rest, Stack, Callbacks, Opts);
array(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> array(Stream, Stack, Callbacks, Opts) end.
value(<<?quote/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
value(<<?start_object/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
value(<<?start_array/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
value(<<$t/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
tr(Rest, Stack, Callbacks, Opts);
value(<<$f/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
fa(Rest, Stack, Callbacks, Opts);
value(<<$n/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
nu(Rest, Stack, Callbacks, Opts);
value(<<?negative/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
negative(Rest, Stack, Callbacks, Opts, "-");
value(<<?zero/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
zero(Rest, Stack, Callbacks, Opts, "0");
value(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S]);
value(<<?solidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> value(Resume, Stack, Callbacks, Opts) end);
value(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
value(Rest, Stack, Callbacks, Opts);
value(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> value(Stream, Stack, Callbacks, Opts) end.
colon(<<?colon/utf16-big, Rest/binary>>, [key|Stack], Callbacks, Opts) ->
value(Rest, [object|Stack], Callbacks, Opts);
colon(<<?solidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> colon(Resume, Stack, Callbacks, Opts) end);
colon(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
colon(Rest, Stack, Callbacks, Opts);
colon(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> colon(Stream, Stack, Callbacks, Opts) end.
key(<<?quote/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
key(<<?solidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> key(Resume, Stack, Callbacks, Opts) end);
key(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
key(Rest, Stack, Callbacks, Opts);
key(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> key(Stream, Stack, Callbacks, Opts) end.
%% string has an additional parameter, an accumulator (Acc) used to hold the intermediate
%% representation of the string being parsed. using a list of integers representing
%% unicode codepoints is faster than constructing binaries, many of which will be
%% converted back to lists by the user anyways.
string(<<?quote/utf16-big, Rest/binary>>, [key|_] = Stack, Callbacks, Opts, Acc) ->
colon(Rest, Stack, callback({key, lists:reverse(Acc)}, Callbacks), Opts);
string(<<?quote/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback({string, lists:reverse(Acc)}, Callbacks), Opts);
string(<<?rsolidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escape(Rest, Stack, Callbacks, Opts, Acc);
string(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_noncontrol(S) ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
string(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> string(Stream, Stack, Callbacks, Opts, Acc) end.
%% only thing to note here is the additional accumulator passed to escaped_unicode used
%% to hold the codepoint sequence. unescessary, but nicer than using the string
%% accumulator.
escape(<<"b"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\b" ++ Acc);
escape(<<"f"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\f" ++ Acc);
escape(<<"n"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\n" ++ Acc);
escape(<<"r"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\r" ++ Acc);
escape(<<"t"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\t" ++ Acc);
escape(<<"u"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escaped_unicode(Rest, Stack, Callbacks, Opts, Acc, []);
escape(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc)
when S =:= ?quote; S =:= ?solidus; S =:= ?rsolidus ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
escape(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> escape(Stream, Stack, Callbacks, Opts, Acc) end.
%% this code is ugly and unfortunate, but so is json's handling of escaped unicode
%% codepoint sequences. if the ascii option is present, the sequence is converted
%% to a codepoint and inserted into the string if it represents an ascii value. if
%% the codepoint option is present the sequence is converted and inserted as long
%% as it represents a valid 16 bit integer value (this is where json's spec gets
%% insane). any other option and the sequence is converted back to an erlang string
%% and appended to the string in place.
escaped_unicode(<<D/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, String, [C, B, A]) ->
X = erlang:list_to_integer([A, B, C, D], 16),
case Opts#opts.escaped_unicode of
ascii when X < 16#0080 ->
string(Rest, Stack, Callbacks, Opts, [X] ++ String)
; codepoint ->
string(Rest, Stack, Callbacks, Opts, [X] ++ String)
; _ ->
string(Rest, Stack, Callbacks, Opts, [?rsolidus, $u, A, B, C, D] ++ String)
end;
escaped_unicode(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, String, Acc) when ?is_hex(S) ->
escaped_unicode(Rest, Stack, Callbacks, Opts, String, [S] ++ Acc);
escaped_unicode(<<>>, Stack, Callbacks, Opts, String, Acc) ->
fun(Stream) -> escaped_unicode(Stream, Stack, Callbacks, Opts, String, Acc) end.
%% like strings, numbers are collected in an intermediate accumulator before
%% being emitted to the callback handler. no processing of numbers is done in
%% process, it's left for the user, though there are convenience functions to
%% convert them into erlang floats/integers in jsx_utils.erl.
%% TODO: actually write that jsx_utils.erl module mentioned above...
negative(<<"0"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
zero(Rest, Stack, Callbacks, Opts, "0" ++ Acc);
negative(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
negative(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> negative(Stream, Stack, Callbacks, Opts, Acc) end.
zero(<<?end_object/utf16-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
zero(<<?end_array/utf16-big, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
zero(<<?comma/utf16-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?comma/utf16-big, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?decimalpoint/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?decimalpoint] ++ Acc);
zero(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?solidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> zero(Resume, Stack, Callbacks, Opts, Acc) end);
zero(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
zero(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> zero(Stream, Stack, Callbacks, Opts, Acc) end.
integer(<<?end_object/utf16-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
integer(<<?end_array/utf16-big, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
integer(<<?comma/utf16-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?comma/utf16-big, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?decimalpoint/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?decimalpoint] ++ Acc);
integer(<<?zero/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
integer(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
integer(<<"e"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
integer(<<"E"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
integer(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
integer(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?solidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> integer(Resume, Stack, Callbacks, Opts, Acc) end);
integer(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
integer(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> integer(Stream, Stack, Callbacks, Opts, Acc) end.
fraction(<<?end_object/utf16-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
fraction(<<?end_array/utf16-big, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
fraction(<<?comma/utf16-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?comma/utf16-big, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?zero/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
fraction(<<"e"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
fraction(<<"E"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
fraction(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
fraction(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
fraction(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?solidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> fraction(Resume, Stack, Callbacks, Opts, Acc) end);
fraction(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
fraction(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> fraction(Stream, Stack, Callbacks, Opts, Acc) end.
e(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?positive; S =:= ?negative ->
ex(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
e(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?zero; ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
e(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> e(Stream, Stack, Callbacks, Opts, Acc) end.
ex(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?zero; ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
ex(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> ex(Stream, Stack, Callbacks, Opts, Acc) end.
exp(<<?end_object/utf16-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
exp(<<?end_array/utf16-big, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
exp(<<?comma/utf16-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?comma/utf16-big, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?zero/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
exp(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
exp(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
exp(<<S/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?rsolidus/utf16-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> exp(Resume, Stack, Callbacks, Opts, Acc) end);
exp(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
exp(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> exp(Stream, Stack, Callbacks, Opts, Acc) end.
tr(<<"r"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
tru(Rest, Stack, Callbacks, Opts);
tr(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> tr(Stream, Stack, Callbacks, Opts) end.
tru(<<"u"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
true(Rest, Stack, Callbacks, Opts);
tru(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> tru(Stream, Stack, Callbacks, Opts) end.
true(<<"e"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, true}, Callbacks), Opts);
true(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> true(Stream, Stack, Callbacks, Opts) end.
fa(<<"a"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
fal(Rest, Stack, Callbacks, Opts);
fa(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fa(Stream, Stack, Callbacks, Opts) end.
fal(<<"l"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
fals(Rest, Stack, Callbacks, Opts);
fal(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fal(Stream, Stack, Callbacks, Opts) end.
fals(<<"s"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
false(Rest, Stack, Callbacks, Opts);
fals(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fals(Stream, Stack, Callbacks, Opts) end.
false(<<"e"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, false}, Callbacks), Opts);
false(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> false(Stream, Stack, Callbacks, Opts) end.
nu(<<"u"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
nul(Rest, Stack, Callbacks, Opts);
nu(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> nu(Stream, Stack, Callbacks, Opts) end.
nul(<<"l"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
null(Rest, Stack, Callbacks, Opts);
nul(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> nul(Stream, Stack, Callbacks, Opts) end.
null(<<"l"/utf16-big, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, null}, Callbacks), Opts);
null(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> null(Stream, Stack, Callbacks, Opts) end.
%% comments are c style, /* blah blah */ and are STRONGLY discouraged. any unicode
%% character is valid in a comment, except, obviously the */ sequence which ends
%% the comment. they're implemented as a closure called when the comment ends that
%% returns execution to the point where the comment began. comments are not
%% recorded in any way, simply parsed.
maybe_comment(<<?star/utf16-big, Rest/binary>>, Resume) ->
comment(Rest, Resume);
maybe_comment(<<>>, Resume) ->
fun(Stream) -> maybe_comment(Stream, Resume) end.
comment(<<?star/utf16-big, Rest/binary>>, Resume) ->
maybe_comment_done(Rest, Resume);
comment(<<_/utf16-big, Rest/binary>>, Resume) ->
comment(Rest, Resume);
comment(<<>>, Resume) ->
fun(Stream) -> comment(Stream, Resume) end.
maybe_comment_done(<<?solidus/utf16-big, Rest/binary>>, Resume) ->
Resume(Rest);
maybe_comment_done(<<>>, Resume) ->
fun(Stream) -> maybe_comment_done(Stream, Resume) end.
%% helper function for dispatching of parser events
callback(eof, {none, Callbacks}) ->
lists:reverse(Callbacks);
callback(Event, {none, Callbacks}) ->
{none, [Event] ++ Callbacks};
callback(Event, {Mod, State}) when is_atom(Mod) ->
{Mod, Mod:jsx_event(Event, State)};
callback(Event, {F, State}) when is_function(F) ->
{F, F(Event, State)}.

View file

@ -1,436 +0,0 @@
-module(jsx_utf16l).
-export([start/4]).
-include("jsx_common.hrl").
%% this code is mostly autogenerated and mostly ugly. apologies. for more insight on
%% Callbacks or Opts, see the comments accompanying decoder/2 (in jsx.erl). Stack
%% is a stack of flags used to track depth and to keep track of whether we are
%% returning from a value or a key inside objects. all pops, peeks and pushes are
%% inlined. the code that handles naked values and comments is not optimized by the
%% compiler for efficient matching, but you shouldn't be using naked values or comments
%% anyways, they are horrible and contrary to the spec.
start(<<?start_object/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
start(<<?start_array/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
start(<<?quote/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
string(Rest, Stack, Callbacks, Opts, []);
start(<<$t/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
tr(Rest, Stack, Callbacks, Opts);
start(<<$f/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
fa(Rest, Stack, Callbacks, Opts);
start(<<$n/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
nu(Rest, Stack, Callbacks, Opts);
start(<<?negative/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
negative(Rest, Stack, Callbacks, Opts, "-");
start(<<?zero/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
zero(Rest, Stack, Callbacks, Opts, "0");
start(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S), Opts#opts.naked_values == true ->
integer(Rest, Stack, Callbacks, Opts, [S]);
start(<<?solidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> start(Resume, Stack, Callbacks, Opts) end);
start(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
start(Rest, Stack, Callbacks, Opts);
start(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> start(Stream, Stack, Callbacks, Opts) end.
maybe_done(<<?end_object/utf16-little, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
maybe_done(<<?end_array/utf16-little, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_array, Callbacks), Opts);
maybe_done(<<?comma/utf16-little, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
key(Rest, [key|Stack], Callbacks, Opts);
maybe_done(<<?comma/utf16-little, Rest/binary>>, [array|_] = Stack, Callbacks, Opts) ->
value(Rest, Stack, Callbacks, Opts);
maybe_done(<<?solidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> maybe_done(Resume, Stack, Callbacks, Opts) end);
maybe_done(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, Callbacks, Opts);
maybe_done(<<>>, [], Callbacks, _Opts) ->
callback(eof, Callbacks);
maybe_done(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> maybe_done(Stream, Stack, Callbacks, Opts) end.
object(<<?end_object/utf16-little, Rest/binary>>, [key|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
object(<<?quote/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
object(<<?solidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> object(Resume, Stack, Callbacks, Opts) end);
object(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
object(Rest, Stack, Callbacks, Opts);
object(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> object(Stream, Stack, Callbacks, Opts) end.
array(<<?quote/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
array(<<?start_object/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
array(<<?start_array/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
array(<<?end_array/utf16-little, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_array, Callbacks), Opts);
array(<<$t/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
tr(Rest, Stack, Callbacks, Opts);
array(<<$f/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
fa(Rest, Stack, Callbacks, Opts);
array(<<$n/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
nu(Rest, Stack, Callbacks, Opts);
array(<<?negative/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
negative(Rest, Stack, Callbacks, Opts, "-");
array(<<?zero/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
zero(Rest, Stack, Callbacks, Opts, "0");
array(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S]);
array(<<?solidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> array(Resume, Stack, Callbacks, Opts) end);
array(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
array(Rest, Stack, Callbacks, Opts);
array(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> array(Stream, Stack, Callbacks, Opts) end.
value(<<?quote/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
value(<<?start_object/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
value(<<?start_array/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
value(<<$t/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
tr(Rest, Stack, Callbacks, Opts);
value(<<$f/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
fa(Rest, Stack, Callbacks, Opts);
value(<<$n/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
nu(Rest, Stack, Callbacks, Opts);
value(<<?negative/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
negative(Rest, Stack, Callbacks, Opts, "-");
value(<<?zero/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
zero(Rest, Stack, Callbacks, Opts, "0");
value(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S]);
value(<<?solidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> value(Resume, Stack, Callbacks, Opts) end);
value(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
value(Rest, Stack, Callbacks, Opts);
value(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> value(Stream, Stack, Callbacks, Opts) end.
colon(<<?colon/utf16-little, Rest/binary>>, [key|Stack], Callbacks, Opts) ->
value(Rest, [object|Stack], Callbacks, Opts);
colon(<<?solidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> colon(Resume, Stack, Callbacks, Opts) end);
colon(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
colon(Rest, Stack, Callbacks, Opts);
colon(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> colon(Stream, Stack, Callbacks, Opts) end.
key(<<?quote/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
key(<<?solidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> key(Resume, Stack, Callbacks, Opts) end);
key(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
key(Rest, Stack, Callbacks, Opts);
key(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> key(Stream, Stack, Callbacks, Opts) end.
%% string has an additional parameter, an accumulator (Acc) used to hold the intermediate
%% representation of the string being parsed. using a list of integers representing
%% unicode codepoints is faster than constructing binaries, many of which will be
%% converted back to lists by the user anyways.
string(<<?quote/utf16-little, Rest/binary>>, [key|_] = Stack, Callbacks, Opts, Acc) ->
colon(Rest, Stack, callback({key, lists:reverse(Acc)}, Callbacks), Opts);
string(<<?quote/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback({string, lists:reverse(Acc)}, Callbacks), Opts);
string(<<?rsolidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escape(Rest, Stack, Callbacks, Opts, Acc);
string(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_noncontrol(S) ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
string(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> string(Stream, Stack, Callbacks, Opts, Acc) end.
%% only thing to note here is the additional accumulator passed to escaped_unicode used
%% to hold the codepoint sequence. unescessary, but nicer than using the string
%% accumulator.
escape(<<"b"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\b" ++ Acc);
escape(<<"f"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\f" ++ Acc);
escape(<<"n"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\n" ++ Acc);
escape(<<"r"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\r" ++ Acc);
escape(<<"t"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\t" ++ Acc);
escape(<<"u"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escaped_unicode(Rest, Stack, Callbacks, Opts, Acc, []);
escape(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc)
when S =:= ?quote; S =:= ?solidus; S =:= ?rsolidus ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
escape(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> escape(Stream, Stack, Callbacks, Opts, Acc) end.
%% this code is ugly and unfortunate, but so is json's handling of escaped unicode
%% codepoint sequences. if the ascii option is present, the sequence is converted
%% to a codepoint and inserted into the string if it represents an ascii value. if
%% the codepoint option is present the sequence is converted and inserted as long
%% as it represents a valid 16 bit integer value (this is where json's spec gets
%% insane). any other option and the sequence is converted back to an erlang string
%% and appended to the string in place.
escaped_unicode(<<D/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, String, [C, B, A]) ->
X = erlang:list_to_integer([A, B, C, D], 16),
case Opts#opts.escaped_unicode of
ascii when X < 16#0080 ->
string(Rest, Stack, Callbacks, Opts, [X] ++ String)
; codepoint ->
string(Rest, Stack, Callbacks, Opts, [X] ++ String)
; _ ->
string(Rest, Stack, Callbacks, Opts, [?rsolidus, $u, A, B, C, D] ++ String)
end;
escaped_unicode(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, String, Acc) when ?is_hex(S) ->
escaped_unicode(Rest, Stack, Callbacks, Opts, String, [S] ++ Acc);
escaped_unicode(<<>>, Stack, Callbacks, Opts, String, Acc) ->
fun(Stream) -> escaped_unicode(Stream, Stack, Callbacks, Opts, String, Acc) end.
%% like strings, numbers are collected in an intermediate accumulator before
%% being emitted to the callback handler. no processing of numbers is done in
%% process, it's left for the user, though there are convenience functions to
%% convert them into erlang floats/integers in jsx_utils.erl.
%% TODO: actually write that jsx_utils.erl module mentioned above...
negative(<<"0"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
zero(Rest, Stack, Callbacks, Opts, "0" ++ Acc);
negative(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
negative(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> negative(Stream, Stack, Callbacks, Opts, Acc) end.
zero(<<?end_object/utf16-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
zero(<<?end_array/utf16-little, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
zero(<<?comma/utf16-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?comma/utf16-little, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?decimalpoint/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?decimalpoint] ++ Acc);
zero(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?solidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> zero(Resume, Stack, Callbacks, Opts, Acc) end);
zero(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
zero(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> zero(Stream, Stack, Callbacks, Opts, Acc) end.
integer(<<?end_object/utf16-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
integer(<<?end_array/utf16-little, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
integer(<<?comma/utf16-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?comma/utf16-little, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?decimalpoint/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?decimalpoint] ++ Acc);
integer(<<?zero/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
integer(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
integer(<<"e"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
integer(<<"E"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
integer(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
integer(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?solidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> integer(Resume, Stack, Callbacks, Opts, Acc) end);
integer(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
integer(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> integer(Stream, Stack, Callbacks, Opts, Acc) end.
fraction(<<?end_object/utf16-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
fraction(<<?end_array/utf16-little, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
fraction(<<?comma/utf16-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?comma/utf16-little, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?zero/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
fraction(<<"e"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
fraction(<<"E"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
fraction(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
fraction(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
fraction(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?solidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> fraction(Resume, Stack, Callbacks, Opts, Acc) end);
fraction(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
fraction(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> fraction(Stream, Stack, Callbacks, Opts, Acc) end.
e(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?positive; S =:= ?negative ->
ex(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
e(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?zero; ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
e(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> e(Stream, Stack, Callbacks, Opts, Acc) end.
ex(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?zero; ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
ex(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> ex(Stream, Stack, Callbacks, Opts, Acc) end.
exp(<<?end_object/utf16-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
exp(<<?end_array/utf16-little, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
exp(<<?comma/utf16-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?comma/utf16-little, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?zero/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
exp(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
exp(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
exp(<<S/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?rsolidus/utf16-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> exp(Resume, Stack, Callbacks, Opts, Acc) end);
exp(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
exp(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> exp(Stream, Stack, Callbacks, Opts, Acc) end.
tr(<<"r"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
tru(Rest, Stack, Callbacks, Opts);
tr(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> tr(Stream, Stack, Callbacks, Opts) end.
tru(<<"u"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
true(Rest, Stack, Callbacks, Opts);
tru(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> tru(Stream, Stack, Callbacks, Opts) end.
true(<<"e"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, true}, Callbacks), Opts);
true(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> true(Stream, Stack, Callbacks, Opts) end.
fa(<<"a"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
fal(Rest, Stack, Callbacks, Opts);
fa(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fa(Stream, Stack, Callbacks, Opts) end.
fal(<<"l"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
fals(Rest, Stack, Callbacks, Opts);
fal(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fal(Stream, Stack, Callbacks, Opts) end.
fals(<<"s"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
false(Rest, Stack, Callbacks, Opts);
fals(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fals(Stream, Stack, Callbacks, Opts) end.
false(<<"e"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, false}, Callbacks), Opts);
false(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> false(Stream, Stack, Callbacks, Opts) end.
nu(<<"u"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
nul(Rest, Stack, Callbacks, Opts);
nu(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> nu(Stream, Stack, Callbacks, Opts) end.
nul(<<"l"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
null(Rest, Stack, Callbacks, Opts);
nul(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> nul(Stream, Stack, Callbacks, Opts) end.
null(<<"l"/utf16-little, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, null}, Callbacks), Opts);
null(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> null(Stream, Stack, Callbacks, Opts) end.
%% comments are c style, /* blah blah */ and are STRONGLY discouraged. any unicode
%% character is valid in a comment, except, obviously the */ sequence which ends
%% the comment. they're implemented as a closure called when the comment ends that
%% returns execution to the point where the comment began. comments are not
%% recorded in any way, simply parsed.
maybe_comment(<<?star/utf16-little, Rest/binary>>, Resume) ->
comment(Rest, Resume);
maybe_comment(<<>>, Resume) ->
fun(Stream) -> maybe_comment(Stream, Resume) end.
comment(<<?star/utf16-little, Rest/binary>>, Resume) ->
maybe_comment_done(Rest, Resume);
comment(<<_/utf16-little, Rest/binary>>, Resume) ->
comment(Rest, Resume);
comment(<<>>, Resume) ->
fun(Stream) -> comment(Stream, Resume) end.
maybe_comment_done(<<?solidus/utf16-little, Rest/binary>>, Resume) ->
Resume(Rest);
maybe_comment_done(<<>>, Resume) ->
fun(Stream) -> maybe_comment_done(Stream, Resume) end.
%% helper function for dispatching of parser events
callback(eof, {none, Callbacks}) ->
lists:reverse(Callbacks);
callback(Event, {none, Callbacks}) ->
{none, [Event] ++ Callbacks};
callback(Event, {Mod, State}) when is_atom(Mod) ->
{Mod, Mod:jsx_event(Event, State)};
callback(Event, {F, State}) when is_function(F) ->
{F, F(Event, State)}.

View file

@ -1,436 +0,0 @@
-module(jsx_utf32b).
-export([start/4]).
-include("jsx_common.hrl").
%% this code is mostly autogenerated and mostly ugly. apologies. for more insight on
%% Callbacks or Opts, see the comments accompanying decoder/2 (in jsx.erl). Stack
%% is a stack of flags used to track depth and to keep track of whether we are
%% returning from a value or a key inside objects. all pops, peeks and pushes are
%% inlined. the code that handles naked values and comments is not optimized by the
%% compiler for efficient matching, but you shouldn't be using naked values or comments
%% anyways, they are horrible and contrary to the spec.
start(<<?start_object/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
start(<<?start_array/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
start(<<?quote/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
string(Rest, Stack, Callbacks, Opts, []);
start(<<$t/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
tr(Rest, Stack, Callbacks, Opts);
start(<<$f/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
fa(Rest, Stack, Callbacks, Opts);
start(<<$n/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
nu(Rest, Stack, Callbacks, Opts);
start(<<?negative/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
negative(Rest, Stack, Callbacks, Opts, "-");
start(<<?zero/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
zero(Rest, Stack, Callbacks, Opts, "0");
start(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S), Opts#opts.naked_values == true ->
integer(Rest, Stack, Callbacks, Opts, [S]);
start(<<?solidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> start(Resume, Stack, Callbacks, Opts) end);
start(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
start(Rest, Stack, Callbacks, Opts);
start(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> start(Stream, Stack, Callbacks, Opts) end.
maybe_done(<<?end_object/utf32-big, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
maybe_done(<<?end_array/utf32-big, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_array, Callbacks), Opts);
maybe_done(<<?comma/utf32-big, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
key(Rest, [key|Stack], Callbacks, Opts);
maybe_done(<<?comma/utf32-big, Rest/binary>>, [array|_] = Stack, Callbacks, Opts) ->
value(Rest, Stack, Callbacks, Opts);
maybe_done(<<?solidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> maybe_done(Resume, Stack, Callbacks, Opts) end);
maybe_done(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, Callbacks, Opts);
maybe_done(<<>>, [], Callbacks, _Opts) ->
callback(eof, Callbacks);
maybe_done(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> maybe_done(Stream, Stack, Callbacks, Opts) end.
object(<<?end_object/utf32-big, Rest/binary>>, [key|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
object(<<?quote/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
object(<<?solidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> object(Resume, Stack, Callbacks, Opts) end);
object(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
object(Rest, Stack, Callbacks, Opts);
object(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> object(Stream, Stack, Callbacks, Opts) end.
array(<<?quote/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
array(<<?start_object/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
array(<<?start_array/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
array(<<?end_array/utf32-big, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_array, Callbacks), Opts);
array(<<$t/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
tr(Rest, Stack, Callbacks, Opts);
array(<<$f/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
fa(Rest, Stack, Callbacks, Opts);
array(<<$n/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
nu(Rest, Stack, Callbacks, Opts);
array(<<?negative/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
negative(Rest, Stack, Callbacks, Opts, "-");
array(<<?zero/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
zero(Rest, Stack, Callbacks, Opts, "0");
array(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S]);
array(<<?solidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> array(Resume, Stack, Callbacks, Opts) end);
array(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
array(Rest, Stack, Callbacks, Opts);
array(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> array(Stream, Stack, Callbacks, Opts) end.
value(<<?quote/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
value(<<?start_object/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
value(<<?start_array/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
value(<<$t/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
tr(Rest, Stack, Callbacks, Opts);
value(<<$f/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
fa(Rest, Stack, Callbacks, Opts);
value(<<$n/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
nu(Rest, Stack, Callbacks, Opts);
value(<<?negative/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
negative(Rest, Stack, Callbacks, Opts, "-");
value(<<?zero/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
zero(Rest, Stack, Callbacks, Opts, "0");
value(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S]);
value(<<?solidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> value(Resume, Stack, Callbacks, Opts) end);
value(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
value(Rest, Stack, Callbacks, Opts);
value(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> value(Stream, Stack, Callbacks, Opts) end.
colon(<<?colon/utf32-big, Rest/binary>>, [key|Stack], Callbacks, Opts) ->
value(Rest, [object|Stack], Callbacks, Opts);
colon(<<?solidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> colon(Resume, Stack, Callbacks, Opts) end);
colon(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
colon(Rest, Stack, Callbacks, Opts);
colon(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> colon(Stream, Stack, Callbacks, Opts) end.
key(<<?quote/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
key(<<?solidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> key(Resume, Stack, Callbacks, Opts) end);
key(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
key(Rest, Stack, Callbacks, Opts);
key(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> key(Stream, Stack, Callbacks, Opts) end.
%% string has an additional parameter, an accumulator (Acc) used to hold the intermediate
%% representation of the string being parsed. using a list of integers representing
%% unicode codepoints is faster than constructing binaries, many of which will be
%% converted back to lists by the user anyways.
string(<<?quote/utf32-big, Rest/binary>>, [key|_] = Stack, Callbacks, Opts, Acc) ->
colon(Rest, Stack, callback({key, lists:reverse(Acc)}, Callbacks), Opts);
string(<<?quote/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback({string, lists:reverse(Acc)}, Callbacks), Opts);
string(<<?rsolidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escape(Rest, Stack, Callbacks, Opts, Acc);
string(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_noncontrol(S) ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
string(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> string(Stream, Stack, Callbacks, Opts, Acc) end.
%% only thing to note here is the additional accumulator passed to escaped_unicode used
%% to hold the codepoint sequence. unescessary, but nicer than using the string
%% accumulator.
escape(<<"b"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\b" ++ Acc);
escape(<<"f"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\f" ++ Acc);
escape(<<"n"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\n" ++ Acc);
escape(<<"r"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\r" ++ Acc);
escape(<<"t"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\t" ++ Acc);
escape(<<"u"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escaped_unicode(Rest, Stack, Callbacks, Opts, Acc, []);
escape(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc)
when S =:= ?quote; S =:= ?solidus; S =:= ?rsolidus ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
escape(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> escape(Stream, Stack, Callbacks, Opts, Acc) end.
%% this code is ugly and unfortunate, but so is json's handling of escaped unicode
%% codepoint sequences. if the ascii option is present, the sequence is converted
%% to a codepoint and inserted into the string if it represents an ascii value. if
%% the codepoint option is present the sequence is converted and inserted as long
%% as it represents a valid 16 bit integer value (this is where json's spec gets
%% insane). any other option and the sequence is converted back to an erlang string
%% and appended to the string in place.
escaped_unicode(<<D/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, String, [C, B, A]) ->
X = erlang:list_to_integer([A, B, C, D], 16),
case Opts#opts.escaped_unicode of
ascii when X < 16#0080 ->
string(Rest, Stack, Callbacks, Opts, [X] ++ String)
; codepoint ->
string(Rest, Stack, Callbacks, Opts, [X] ++ String)
; _ ->
string(Rest, Stack, Callbacks, Opts, [?rsolidus, $u, A, B, C, D] ++ String)
end;
escaped_unicode(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, String, Acc) when ?is_hex(S) ->
escaped_unicode(Rest, Stack, Callbacks, Opts, String, [S] ++ Acc);
escaped_unicode(<<>>, Stack, Callbacks, Opts, String, Acc) ->
fun(Stream) -> escaped_unicode(Stream, Stack, Callbacks, Opts, String, Acc) end.
%% like strings, numbers are collected in an intermediate accumulator before
%% being emitted to the callback handler. no processing of numbers is done in
%% process, it's left for the user, though there are convenience functions to
%% convert them into erlang floats/integers in jsx_utils.erl.
%% TODO: actually write that jsx_utils.erl module mentioned above...
negative(<<"0"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
zero(Rest, Stack, Callbacks, Opts, "0" ++ Acc);
negative(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
negative(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> negative(Stream, Stack, Callbacks, Opts, Acc) end.
zero(<<?end_object/utf32-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
zero(<<?end_array/utf32-big, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
zero(<<?comma/utf32-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?comma/utf32-big, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?decimalpoint/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?decimalpoint] ++ Acc);
zero(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?solidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> zero(Resume, Stack, Callbacks, Opts, Acc) end);
zero(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
zero(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> zero(Stream, Stack, Callbacks, Opts, Acc) end.
integer(<<?end_object/utf32-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
integer(<<?end_array/utf32-big, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
integer(<<?comma/utf32-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?comma/utf32-big, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?decimalpoint/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?decimalpoint] ++ Acc);
integer(<<?zero/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
integer(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
integer(<<"e"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
integer(<<"E"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
integer(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
integer(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?solidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> integer(Resume, Stack, Callbacks, Opts, Acc) end);
integer(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
integer(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> integer(Stream, Stack, Callbacks, Opts, Acc) end.
fraction(<<?end_object/utf32-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
fraction(<<?end_array/utf32-big, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
fraction(<<?comma/utf32-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?comma/utf32-big, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?zero/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
fraction(<<"e"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
fraction(<<"E"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
fraction(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
fraction(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
fraction(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?solidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> fraction(Resume, Stack, Callbacks, Opts, Acc) end);
fraction(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
fraction(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> fraction(Stream, Stack, Callbacks, Opts, Acc) end.
e(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?positive; S =:= ?negative ->
ex(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
e(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?zero; ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
e(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> e(Stream, Stack, Callbacks, Opts, Acc) end.
ex(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?zero; ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
ex(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> ex(Stream, Stack, Callbacks, Opts, Acc) end.
exp(<<?end_object/utf32-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
exp(<<?end_array/utf32-big, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
exp(<<?comma/utf32-big, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?comma/utf32-big, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?zero/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
exp(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
exp(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
exp(<<S/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?rsolidus/utf32-big, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> exp(Resume, Stack, Callbacks, Opts, Acc) end);
exp(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
exp(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> exp(Stream, Stack, Callbacks, Opts, Acc) end.
tr(<<"r"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
tru(Rest, Stack, Callbacks, Opts);
tr(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> tr(Stream, Stack, Callbacks, Opts) end.
tru(<<"u"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
true(Rest, Stack, Callbacks, Opts);
tru(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> tru(Stream, Stack, Callbacks, Opts) end.
true(<<"e"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, true}, Callbacks), Opts);
true(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> true(Stream, Stack, Callbacks, Opts) end.
fa(<<"a"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
fal(Rest, Stack, Callbacks, Opts);
fa(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fa(Stream, Stack, Callbacks, Opts) end.
fal(<<"l"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
fals(Rest, Stack, Callbacks, Opts);
fal(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fal(Stream, Stack, Callbacks, Opts) end.
fals(<<"s"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
false(Rest, Stack, Callbacks, Opts);
fals(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fals(Stream, Stack, Callbacks, Opts) end.
false(<<"e"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, false}, Callbacks), Opts);
false(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> false(Stream, Stack, Callbacks, Opts) end.
nu(<<"u"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
nul(Rest, Stack, Callbacks, Opts);
nu(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> nu(Stream, Stack, Callbacks, Opts) end.
nul(<<"l"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
null(Rest, Stack, Callbacks, Opts);
nul(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> nul(Stream, Stack, Callbacks, Opts) end.
null(<<"l"/utf32-big, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, null}, Callbacks), Opts);
null(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> null(Stream, Stack, Callbacks, Opts) end.
%% comments are c style, /* blah blah */ and are STRONGLY discouraged. any unicode
%% character is valid in a comment, except, obviously the */ sequence which ends
%% the comment. they're implemented as a closure called when the comment ends that
%% returns execution to the point where the comment began. comments are not
%% recorded in any way, simply parsed.
maybe_comment(<<?star/utf32-big, Rest/binary>>, Resume) ->
comment(Rest, Resume);
maybe_comment(<<>>, Resume) ->
fun(Stream) -> maybe_comment(Stream, Resume) end.
comment(<<?star/utf32-big, Rest/binary>>, Resume) ->
maybe_comment_done(Rest, Resume);
comment(<<_/utf32-big, Rest/binary>>, Resume) ->
comment(Rest, Resume);
comment(<<>>, Resume) ->
fun(Stream) -> comment(Stream, Resume) end.
maybe_comment_done(<<?solidus/utf32-big, Rest/binary>>, Resume) ->
Resume(Rest);
maybe_comment_done(<<>>, Resume) ->
fun(Stream) -> maybe_comment_done(Stream, Resume) end.
%% helper function for dispatching of parser events
callback(eof, {none, Callbacks}) ->
lists:reverse(Callbacks);
callback(Event, {none, Callbacks}) ->
{none, [Event] ++ Callbacks};
callback(Event, {Mod, State}) when is_atom(Mod) ->
{Mod, Mod:jsx_event(Event, State)};
callback(Event, {F, State}) when is_function(F) ->
{F, F(Event, State)}.

View file

@ -1,436 +0,0 @@
-module(jsx_utf32l).
-export([start/4]).
-include("jsx_common.hrl").
%% this code is mostly autogenerated and mostly ugly. apologies. for more insight on
%% Callbacks or Opts, see the comments accompanying decoder/2 (in jsx.erl). Stack
%% is a stack of flags used to track depth and to keep track of whether we are
%% returning from a value or a key inside objects. all pops, peeks and pushes are
%% inlined. the code that handles naked values and comments is not optimized by the
%% compiler for efficient matching, but you shouldn't be using naked values or comments
%% anyways, they are horrible and contrary to the spec.
start(<<?start_object/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
start(<<?start_array/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
start(<<?quote/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
string(Rest, Stack, Callbacks, Opts, []);
start(<<$t/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
tr(Rest, Stack, Callbacks, Opts);
start(<<$f/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
fa(Rest, Stack, Callbacks, Opts);
start(<<$n/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
nu(Rest, Stack, Callbacks, Opts);
start(<<?negative/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
negative(Rest, Stack, Callbacks, Opts, "-");
start(<<?zero/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
zero(Rest, Stack, Callbacks, Opts, "0");
start(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S), Opts#opts.naked_values == true ->
integer(Rest, Stack, Callbacks, Opts, [S]);
start(<<?solidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> start(Resume, Stack, Callbacks, Opts) end);
start(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
start(Rest, Stack, Callbacks, Opts);
start(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> start(Stream, Stack, Callbacks, Opts) end.
maybe_done(<<?end_object/utf32-little, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
maybe_done(<<?end_array/utf32-little, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_array, Callbacks), Opts);
maybe_done(<<?comma/utf32-little, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
key(Rest, [key|Stack], Callbacks, Opts);
maybe_done(<<?comma/utf32-little, Rest/binary>>, [array|_] = Stack, Callbacks, Opts) ->
value(Rest, Stack, Callbacks, Opts);
maybe_done(<<?solidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> maybe_done(Resume, Stack, Callbacks, Opts) end);
maybe_done(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, Callbacks, Opts);
maybe_done(<<>>, [], Callbacks, _Opts) ->
callback(eof, Callbacks);
maybe_done(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> maybe_done(Stream, Stack, Callbacks, Opts) end.
object(<<?end_object/utf32-little, Rest/binary>>, [key|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
object(<<?quote/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
object(<<?solidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> object(Resume, Stack, Callbacks, Opts) end);
object(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
object(Rest, Stack, Callbacks, Opts);
object(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> object(Stream, Stack, Callbacks, Opts) end.
array(<<?quote/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
array(<<?start_object/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
array(<<?start_array/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
array(<<?end_array/utf32-little, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_array, Callbacks), Opts);
array(<<$t/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
tr(Rest, Stack, Callbacks, Opts);
array(<<$f/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
fa(Rest, Stack, Callbacks, Opts);
array(<<$n/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
nu(Rest, Stack, Callbacks, Opts);
array(<<?negative/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
negative(Rest, Stack, Callbacks, Opts, "-");
array(<<?zero/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
zero(Rest, Stack, Callbacks, Opts, "0");
array(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S]);
array(<<?solidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> array(Resume, Stack, Callbacks, Opts) end);
array(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
array(Rest, Stack, Callbacks, Opts);
array(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> array(Stream, Stack, Callbacks, Opts) end.
value(<<?quote/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
value(<<?start_object/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
value(<<?start_array/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
value(<<$t/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
tr(Rest, Stack, Callbacks, Opts);
value(<<$f/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
fa(Rest, Stack, Callbacks, Opts);
value(<<$n/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
nu(Rest, Stack, Callbacks, Opts);
value(<<?negative/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
negative(Rest, Stack, Callbacks, Opts, "-");
value(<<?zero/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
zero(Rest, Stack, Callbacks, Opts, "0");
value(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S]);
value(<<?solidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> value(Resume, Stack, Callbacks, Opts) end);
value(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
value(Rest, Stack, Callbacks, Opts);
value(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> value(Stream, Stack, Callbacks, Opts) end.
colon(<<?colon/utf32-little, Rest/binary>>, [key|Stack], Callbacks, Opts) ->
value(Rest, [object|Stack], Callbacks, Opts);
colon(<<?solidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> colon(Resume, Stack, Callbacks, Opts) end);
colon(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
colon(Rest, Stack, Callbacks, Opts);
colon(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> colon(Stream, Stack, Callbacks, Opts) end.
key(<<?quote/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
key(<<?solidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> key(Resume, Stack, Callbacks, Opts) end);
key(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
key(Rest, Stack, Callbacks, Opts);
key(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> key(Stream, Stack, Callbacks, Opts) end.
%% string has an additional parameter, an accumulator (Acc) used to hold the intermediate
%% representation of the string being parsed. using a list of integers representing
%% unicode codepoints is faster than constructing binaries, many of which will be
%% converted back to lists by the user anyways.
string(<<?quote/utf32-little, Rest/binary>>, [key|_] = Stack, Callbacks, Opts, Acc) ->
colon(Rest, Stack, callback({key, lists:reverse(Acc)}, Callbacks), Opts);
string(<<?quote/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback({string, lists:reverse(Acc)}, Callbacks), Opts);
string(<<?rsolidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escape(Rest, Stack, Callbacks, Opts, Acc);
string(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_noncontrol(S) ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
string(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> string(Stream, Stack, Callbacks, Opts, Acc) end.
%% only thing to note here is the additional accumulator passed to escaped_unicode used
%% to hold the codepoint sequence. unescessary, but nicer than using the string
%% accumulator.
escape(<<"b"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\b" ++ Acc);
escape(<<"f"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\f" ++ Acc);
escape(<<"n"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\n" ++ Acc);
escape(<<"r"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\r" ++ Acc);
escape(<<"t"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\t" ++ Acc);
escape(<<"u"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escaped_unicode(Rest, Stack, Callbacks, Opts, Acc, []);
escape(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc)
when S =:= ?quote; S =:= ?solidus; S =:= ?rsolidus ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
escape(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> escape(Stream, Stack, Callbacks, Opts, Acc) end.
%% this code is ugly and unfortunate, but so is json's handling of escaped unicode
%% codepoint sequences. if the ascii option is present, the sequence is converted
%% to a codepoint and inserted into the string if it represents an ascii value. if
%% the codepoint option is present the sequence is converted and inserted as long
%% as it represents a valid 16 bit integer value (this is where json's spec gets
%% insane). any other option and the sequence is converted back to an erlang string
%% and appended to the string in place.
escaped_unicode(<<D/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, String, [C, B, A]) ->
X = erlang:list_to_integer([A, B, C, D], 16),
case Opts#opts.escaped_unicode of
ascii when X < 16#0080 ->
string(Rest, Stack, Callbacks, Opts, [X] ++ String)
; codepoint ->
string(Rest, Stack, Callbacks, Opts, [X] ++ String)
; _ ->
string(Rest, Stack, Callbacks, Opts, [?rsolidus, $u, A, B, C, D] ++ String)
end;
escaped_unicode(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, String, Acc) when ?is_hex(S) ->
escaped_unicode(Rest, Stack, Callbacks, Opts, String, [S] ++ Acc);
escaped_unicode(<<>>, Stack, Callbacks, Opts, String, Acc) ->
fun(Stream) -> escaped_unicode(Stream, Stack, Callbacks, Opts, String, Acc) end.
%% like strings, numbers are collected in an intermediate accumulator before
%% being emitted to the callback handler. no processing of numbers is done in
%% process, it's left for the user, though there are convenience functions to
%% convert them into erlang floats/integers in jsx_utils.erl.
%% TODO: actually write that jsx_utils.erl module mentioned above...
negative(<<"0"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
zero(Rest, Stack, Callbacks, Opts, "0" ++ Acc);
negative(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
negative(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> negative(Stream, Stack, Callbacks, Opts, Acc) end.
zero(<<?end_object/utf32-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
zero(<<?end_array/utf32-little, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
zero(<<?comma/utf32-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?comma/utf32-little, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?decimalpoint/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?decimalpoint] ++ Acc);
zero(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?solidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> zero(Resume, Stack, Callbacks, Opts, Acc) end);
zero(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
zero(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> zero(Stream, Stack, Callbacks, Opts, Acc) end.
integer(<<?end_object/utf32-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
integer(<<?end_array/utf32-little, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
integer(<<?comma/utf32-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?comma/utf32-little, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?decimalpoint/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?decimalpoint] ++ Acc);
integer(<<?zero/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
integer(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
integer(<<"e"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
integer(<<"E"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
integer(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
integer(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?solidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> integer(Resume, Stack, Callbacks, Opts, Acc) end);
integer(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
integer(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> integer(Stream, Stack, Callbacks, Opts, Acc) end.
fraction(<<?end_object/utf32-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
fraction(<<?end_array/utf32-little, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
fraction(<<?comma/utf32-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?comma/utf32-little, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?zero/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
fraction(<<"e"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
fraction(<<"E"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
fraction(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
fraction(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
fraction(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?solidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> fraction(Resume, Stack, Callbacks, Opts, Acc) end);
fraction(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
fraction(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> fraction(Stream, Stack, Callbacks, Opts, Acc) end.
e(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?positive; S =:= ?negative ->
ex(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
e(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?zero; ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
e(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> e(Stream, Stack, Callbacks, Opts, Acc) end.
ex(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?zero; ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
ex(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> ex(Stream, Stack, Callbacks, Opts, Acc) end.
exp(<<?end_object/utf32-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
exp(<<?end_array/utf32-little, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
exp(<<?comma/utf32-little, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?comma/utf32-little, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?zero/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
exp(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
exp(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
exp(<<S/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?rsolidus/utf32-little, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> exp(Resume, Stack, Callbacks, Opts, Acc) end);
exp(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
exp(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> exp(Stream, Stack, Callbacks, Opts, Acc) end.
tr(<<"r"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
tru(Rest, Stack, Callbacks, Opts);
tr(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> tr(Stream, Stack, Callbacks, Opts) end.
tru(<<"u"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
true(Rest, Stack, Callbacks, Opts);
tru(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> tru(Stream, Stack, Callbacks, Opts) end.
true(<<"e"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, true}, Callbacks), Opts);
true(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> true(Stream, Stack, Callbacks, Opts) end.
fa(<<"a"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
fal(Rest, Stack, Callbacks, Opts);
fa(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fa(Stream, Stack, Callbacks, Opts) end.
fal(<<"l"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
fals(Rest, Stack, Callbacks, Opts);
fal(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fal(Stream, Stack, Callbacks, Opts) end.
fals(<<"s"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
false(Rest, Stack, Callbacks, Opts);
fals(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fals(Stream, Stack, Callbacks, Opts) end.
false(<<"e"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, false}, Callbacks), Opts);
false(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> false(Stream, Stack, Callbacks, Opts) end.
nu(<<"u"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
nul(Rest, Stack, Callbacks, Opts);
nu(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> nu(Stream, Stack, Callbacks, Opts) end.
nul(<<"l"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
null(Rest, Stack, Callbacks, Opts);
nul(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> nul(Stream, Stack, Callbacks, Opts) end.
null(<<"l"/utf32-little, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, null}, Callbacks), Opts);
null(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> null(Stream, Stack, Callbacks, Opts) end.
%% comments are c style, /* blah blah */ and are STRONGLY discouraged. any unicode
%% character is valid in a comment, except, obviously the */ sequence which ends
%% the comment. they're implemented as a closure called when the comment ends that
%% returns execution to the point where the comment began. comments are not
%% recorded in any way, simply parsed.
maybe_comment(<<?star/utf32-little, Rest/binary>>, Resume) ->
comment(Rest, Resume);
maybe_comment(<<>>, Resume) ->
fun(Stream) -> maybe_comment(Stream, Resume) end.
comment(<<?star/utf32-little, Rest/binary>>, Resume) ->
maybe_comment_done(Rest, Resume);
comment(<<_/utf32-little, Rest/binary>>, Resume) ->
comment(Rest, Resume);
comment(<<>>, Resume) ->
fun(Stream) -> comment(Stream, Resume) end.
maybe_comment_done(<<?solidus/utf32-little, Rest/binary>>, Resume) ->
Resume(Rest);
maybe_comment_done(<<>>, Resume) ->
fun(Stream) -> maybe_comment_done(Stream, Resume) end.
%% helper function for dispatching of parser events
callback(eof, {none, Callbacks}) ->
lists:reverse(Callbacks);
callback(Event, {none, Callbacks}) ->
{none, [Event] ++ Callbacks};
callback(Event, {Mod, State}) when is_atom(Mod) ->
{Mod, Mod:jsx_event(Event, State)};
callback(Event, {F, State}) when is_function(F) ->
{F, F(Event, State)}.

View file

@ -1,436 +0,0 @@
-module(jsx_utf8).
-export([start/4]).
-include("jsx_common.hrl").
%% this code is mostly autogenerated and mostly ugly. apologies. for more insight on
%% Callbacks or Opts, see the comments accompanying decoder/2 (in jsx.erl). Stack
%% is a stack of flags used to track depth and to keep track of whether we are
%% returning from a value or a key inside objects. all pops, peeks and pushes are
%% inlined. the code that handles naked values and comments is not optimized by the
%% compiler for efficient matching, but you shouldn't be using naked values or comments
%% anyways, they are horrible and contrary to the spec.
start(<<?start_object/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
start(<<?start_array/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
start(<<?quote/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
string(Rest, Stack, Callbacks, Opts, []);
start(<<$t/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
tr(Rest, Stack, Callbacks, Opts);
start(<<$f/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
fa(Rest, Stack, Callbacks, Opts);
start(<<$n/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
nu(Rest, Stack, Callbacks, Opts);
start(<<?negative/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
negative(Rest, Stack, Callbacks, Opts, "-");
start(<<?zero/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.naked_values == true ->
zero(Rest, Stack, Callbacks, Opts, "0");
start(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S), Opts#opts.naked_values == true ->
integer(Rest, Stack, Callbacks, Opts, [S]);
start(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> start(Resume, Stack, Callbacks, Opts) end);
start(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
start(Rest, Stack, Callbacks, Opts);
start(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> start(Stream, Stack, Callbacks, Opts) end.
maybe_done(<<?end_object/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
maybe_done(<<?end_array/utf8, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_array, Callbacks), Opts);
maybe_done(<<?comma/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts) ->
key(Rest, [key|Stack], Callbacks, Opts);
maybe_done(<<?comma/utf8, Rest/binary>>, [array|_] = Stack, Callbacks, Opts) ->
value(Rest, Stack, Callbacks, Opts);
maybe_done(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> maybe_done(Resume, Stack, Callbacks, Opts) end);
maybe_done(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, Callbacks, Opts);
maybe_done(<<>>, [], Callbacks, _Opts) ->
callback(eof, Callbacks);
maybe_done(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> maybe_done(Stream, Stack, Callbacks, Opts) end.
object(<<?end_object/utf8, Rest/binary>>, [key|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_object, Callbacks), Opts);
object(<<?quote/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
object(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> object(Resume, Stack, Callbacks, Opts) end);
object(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
object(Rest, Stack, Callbacks, Opts);
object(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> object(Stream, Stack, Callbacks, Opts) end.
array(<<?quote/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
array(<<?start_object/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
array(<<?start_array/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
array(<<?end_array/utf8, Rest/binary>>, [array|Stack], Callbacks, Opts) ->
maybe_done(Rest, Stack, callback(end_array, Callbacks), Opts);
array(<<$t/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
tr(Rest, Stack, Callbacks, Opts);
array(<<$f/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
fa(Rest, Stack, Callbacks, Opts);
array(<<$n/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
nu(Rest, Stack, Callbacks, Opts);
array(<<?negative/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
negative(Rest, Stack, Callbacks, Opts, "-");
array(<<?zero/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
zero(Rest, Stack, Callbacks, Opts, "0");
array(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S]);
array(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> array(Resume, Stack, Callbacks, Opts) end);
array(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
array(Rest, Stack, Callbacks, Opts);
array(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> array(Stream, Stack, Callbacks, Opts) end.
value(<<?quote/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
value(<<?start_object/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
object(Rest, [key|Stack], callback(start_object, Callbacks), Opts);
value(<<?start_array/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
array(Rest, [array|Stack], callback(start_array, Callbacks), Opts);
value(<<$t/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
tr(Rest, Stack, Callbacks, Opts);
value(<<$f/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
fa(Rest, Stack, Callbacks, Opts);
value(<<$n/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
nu(Rest, Stack, Callbacks, Opts);
value(<<?negative/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
negative(Rest, Stack, Callbacks, Opts, "-");
value(<<?zero/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
zero(Rest, Stack, Callbacks, Opts, "0");
value(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S]);
value(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> value(Resume, Stack, Callbacks, Opts) end);
value(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
value(Rest, Stack, Callbacks, Opts);
value(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> value(Stream, Stack, Callbacks, Opts) end.
colon(<<?colon/utf8, Rest/binary>>, [key|Stack], Callbacks, Opts) ->
value(Rest, [object|Stack], Callbacks, Opts);
colon(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> colon(Resume, Stack, Callbacks, Opts) end);
colon(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
colon(Rest, Stack, Callbacks, Opts);
colon(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> colon(Stream, Stack, Callbacks, Opts) end.
key(<<?quote/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
string(Rest, Stack, Callbacks, Opts, []);
key(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> key(Resume, Stack, Callbacks, Opts) end);
key(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts) when ?is_whitespace(S) ->
key(Rest, Stack, Callbacks, Opts);
key(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> key(Stream, Stack, Callbacks, Opts) end.
%% string has an additional parameter, an accumulator (Acc) used to hold the intermediate
%% representation of the string being parsed. using a list of integers representing
%% unicode codepoints is faster than constructing binaries, many of which will be
%% converted back to lists by the user anyways.
string(<<?quote/utf8, Rest/binary>>, [key|_] = Stack, Callbacks, Opts, Acc) ->
colon(Rest, Stack, callback({key, lists:reverse(Acc)}, Callbacks), Opts);
string(<<?quote/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback({string, lists:reverse(Acc)}, Callbacks), Opts);
string(<<?rsolidus/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escape(Rest, Stack, Callbacks, Opts, Acc);
string(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_noncontrol(S) ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
string(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> string(Stream, Stack, Callbacks, Opts, Acc) end.
%% only thing to note here is the additional accumulator passed to escaped_unicode used
%% to hold the codepoint sequence. unescessary, but nicer than using the string
%% accumulator.
escape(<<"b"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\b" ++ Acc);
escape(<<"f"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\f" ++ Acc);
escape(<<"n"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\n" ++ Acc);
escape(<<"r"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\r" ++ Acc);
escape(<<"t"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
string(Rest, Stack, Callbacks, Opts, "\t" ++ Acc);
escape(<<"u"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escaped_unicode(Rest, Stack, Callbacks, Opts, Acc, []);
escape(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc)
when S =:= ?quote; S =:= ?solidus; S =:= ?rsolidus ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
escape(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> escape(Stream, Stack, Callbacks, Opts, Acc) end.
%% this code is ugly and unfortunate, but so is json's handling of escaped unicode
%% codepoint sequences. if the ascii option is present, the sequence is converted
%% to a codepoint and inserted into the string if it represents an ascii value. if
%% the codepoint option is present the sequence is converted and inserted as long
%% as it represents a valid 16 bit integer value (this is where json's spec gets
%% insane). any other option and the sequence is converted back to an erlang string
%% and appended to the string in place.
escaped_unicode(<<D/utf8, Rest/binary>>, Stack, Callbacks, Opts, String, [C, B, A]) ->
X = erlang:list_to_integer([A, B, C, D], 16),
case Opts#opts.escaped_unicode of
ascii when X < 16#0080 ->
string(Rest, Stack, Callbacks, Opts, [X] ++ String)
; codepoint ->
string(Rest, Stack, Callbacks, Opts, [X] ++ String)
; _ ->
string(Rest, Stack, Callbacks, Opts, [?rsolidus, $u, A, B, C, D] ++ String)
end;
escaped_unicode(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, String, Acc) when ?is_hex(S) ->
escaped_unicode(Rest, Stack, Callbacks, Opts, String, [S] ++ Acc);
escaped_unicode(<<>>, Stack, Callbacks, Opts, String, Acc) ->
fun(Stream) -> escaped_unicode(Stream, Stack, Callbacks, Opts, String, Acc) end.
%% like strings, numbers are collected in an intermediate accumulator before
%% being emitted to the callback handler. no processing of numbers is done in
%% process, it's left for the user, though there are convenience functions to
%% convert them into erlang floats/integers in jsx_utils.erl.
%% TODO: actually write that jsx_utils.erl module mentioned above...
negative(<<"0"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
zero(Rest, Stack, Callbacks, Opts, "0" ++ Acc);
negative(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
negative(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> negative(Stream, Stack, Callbacks, Opts, Acc) end.
zero(<<?end_object/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
zero(<<?end_array/utf8, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
zero(<<?comma/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?comma/utf8, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?decimalpoint/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?decimalpoint] ++ Acc);
zero(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
zero(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> zero(Resume, Stack, Callbacks, Opts, Acc) end);
zero(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
zero(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> zero(Stream, Stack, Callbacks, Opts, Acc) end.
integer(<<?end_object/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
integer(<<?end_array/utf8, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
integer(<<?comma/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?comma/utf8, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?decimalpoint/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?decimalpoint] ++ Acc);
integer(<<?zero/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
integer(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
integer(<<"e"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
integer(<<"E"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
integer(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
integer(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
integer(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
integer(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> integer(Resume, Stack, Callbacks, Opts, Acc) end);
integer(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
integer(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> integer(Stream, Stack, Callbacks, Opts, Acc) end.
fraction(<<?end_object/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
fraction(<<?end_array/utf8, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
fraction(<<?comma/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?comma/utf8, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?zero/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
fraction(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
fraction(<<"e"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
fraction(<<"E"/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
e(Rest, Stack, Callbacks, Opts, "e" ++ Acc);
fraction(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
fraction(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
fraction(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
fraction(<<?solidus/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> fraction(Resume, Stack, Callbacks, Opts, Acc) end);
fraction(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
fraction(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> fraction(Stream, Stack, Callbacks, Opts, Acc) end.
e(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?positive; S =:= ?negative ->
ex(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
e(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?zero; ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
e(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> e(Stream, Stack, Callbacks, Opts, Acc) end.
ex(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when S =:= ?zero; ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
ex(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> ex(Stream, Stack, Callbacks, Opts, Acc) end.
exp(<<?end_object/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_object, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
exp(<<?end_array/utf8, Rest/binary>>, [array|Stack], Callbacks, Opts, Acc) ->
maybe_done(Rest, Stack, callback(end_array, callback({number, lists:reverse(Acc)}, Callbacks)), Opts);
exp(<<?comma/utf8, Rest/binary>>, [object|Stack], Callbacks, Opts, Acc) ->
key(Rest, [key|Stack], callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?comma/utf8, Rest/binary>>, [array|_] = Stack, Callbacks, Opts, Acc) ->
value(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?zero/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
exp(Rest, Stack, Callbacks, Opts, [?zero] ++ Acc);
exp(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_nonzero(S) ->
exp(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
exp(<<S/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_whitespace(S) ->
maybe_done(Rest, Stack, callback({number, lists:reverse(Acc)}, Callbacks), Opts);
exp(<<?rsolidus/utf8, Rest/binary>>, Stack, Callbacks, Opts, Acc) when Opts#opts.comments == true ->
maybe_comment(Rest, fun(Resume) -> exp(Resume, Stack, Callbacks, Opts, Acc) end);
exp(<<>>, [], Callbacks, _Opts, Acc) ->
callback(eof, callback({number, lists:reverse(Acc)}, Callbacks));
exp(<<>>, Stack, Callbacks, Opts, Acc) ->
fun(Stream) -> exp(Stream, Stack, Callbacks, Opts, Acc) end.
tr(<<"r"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
tru(Rest, Stack, Callbacks, Opts);
tr(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> tr(Stream, Stack, Callbacks, Opts) end.
tru(<<"u"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
true(Rest, Stack, Callbacks, Opts);
tru(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> tru(Stream, Stack, Callbacks, Opts) end.
true(<<"e"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, true}, Callbacks), Opts);
true(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> true(Stream, Stack, Callbacks, Opts) end.
fa(<<"a"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
fal(Rest, Stack, Callbacks, Opts);
fa(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fa(Stream, Stack, Callbacks, Opts) end.
fal(<<"l"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
fals(Rest, Stack, Callbacks, Opts);
fal(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fal(Stream, Stack, Callbacks, Opts) end.
fals(<<"s"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
false(Rest, Stack, Callbacks, Opts);
fals(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> fals(Stream, Stack, Callbacks, Opts) end.
false(<<"e"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, false}, Callbacks), Opts);
false(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> false(Stream, Stack, Callbacks, Opts) end.
nu(<<"u"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
nul(Rest, Stack, Callbacks, Opts);
nu(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> nu(Stream, Stack, Callbacks, Opts) end.
nul(<<"l"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
null(Rest, Stack, Callbacks, Opts);
nul(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> nul(Stream, Stack, Callbacks, Opts) end.
null(<<"l"/utf8, Rest/binary>>, Stack, Callbacks, Opts) ->
maybe_done(Rest, Stack, callback({literal, null}, Callbacks), Opts);
null(<<>>, Stack, Callbacks, Opts) ->
fun(Stream) -> null(Stream, Stack, Callbacks, Opts) end.
%% comments are c style, /* blah blah */ and are STRONGLY discouraged. any unicode
%% character is valid in a comment, except, obviously the */ sequence which ends
%% the comment. they're implemented as a closure called when the comment ends that
%% returns execution to the point where the comment began. comments are not
%% recorded in any way, simply parsed.
maybe_comment(<<?star/utf8, Rest/binary>>, Resume) ->
comment(Rest, Resume);
maybe_comment(<<>>, Resume) ->
fun(Stream) -> maybe_comment(Stream, Resume) end.
comment(<<?star/utf8, Rest/binary>>, Resume) ->
maybe_comment_done(Rest, Resume);
comment(<<_/utf8, Rest/binary>>, Resume) ->
comment(Rest, Resume);
comment(<<>>, Resume) ->
fun(Stream) -> comment(Stream, Resume) end.
maybe_comment_done(<<?solidus/utf8, Rest/binary>>, Resume) ->
Resume(Rest);
maybe_comment_done(<<>>, Resume) ->
fun(Stream) -> maybe_comment_done(Stream, Resume) end.
%% helper function for dispatching of parser events
callback(eof, {none, Callbacks}) ->
lists:reverse(Callbacks);
callback(Event, {none, Callbacks}) ->
{none, [Event] ++ Callbacks};
callback(Event, {Mod, State}) when is_atom(Mod) ->
{Mod, Mod:jsx_event(Event, State)};
callback(Event, {F, State}) when is_function(F) ->
{F, F(Event, State)}.

View file

@ -1 +0,0 @@
42

View file

@ -1,2 +0,0 @@
[{number, "42"}].
[{naked_values, true}].

View file

@ -26,11 +26,22 @@ test_body(TestSpec, Dir) ->
case file:consult(Dir ++ "/" ++ TestSpec) of
{ok, [Events]} ->
Decoder = jsx:decoder(),
[{TestName, ?_assertEqual(Decoder(JSON), Events)}]
[{TestName, ?_assertEqual(decode(Decoder, JSON), Events)}]
; {ok, [Events, Flags]} ->
Decoder = jsx:decoder(none, Flags),
[{TestName, ?_assertEqual(Decoder(JSON), Events)}]
[{TestName, ?_assertEqual(decode(Decoder, JSON), Events)}]
end
catch _:_ -> []
end.
decode(F, <<>>) ->
{Result, _} = F(eof),
Result;
decode(F, <<A/utf8, Rest/binary>>) ->
case F(<<A/utf8>>) of
G when is_function(G) ->
decode(G, Rest)
; {Result, _} ->
Result
end.