umgeher/jsx
0
Fork 0
Code Issues Pull requests Activity

fixed bug related to partial utfx sequences

Browse source
This commit is contained in:
alisdair sullivan 2010-06-01 19:42:43 -07:00
parent c287315ebf
commit ce503823c8
4 changed files with 61 additions and 11 deletions
Download patch file Download diff file Expand all files Collapse all files

17
src/jsx_utf16.erl
View file

@ -193,8 +193,21 @@ string(<<?rsolidus/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) ->
escape(Rest, Stack, Callbacks, Opts, Acc);
string(<<S/?encoding, Rest/binary>>, Stack, Callbacks, Opts, Acc) when ?is_noncontrol(S) ->
string(Rest, Stack, Callbacks, Opts, [S] ++ Acc);
string(Bin, Stack, Callbacks, Opts, Acc) when byte_size(Bin) < 2 ->
{incomplete, fun(Stream) -> string(<<Bin/binary, Stream/binary>>, Stack, Callbacks, Opts, Acc) end}.
string(Bin, Stack, Callbacks, Opts, Acc) ->
case partial_utf16(Bin) of
true ->
{incomplete, fun(Stream) -> string(<<Bin/binary, Stream/binary>>, Stack, Callbacks, Opts, Acc) end}
; false ->
erlang:error(function_clause)
end.
partial_utf16(<<>>) -> true;
%% this case is not strictly true, there are single bytes that should be rejected, but
%% they're rare enough they can be ignored
partial_utf16(<<_X>>) -> true;
partial_utf16(<<X, _Y>>) when X >= 16#d8, X =< 16#df -> true;
partial_utf16(<<X, _Y, Z>>) when X >= 16#d8, X =< 16#df, Z >= 16#dc, Z =< 16#df -> true;
partial_utf16(_) -> false.
%% only thing to note here is the additional accumulator passed to escaped_unicode used